futriix/src/script_lua.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1783 lines
64 KiB
C
Raw Normal View History

/*
* Copyright (c) 2009-2021, Redis Ltd.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* * Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* * Neither the name of Redis nor the names of its contributors may be used
* to endorse or promote products derived from this software without
* specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include "script_lua.h"
#include "fpconv_dtoa.h"
#include "server.h"
#include "sha1.h"
#include "rand.h"
#include "cluster.h"
#include "monotonic.h"
#include "resp_parser.h"
Function Flags support (no-writes, no-cluster, allow-state, allow-oom) (#10066) # Redis Functions Flags Following the discussion on #10025 Added Functions Flags support. The PR is divided to 2 sections: * Add named argument support to `redis.register_function` API. * Add support for function flags ## `redis.register_function` named argument support The first part of the PR adds support for named argument on `redis.register_function`, example: ``` redis.register_function{ function_name='f1', callback=function() return 'hello' end, description='some desc' } ``` The positional arguments is also kept, which means that it still possible to write: ``` redis.register_function('f1', function() return 'hello' end) ``` But notice that it is no longer possible to pass the optional description argument on the positional argument version. Positional argument was change to allow passing only the mandatory arguments (function name and callback). To pass more arguments the user must use the named argument version. As with positional arguments, the `function_name` and `callback` is mandatory and an error will be raise if those are missing. Also, an error will be raise if an unknown argument name is given or the arguments type is wrong. Tests was added to verify the new syntax. ## Functions Flags The second part of the PR is adding functions flags support. Flags are given to Redis when the engine calls `functionLibCreateFunction`, supported flags are: * `no-writes` - indicating the function perform no writes which means that it is OK to run it on: * read-only replica * Using FCALL_RO * If disk error detected It will not be possible to run a function in those situations unless the function turns on the `no-writes` flag * `allow-oom` - indicate that its OK to run the function even if Redis is in OOM state, if the function will not turn on this flag it will not be possible to run it if OOM reached (even if the function declares `no-writes` and even if `fcall_ro` is used). If this flag is set, any command will be allow on OOM (even those that is marked with CMD_DENYOOM). The assumption is that this flag is for advance users that knows its meaning and understand what they are doing, and Redis trust them to not increase the memory usage. (e.g. it could be an INCR or a modification on an existing key, or a DEL command) * `allow-state` - indicate that its OK to run the function on stale replica, in this case we will also make sure the function is only perform `stale` commands and raise an error if not. * `no-cluster` - indicate to disallow running the function if cluster is enabled. Default behaviure of functions (if no flags is given): 1. Allow functions to read and write 2. Do not run functions on OOM 3. Do not run functions on stale replica 4. Allow functions on cluster ### Lua API for functions flags On Lua engine, it is possible to give functions flags as `flags` named argument: ``` redis.register_function{function_name='f1', callback=function() return 1 end, flags={'no-writes', 'allow-oom'}, description='description'} ``` The function flags argument must be a Lua table that contains all the requested flags, The following will result in an error: * Unknown flag * Wrong flag type Default behaviour is the same as if no flags are used. Tests were added to verify all flags functionality ## Additional changes * mark FCALL and FCALL_RO with CMD_STALE flag (unlike EVAL), so that they can run if the function was registered with the `allow-stale` flag. * Verify `CMD_STALE` on `scriptCall` (`redis.call`), so it will not be possible to call commands from script while stale unless the command is marked with the `CMD_STALE` flags. so that even if the function is allowed while stale we do not allow it to bypass the `CMD_STALE` flag of commands. * Flags section was added to `FUNCTION LIST` command to provide the set of flags for each function: ``` > FUNCTION list withcode 1) 1) "library_name" 2) "test" 3) "engine" 4) "LUA" 5) "description" 6) (nil) 7) "functions" 8) 1) 1) "name" 2) "f1" 3) "description" 4) (nil) 5) "flags" 6) (empty array) 9) "library_code" 10) "redis.register_function{function_name='f1', callback=function() return 1 end}" ``` * Added API to get Redis version from within a script, The redis version can be provided using: 1. `redis.REDIS_VERSION` - string representation of the redis version in the format of MAJOR.MINOR.PATH 2. `redis.REDIS_VERSION_NUM` - number representation of the redis version in the format of `0x00MMmmpp` (`MM` - major, `mm` - minor, `pp` - patch). The number version can be used to check if version is greater or less another version. The string version can be used to return to the user or print as logs. This new API is provided to eval scripts and functions, it also possible to use this API during functions loading phase.
2022-01-14 14:02:02 +02:00
#include "version.h"
#include <lauxlib.h>
#include <lualib.h>
#include <ctype.h>
#include <math.h>
/* Globals that are added by the Lua libraries */
static char *libraries_allow_list[] = {
"string",
"cjson",
"bit",
"cmsgpack",
"math",
"table",
"struct",
"os",
NULL,
};
/* Lua API globals */
static char *server_api_allow_list[] = {
SERVER_API_NAME,
REDIS_API_NAME,
"__redis__err__handler", /* Backwards compatible error handler */
"__server__err__handler", /* error handler for eval, currently located on globals.
Should move to registry. */
NULL,
};
/* Lua builtins */
static char *lua_builtins_allow_list[] = {
"xpcall",
"tostring",
"getfenv",
"setmetatable",
"next",
"assert",
"tonumber",
"rawequal",
"collectgarbage",
"getmetatable",
"rawset",
"pcall",
"coroutine",
"type",
"_G",
"select",
"unpack",
"gcinfo",
"pairs",
"rawget",
"loadstring",
"ipairs",
"_VERSION",
"setfenv",
"load",
"error",
NULL,
};
/* Lua builtins which are not documented on the Lua documentation */
static char *lua_builtins_not_documented_allow_list[] = {
"newproxy",
NULL,
};
/* Lua builtins which are allowed on initialization but will be removed right after */
static char *lua_builtins_removed_after_initialization_allow_list[] = {
"debug", /* debug will be set to nil after the error handler will be created */
NULL,
};
/* Those allow lists was created from the globals that was
* available to the user when the allow lists was first introduce.
* Because we do not want to break backward compatibility we keep
* all the globals. The allow lists will prevent us from accidentally
* creating unwanted globals in the future.
*
* Also notice that the allow list is only checked on start time,
* after that the global table is locked so not need to check anything.*/
static char **allow_lists[] = {
libraries_allow_list,
server_api_allow_list,
lua_builtins_allow_list,
lua_builtins_not_documented_allow_list,
lua_builtins_removed_after_initialization_allow_list,
NULL,
};
/* Deny list contains elements which we know we do not want to add to globals
* and there is no need to print a warning message form them. We will print a
* log message only if an element was added to the globals and the element is
* not on the allow list nor on the back list. */
static char *deny_list[] = {
"dofile",
"loadfile",
"print",
NULL,
};
static int server_math_random(lua_State *L);
static int server_math_randomseed(lua_State *L);
static void redisProtocolToLuaType_Int(void *ctx, long long val, const char *proto, size_t proto_len);
static void
redisProtocolToLuaType_BulkString(void *ctx, const char *str, size_t len, const char *proto, size_t proto_len);
static void redisProtocolToLuaType_NullBulkString(void *ctx, const char *proto, size_t proto_len);
static void redisProtocolToLuaType_NullArray(void *ctx, const char *proto, size_t proto_len);
static void redisProtocolToLuaType_Status(void *ctx, const char *str, size_t len, const char *proto, size_t proto_len);
static void redisProtocolToLuaType_Error(void *ctx, const char *str, size_t len, const char *proto, size_t proto_len);
static void redisProtocolToLuaType_Array(struct ReplyParser *parser, void *ctx, size_t len, const char *proto);
static void redisProtocolToLuaType_Map(struct ReplyParser *parser, void *ctx, size_t len, const char *proto);
static void redisProtocolToLuaType_Set(struct ReplyParser *parser, void *ctx, size_t len, const char *proto);
static void redisProtocolToLuaType_Null(void *ctx, const char *proto, size_t proto_len);
static void redisProtocolToLuaType_Bool(void *ctx, int val, const char *proto, size_t proto_len);
static void redisProtocolToLuaType_Double(void *ctx, double d, const char *proto, size_t proto_len);
static void
redisProtocolToLuaType_BigNumber(void *ctx, const char *str, size_t len, const char *proto, size_t proto_len);
static void redisProtocolToLuaType_VerbatimString(void *ctx,
const char *format,
const char *str,
size_t len,
const char *proto,
size_t proto_len);
static void redisProtocolToLuaType_Attribute(struct ReplyParser *parser, void *ctx, size_t len, const char *proto);
static void luaReplyToServerReply(client *c, client *script_client, lua_State *lua);
/*
* Save the give pointer on Lua registry, used to save the Lua context and
* function context so we can retrieve them from lua_State.
*/
void luaSaveOnRegistry(lua_State *lua, const char *name, void *ptr) {
lua_pushstring(lua, name);
if (ptr) {
lua_pushlightuserdata(lua, ptr);
} else {
lua_pushnil(lua);
}
lua_settable(lua, LUA_REGISTRYINDEX);
}
/*
* Get a saved pointer from registry
*/
void *luaGetFromRegistry(lua_State *lua, const char *name) {
lua_pushstring(lua, name);
lua_gettable(lua, LUA_REGISTRYINDEX);
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
if (lua_isnil(lua, -1)) {
lua_pop(lua, 1); /* pops the value */
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
return NULL;
}
/* must be light user data */
serverAssert(lua_islightuserdata(lua, -1));
void *ptr = (void *)lua_topointer(lua, -1);
serverAssert(ptr);
/* pops the value */
lua_pop(lua, 1);
return ptr;
}
/* ---------------------------------------------------------------------------
* Server reply to Lua type conversion functions.
* ------------------------------------------------------------------------- */
/* Take a server reply in the RESP format and convert it into a
* Lua type. Thanks to this function, and the introduction of not connected
* clients, it is trivial to implement the server() lua function.
*
* Basically we take the arguments, execute the command in the context
* of a non connected client, then take the generated reply and convert it
* into a suitable Lua type. With this trick the scripting feature does not
* need the introduction of a full server internals API. The script
* is like a normal client that bypasses all the slow I/O paths.
*
* Note: in this function we do not do any sanity check as the reply is
* generated by the server directly. This allows us to go faster.
*
* Errors are returned as a table with a single 'err' field set to the
* error string.
*/
static const ReplyParserCallbacks DefaultLuaTypeParserCallbacks = {
.null_array_callback = redisProtocolToLuaType_NullArray,
.bulk_string_callback = redisProtocolToLuaType_BulkString,
.null_bulk_string_callback = redisProtocolToLuaType_NullBulkString,
.error_callback = redisProtocolToLuaType_Error,
.simple_str_callback = redisProtocolToLuaType_Status,
.long_callback = redisProtocolToLuaType_Int,
.array_callback = redisProtocolToLuaType_Array,
.set_callback = redisProtocolToLuaType_Set,
.map_callback = redisProtocolToLuaType_Map,
.bool_callback = redisProtocolToLuaType_Bool,
.double_callback = redisProtocolToLuaType_Double,
.null_callback = redisProtocolToLuaType_Null,
.big_number_callback = redisProtocolToLuaType_BigNumber,
.verbatim_string_callback = redisProtocolToLuaType_VerbatimString,
.attribute_callback = redisProtocolToLuaType_Attribute,
.error = NULL,
};
static void redisProtocolToLuaType(lua_State *lua, char *reply) {
ReplyParser parser = {.curr_location = reply, .callbacks = DefaultLuaTypeParserCallbacks};
parseReply(&parser, lua);
}
static void redisProtocolToLuaType_Int(void *ctx, long long val, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 1)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_pushnumber(lua, (lua_Number)val);
}
static void redisProtocolToLuaType_NullBulkString(void *ctx, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 1)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_pushboolean(lua, 0);
}
static void redisProtocolToLuaType_NullArray(void *ctx, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 1)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_pushboolean(lua, 0);
}
static void
redisProtocolToLuaType_BulkString(void *ctx, const char *str, size_t len, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 1)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_pushlstring(lua, str, len);
}
static void redisProtocolToLuaType_Status(void *ctx, const char *str, size_t len, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 3)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_newtable(lua);
lua_pushstring(lua, "ok");
lua_pushlstring(lua, str, len);
lua_settable(lua, -3);
}
static void redisProtocolToLuaType_Error(void *ctx, const char *str, size_t len, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 3)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
sds err_msg = sdscatlen(sdsnew("-"), str, len);
luaPushErrorBuff(lua, err_msg);
/* push a field indicate to ignore updating the stats on this error
* because it was already updated when executing the command. */
lua_pushstring(lua, "ignore_error_stats_update");
lua_pushboolean(lua, 1);
lua_settable(lua, -3);
}
static void redisProtocolToLuaType_Map(struct ReplyParser *parser, void *ctx, size_t len, const char *proto) {
UNUSED(proto);
lua_State *lua = ctx;
if (lua) {
if (!lua_checkstack(lua, 3)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_newtable(lua);
lua_pushstring(lua, "map");
lua_createtable(lua, 0, len);
}
for (size_t j = 0; j < len; j++) {
parseReply(parser, lua);
parseReply(parser, lua);
if (lua) lua_settable(lua, -3);
}
if (lua) lua_settable(lua, -3);
}
static void redisProtocolToLuaType_Set(struct ReplyParser *parser, void *ctx, size_t len, const char *proto) {
UNUSED(proto);
lua_State *lua = ctx;
if (lua) {
if (!lua_checkstack(lua, 3)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_newtable(lua);
lua_pushstring(lua, "set");
lua_createtable(lua, 0, len);
}
for (size_t j = 0; j < len; j++) {
parseReply(parser, lua);
if (lua) {
if (!lua_checkstack(lua, 1)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic.
* Notice that here we need to check the stack again because the recursive
* call to redisProtocolToLuaType might have use the room allocated in the stack*/
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_pushboolean(lua, 1);
lua_settable(lua, -3);
}
}
if (lua) lua_settable(lua, -3);
}
static void redisProtocolToLuaType_Array(struct ReplyParser *parser, void *ctx, size_t len, const char *proto) {
UNUSED(proto);
lua_State *lua = ctx;
if (lua) {
if (!lua_checkstack(lua, 2)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_createtable(lua, len, 0);
}
for (size_t j = 0; j < len; j++) {
if (lua) lua_pushnumber(lua, j + 1);
parseReply(parser, lua);
if (lua) lua_settable(lua, -3);
}
}
static void redisProtocolToLuaType_Attribute(struct ReplyParser *parser, void *ctx, size_t len, const char *proto) {
UNUSED(proto);
/* Parse the attribute reply.
* Currently, we do not expose the attribute to the Lua script so
* we just need to continue parsing and ignore it (the NULL ensures that the
* reply will be ignored). */
for (size_t j = 0; j < len; j++) {
parseReply(parser, NULL);
parseReply(parser, NULL);
}
/* Parse the reply itself. */
parseReply(parser, ctx);
}
static void redisProtocolToLuaType_VerbatimString(void *ctx,
const char *format,
const char *str,
size_t len,
const char *proto,
size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 5)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_newtable(lua);
lua_pushstring(lua, "verbatim_string");
lua_newtable(lua);
lua_pushstring(lua, "string");
lua_pushlstring(lua, str, len);
lua_settable(lua, -3);
lua_pushstring(lua, "format");
lua_pushlstring(lua, format, 3);
lua_settable(lua, -3);
lua_settable(lua, -3);
}
static void
redisProtocolToLuaType_BigNumber(void *ctx, const char *str, size_t len, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 3)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_newtable(lua);
lua_pushstring(lua, "big_number");
lua_pushlstring(lua, str, len);
lua_settable(lua, -3);
}
static void redisProtocolToLuaType_Null(void *ctx, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 1)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_pushnil(lua);
}
static void redisProtocolToLuaType_Bool(void *ctx, int val, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 1)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_pushboolean(lua, val);
}
static void redisProtocolToLuaType_Double(void *ctx, double d, const char *proto, size_t proto_len) {
UNUSED(proto);
UNUSED(proto_len);
if (!ctx) {
return;
}
lua_State *lua = ctx;
if (!lua_checkstack(lua, 3)) {
/* Increase the Lua stack if needed, to make sure there is enough room
* to push elements to the stack. On failure, exit with panic. */
serverPanic("lua stack limit reach when parsing server.call reply");
}
lua_newtable(lua);
lua_pushstring(lua, "double");
lua_pushnumber(lua, d);
lua_settable(lua, -3);
}
/* This function is used in order to push an error on the Lua stack in the
* format used by server.pcall to return errors, which is a lua table
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
* with an "err" field set to the error string including the error code.
* Note that this table is never a valid reply by proper commands,
* since the returned tables are otherwise always indexed by integers, never by strings.
*
* The function takes ownership on the given err_buffer. */
void luaPushErrorBuff(lua_State *lua, sds err_buffer) {
Consistent erros returned from EVAL scripts (#10218) This PR handles inconsistencies in errors returned from lua scripts. Details of the problem can be found in #10165. ### Changes - Remove double stack trace. It's enough that a stack trace is automatically added by the engine's error handler see https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/function_lua.c#L472-L485 and https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/eval.c#L243-L255 - Make sure all errors a preceded with an error code. Passing a simple string to `luaPushError()` will prepend it with a generic `ERR` error code. - Make sure lua error table doesn't include a RESP `-` error status. Lua stores redis error's as a lua table with a single `err` field and a string. When the string is translated back to RESP we add a `-` to it. See https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/script_lua.c#L510-L517 So there's no need to store it in the lua table. ### Before & After ```diff --- <unnamed> +++ <unnamed> @@ -1,14 +1,14 @@ 1: config set maxmemory 1 2: +OK 3: eval "return redis.call('set','x','y')" 0 - 4: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: @user_script: 1: -OOM command not allowed when used memory > 'maxmemory'. + 4: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. 5: eval "return redis.pcall('set','x','y')" 0 - 6: -@user_script: 1: -OOM command not allowed when used memory > 'maxmemory'. + 6: -OOM command not allowed when used memory > 'maxmemory'. 7: eval "return redis.call('select',99)" 0 8: -ERR Error running script (call to 4ad5abfc50bbccb484223905f9a16f09cd043ba8): @user_script:1: ERR DB index is out of range 9: eval "return redis.pcall('select',99)" 0 10: -ERR DB index is out of range 11: eval_ro "return redis.call('set','x','y')" 0 -12: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: @user_script: 1: Write commands are not allowed from read-only scripts. +12: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: ERR Write commands are not allowed from read-only scripts. 13: eval_ro "return redis.pcall('set','x','y')" 0 -14: -@user_script: 1: Write commands are not allowed from read-only scripts. +14: -ERR Write commands are not allowed from read-only scripts. ```
2022-02-08 11:44:40 +02:00
sds msg;
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
sds error_code;
/* If debugging is active and in step mode, log errors resulting from
* server commands. */
if (ldbIsEnabled()) {
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
ldbLog(sdscatprintf(sdsempty(), "<error> %s", err_buffer));
}
Consistent erros returned from EVAL scripts (#10218) This PR handles inconsistencies in errors returned from lua scripts. Details of the problem can be found in #10165. ### Changes - Remove double stack trace. It's enough that a stack trace is automatically added by the engine's error handler see https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/function_lua.c#L472-L485 and https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/eval.c#L243-L255 - Make sure all errors a preceded with an error code. Passing a simple string to `luaPushError()` will prepend it with a generic `ERR` error code. - Make sure lua error table doesn't include a RESP `-` error status. Lua stores redis error's as a lua table with a single `err` field and a string. When the string is translated back to RESP we add a `-` to it. See https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/script_lua.c#L510-L517 So there's no need to store it in the lua table. ### Before & After ```diff --- <unnamed> +++ <unnamed> @@ -1,14 +1,14 @@ 1: config set maxmemory 1 2: +OK 3: eval "return redis.call('set','x','y')" 0 - 4: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: @user_script: 1: -OOM command not allowed when used memory > 'maxmemory'. + 4: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. 5: eval "return redis.pcall('set','x','y')" 0 - 6: -@user_script: 1: -OOM command not allowed when used memory > 'maxmemory'. + 6: -OOM command not allowed when used memory > 'maxmemory'. 7: eval "return redis.call('select',99)" 0 8: -ERR Error running script (call to 4ad5abfc50bbccb484223905f9a16f09cd043ba8): @user_script:1: ERR DB index is out of range 9: eval "return redis.pcall('select',99)" 0 10: -ERR DB index is out of range 11: eval_ro "return redis.call('set','x','y')" 0 -12: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: @user_script: 1: Write commands are not allowed from read-only scripts. +12: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: ERR Write commands are not allowed from read-only scripts. 13: eval_ro "return redis.pcall('set','x','y')" 0 -14: -@user_script: 1: Write commands are not allowed from read-only scripts. +14: -ERR Write commands are not allowed from read-only scripts. ```
2022-02-08 11:44:40 +02:00
/* There are two possible formats for the received `error` string:
* 1) "-CODE msg": in this case we remove the leading '-' since we don't store it as part of the lua error format.
* 2) "msg": in this case we prepend a generic 'ERR' code since all error statuses need some error code.
* We support format (1) so this function can reuse the error messages used in other places.
Consistent erros returned from EVAL scripts (#10218) This PR handles inconsistencies in errors returned from lua scripts. Details of the problem can be found in #10165. ### Changes - Remove double stack trace. It's enough that a stack trace is automatically added by the engine's error handler see https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/function_lua.c#L472-L485 and https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/eval.c#L243-L255 - Make sure all errors a preceded with an error code. Passing a simple string to `luaPushError()` will prepend it with a generic `ERR` error code. - Make sure lua error table doesn't include a RESP `-` error status. Lua stores redis error's as a lua table with a single `err` field and a string. When the string is translated back to RESP we add a `-` to it. See https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/script_lua.c#L510-L517 So there's no need to store it in the lua table. ### Before & After ```diff --- <unnamed> +++ <unnamed> @@ -1,14 +1,14 @@ 1: config set maxmemory 1 2: +OK 3: eval "return redis.call('set','x','y')" 0 - 4: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: @user_script: 1: -OOM command not allowed when used memory > 'maxmemory'. + 4: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. 5: eval "return redis.pcall('set','x','y')" 0 - 6: -@user_script: 1: -OOM command not allowed when used memory > 'maxmemory'. + 6: -OOM command not allowed when used memory > 'maxmemory'. 7: eval "return redis.call('select',99)" 0 8: -ERR Error running script (call to 4ad5abfc50bbccb484223905f9a16f09cd043ba8): @user_script:1: ERR DB index is out of range 9: eval "return redis.pcall('select',99)" 0 10: -ERR DB index is out of range 11: eval_ro "return redis.call('set','x','y')" 0 -12: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: @user_script: 1: Write commands are not allowed from read-only scripts. +12: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: ERR Write commands are not allowed from read-only scripts. 13: eval_ro "return redis.pcall('set','x','y')" 0 -14: -@user_script: 1: Write commands are not allowed from read-only scripts. +14: -ERR Write commands are not allowed from read-only scripts. ```
2022-02-08 11:44:40 +02:00
* We support format (2) so it'll be easy to pass descriptive errors to this function without worrying about format.
*/
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
if (err_buffer[0] == '-') {
/* derive error code from the message */
char *err_msg = strstr(err_buffer, " ");
if (!err_msg) {
msg = sdsnew(err_buffer + 1);
error_code = sdsnew("ERR");
} else {
*err_msg = '\0';
msg = sdsnew(err_msg + 1);
error_code = sdsnew(err_buffer + 1);
}
sdsfree(err_buffer);
} else {
msg = err_buffer;
error_code = sdsnew("ERR");
}
/* Trim newline at end of string. If we reuse the ready-made error objects (case 1 above) then we might
* have a newline that needs to be trimmed. In any case the lua server error table shouldn't end with a newline. */
Consistent erros returned from EVAL scripts (#10218) This PR handles inconsistencies in errors returned from lua scripts. Details of the problem can be found in #10165. ### Changes - Remove double stack trace. It's enough that a stack trace is automatically added by the engine's error handler see https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/function_lua.c#L472-L485 and https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/eval.c#L243-L255 - Make sure all errors a preceded with an error code. Passing a simple string to `luaPushError()` will prepend it with a generic `ERR` error code. - Make sure lua error table doesn't include a RESP `-` error status. Lua stores redis error's as a lua table with a single `err` field and a string. When the string is translated back to RESP we add a `-` to it. See https://github.com/redis/redis/blob/d0bc4fff18afdf9e5421cc88e23ffbb876ecaec3/src/script_lua.c#L510-L517 So there's no need to store it in the lua table. ### Before & After ```diff --- <unnamed> +++ <unnamed> @@ -1,14 +1,14 @@ 1: config set maxmemory 1 2: +OK 3: eval "return redis.call('set','x','y')" 0 - 4: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: @user_script: 1: -OOM command not allowed when used memory > 'maxmemory'. + 4: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. 5: eval "return redis.pcall('set','x','y')" 0 - 6: -@user_script: 1: -OOM command not allowed when used memory > 'maxmemory'. + 6: -OOM command not allowed when used memory > 'maxmemory'. 7: eval "return redis.call('select',99)" 0 8: -ERR Error running script (call to 4ad5abfc50bbccb484223905f9a16f09cd043ba8): @user_script:1: ERR DB index is out of range 9: eval "return redis.pcall('select',99)" 0 10: -ERR DB index is out of range 11: eval_ro "return redis.call('set','x','y')" 0 -12: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: @user_script: 1: Write commands are not allowed from read-only scripts. +12: -ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: ERR Write commands are not allowed from read-only scripts. 13: eval_ro "return redis.pcall('set','x','y')" 0 -14: -@user_script: 1: Write commands are not allowed from read-only scripts. +14: -ERR Write commands are not allowed from read-only scripts. ```
2022-02-08 11:44:40 +02:00
msg = sdstrim(msg, "\r\n");
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
sds final_msg = sdscatfmt(error_code, " %s", msg);
lua_newtable(lua);
lua_pushstring(lua, "err");
lua_pushstring(lua, final_msg);
lua_settable(lua, -3);
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
sdsfree(msg);
sdsfree(final_msg);
}
void luaPushError(lua_State *lua, const char *error) {
luaPushErrorBuff(lua, sdsnew(error));
}
/* In case the error set into the Lua stack by luaPushError() was generated
* by the non-error-trapping version of server.pcall(), which is server.call(),
* this function will raise the Lua error so that the execution of the
* script will be halted. */
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
int luaError(lua_State *lua) {
return lua_error(lua);
}
/* ---------------------------------------------------------------------------
* Lua reply to server reply conversion functions.
* ------------------------------------------------------------------------- */
/* Reply to client 'c' converting the top element in the Lua stack to a
* server reply. As a side effect the element is consumed from the stack. */
static void luaReplyToServerReply(client *c, client *script_client, lua_State *lua) {
int t = lua_type(lua, -1);
if (!lua_checkstack(lua, 4)) {
/* Increase the Lua stack if needed to make sure there is enough room
* to push 4 elements to the stack. On failure, return error.
* Notice that we need, in the worst case, 4 elements because returning a map might
* require push 4 elements to the Lua stack.*/
addReplyError(c, "reached lua stack limit");
lua_pop(lua, 1); /* pop the element from the stack */
return;
}
switch (t) {
case LUA_TSTRING: addReplyBulkCBuffer(c, (char *)lua_tostring(lua, -1), lua_strlen(lua, -1)); break;
case LUA_TBOOLEAN:
if (script_client->resp == 2)
addReply(c, lua_toboolean(lua, -1) ? shared.cone : shared.null[c->resp]);
else
addReplyBool(c, lua_toboolean(lua, -1));
break;
case LUA_TNUMBER: addReplyLongLong(c, (long long)lua_tonumber(lua, -1)); break;
case LUA_TTABLE:
/* We need to check if it is an array, an error, or a status reply.
* Error are returned as a single element table with 'err' field.
* Status replies are returned as single element table with 'ok'
* field. */
/* Handle error reply. */
/* we took care of the stack size on function start */
lua_pushstring(lua, "err");
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -2);
t = lua_type(lua, -1);
if (t == LUA_TSTRING) {
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
lua_pop(lua,
1); /* pop the error message, we will use luaExtractErrorInformation to get error information */
errorInfo err_info = {0};
luaExtractErrorInformation(lua, &err_info);
addReplyErrorFormatEx(
c, ERR_REPLY_FLAG_CUSTOM | (err_info.ignore_err_stats_update ? ERR_REPLY_FLAG_NO_STATS_UPDATE : 0),
"-%s", err_info.msg);
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
luaErrorInformationDiscard(&err_info);
lua_pop(lua, 1); /* pop the result table */
return;
}
lua_pop(lua, 1); /* Discard field name pushed before. */
/* Handle status reply. */
lua_pushstring(lua, "ok");
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -2);
t = lua_type(lua, -1);
if (t == LUA_TSTRING) {
sds ok = sdsnew(lua_tostring(lua, -1));
sdsmapchars(ok, "\r\n", " ", 2);
addReplyStatusLength(c, ok, sdslen(ok));
sdsfree(ok);
lua_pop(lua, 2);
return;
}
lua_pop(lua, 1); /* Discard field name pushed before. */
/* Handle double reply. */
lua_pushstring(lua, "double");
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -2);
t = lua_type(lua, -1);
if (t == LUA_TNUMBER) {
addReplyDouble(c, lua_tonumber(lua, -1));
lua_pop(lua, 2);
return;
}
lua_pop(lua, 1); /* Discard field name pushed before. */
/* Handle big number reply. */
lua_pushstring(lua, "big_number");
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -2);
t = lua_type(lua, -1);
if (t == LUA_TSTRING) {
sds big_num = sdsnewlen(lua_tostring(lua, -1), lua_strlen(lua, -1));
sdsmapchars(big_num, "\r\n", " ", 2);
addReplyBigNum(c, big_num, sdslen(big_num));
sdsfree(big_num);
lua_pop(lua, 2);
return;
}
lua_pop(lua, 1); /* Discard field name pushed before. */
/* Handle verbatim reply. */
lua_pushstring(lua, "verbatim_string");
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -2);
t = lua_type(lua, -1);
if (t == LUA_TTABLE) {
lua_pushstring(lua, "format");
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -2);
t = lua_type(lua, -1);
if (t == LUA_TSTRING) {
char *format = (char *)lua_tostring(lua, -1);
lua_pushstring(lua, "string");
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -3);
t = lua_type(lua, -1);
if (t == LUA_TSTRING) {
size_t len;
char *str = (char *)lua_tolstring(lua, -1, &len);
addReplyVerbatim(c, str, len, format);
lua_pop(lua, 4);
return;
}
lua_pop(lua, 1);
}
lua_pop(lua, 1);
}
lua_pop(lua, 1); /* Discard field name pushed before. */
/* Handle map reply. */
lua_pushstring(lua, "map");
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -2);
t = lua_type(lua, -1);
if (t == LUA_TTABLE) {
int maplen = 0;
void *replylen = addReplyDeferredLen(c);
/* we took care of the stack size on function start */
lua_pushnil(lua); /* Use nil to start iteration. */
while (lua_next(lua, -2)) {
/* Stack now: table, key, value */
lua_pushvalue(lua, -2); /* Dup key before consuming. */
luaReplyToServerReply(c, script_client, lua); /* Return key. */
luaReplyToServerReply(c, script_client, lua); /* Return value. */
/* Stack now: table, key. */
maplen++;
}
setDeferredMapLen(c, replylen, maplen);
lua_pop(lua, 2);
return;
}
lua_pop(lua, 1); /* Discard field name pushed before. */
/* Handle set reply. */
lua_pushstring(lua, "set");
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -2);
t = lua_type(lua, -1);
if (t == LUA_TTABLE) {
int setlen = 0;
void *replylen = addReplyDeferredLen(c);
/* we took care of the stack size on function start */
lua_pushnil(lua); /* Use nil to start iteration. */
while (lua_next(lua, -2)) {
/* Stack now: table, key, true */
lua_pop(lua, 1); /* Discard the boolean value. */
lua_pushvalue(lua, -1); /* Dup key before consuming. */
luaReplyToServerReply(c, script_client, lua); /* Return key. */
/* Stack now: table, key. */
setlen++;
}
setDeferredSetLen(c, replylen, setlen);
lua_pop(lua, 2);
return;
}
lua_pop(lua, 1); /* Discard field name pushed before. */
/* Handle the array reply. */
void *replylen = addReplyDeferredLen(c);
int j = 1, mbulklen = 0;
while (1) {
/* we took care of the stack size on function start */
lua_pushnumber(lua, j++);
Fix #11030, use lua_rawget to avoid triggering metatables and crash. (#11032) Fix #11030, use lua_rawget to avoid triggering metatables. #11030 shows how return `_G` from the Lua script (either function or eval), cause the Lua interpreter to Panic and the Redis processes to exit with error code 1. Though return `_G` only panic on Redis 7 and 6.2.7, the underline issue exists on older versions as well (6.0 and 6.2). The underline issue is returning a table with a metatable such that the metatable raises an error. The following example demonstrate the issue: ``` 127.0.0.1:6379> eval "local a = {}; setmetatable(a,{__index=function() foo() end}) return a" 0 Error: Server closed the connection ``` ``` PANIC: unprotected error in call to Lua API (user_script:1: Script attempted to access nonexistent global variable 'foo') ``` The Lua panic happened because when returning the result to the client, Redis needs to introspect the returning table and transform the table into a resp. In order to scan the table, Redis uses `lua_gettable` api which might trigger the metatable (if exists) and might raise an error. This code is not running inside `pcall` (Lua protected call), so raising an error causes the Lua to panic and exit. Notice that this is not a crash, its a Lua panic that exit with error code 1. Returning `_G` panics on Redis 7 and 6.2.7 because on those versions `_G` has a metatable that raises error when trying to fetch a none existing key. ### Solution Instead of using `lua_gettable` that might raise error and cause the issue, use `lua_rawget` that simply return the value from the table without triggering any metatable logic. This is promised not to raise and error. The downside of this solution is that it might be considered as breaking change, if someone rely on metatable in the returned value. An alternative solution is to wrap this entire logic with `pcall` (Lua protected call), this alternative require a much bigger refactoring. ### Back Porting The same fix will work on older versions as well (6.2, 6.0). Notice that on those version, the issue can cause Redis to crash if inside the metatable logic there is an attempt to accesses Redis (`redis.call`). On 7.0, there is not crash and the `redis.call` is executed as if it was done from inside the script itself. ### Tests Tests was added the verify the fix
2022-07-26 10:33:50 +03:00
lua_rawget(lua, -2);
t = lua_type(lua, -1);
if (t == LUA_TNIL) {
lua_pop(lua, 1);
break;
}
luaReplyToServerReply(c, script_client, lua);
mbulklen++;
}
setDeferredArrayLen(c, replylen, mbulklen);
break;
default: addReplyNull(c);
}
lua_pop(lua, 1);
}
/* ---------------------------------------------------------------------------
* Lua server.* functions implementations.
* ------------------------------------------------------------------------- */
void freeLuaServerArgv(robj **argv, int argc, int argv_len);
/* Cached argv array across calls. */
static robj **lua_argv = NULL;
static int lua_argv_size = 0;
/* Cache of recently used small arguments to avoid malloc calls. */
static robj *lua_args_cached_objects[LUA_CMD_OBJCACHE_SIZE];
static size_t lua_args_cached_objects_len[LUA_CMD_OBJCACHE_SIZE];
static robj **luaArgsToServerArgv(lua_State *lua, int *argc, int *argv_len) {
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
int j;
/* Require at least one argument */
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
*argc = lua_gettop(lua);
if (*argc == 0) {
luaPushError(lua, "Please specify at least one argument for this call");
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
return NULL;
}
/* Build the arguments vector (reuse a cached argv from last call) */
if (lua_argv_size < *argc) {
lua_argv = zrealloc(lua_argv, sizeof(robj *) * *argc);
lua_argv_size = *argc;
}
*argv_len = lua_argv_size;
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
for (j = 0; j < *argc; j++) {
char *obj_s;
size_t obj_len;
char dbuf[64];
if (lua_type(lua, j + 1) == LUA_TNUMBER) {
/* We can't use lua_tolstring() for number -> string conversion
* since Lua uses a format specifier that loses precision. */
lua_Number num = lua_tonumber(lua, j + 1);
/* Integer printing function is much faster, check if we can safely use it.
* Since lua_Number is not explicitly an integer or a double, we need to make an effort
* to convert it as an integer when that's possible, since the string could later be used
* in a context that doesn't support scientific notation (e.g. 1e9 instead of 100000000). */
long long lvalue;
if (double2ll((double)num, &lvalue))
obj_len = ll2string(dbuf, sizeof(dbuf), lvalue);
else {
obj_len = fpconv_dtoa((double)num, dbuf);
dbuf[obj_len] = '\0';
}
obj_s = dbuf;
} else {
obj_s = (char *)lua_tolstring(lua, j + 1, &obj_len);
if (obj_s == NULL) break; /* Not a string. */
}
/* Try to use a cached object. */
if (j < LUA_CMD_OBJCACHE_SIZE && lua_args_cached_objects[j] && lua_args_cached_objects_len[j] >= obj_len) {
sds s = lua_args_cached_objects[j]->ptr;
lua_argv[j] = lua_args_cached_objects[j];
lua_args_cached_objects[j] = NULL;
memcpy(s, obj_s, obj_len + 1);
sdssetlen(s, obj_len);
} else {
lua_argv[j] = createStringObject(obj_s, obj_len);
}
}
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
/* Pop all arguments from the stack, we do not need them anymore
* and this way we guaranty we will have room on the stack for the result. */
lua_pop(lua, *argc);
/* Check if one of the arguments passed by the Lua script
* is not a string or an integer (lua_isstring() return true for
* integers as well). */
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
if (j != *argc) {
freeLuaServerArgv(lua_argv, j, lua_argv_size);
luaPushError(lua, "Command arguments must be strings or integers");
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
return NULL;
}
return lua_argv;
}
void freeLuaServerArgv(robj **argv, int argc, int argv_len) {
int j;
for (j = 0; j < argc; j++) {
robj *o = argv[j];
/* Try to cache the object in the lua_args_cached_objects array.
* The object must be small, SDS-encoded, and with refcount = 1
* (we must be the only owner) for us to cache it. */
if (j < LUA_CMD_OBJCACHE_SIZE && o->refcount == 1 &&
(o->encoding == OBJ_ENCODING_RAW || o->encoding == OBJ_ENCODING_EMBSTR) &&
sdslen(o->ptr) <= LUA_CMD_OBJCACHE_MAX_LEN) {
sds s = o->ptr;
if (lua_args_cached_objects[j]) decrRefCount(lua_args_cached_objects[j]);
lua_args_cached_objects[j] = o;
lua_args_cached_objects_len[j] = sdsalloc(s);
} else {
decrRefCount(o);
}
}
if (argv != lua_argv || argv_len != lua_argv_size) {
/* The command changed argv, scrap the cache and start over. */
zfree(argv);
lua_argv = NULL;
lua_argv_size = 0;
}
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
}
static int luaServerGenericCommand(lua_State *lua, int raise_error) {
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
int j;
scriptRunCtx *rctx = luaGetFromRegistry(lua, REGISTRY_RUN_CTX_NAME);
serverAssert(rctx); /* Only supported inside script invocation */
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
sds err = NULL;
client *c = rctx->c;
sds reply;
c->argv = luaArgsToServerArgv(lua, &c->argc, &c->argv_len);
if (c->argv == NULL) {
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
return raise_error ? luaError(lua) : 1;
}
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
static int inuse = 0; /* Recursive calls detection. */
/* By using Lua debug hooks it is possible to trigger a recursive call
* to luaServerGenericCommand(), which normally should never happen.
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
* To make this function reentrant is futile and makes it slower, but
* we should at least detect such a misuse, and abort. */
if (inuse) {
char *recursion_warning = "luaRedisGenericCommand() recursive call detected. "
"Are you doing funny stuff with Lua debug hooks?";
serverLog(LL_WARNING, "%s", recursion_warning);
luaPushError(lua, recursion_warning);
return 1;
}
inuse++;
/* Log the command if debugging is active. */
if (ldbIsEnabled()) {
sds cmdlog = sdsnew("<command>");
for (j = 0; j < c->argc; j++) {
if (j == 10) {
cmdlog = sdscatprintf(cmdlog, " ... (%d more)", c->argc - j - 1);
break;
} else {
cmdlog = sdscatlen(cmdlog, " ", 1);
cmdlog = sdscatsds(cmdlog, c->argv[j]->ptr);
}
}
ldbLog(cmdlog);
}
scriptCall(rctx, &err);
if (err) {
luaPushError(lua, err);
sdsfree(err);
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
/* push a field indicate to ignore updating the stats on this error
* because it was already updated when executing the command. */
lua_pushstring(lua, "ignore_error_stats_update");
lua_pushboolean(lua, 1);
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
lua_settable(lua, -3);
goto cleanup;
}
/* Convert the result of the command into a suitable Lua type.
* The first thing we need is to create a single string from the client
* output buffers. */
if (listLength(c->reply) == 0 && (size_t)c->bufpos < c->buf_usable_size) {
/* This is a fast path for the common case of a reply inside the
* client static buffer. Don't create an SDS string but just use
* the client buffer directly. */
c->buf[c->bufpos] = '\0';
reply = c->buf;
c->bufpos = 0;
} else {
reply = sdsnewlen(c->buf, c->bufpos);
c->bufpos = 0;
while (listLength(c->reply)) {
clientReplyBlock *o = listNodeValue(listFirst(c->reply));
reply = sdscatlen(reply, o->buf, o->used);
listDelNode(c->reply, listFirst(c->reply));
}
}
if (raise_error && reply[0] != '-') raise_error = 0;
redisProtocolToLuaType(lua, reply);
/* If the debugger is active, log the reply from the server. */
if (ldbIsEnabled())
ldbLogRespReply(reply);
if (reply != c->buf) sdsfree(reply);
c->reply_bytes = 0;
cleanup:
/* Clean up. Command code may have changed argv/argc so we use the
* argv/argc of the client instead of the local variables. */
freeLuaServerArgv(c->argv, c->argc, c->argv_len);
c->argc = c->argv_len = 0;
c->user = NULL;
c->argv = NULL;
reprocess command when client is unblocked on keys (#11012) *TL;DR* --------------------------------------- Following the discussion over the issue [#7551](https://github.com/redis/redis/issues/7551) We decided to refactor the client blocking code to eliminate some of the code duplications and to rebuild the infrastructure better for future key blocking cases. *In this PR* --------------------------------------- 1. reprocess the command once a client becomes unblocked on key (instead of running custom code for the unblocked path that's different than the one that would have run if blocking wasn't needed) 2. eliminate some (now) irrelevant code for handling unblocking lists/zsets/streams etc... 3. modify some tests to intercept the error in cases of error on reprocess after unblock (see details in the notes section below) 4. replace '$' on the client argv with current stream id. Since once we reprocess the stream XREAD we need to read from the last msg and not wait for new msg in order to prevent endless block loop. 5. Added statistics to the info "Clients" section to report the: * `total_blocking_keys` - number of blocking keys * `total_blocking_keys_on_nokey` - number of blocking keys which have at least 1 client which would like to be unblocked on when the key is deleted. 6. Avoid expiring unblocked key during unblock. Previously we used to lookup the unblocked key which might have been expired during the lookup. Now we lookup the key using NOTOUCH and NOEXPIRE to avoid deleting it at this point, so propagating commands in blocked.c is no longer needed. 7. deprecated command flags. We decided to remove the CMD_CALL_STATS and CMD_CALL_SLOWLOG and make an explicit verification in the call() function in order to decide if stats update should take place. This should simplify the logic and also mitigate existing issues: for example module calls which are triggered as part of AOF loading might still report stats even though they are called during AOF loading. *Behavior changes* --------------------------------------------------- 1. As this implementation prevents writing dedicated code handling unblocked streams/lists/zsets, since we now re-process the command once the client is unblocked some errors will be reported differently. The old implementation used to issue ``UNBLOCKED the stream key no longer exists`` in the following cases: - The stream key has been deleted (ie. calling DEL) - The stream and group existed but the key type was changed by overriding it (ie. with set command) - The key not longer exists after we swapdb with a db which does not contains this key - After swapdb when the new db has this key but with different type. In the new implementation the reported errors will be the same as if the command was processed after effect: **NOGROUP** - in case key no longer exists, or **WRONGTYPE** in case the key was overridden with a different type. 2. Reprocessing the command means that some checks will be reevaluated once the client is unblocked. For example, ACL rules might change since the command originally was executed and will fail once the client is unblocked. Another example is OOM condition checks which might enable the command to run and block but fail the command reprocess once the client is unblocked. 3. One of the changes in this PR is that no command stats are being updated once the command is blocked (all stats will be updated once the client is unblocked). This implies that when we have many clients blocked, users will no longer be able to get that information from the command stats. However the information can still be gathered from the client list. **Client blocking** --------------------------------------------------- the blocking on key will still be triggered the same way as it is done today. in order to block the current client on list of keys, the call to blockForKeys will still need to be made which will perform the same as it is today: * add the client to the list of blocked clients on each key * keep the key with a matching list node (position in the global blocking clients list for that key) in the client private blocking key dict. * flag the client with CLIENT_BLOCKED * update blocking statistics * register the client on the timeout table **Key Unblock** --------------------------------------------------- Unblocking a specific key will be triggered (same as today) by calling signalKeyAsReady. the implementation in that part will stay the same as today - adding the key to the global readyList. The reason to maintain the readyList (as apposed to iterating over all clients blocked on the specific key) is in order to keep the signal operation as short as possible, since it is called during the command processing. The main change is that instead of going through a dedicated code path that operates the blocked command we will just call processPendingCommandsAndResetClient. **ClientUnblock (keys)** --------------------------------------------------- 1. Unblocking clients on keys will be triggered after command is processed and during the beforeSleep 8. the general schema is: 9. For each key *k* in the readyList: ``` For each client *c* which is blocked on *k*: in case either: 1. *k* exists AND the *k* type matches the current client blocking type OR 2. *k* exists and *c* is blocked on module command OR 3. *k* does not exists and *c* was blocked with the flag unblock_on_deleted_key do: 1. remove the client from the list of clients blocked on this key 2. remove the blocking list node from the client blocking key dict 3. remove the client from the timeout list 10. queue the client on the unblocked_clients list 11. *NEW*: call processCommandAndResetClient(c); ``` *NOTE:* for module blocked clients we will still call the moduleUnblockClientByHandle which will queue the client for processing in moduleUnblockedClients list. **Process Unblocked clients** --------------------------------------------------- The process of all unblocked clients is done in the beforeSleep and no change is planned in that part. The general schema will be: For each client *c* in server.unblocked_clients: * remove client from the server.unblocked_clients * set back the client readHandler * continue processing the pending command and input buffer. *Some notes regarding the new implementation* --------------------------------------------------- 1. Although it was proposed, it is currently difficult to remove the read handler from the client while it is blocked. The reason is that a blocked client should be unblocked when it is disconnected, or we might consume data into void. 2. While this PR mainly keep the current blocking logic as-is, there might be some future additions to the infrastructure that we would like to have: - allow non-preemptive blocking of client - sometimes we can think that a new kind of blocking can be expected to not be preempt. for example lets imagine we hold some keys on disk and when a command needs to process them it will block until the keys are uploaded. in this case we will want the client to not disconnect or be unblocked until the process is completed (remove the client read handler, prevent client timeout, disable unblock via debug command etc...). - allow generic blocking based on command declared keys - we might want to add a hook before command processing to check if any of the declared keys require the command to block. this way it would be easier to add new kinds of key-based blocking mechanisms. Co-authored-by: Oran Agra <oran@redislabs.com> Signed-off-by: Ran Shidlansik <ranshid@amazon.com>
2023-01-01 23:35:42 +02:00
resetClient(c);
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
inuse--;
if (raise_error) {
/* If we are here we should have an error in the stack, in the
* form of a table with an "err" field. Extract the string to
* return the plain error. */
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
return luaError(lua);
}
return 1;
}
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
/* Our implementation to lua pcall.
* We need this implementation for backward
* comparability with older Redis OSS versions.
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
*
* On Redis OSS 7, the error object is a table,
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
* compare to older version where the error
* object is a string. To keep backward
* comparability we catch the table object
* and just return the error message. */
static int luaRedisPcall(lua_State *lua) {
int argc = lua_gettop(lua);
lua_pushboolean(lua, 1); /* result place holder */
lua_insert(lua, 1);
if (lua_pcall(lua, argc - 1, LUA_MULTRET, 0)) {
/* Error */
lua_remove(lua, 1); /* remove the result place holder, now we have room for at least one element */
if (lua_istable(lua, -1)) {
lua_getfield(lua, -1, "err");
if (lua_isstring(lua, -1)) {
lua_replace(lua, -2); /* replace the error message with the table */
}
}
lua_pushboolean(lua, 0); /* push result */
lua_insert(lua, 1);
}
return lua_gettop(lua);
}
/* server.call() */
static int luaRedisCallCommand(lua_State *lua) {
return luaServerGenericCommand(lua, 1);
}
/* server.pcall() */
static int luaRedisPCallCommand(lua_State *lua) {
return luaServerGenericCommand(lua, 0);
}
/* This adds server.sha1hex(string) to Lua scripts using the same hashing
* function used for sha1ing lua scripts. */
static int luaRedisSha1hexCommand(lua_State *lua) {
int argc = lua_gettop(lua);
char digest[41];
size_t len;
char *s;
if (argc != 1) {
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
luaPushError(lua, "wrong number of arguments");
return luaError(lua);
}
s = (char *)lua_tolstring(lua, 1, &len);
sha1hex(digest, s, len);
lua_pushstring(lua, digest);
return 1;
}
/* Returns a table with a single field 'field' set to the string value
* passed as argument. This helper function is handy when returning
* a RESP error or status reply from Lua:
*
* return server.error_reply("ERR Some Error")
* return server.status_reply("ERR Some Error")
*/
static int luaRedisReturnSingleFieldTable(lua_State *lua, char *field) {
if (lua_gettop(lua) != 1 || lua_type(lua, -1) != LUA_TSTRING) {
luaPushError(lua, "wrong number or type of arguments");
return 1;
}
lua_newtable(lua);
lua_pushstring(lua, field);
lua_pushvalue(lua, -3);
lua_settable(lua, -3);
return 1;
}
/* server.error_reply() */
static int luaRedisErrorReplyCommand(lua_State *lua) {
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
if (lua_gettop(lua) != 1 || lua_type(lua, -1) != LUA_TSTRING) {
luaPushError(lua, "wrong number or type of arguments");
return 1;
}
/* add '-' if not exists */
const char *err = lua_tostring(lua, -1);
sds err_buff = NULL;
if (err[0] != '-') {
err_buff = sdscatfmt(sdsempty(), "-%s", err);
} else {
err_buff = sdsnew(err);
}
luaPushErrorBuff(lua, err_buff);
return 1;
}
/* server.status_reply() */
static int luaRedisStatusReplyCommand(lua_State *lua) {
return luaRedisReturnSingleFieldTable(lua, "ok");
}
/* server.set_repl()
*
* Set the propagation of write commands executed in the context of the
* script to on/off for AOF and replicas. */
static int luaRedisSetReplCommand(lua_State *lua) {
Remove EVAL script verbatim replication, propagation, and deterministic execution logic (#9812) # Background The main goal of this PR is to remove relevant logics on Lua script verbatim replication, only keeping effects replication logic, which has been set as default since Redis 5.0. As a result, Lua in Redis 7.0 would be acting the same as Redis 6.0 with default configuration from users' point of view. There are lots of reasons to remove verbatim replication. Antirez has listed some of the benefits in Issue #5292: >1. No longer need to explain to users side effects into scripts. They can do whatever they want. >2. No need for a cache about scripts that we sent or not to the slaves. >3. No need to sort the output of certain commands inside scripts (SMEMBERS and others): this both simplifies and gains speed. >4. No need to store scripts inside the RDB file in order to startup correctly. >5. No problems about evicting keys during the script execution. When looking back at Redis 5.0, antirez and core team decided to set the config `lua-replicate-commands yes` by default instead of removing verbatim replication directly, in case some bad situations happened. 3 years later now before Redis 7.0, it's time to remove it formally. # Changes - configuration for lua-replicate-commands removed - created config file stub for backward compatibility - Replication script cache removed - this is useless under script effects replication - relevant statistics also removed - script persistence in RDB files is also removed - Propagation of SCRIPT LOAD and SCRIPT FLUSH to replica / AOF removed - Deterministic execution logic in scripts removed (i.e. don't run write commands after random ones, and sorting output of commands with random order) - the flags indicating which commands have non-deterministic results are kept as hints to clients. - `redis.replicate_commands()` & `redis.set_repl()` changed - now `redis.replicate_commands()` does nothing and return an 1 - ...and then `redis.set_repl()` can be issued before `redis.replicate_commands()` now - Relevant TCL cases adjusted - DEBUG lua-always-replicate-commands removed # Other changes - Fix a recent bug comparing CLIENT_ID_AOF to original_client->flags instead of id. (introduced in #9780) Co-authored-by: Oran Agra <oran@redislabs.com>
2021-12-21 14:32:42 +08:00
int flags, argc = lua_gettop(lua);
scriptRunCtx *rctx = luaGetFromRegistry(lua, REGISTRY_RUN_CTX_NAME);
serverAssert(rctx); /* Only supported inside script invocation */
Remove EVAL script verbatim replication, propagation, and deterministic execution logic (#9812) # Background The main goal of this PR is to remove relevant logics on Lua script verbatim replication, only keeping effects replication logic, which has been set as default since Redis 5.0. As a result, Lua in Redis 7.0 would be acting the same as Redis 6.0 with default configuration from users' point of view. There are lots of reasons to remove verbatim replication. Antirez has listed some of the benefits in Issue #5292: >1. No longer need to explain to users side effects into scripts. They can do whatever they want. >2. No need for a cache about scripts that we sent or not to the slaves. >3. No need to sort the output of certain commands inside scripts (SMEMBERS and others): this both simplifies and gains speed. >4. No need to store scripts inside the RDB file in order to startup correctly. >5. No problems about evicting keys during the script execution. When looking back at Redis 5.0, antirez and core team decided to set the config `lua-replicate-commands yes` by default instead of removing verbatim replication directly, in case some bad situations happened. 3 years later now before Redis 7.0, it's time to remove it formally. # Changes - configuration for lua-replicate-commands removed - created config file stub for backward compatibility - Replication script cache removed - this is useless under script effects replication - relevant statistics also removed - script persistence in RDB files is also removed - Propagation of SCRIPT LOAD and SCRIPT FLUSH to replica / AOF removed - Deterministic execution logic in scripts removed (i.e. don't run write commands after random ones, and sorting output of commands with random order) - the flags indicating which commands have non-deterministic results are kept as hints to clients. - `redis.replicate_commands()` & `redis.set_repl()` changed - now `redis.replicate_commands()` does nothing and return an 1 - ...and then `redis.set_repl()` can be issued before `redis.replicate_commands()` now - Relevant TCL cases adjusted - DEBUG lua-always-replicate-commands removed # Other changes - Fix a recent bug comparing CLIENT_ID_AOF to original_client->flags instead of id. (introduced in #9780) Co-authored-by: Oran Agra <oran@redislabs.com>
2021-12-21 14:32:42 +08:00
if (argc != 1) {
luaPushError(lua, "server.set_repl() requires one argument.");
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
return luaError(lua);
}
flags = lua_tonumber(lua, -1);
if ((flags & ~(PROPAGATE_AOF | PROPAGATE_REPL)) != 0) {
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
luaPushError(lua, "Invalid replication flags. Use REPL_AOF, REPL_REPLICA, REPL_ALL or REPL_NONE.");
return luaError(lua);
}
scriptSetRepl(rctx, flags);
return 0;
}
/* server.acl_check_cmd()
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
*
* Checks ACL permissions for given command for the current user. */
static int luaRedisAclCheckCmdPermissionsCommand(lua_State *lua) {
scriptRunCtx *rctx = luaGetFromRegistry(lua, REGISTRY_RUN_CTX_NAME);
serverAssert(rctx); /* Only supported inside script invocation */
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
int raise_error = 0;
int argc, argv_len;
robj **argv = luaArgsToServerArgv(lua, &argc, &argv_len);
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
/* Require at least one argument */
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
if (argv == NULL) return luaError(lua);
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
/* Find command */
struct serverCommand *cmd;
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
if ((cmd = lookupCommand(argv, argc)) == NULL) {
luaPushError(lua, "Invalid command passed to server.acl_check_cmd()");
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
raise_error = 1;
} else {
int keyidxptr;
if (ACLCheckAllUserCommandPerm(rctx->original_client->user, cmd, argv, argc, &keyidxptr) != ACL_OK) {
lua_pushboolean(lua, 0);
} else {
lua_pushboolean(lua, 1);
}
}
freeLuaServerArgv(argv, argc, argv_len);
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
if (raise_error)
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
return luaError(lua);
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
else
return 1;
}
/* server.log() */
static int luaLogCommand(lua_State *lua) {
int j, argc = lua_gettop(lua);
int level;
sds log;
if (argc < 2) {
luaPushError(lua, "server.log() requires two arguments or more.");
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
return luaError(lua);
} else if (!lua_isnumber(lua, -argc)) {
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
luaPushError(lua, "First argument must be a number (log level).");
return luaError(lua);
}
level = lua_tonumber(lua, -argc);
if (level < LL_DEBUG || level > LL_WARNING) {
luaPushError(lua, "Invalid log level.");
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
return luaError(lua);
}
if (level < server.verbosity) return 0;
/* Glue together all the arguments */
log = sdsempty();
for (j = 1; j < argc; j++) {
size_t len;
char *s;
s = (char *)lua_tolstring(lua, (-argc) + j, &len);
if (s) {
if (j != 1) log = sdscatlen(log, " ", 1);
log = sdscatlen(log, s, len);
}
}
serverLogRaw(level, log);
sdsfree(log);
return 0;
}
/* server.setresp() */
static int luaSetResp(lua_State *lua) {
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
scriptRunCtx *rctx = luaGetFromRegistry(lua, REGISTRY_RUN_CTX_NAME);
serverAssert(rctx); /* Only supported inside script invocation */
int argc = lua_gettop(lua);
if (argc != 1) {
luaPushError(lua, "server.setresp() requires one argument.");
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
return luaError(lua);
}
int resp = lua_tonumber(lua, -argc);
if (resp != 2 && resp != 3) {
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
luaPushError(lua, "RESP version must be 2 or 3.");
return luaError(lua);
}
scriptSetResp(rctx, resp);
return 0;
}
/* ---------------------------------------------------------------------------
* Lua engine initialization and reset.
* ------------------------------------------------------------------------- */
static void luaLoadLib(lua_State *lua, const char *libname, lua_CFunction luafunc) {
lua_pushcfunction(lua, luafunc);
lua_pushstring(lua, libname);
lua_call(lua, 1, 0);
}
LUALIB_API int(luaopen_cjson)(lua_State *L);
LUALIB_API int(luaopen_struct)(lua_State *L);
LUALIB_API int(luaopen_cmsgpack)(lua_State *L);
LUALIB_API int(luaopen_bit)(lua_State *L);
static void luaLoadLibraries(lua_State *lua) {
luaLoadLib(lua, "", luaopen_base);
luaLoadLib(lua, LUA_TABLIBNAME, luaopen_table);
luaLoadLib(lua, LUA_STRLIBNAME, luaopen_string);
luaLoadLib(lua, LUA_MATHLIBNAME, luaopen_math);
luaLoadLib(lua, LUA_DBLIBNAME, luaopen_debug);
luaLoadLib(lua, LUA_OSLIBNAME, luaopen_os);
luaLoadLib(lua, "cjson", luaopen_cjson);
luaLoadLib(lua, "struct", luaopen_struct);
luaLoadLib(lua, "cmsgpack", luaopen_cmsgpack);
luaLoadLib(lua, "bit", luaopen_bit);
#if 0 /* Stuff that we don't load currently, for sandboxing concerns. */
luaLoadLib(lua, LUA_LOADLIBNAME, luaopen_package);
#endif
}
Function Flags support (no-writes, no-cluster, allow-state, allow-oom) (#10066) # Redis Functions Flags Following the discussion on #10025 Added Functions Flags support. The PR is divided to 2 sections: * Add named argument support to `redis.register_function` API. * Add support for function flags ## `redis.register_function` named argument support The first part of the PR adds support for named argument on `redis.register_function`, example: ``` redis.register_function{ function_name='f1', callback=function() return 'hello' end, description='some desc' } ``` The positional arguments is also kept, which means that it still possible to write: ``` redis.register_function('f1', function() return 'hello' end) ``` But notice that it is no longer possible to pass the optional description argument on the positional argument version. Positional argument was change to allow passing only the mandatory arguments (function name and callback). To pass more arguments the user must use the named argument version. As with positional arguments, the `function_name` and `callback` is mandatory and an error will be raise if those are missing. Also, an error will be raise if an unknown argument name is given or the arguments type is wrong. Tests was added to verify the new syntax. ## Functions Flags The second part of the PR is adding functions flags support. Flags are given to Redis when the engine calls `functionLibCreateFunction`, supported flags are: * `no-writes` - indicating the function perform no writes which means that it is OK to run it on: * read-only replica * Using FCALL_RO * If disk error detected It will not be possible to run a function in those situations unless the function turns on the `no-writes` flag * `allow-oom` - indicate that its OK to run the function even if Redis is in OOM state, if the function will not turn on this flag it will not be possible to run it if OOM reached (even if the function declares `no-writes` and even if `fcall_ro` is used). If this flag is set, any command will be allow on OOM (even those that is marked with CMD_DENYOOM). The assumption is that this flag is for advance users that knows its meaning and understand what they are doing, and Redis trust them to not increase the memory usage. (e.g. it could be an INCR or a modification on an existing key, or a DEL command) * `allow-state` - indicate that its OK to run the function on stale replica, in this case we will also make sure the function is only perform `stale` commands and raise an error if not. * `no-cluster` - indicate to disallow running the function if cluster is enabled. Default behaviure of functions (if no flags is given): 1. Allow functions to read and write 2. Do not run functions on OOM 3. Do not run functions on stale replica 4. Allow functions on cluster ### Lua API for functions flags On Lua engine, it is possible to give functions flags as `flags` named argument: ``` redis.register_function{function_name='f1', callback=function() return 1 end, flags={'no-writes', 'allow-oom'}, description='description'} ``` The function flags argument must be a Lua table that contains all the requested flags, The following will result in an error: * Unknown flag * Wrong flag type Default behaviour is the same as if no flags are used. Tests were added to verify all flags functionality ## Additional changes * mark FCALL and FCALL_RO with CMD_STALE flag (unlike EVAL), so that they can run if the function was registered with the `allow-stale` flag. * Verify `CMD_STALE` on `scriptCall` (`redis.call`), so it will not be possible to call commands from script while stale unless the command is marked with the `CMD_STALE` flags. so that even if the function is allowed while stale we do not allow it to bypass the `CMD_STALE` flag of commands. * Flags section was added to `FUNCTION LIST` command to provide the set of flags for each function: ``` > FUNCTION list withcode 1) 1) "library_name" 2) "test" 3) "engine" 4) "LUA" 5) "description" 6) (nil) 7) "functions" 8) 1) 1) "name" 2) "f1" 3) "description" 4) (nil) 5) "flags" 6) (empty array) 9) "library_code" 10) "redis.register_function{function_name='f1', callback=function() return 1 end}" ``` * Added API to get Redis version from within a script, The redis version can be provided using: 1. `redis.REDIS_VERSION` - string representation of the redis version in the format of MAJOR.MINOR.PATH 2. `redis.REDIS_VERSION_NUM` - number representation of the redis version in the format of `0x00MMmmpp` (`MM` - major, `mm` - minor, `pp` - patch). The number version can be used to check if version is greater or less another version. The string version can be used to return to the user or print as logs. This new API is provided to eval scripts and functions, it also possible to use this API during functions loading phase.
2022-01-14 14:02:02 +02:00
/* Return sds of the string value located on stack at the given index.
* Return NULL if the value is not a string. */
Adds support for scripting engines as Valkey modules (#1277) This PR extends the module API to support the addition of different scripting engines to execute user defined functions. The scripting engine can be implemented as a Valkey module, and can be dynamically loaded with the `loadmodule` config directive, or with the `MODULE LOAD` command. This PR also adds an example of a dummy scripting engine module, to show how to use the new module API. The dummy module is implemented in `tests/modules/helloscripting.c`. The current module API support, only allows to load scripting engines to run functions using `FCALL` command. The additions to the module API are the following: ```c /* This struct represents a scripting engine function that results from the * compilation of a script by the engine implementation. */ struct ValkeyModuleScriptingEngineCompiledFunction typedef ValkeyModuleScriptingEngineCompiledFunction **(*ValkeyModuleScriptingEngineCreateFunctionsLibraryFunc)( ValkeyModuleScriptingEngineCtx *engine_ctx, const char *code, size_t timeout, size_t *out_num_compiled_functions, char **err); typedef void (*ValkeyModuleScriptingEngineCallFunctionFunc)( ValkeyModuleCtx *module_ctx, ValkeyModuleScriptingEngineCtx *engine_ctx, ValkeyModuleScriptingEngineFunctionCtx *func_ctx, void *compiled_function, ValkeyModuleString **keys, size_t nkeys, ValkeyModuleString **args, size_t nargs); typedef size_t (*ValkeyModuleScriptingEngineGetUsedMemoryFunc)( ValkeyModuleScriptingEngineCtx *engine_ctx); typedef size_t (*ValkeyModuleScriptingEngineGetFunctionMemoryOverheadFunc)( void *compiled_function); typedef size_t (*ValkeyModuleScriptingEngineGetEngineMemoryOverheadFunc)( ValkeyModuleScriptingEngineCtx *engine_ctx); typedef void (*ValkeyModuleScriptingEngineFreeFunctionFunc)( ValkeyModuleScriptingEngineCtx *engine_ctx, void *compiled_function); /* This struct stores the callback functions implemented by the scripting * engine to provide the functionality for the `FUNCTION *` commands. */ typedef struct ValkeyModuleScriptingEngineMethodsV1 { uint64_t version; /* Version of this structure for ABI compat. */ /* Library create function callback. When a new script is loaded, this * callback will be called with the script code, and returns a list of * ValkeyModuleScriptingEngineCompiledFunc objects. */ ValkeyModuleScriptingEngineCreateFunctionsLibraryFunc create_functions_library; /* The callback function called when `FCALL` command is called on a function * registered in this engine. */ ValkeyModuleScriptingEngineCallFunctionFunc call_function; /* Function callback to get current used memory by the engine. */ ValkeyModuleScriptingEngineGetUsedMemoryFunc get_used_memory; /* Function callback to return memory overhead for a given function. */ ValkeyModuleScriptingEngineGetFunctionMemoryOverheadFunc get_function_memory_overhead; /* Function callback to return memory overhead of the engine. */ ValkeyModuleScriptingEngineGetEngineMemoryOverheadFunc get_engine_memory_overhead; /* Function callback to free the memory of a registered engine function. */ ValkeyModuleScriptingEngineFreeFunctionFunc free_function; } ValkeyModuleScriptingEngineMethodsV1; /* Registers a new scripting engine in the server. * * - `engine_name`: the name of the scripting engine. This name will match * against the engine name specified in the script header using a shebang. * * - `engine_ctx`: engine specific context pointer. * * - `engine_methods`: the struct with the scripting engine callback functions * pointers. */ int ValkeyModule_RegisterScriptingEngine(ValkeyModuleCtx *ctx, const char *engine_name, void *engine_ctx, ValkeyModuleScriptingEngineMethods engine_methods); /* Removes the scripting engine from the server. * * `engine_name` is the name of the scripting engine. * */ int ValkeyModule_UnregisterScriptingEngine(ValkeyModuleCtx *ctx, const char *engine_name); ``` --------- Signed-off-by: Ricardo Dias <ricardo.dias@percona.com>
2024-12-21 22:09:35 +00:00
robj *luaGetStringObject(lua_State *lua, int index) {
Function Flags support (no-writes, no-cluster, allow-state, allow-oom) (#10066) # Redis Functions Flags Following the discussion on #10025 Added Functions Flags support. The PR is divided to 2 sections: * Add named argument support to `redis.register_function` API. * Add support for function flags ## `redis.register_function` named argument support The first part of the PR adds support for named argument on `redis.register_function`, example: ``` redis.register_function{ function_name='f1', callback=function() return 'hello' end, description='some desc' } ``` The positional arguments is also kept, which means that it still possible to write: ``` redis.register_function('f1', function() return 'hello' end) ``` But notice that it is no longer possible to pass the optional description argument on the positional argument version. Positional argument was change to allow passing only the mandatory arguments (function name and callback). To pass more arguments the user must use the named argument version. As with positional arguments, the `function_name` and `callback` is mandatory and an error will be raise if those are missing. Also, an error will be raise if an unknown argument name is given or the arguments type is wrong. Tests was added to verify the new syntax. ## Functions Flags The second part of the PR is adding functions flags support. Flags are given to Redis when the engine calls `functionLibCreateFunction`, supported flags are: * `no-writes` - indicating the function perform no writes which means that it is OK to run it on: * read-only replica * Using FCALL_RO * If disk error detected It will not be possible to run a function in those situations unless the function turns on the `no-writes` flag * `allow-oom` - indicate that its OK to run the function even if Redis is in OOM state, if the function will not turn on this flag it will not be possible to run it if OOM reached (even if the function declares `no-writes` and even if `fcall_ro` is used). If this flag is set, any command will be allow on OOM (even those that is marked with CMD_DENYOOM). The assumption is that this flag is for advance users that knows its meaning and understand what they are doing, and Redis trust them to not increase the memory usage. (e.g. it could be an INCR or a modification on an existing key, or a DEL command) * `allow-state` - indicate that its OK to run the function on stale replica, in this case we will also make sure the function is only perform `stale` commands and raise an error if not. * `no-cluster` - indicate to disallow running the function if cluster is enabled. Default behaviure of functions (if no flags is given): 1. Allow functions to read and write 2. Do not run functions on OOM 3. Do not run functions on stale replica 4. Allow functions on cluster ### Lua API for functions flags On Lua engine, it is possible to give functions flags as `flags` named argument: ``` redis.register_function{function_name='f1', callback=function() return 1 end, flags={'no-writes', 'allow-oom'}, description='description'} ``` The function flags argument must be a Lua table that contains all the requested flags, The following will result in an error: * Unknown flag * Wrong flag type Default behaviour is the same as if no flags are used. Tests were added to verify all flags functionality ## Additional changes * mark FCALL and FCALL_RO with CMD_STALE flag (unlike EVAL), so that they can run if the function was registered with the `allow-stale` flag. * Verify `CMD_STALE` on `scriptCall` (`redis.call`), so it will not be possible to call commands from script while stale unless the command is marked with the `CMD_STALE` flags. so that even if the function is allowed while stale we do not allow it to bypass the `CMD_STALE` flag of commands. * Flags section was added to `FUNCTION LIST` command to provide the set of flags for each function: ``` > FUNCTION list withcode 1) 1) "library_name" 2) "test" 3) "engine" 4) "LUA" 5) "description" 6) (nil) 7) "functions" 8) 1) 1) "name" 2) "f1" 3) "description" 4) (nil) 5) "flags" 6) (empty array) 9) "library_code" 10) "redis.register_function{function_name='f1', callback=function() return 1 end}" ``` * Added API to get Redis version from within a script, The redis version can be provided using: 1. `redis.REDIS_VERSION` - string representation of the redis version in the format of MAJOR.MINOR.PATH 2. `redis.REDIS_VERSION_NUM` - number representation of the redis version in the format of `0x00MMmmpp` (`MM` - major, `mm` - minor, `pp` - patch). The number version can be used to check if version is greater or less another version. The string version can be used to return to the user or print as logs. This new API is provided to eval scripts and functions, it also possible to use this API during functions loading phase.
2022-01-14 14:02:02 +02:00
if (!lua_isstring(lua, index)) {
return NULL;
}
size_t len;
const char *str = lua_tolstring(lua, index, &len);
Adds support for scripting engines as Valkey modules (#1277) This PR extends the module API to support the addition of different scripting engines to execute user defined functions. The scripting engine can be implemented as a Valkey module, and can be dynamically loaded with the `loadmodule` config directive, or with the `MODULE LOAD` command. This PR also adds an example of a dummy scripting engine module, to show how to use the new module API. The dummy module is implemented in `tests/modules/helloscripting.c`. The current module API support, only allows to load scripting engines to run functions using `FCALL` command. The additions to the module API are the following: ```c /* This struct represents a scripting engine function that results from the * compilation of a script by the engine implementation. */ struct ValkeyModuleScriptingEngineCompiledFunction typedef ValkeyModuleScriptingEngineCompiledFunction **(*ValkeyModuleScriptingEngineCreateFunctionsLibraryFunc)( ValkeyModuleScriptingEngineCtx *engine_ctx, const char *code, size_t timeout, size_t *out_num_compiled_functions, char **err); typedef void (*ValkeyModuleScriptingEngineCallFunctionFunc)( ValkeyModuleCtx *module_ctx, ValkeyModuleScriptingEngineCtx *engine_ctx, ValkeyModuleScriptingEngineFunctionCtx *func_ctx, void *compiled_function, ValkeyModuleString **keys, size_t nkeys, ValkeyModuleString **args, size_t nargs); typedef size_t (*ValkeyModuleScriptingEngineGetUsedMemoryFunc)( ValkeyModuleScriptingEngineCtx *engine_ctx); typedef size_t (*ValkeyModuleScriptingEngineGetFunctionMemoryOverheadFunc)( void *compiled_function); typedef size_t (*ValkeyModuleScriptingEngineGetEngineMemoryOverheadFunc)( ValkeyModuleScriptingEngineCtx *engine_ctx); typedef void (*ValkeyModuleScriptingEngineFreeFunctionFunc)( ValkeyModuleScriptingEngineCtx *engine_ctx, void *compiled_function); /* This struct stores the callback functions implemented by the scripting * engine to provide the functionality for the `FUNCTION *` commands. */ typedef struct ValkeyModuleScriptingEngineMethodsV1 { uint64_t version; /* Version of this structure for ABI compat. */ /* Library create function callback. When a new script is loaded, this * callback will be called with the script code, and returns a list of * ValkeyModuleScriptingEngineCompiledFunc objects. */ ValkeyModuleScriptingEngineCreateFunctionsLibraryFunc create_functions_library; /* The callback function called when `FCALL` command is called on a function * registered in this engine. */ ValkeyModuleScriptingEngineCallFunctionFunc call_function; /* Function callback to get current used memory by the engine. */ ValkeyModuleScriptingEngineGetUsedMemoryFunc get_used_memory; /* Function callback to return memory overhead for a given function. */ ValkeyModuleScriptingEngineGetFunctionMemoryOverheadFunc get_function_memory_overhead; /* Function callback to return memory overhead of the engine. */ ValkeyModuleScriptingEngineGetEngineMemoryOverheadFunc get_engine_memory_overhead; /* Function callback to free the memory of a registered engine function. */ ValkeyModuleScriptingEngineFreeFunctionFunc free_function; } ValkeyModuleScriptingEngineMethodsV1; /* Registers a new scripting engine in the server. * * - `engine_name`: the name of the scripting engine. This name will match * against the engine name specified in the script header using a shebang. * * - `engine_ctx`: engine specific context pointer. * * - `engine_methods`: the struct with the scripting engine callback functions * pointers. */ int ValkeyModule_RegisterScriptingEngine(ValkeyModuleCtx *ctx, const char *engine_name, void *engine_ctx, ValkeyModuleScriptingEngineMethods engine_methods); /* Removes the scripting engine from the server. * * `engine_name` is the name of the scripting engine. * */ int ValkeyModule_UnregisterScriptingEngine(ValkeyModuleCtx *ctx, const char *engine_name); ``` --------- Signed-off-by: Ricardo Dias <ricardo.dias@percona.com>
2024-12-21 22:09:35 +00:00
robj *str_obj = createStringObject(str, len);
return str_obj;
Function Flags support (no-writes, no-cluster, allow-state, allow-oom) (#10066) # Redis Functions Flags Following the discussion on #10025 Added Functions Flags support. The PR is divided to 2 sections: * Add named argument support to `redis.register_function` API. * Add support for function flags ## `redis.register_function` named argument support The first part of the PR adds support for named argument on `redis.register_function`, example: ``` redis.register_function{ function_name='f1', callback=function() return 'hello' end, description='some desc' } ``` The positional arguments is also kept, which means that it still possible to write: ``` redis.register_function('f1', function() return 'hello' end) ``` But notice that it is no longer possible to pass the optional description argument on the positional argument version. Positional argument was change to allow passing only the mandatory arguments (function name and callback). To pass more arguments the user must use the named argument version. As with positional arguments, the `function_name` and `callback` is mandatory and an error will be raise if those are missing. Also, an error will be raise if an unknown argument name is given or the arguments type is wrong. Tests was added to verify the new syntax. ## Functions Flags The second part of the PR is adding functions flags support. Flags are given to Redis when the engine calls `functionLibCreateFunction`, supported flags are: * `no-writes` - indicating the function perform no writes which means that it is OK to run it on: * read-only replica * Using FCALL_RO * If disk error detected It will not be possible to run a function in those situations unless the function turns on the `no-writes` flag * `allow-oom` - indicate that its OK to run the function even if Redis is in OOM state, if the function will not turn on this flag it will not be possible to run it if OOM reached (even if the function declares `no-writes` and even if `fcall_ro` is used). If this flag is set, any command will be allow on OOM (even those that is marked with CMD_DENYOOM). The assumption is that this flag is for advance users that knows its meaning and understand what they are doing, and Redis trust them to not increase the memory usage. (e.g. it could be an INCR or a modification on an existing key, or a DEL command) * `allow-state` - indicate that its OK to run the function on stale replica, in this case we will also make sure the function is only perform `stale` commands and raise an error if not. * `no-cluster` - indicate to disallow running the function if cluster is enabled. Default behaviure of functions (if no flags is given): 1. Allow functions to read and write 2. Do not run functions on OOM 3. Do not run functions on stale replica 4. Allow functions on cluster ### Lua API for functions flags On Lua engine, it is possible to give functions flags as `flags` named argument: ``` redis.register_function{function_name='f1', callback=function() return 1 end, flags={'no-writes', 'allow-oom'}, description='description'} ``` The function flags argument must be a Lua table that contains all the requested flags, The following will result in an error: * Unknown flag * Wrong flag type Default behaviour is the same as if no flags are used. Tests were added to verify all flags functionality ## Additional changes * mark FCALL and FCALL_RO with CMD_STALE flag (unlike EVAL), so that they can run if the function was registered with the `allow-stale` flag. * Verify `CMD_STALE` on `scriptCall` (`redis.call`), so it will not be possible to call commands from script while stale unless the command is marked with the `CMD_STALE` flags. so that even if the function is allowed while stale we do not allow it to bypass the `CMD_STALE` flag of commands. * Flags section was added to `FUNCTION LIST` command to provide the set of flags for each function: ``` > FUNCTION list withcode 1) 1) "library_name" 2) "test" 3) "engine" 4) "LUA" 5) "description" 6) (nil) 7) "functions" 8) 1) 1) "name" 2) "f1" 3) "description" 4) (nil) 5) "flags" 6) (empty array) 9) "library_code" 10) "redis.register_function{function_name='f1', callback=function() return 1 end}" ``` * Added API to get Redis version from within a script, The redis version can be provided using: 1. `redis.REDIS_VERSION` - string representation of the redis version in the format of MAJOR.MINOR.PATH 2. `redis.REDIS_VERSION_NUM` - number representation of the redis version in the format of `0x00MMmmpp` (`MM` - major, `mm` - minor, `pp` - patch). The number version can be used to check if version is greater or less another version. The string version can be used to return to the user or print as logs. This new API is provided to eval scripts and functions, it also possible to use this API during functions loading phase.
2022-01-14 14:02:02 +02:00
}
static int luaProtectedTableError(lua_State *lua) {
int argc = lua_gettop(lua);
if (argc != 2) {
serverLog(LL_WARNING, "malicious code trying to call luaProtectedTableError with wrong arguments");
luaL_error(lua, "Wrong number of arguments to luaProtectedTableError");
}
if (!lua_isstring(lua, -1) && !lua_isnumber(lua, -1)) {
luaL_error(lua, "Second argument to luaProtectedTableError must be a string or number");
}
const char *variable_name = lua_tostring(lua, -1);
luaL_error(lua, "Script attempted to access nonexistent global variable '%s'", variable_name);
return 0;
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
}
/* Set a special metatable on the table on the top of the stack.
* The metatable will raise an error if the user tries to fetch
* an un-existing value.
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
*
* The function assumes the Lua stack have a least enough
* space to push 2 element, its up to the caller to verify
* this before calling this function. */
void luaSetErrorMetatable(lua_State *lua) {
lua_newtable(lua); /* push metatable */
lua_pushcfunction(lua, luaProtectedTableError); /* push get error handler */
lua_setfield(lua, -2, "__index");
lua_setmetatable(lua, -2);
}
static int luaNewIndexAllowList(lua_State *lua) {
int argc = lua_gettop(lua);
if (argc != 3) {
serverLog(LL_WARNING, "malicious code trying to call luaNewIndexAllowList with wrong arguments");
luaL_error(lua, "Wrong number of arguments to luaNewIndexAllowList");
}
if (!lua_istable(lua, -3)) {
luaL_error(lua, "first argument to luaNewIndexAllowList must be a table");
}
if (!lua_isstring(lua, -2) && !lua_isnumber(lua, -2)) {
luaL_error(lua, "Second argument to luaNewIndexAllowList must be a string or number");
}
const char *variable_name = lua_tostring(lua, -2);
/* check if the key is in our allow list */
char ***allow_l = allow_lists;
for (; *allow_l; ++allow_l) {
char **c = *allow_l;
for (; *c; ++c) {
if (strcmp(*c, variable_name) == 0) {
break;
}
}
if (*c) {
break;
}
}
if (!*allow_l) {
/* Search the value on the back list, if its there we know that it was removed
* on purpose and there is no need to print a warning. */
char **c = deny_list;
for (; *c; ++c) {
if (strcmp(*c, variable_name) == 0) {
break;
}
}
if (!*c) {
serverLog(LL_WARNING,
"A key '%s' was added to Lua globals which is not on the globals allow list nor listed on the "
"deny list.",
variable_name);
}
} else {
lua_rawset(lua, -3);
}
return 0;
}
/* Set a metatable with '__newindex' function that verify that
* the new index appears on our globals while list.
*
* The metatable is set on the table which located on the top
* of the stack.
*/
void luaSetAllowListProtection(lua_State *lua) {
lua_newtable(lua); /* push metatable */
lua_pushcfunction(lua, luaNewIndexAllowList); /* push get error handler */
lua_setfield(lua, -2, "__newindex");
lua_setmetatable(lua, -2);
}
/* Set the readonly flag on the table located on the top of the stack
* and recursively call this function on each table located on the original
* table. Also, recursively call this function on the metatables.*/
void luaSetTableProtectionRecursively(lua_State *lua) {
/* This protect us from a loop in case we already visited the table
* For example, globals has '_G' key which is pointing back to globals. */
if (lua_isreadonlytable(lua, -1)) {
return;
}
/* protect the current table */
lua_enablereadonlytable(lua, -1, 1);
lua_checkstack(lua, 2);
lua_pushnil(lua); /* Use nil to start iteration. */
while (lua_next(lua, -2)) {
/* Stack now: table, key, value */
if (lua_istable(lua, -1)) {
luaSetTableProtectionRecursively(lua);
}
lua_pop(lua, 1);
}
/* protect the metatable if exists */
if (lua_getmetatable(lua, -1)) {
luaSetTableProtectionRecursively(lua);
lua_pop(lua, 1); /* pop the metatable */
}
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
}
Function Flags support (no-writes, no-cluster, allow-state, allow-oom) (#10066) # Redis Functions Flags Following the discussion on #10025 Added Functions Flags support. The PR is divided to 2 sections: * Add named argument support to `redis.register_function` API. * Add support for function flags ## `redis.register_function` named argument support The first part of the PR adds support for named argument on `redis.register_function`, example: ``` redis.register_function{ function_name='f1', callback=function() return 'hello' end, description='some desc' } ``` The positional arguments is also kept, which means that it still possible to write: ``` redis.register_function('f1', function() return 'hello' end) ``` But notice that it is no longer possible to pass the optional description argument on the positional argument version. Positional argument was change to allow passing only the mandatory arguments (function name and callback). To pass more arguments the user must use the named argument version. As with positional arguments, the `function_name` and `callback` is mandatory and an error will be raise if those are missing. Also, an error will be raise if an unknown argument name is given or the arguments type is wrong. Tests was added to verify the new syntax. ## Functions Flags The second part of the PR is adding functions flags support. Flags are given to Redis when the engine calls `functionLibCreateFunction`, supported flags are: * `no-writes` - indicating the function perform no writes which means that it is OK to run it on: * read-only replica * Using FCALL_RO * If disk error detected It will not be possible to run a function in those situations unless the function turns on the `no-writes` flag * `allow-oom` - indicate that its OK to run the function even if Redis is in OOM state, if the function will not turn on this flag it will not be possible to run it if OOM reached (even if the function declares `no-writes` and even if `fcall_ro` is used). If this flag is set, any command will be allow on OOM (even those that is marked with CMD_DENYOOM). The assumption is that this flag is for advance users that knows its meaning and understand what they are doing, and Redis trust them to not increase the memory usage. (e.g. it could be an INCR or a modification on an existing key, or a DEL command) * `allow-state` - indicate that its OK to run the function on stale replica, in this case we will also make sure the function is only perform `stale` commands and raise an error if not. * `no-cluster` - indicate to disallow running the function if cluster is enabled. Default behaviure of functions (if no flags is given): 1. Allow functions to read and write 2. Do not run functions on OOM 3. Do not run functions on stale replica 4. Allow functions on cluster ### Lua API for functions flags On Lua engine, it is possible to give functions flags as `flags` named argument: ``` redis.register_function{function_name='f1', callback=function() return 1 end, flags={'no-writes', 'allow-oom'}, description='description'} ``` The function flags argument must be a Lua table that contains all the requested flags, The following will result in an error: * Unknown flag * Wrong flag type Default behaviour is the same as if no flags are used. Tests were added to verify all flags functionality ## Additional changes * mark FCALL and FCALL_RO with CMD_STALE flag (unlike EVAL), so that they can run if the function was registered with the `allow-stale` flag. * Verify `CMD_STALE` on `scriptCall` (`redis.call`), so it will not be possible to call commands from script while stale unless the command is marked with the `CMD_STALE` flags. so that even if the function is allowed while stale we do not allow it to bypass the `CMD_STALE` flag of commands. * Flags section was added to `FUNCTION LIST` command to provide the set of flags for each function: ``` > FUNCTION list withcode 1) 1) "library_name" 2) "test" 3) "engine" 4) "LUA" 5) "description" 6) (nil) 7) "functions" 8) 1) 1) "name" 2) "f1" 3) "description" 4) (nil) 5) "flags" 6) (empty array) 9) "library_code" 10) "redis.register_function{function_name='f1', callback=function() return 1 end}" ``` * Added API to get Redis version from within a script, The redis version can be provided using: 1. `redis.REDIS_VERSION` - string representation of the redis version in the format of MAJOR.MINOR.PATH 2. `redis.REDIS_VERSION_NUM` - number representation of the redis version in the format of `0x00MMmmpp` (`MM` - major, `mm` - minor, `pp` - patch). The number version can be used to check if version is greater or less another version. The string version can be used to return to the user or print as logs. This new API is provided to eval scripts and functions, it also possible to use this API during functions loading phase.
2022-01-14 14:02:02 +02:00
void luaRegisterVersion(lua_State *lua) {
/* For legacy compatibility reasons include Redis versions. */
Function Flags support (no-writes, no-cluster, allow-state, allow-oom) (#10066) # Redis Functions Flags Following the discussion on #10025 Added Functions Flags support. The PR is divided to 2 sections: * Add named argument support to `redis.register_function` API. * Add support for function flags ## `redis.register_function` named argument support The first part of the PR adds support for named argument on `redis.register_function`, example: ``` redis.register_function{ function_name='f1', callback=function() return 'hello' end, description='some desc' } ``` The positional arguments is also kept, which means that it still possible to write: ``` redis.register_function('f1', function() return 'hello' end) ``` But notice that it is no longer possible to pass the optional description argument on the positional argument version. Positional argument was change to allow passing only the mandatory arguments (function name and callback). To pass more arguments the user must use the named argument version. As with positional arguments, the `function_name` and `callback` is mandatory and an error will be raise if those are missing. Also, an error will be raise if an unknown argument name is given or the arguments type is wrong. Tests was added to verify the new syntax. ## Functions Flags The second part of the PR is adding functions flags support. Flags are given to Redis when the engine calls `functionLibCreateFunction`, supported flags are: * `no-writes` - indicating the function perform no writes which means that it is OK to run it on: * read-only replica * Using FCALL_RO * If disk error detected It will not be possible to run a function in those situations unless the function turns on the `no-writes` flag * `allow-oom` - indicate that its OK to run the function even if Redis is in OOM state, if the function will not turn on this flag it will not be possible to run it if OOM reached (even if the function declares `no-writes` and even if `fcall_ro` is used). If this flag is set, any command will be allow on OOM (even those that is marked with CMD_DENYOOM). The assumption is that this flag is for advance users that knows its meaning and understand what they are doing, and Redis trust them to not increase the memory usage. (e.g. it could be an INCR or a modification on an existing key, or a DEL command) * `allow-state` - indicate that its OK to run the function on stale replica, in this case we will also make sure the function is only perform `stale` commands and raise an error if not. * `no-cluster` - indicate to disallow running the function if cluster is enabled. Default behaviure of functions (if no flags is given): 1. Allow functions to read and write 2. Do not run functions on OOM 3. Do not run functions on stale replica 4. Allow functions on cluster ### Lua API for functions flags On Lua engine, it is possible to give functions flags as `flags` named argument: ``` redis.register_function{function_name='f1', callback=function() return 1 end, flags={'no-writes', 'allow-oom'}, description='description'} ``` The function flags argument must be a Lua table that contains all the requested flags, The following will result in an error: * Unknown flag * Wrong flag type Default behaviour is the same as if no flags are used. Tests were added to verify all flags functionality ## Additional changes * mark FCALL and FCALL_RO with CMD_STALE flag (unlike EVAL), so that they can run if the function was registered with the `allow-stale` flag. * Verify `CMD_STALE` on `scriptCall` (`redis.call`), so it will not be possible to call commands from script while stale unless the command is marked with the `CMD_STALE` flags. so that even if the function is allowed while stale we do not allow it to bypass the `CMD_STALE` flag of commands. * Flags section was added to `FUNCTION LIST` command to provide the set of flags for each function: ``` > FUNCTION list withcode 1) 1) "library_name" 2) "test" 3) "engine" 4) "LUA" 5) "description" 6) (nil) 7) "functions" 8) 1) 1) "name" 2) "f1" 3) "description" 4) (nil) 5) "flags" 6) (empty array) 9) "library_code" 10) "redis.register_function{function_name='f1', callback=function() return 1 end}" ``` * Added API to get Redis version from within a script, The redis version can be provided using: 1. `redis.REDIS_VERSION` - string representation of the redis version in the format of MAJOR.MINOR.PATH 2. `redis.REDIS_VERSION_NUM` - number representation of the redis version in the format of `0x00MMmmpp` (`MM` - major, `mm` - minor, `pp` - patch). The number version can be used to check if version is greater or less another version. The string version can be used to return to the user or print as logs. This new API is provided to eval scripts and functions, it also possible to use this API during functions loading phase.
2022-01-14 14:02:02 +02:00
lua_pushstring(lua, "REDIS_VERSION_NUM");
lua_pushnumber(lua, REDIS_VERSION_NUM);
lua_settable(lua, -3);
lua_pushstring(lua, "REDIS_VERSION");
lua_pushstring(lua, REDIS_VERSION);
lua_settable(lua, -3);
/* Now push the Valkey version information. */
lua_pushstring(lua, "VALKEY_VERSION_NUM");
lua_pushnumber(lua, VALKEY_VERSION_NUM);
lua_settable(lua, -3);
lua_pushstring(lua, "VALKEY_VERSION");
lua_pushstring(lua, VALKEY_VERSION);
lua_settable(lua, -3);
lua_pushstring(lua, "SERVER_NAME");
lua_pushstring(lua, SERVER_NAME);
lua_settable(lua, -3);
Function Flags support (no-writes, no-cluster, allow-state, allow-oom) (#10066) # Redis Functions Flags Following the discussion on #10025 Added Functions Flags support. The PR is divided to 2 sections: * Add named argument support to `redis.register_function` API. * Add support for function flags ## `redis.register_function` named argument support The first part of the PR adds support for named argument on `redis.register_function`, example: ``` redis.register_function{ function_name='f1', callback=function() return 'hello' end, description='some desc' } ``` The positional arguments is also kept, which means that it still possible to write: ``` redis.register_function('f1', function() return 'hello' end) ``` But notice that it is no longer possible to pass the optional description argument on the positional argument version. Positional argument was change to allow passing only the mandatory arguments (function name and callback). To pass more arguments the user must use the named argument version. As with positional arguments, the `function_name` and `callback` is mandatory and an error will be raise if those are missing. Also, an error will be raise if an unknown argument name is given or the arguments type is wrong. Tests was added to verify the new syntax. ## Functions Flags The second part of the PR is adding functions flags support. Flags are given to Redis when the engine calls `functionLibCreateFunction`, supported flags are: * `no-writes` - indicating the function perform no writes which means that it is OK to run it on: * read-only replica * Using FCALL_RO * If disk error detected It will not be possible to run a function in those situations unless the function turns on the `no-writes` flag * `allow-oom` - indicate that its OK to run the function even if Redis is in OOM state, if the function will not turn on this flag it will not be possible to run it if OOM reached (even if the function declares `no-writes` and even if `fcall_ro` is used). If this flag is set, any command will be allow on OOM (even those that is marked with CMD_DENYOOM). The assumption is that this flag is for advance users that knows its meaning and understand what they are doing, and Redis trust them to not increase the memory usage. (e.g. it could be an INCR or a modification on an existing key, or a DEL command) * `allow-state` - indicate that its OK to run the function on stale replica, in this case we will also make sure the function is only perform `stale` commands and raise an error if not. * `no-cluster` - indicate to disallow running the function if cluster is enabled. Default behaviure of functions (if no flags is given): 1. Allow functions to read and write 2. Do not run functions on OOM 3. Do not run functions on stale replica 4. Allow functions on cluster ### Lua API for functions flags On Lua engine, it is possible to give functions flags as `flags` named argument: ``` redis.register_function{function_name='f1', callback=function() return 1 end, flags={'no-writes', 'allow-oom'}, description='description'} ``` The function flags argument must be a Lua table that contains all the requested flags, The following will result in an error: * Unknown flag * Wrong flag type Default behaviour is the same as if no flags are used. Tests were added to verify all flags functionality ## Additional changes * mark FCALL and FCALL_RO with CMD_STALE flag (unlike EVAL), so that they can run if the function was registered with the `allow-stale` flag. * Verify `CMD_STALE` on `scriptCall` (`redis.call`), so it will not be possible to call commands from script while stale unless the command is marked with the `CMD_STALE` flags. so that even if the function is allowed while stale we do not allow it to bypass the `CMD_STALE` flag of commands. * Flags section was added to `FUNCTION LIST` command to provide the set of flags for each function: ``` > FUNCTION list withcode 1) 1) "library_name" 2) "test" 3) "engine" 4) "LUA" 5) "description" 6) (nil) 7) "functions" 8) 1) 1) "name" 2) "f1" 3) "description" 4) (nil) 5) "flags" 6) (empty array) 9) "library_code" 10) "redis.register_function{function_name='f1', callback=function() return 1 end}" ``` * Added API to get Redis version from within a script, The redis version can be provided using: 1. `redis.REDIS_VERSION` - string representation of the redis version in the format of MAJOR.MINOR.PATH 2. `redis.REDIS_VERSION_NUM` - number representation of the redis version in the format of `0x00MMmmpp` (`MM` - major, `mm` - minor, `pp` - patch). The number version can be used to check if version is greater or less another version. The string version can be used to return to the user or print as logs. This new API is provided to eval scripts and functions, it also possible to use this API during functions loading phase.
2022-01-14 14:02:02 +02:00
}
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
void luaRegisterLogFunction(lua_State *lua) {
/* server.log and log levels. */
lua_pushstring(lua, "log");
lua_pushcfunction(lua, luaLogCommand);
lua_settable(lua, -3);
lua_pushstring(lua, "LOG_DEBUG");
lua_pushnumber(lua, LL_DEBUG);
lua_settable(lua, -3);
lua_pushstring(lua, "LOG_VERBOSE");
lua_pushnumber(lua, LL_VERBOSE);
lua_settable(lua, -3);
lua_pushstring(lua, "LOG_NOTICE");
lua_pushnumber(lua, LL_NOTICE);
lua_settable(lua, -3);
lua_pushstring(lua, "LOG_WARNING");
lua_pushnumber(lua, LL_WARNING);
lua_settable(lua, -3);
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
}
/*
* Adds server.* functions/fields to lua such as server.call etc.
* This function only handles fields common between Functions and LUA scripting.
* scriptingInit() and functionsInit() may add additional fields specific to each.
*/
void luaRegisterServerAPI(lua_State *lua) {
/* In addition to registering server.call/pcall API, we will throw a custom message when a script accesses
* undefined global variable. LUA stores global variables in the global table, accessible to us on stack at virtual
* index = LUA_GLOBALSINDEX. We will set __index handler in global table's metatable to a custom C function to
* achieve this - handled by luaSetAllowListProtection. Refer to https://www.lua.org/pil/13.4.1.html for
* documentation on __index and https://www.lua.org/pil/contents.html#13 for documentation on metatables. We need to
* pass global table to lua invocations as parameters. To achieve this, lua_pushvalue invocation brings global
* variable table to the top of the stack by pushing value from global index onto the stack. And lua_pop invocation
* after luaSetAllowListProtection removes it - resetting the stack to its original state. */
lua_pushvalue(lua, LUA_GLOBALSINDEX);
luaSetAllowListProtection(lua);
lua_pop(lua, 1);
/* Add default C functions provided in deps/lua codebase to handle basic data types such as table, string etc. */
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
luaLoadLibraries(lua);
/* Before Redis OSS 7, Lua used to return error messages as strings from pcall function. With Valkey (or Redis OSS 7), Lua now returns
* error messages as tables. To keep backwards compatibility, we wrap the Lua pcall function with our own
* implementation of C function that converts table to string. */
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
lua_pushcfunction(lua, luaRedisPcall);
lua_setglobal(lua, "pcall");
/* Create a top level table object on the stack to temporarily hold fields for 'server' table. We will name it as
* 'server' and send it to LUA at the very end. Also add 'call' and 'pcall' functions to the table. */
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
lua_newtable(lua);
lua_pushstring(lua, "call");
lua_pushcfunction(lua, luaRedisCallCommand);
lua_settable(lua, -3);
lua_pushstring(lua, "pcall");
lua_pushcfunction(lua, luaRedisPCallCommand);
lua_settable(lua, -3);
/* Add server.log function and debug level constants. LUA scripts use it to print messages to server log. */
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
luaRegisterLogFunction(lua);
/* Add SERVER_VERSION_NUM, SERVER_VERSION and SERVER_NAME fields with appropriate values. */
Function Flags support (no-writes, no-cluster, allow-state, allow-oom) (#10066) # Redis Functions Flags Following the discussion on #10025 Added Functions Flags support. The PR is divided to 2 sections: * Add named argument support to `redis.register_function` API. * Add support for function flags ## `redis.register_function` named argument support The first part of the PR adds support for named argument on `redis.register_function`, example: ``` redis.register_function{ function_name='f1', callback=function() return 'hello' end, description='some desc' } ``` The positional arguments is also kept, which means that it still possible to write: ``` redis.register_function('f1', function() return 'hello' end) ``` But notice that it is no longer possible to pass the optional description argument on the positional argument version. Positional argument was change to allow passing only the mandatory arguments (function name and callback). To pass more arguments the user must use the named argument version. As with positional arguments, the `function_name` and `callback` is mandatory and an error will be raise if those are missing. Also, an error will be raise if an unknown argument name is given or the arguments type is wrong. Tests was added to verify the new syntax. ## Functions Flags The second part of the PR is adding functions flags support. Flags are given to Redis when the engine calls `functionLibCreateFunction`, supported flags are: * `no-writes` - indicating the function perform no writes which means that it is OK to run it on: * read-only replica * Using FCALL_RO * If disk error detected It will not be possible to run a function in those situations unless the function turns on the `no-writes` flag * `allow-oom` - indicate that its OK to run the function even if Redis is in OOM state, if the function will not turn on this flag it will not be possible to run it if OOM reached (even if the function declares `no-writes` and even if `fcall_ro` is used). If this flag is set, any command will be allow on OOM (even those that is marked with CMD_DENYOOM). The assumption is that this flag is for advance users that knows its meaning and understand what they are doing, and Redis trust them to not increase the memory usage. (e.g. it could be an INCR or a modification on an existing key, or a DEL command) * `allow-state` - indicate that its OK to run the function on stale replica, in this case we will also make sure the function is only perform `stale` commands and raise an error if not. * `no-cluster` - indicate to disallow running the function if cluster is enabled. Default behaviure of functions (if no flags is given): 1. Allow functions to read and write 2. Do not run functions on OOM 3. Do not run functions on stale replica 4. Allow functions on cluster ### Lua API for functions flags On Lua engine, it is possible to give functions flags as `flags` named argument: ``` redis.register_function{function_name='f1', callback=function() return 1 end, flags={'no-writes', 'allow-oom'}, description='description'} ``` The function flags argument must be a Lua table that contains all the requested flags, The following will result in an error: * Unknown flag * Wrong flag type Default behaviour is the same as if no flags are used. Tests were added to verify all flags functionality ## Additional changes * mark FCALL and FCALL_RO with CMD_STALE flag (unlike EVAL), so that they can run if the function was registered with the `allow-stale` flag. * Verify `CMD_STALE` on `scriptCall` (`redis.call`), so it will not be possible to call commands from script while stale unless the command is marked with the `CMD_STALE` flags. so that even if the function is allowed while stale we do not allow it to bypass the `CMD_STALE` flag of commands. * Flags section was added to `FUNCTION LIST` command to provide the set of flags for each function: ``` > FUNCTION list withcode 1) 1) "library_name" 2) "test" 3) "engine" 4) "LUA" 5) "description" 6) (nil) 7) "functions" 8) 1) 1) "name" 2) "f1" 3) "description" 4) (nil) 5) "flags" 6) (empty array) 9) "library_code" 10) "redis.register_function{function_name='f1', callback=function() return 1 end}" ``` * Added API to get Redis version from within a script, The redis version can be provided using: 1. `redis.REDIS_VERSION` - string representation of the redis version in the format of MAJOR.MINOR.PATH 2. `redis.REDIS_VERSION_NUM` - number representation of the redis version in the format of `0x00MMmmpp` (`MM` - major, `mm` - minor, `pp` - patch). The number version can be used to check if version is greater or less another version. The string version can be used to return to the user or print as logs. This new API is provided to eval scripts and functions, it also possible to use this API during functions loading phase.
2022-01-14 14:02:02 +02:00
luaRegisterVersion(lua);
/* Add server.setresp function to allow LUA scripts to change the RESP version for server.call and server.pcall
* invocations. */
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
lua_pushstring(lua, "setresp");
lua_pushcfunction(lua, luaSetResp);
lua_settable(lua, -3);
/* Add server.sha1hex function. */
lua_pushstring(lua, "sha1hex");
lua_pushcfunction(lua, luaRedisSha1hexCommand);
lua_settable(lua, -3);
/* Add server.error_reply and server.status_reply functions. */
lua_pushstring(lua, "error_reply");
lua_pushcfunction(lua, luaRedisErrorReplyCommand);
lua_settable(lua, -3);
lua_pushstring(lua, "status_reply");
lua_pushcfunction(lua, luaRedisStatusReplyCommand);
lua_settable(lua, -3);
/* Add server.set_repl function and associated flags. */
lua_pushstring(lua, "set_repl");
lua_pushcfunction(lua, luaRedisSetReplCommand);
lua_settable(lua, -3);
lua_pushstring(lua, "REPL_NONE");
lua_pushnumber(lua, PROPAGATE_NONE);
lua_settable(lua, -3);
lua_pushstring(lua, "REPL_AOF");
lua_pushnumber(lua, PROPAGATE_AOF);
lua_settable(lua, -3);
lua_pushstring(lua, "REPL_SLAVE");
lua_pushnumber(lua, PROPAGATE_REPL);
lua_settable(lua, -3);
lua_pushstring(lua, "REPL_REPLICA");
lua_pushnumber(lua, PROPAGATE_REPL);
lua_settable(lua, -3);
lua_pushstring(lua, "REPL_ALL");
lua_pushnumber(lua, PROPAGATE_AOF | PROPAGATE_REPL);
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
lua_settable(lua, -3);
/* Add server.acl_check_cmd function. */
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
lua_pushstring(lua, "acl_check_cmd");
lua_pushcfunction(lua, luaRedisAclCheckCmdPermissionsCommand);
lua_settable(lua, -3);
acl check api for functions and eval (#10220) Changes: 1. Adds the `redis.acl_check_cmd()` api to lua scripts. It can be used to check if the current user has permissions to execute a given command. The new function receives the command to check as an argument exactly like `redis.call()` receives the command to execute as an argument. 2. In the PR I unified the code used to convert lua arguments to redis argv arguments from both the new `redis.acl_check_cmd()` API and the `redis.[p]call()` API. This cleans up potential duplicate code. 3. While doing the refactoring in 2 I noticed there's an optimization to reduce allocation calls when parsing lua arguments into an `argv` array in the `redis.[p]call()` implementation. These optimizations were introduced years ago in 48c49c485155ba9e4a7851fd1644c171674c6f0f and 4f686555ce962e6632235d824512ea8fdeda003c. It is unclear why this was added. The original commit message claims a 4% performance increase which I couldn't recreate and might not be worth it even if it did recreate. This PR removes that optimization. Following are details of the benchmark I did that couldn't reveal any performance improvements due to this optimization: ``` benchmark 1: src/redis-benchmark -P 500 -n 10000000 eval 'return redis.call("ping")' 0 benchmark 2: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__")' 0 benchmark 3: src/redis-benchmark -P 500 -r 1000 -n 100000 eval "for i=1,100,1 do redis.call('set','kk'..i,'vv'..__rand_int__) end return redis.call('get','kk5')" 0 benchmark 4: src/redis-benchmark -P 500 -r 1000 -n 1000000 eval 'return redis.call("mset","k1__rand_int__","v1__rand_int__","k2__rand_int__","v2__rand_int__","k3__rand_int__","v3__rand_int__","k4__rand_int__","v4__rand_int__xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx")' ``` I ran the benchmark on this branch with and without commit 68b71680a4d3bb8f0509e06578a9f15d05b92a47 Results in requests per second: cmd | without optimization | without optimization 2nd run | with original optimization | with original optimization 2nd run -- | -- | -- | -- | -- 1 | 461233.34 | 477395.31 | 471098.16 | 469946.91 2 | 34774.14 | 35469.8 | 35149.38 | 34464.93 3 | 6390.59 | 6281.41 | 6146.28 | 6464.12 4 | 28005.71 |   | 27965.77 |   As you can see, different use cases showed identical or negligible performance differences. So finally I decided to chuck the original optimization and simplify the code.
2022-02-07 08:04:01 +02:00
/* Finally set the table as 'server' global var.
* We will also alias it to 'redis' global var for backwards compatibility. */
lua_setglobal(lua, SERVER_API_NAME);
/* lua_getglobal invocation retrieves the 'server' variable value to the stack.
* lua_setglobal invocation uses the value from stack to set 'redis' global variable.
* This is not a deep copy but is enough for our purpose here. */
lua_getglobal(lua, SERVER_API_NAME);
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
lua_setglobal(lua, REDIS_API_NAME);
/* Replace math.random and math.randomseed with custom implementations. */
lua_getglobal(lua, "math");
lua_pushstring(lua, "random");
lua_pushcfunction(lua, server_math_random);
lua_settable(lua, -3);
lua_pushstring(lua, "randomseed");
lua_pushcfunction(lua, server_math_randomseed);
lua_settable(lua, -3);
/* overwrite value of global variable 'math' with this modified math table present on top of stack. */
lua_setglobal(lua, "math");
}
/* Set an array of String Objects as a Lua array (table) stored into a
* global variable. */
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
static void luaCreateArray(lua_State *lua, robj **elev, int elec) {
int j;
lua_createtable(lua, elec, 0);
for (j = 0; j < elec; j++) {
lua_pushlstring(lua, (char *)elev[j]->ptr, sdslen(elev[j]->ptr));
lua_rawseti(lua, -2, j + 1);
}
}
/* ---------------------------------------------------------------------------
* Custom provided math.random
* ------------------------------------------------------------------------- */
/* We replace math.random() with our implementation that is not affected
* by specific libc random() implementations and will output the same sequence
* (for the same seed) in every arch. */
/* The following implementation is the one shipped with Lua itself but with
* rand() replaced by serverLrand48(). */
static int server_math_random(lua_State *L) {
/* the `%' avoids the (rare) case of r==1, and is needed also because on
some systems (SunOS!) `rand()' may return a value larger than RAND_MAX */
lua_Number r = (lua_Number)(serverLrand48() % SERVER_LRAND48_MAX) /
(lua_Number)SERVER_LRAND48_MAX;
switch (lua_gettop(L)) { /* check number of arguments */
case 0: { /* no arguments */
lua_pushnumber(L, r); /* Number between 0 and 1 */
break;
}
case 1: { /* only upper limit */
int u = luaL_checkint(L, 1);
luaL_argcheck(L, 1 <= u, 1, "interval is empty");
lua_pushnumber(L, floor(r * u) + 1); /* int between 1 and `u' */
break;
}
case 2: { /* lower and upper limits */
int l = luaL_checkint(L, 1);
int u = luaL_checkint(L, 2);
luaL_argcheck(L, l <= u, 2, "interval is empty");
lua_pushnumber(L, floor(r * (u - l + 1)) + l); /* int between `l' and `u' */
break;
}
default: return luaL_error(L, "wrong number of arguments");
}
return 1;
}
static int server_math_randomseed(lua_State *L) {
serverSrand48(luaL_checkint(L, 1));
return 0;
}
/* This is the Lua script "count" hook that we use to detect scripts timeout. */
static void luaMaskCountHook(lua_State *lua, lua_Debug *ar) {
UNUSED(ar);
scriptRunCtx *rctx = luaGetFromRegistry(lua, REGISTRY_RUN_CTX_NAME);
serverAssert(rctx); /* Only supported inside script invocation */
if (scriptInterrupt(rctx) == SCRIPT_KILL) {
char *err = NULL;
if (rctx->flags & SCRIPT_EVAL_MODE) {
err = "Script killed by user with SCRIPT KILL.";
} else {
err = "Script killed by user with FUNCTION KILL.";
}
serverLog(LL_NOTICE, "%s", err);
/*
* Set the hook to invoke all the time so the user
* will not be able to catch the error with pcall and invoke
* pcall again which will prevent the script from ever been killed
*/
lua_sethook(lua, luaMaskCountHook, LUA_MASKLINE, 0);
luaPushError(lua, err);
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
luaError(lua);
}
}
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
void luaErrorInformationDiscard(errorInfo *err_info) {
if (err_info->msg) sdsfree(err_info->msg);
if (err_info->source) sdsfree(err_info->source);
if (err_info->line) sdsfree(err_info->line);
}
void luaExtractErrorInformation(lua_State *lua, errorInfo *err_info) {
if (lua_isstring(lua, -1)) {
err_info->msg = sdscatfmt(sdsempty(), "ERR %s", lua_tostring(lua, -1));
err_info->line = NULL;
err_info->source = NULL;
err_info->ignore_err_stats_update = 0;
return;
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
}
lua_getfield(lua, -1, "err");
if (lua_isstring(lua, -1)) {
err_info->msg = sdsnew(lua_tostring(lua, -1));
}
lua_pop(lua, 1);
lua_getfield(lua, -1, "source");
if (lua_isstring(lua, -1)) {
err_info->source = sdsnew(lua_tostring(lua, -1));
}
lua_pop(lua, 1);
lua_getfield(lua, -1, "line");
if (lua_isstring(lua, -1)) {
err_info->line = sdsnew(lua_tostring(lua, -1));
}
lua_pop(lua, 1);
lua_getfield(lua, -1, "ignore_error_stats_update");
if (lua_isboolean(lua, -1)) {
err_info->ignore_err_stats_update = lua_toboolean(lua, -1);
}
lua_pop(lua, 1);
}
void luaCallFunction(scriptRunCtx *run_ctx,
lua_State *lua,
robj **keys,
size_t nkeys,
robj **args,
size_t nargs,
int debug_enabled) {
client *c = run_ctx->original_client;
int delhook = 0;
/* We must set it before we set the Lua hook, theoretically the
* Lua hook might be called wheneven we run any Lua instruction
* such as 'luaSetGlobalArray' and we want the run_ctx to be available
* each time the Lua hook is invoked. */
luaSaveOnRegistry(lua, REGISTRY_RUN_CTX_NAME, run_ctx);
if (server.busy_reply_threshold > 0 && !debug_enabled) {
lua_sethook(lua, luaMaskCountHook, LUA_MASKCOUNT, 100000);
delhook = 1;
} else if (debug_enabled) {
lua_sethook(lua, luaLdbLineHook, LUA_MASKLINE | LUA_MASKCOUNT, 100000);
delhook = 1;
}
/* Populate the argv and keys table accordingly to the arguments that
* EVAL received. */
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
luaCreateArray(lua, keys, nkeys);
/* On eval, keys and arguments are globals. */
if (run_ctx->flags & SCRIPT_EVAL_MODE) {
/* open global protection to set KEYS */
lua_enablereadonlytable(lua, LUA_GLOBALSINDEX, 0);
lua_setglobal(lua, "KEYS");
lua_enablereadonlytable(lua, LUA_GLOBALSINDEX, 1);
}
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
luaCreateArray(lua, args, nargs);
if (run_ctx->flags & SCRIPT_EVAL_MODE) {
/* open global protection to set ARGV */
lua_enablereadonlytable(lua, LUA_GLOBALSINDEX, 0);
lua_setglobal(lua, "ARGV");
lua_enablereadonlytable(lua, LUA_GLOBALSINDEX, 1);
}
/* At this point whether this script was never seen before or if it was
Redis Function Libraries (#10004) # Redis Function Libraries This PR implements Redis Functions Libraries as describe on: https://github.com/redis/redis/issues/9906. Libraries purpose is to provide a better code sharing between functions by allowing to create multiple functions in a single command. Functions that were created together can safely share code between each other without worrying about compatibility issues and versioning. Creating a new library is done using 'FUNCTION LOAD' command (full API is described below) This PR introduces a new struct called libraryInfo, libraryInfo holds information about a library: * name - name of the library * engine - engine used to create the library * code - library code * description - library description * functions - the functions exposed by the library When Redis gets the `FUNCTION LOAD` command it creates a new empty libraryInfo. Redis passes the `CODE` to the relevant engine alongside the empty libraryInfo. As a result, the engine will create one or more functions by calling 'libraryCreateFunction'. The new funcion will be added to the newly created libraryInfo. So far Everything is happening locally on the libraryInfo so it is easy to abort the operation (in case of an error) by simply freeing the libraryInfo. After the library info is fully constructed we start the joining phase by which we will join the new library to the other libraries currently exist on Redis. The joining phase make sure there is no function collision and add the library to the librariesCtx (renamed from functionCtx). LibrariesCtx is used all around the code in the exact same way as functionCtx was used (with respect to RDB loading, replicatio, ...). The only difference is that apart from function dictionary (maps function name to functionInfo object), the librariesCtx contains also a libraries dictionary that maps library name to libraryInfo object. ## New API ### FUNCTION LOAD `FUNCTION LOAD <ENGINE> <LIBRARY NAME> [REPLACE] [DESCRIPTION <DESCRIPTION>] <CODE>` Create a new library with the given parameters: * ENGINE - REPLACE Engine name to use to create the library. * LIBRARY NAME - The new library name. * REPLACE - If the library already exists, replace it. * DESCRIPTION - Library description. * CODE - Library code. Return "OK" on success, or error on the following cases: * Library name already taken and REPLACE was not used * Name collision with another existing library (even if replace was uses) * Library registration failed by the engine (usually compilation error) ## Changed API ### FUNCTION LIST `FUNCTION LIST [LIBRARYNAME <LIBRARY NAME PATTERN>] [WITHCODE]` Command was modified to also allow getting libraries code (so `FUNCTION INFO` command is no longer needed and removed). In addition the command gets an option argument, `LIBRARYNAME` allows you to only get libraries that match the given `LIBRARYNAME` pattern. By default, it returns all libraries. ### INFO MEMORY Added number of libraries to `INFO MEMORY` ### Commands flags `DENYOOM` flag was set on `FUNCTION LOAD` and `FUNCTION RESTORE`. We consider those commands as commands that add new data to the dateset (functions are data) and so we want to disallows to run those commands on OOM. ## Removed API * FUNCTION CREATE - Decided on https://github.com/redis/redis/issues/9906 * FUNCTION INFO - Decided on https://github.com/redis/redis/issues/9899 ## Lua engine changes When the Lua engine gets the code given on `FUNCTION LOAD` command, it immediately runs it, we call this run the loading run. Loading run is not a usual script run, it is not possible to invoke any Redis command from within the load run. Instead there is a new API provided by `library` object. The new API's: * `redis.log` - behave the same as `redis.log` * `redis.register_function` - register a new function to the library The loading run purpose is to register functions using the new `redis.register_function` API. Any attempt to use any other API will result in an error. In addition, the load run is has a time limit of 500ms, error is raise on timeout and the entire operation is aborted. ### `redis.register_function` `redis.register_function(<function_name>, <callback>, [<description>])` This new API allows users to register a new function that will be linked to the newly created library. This API can only be called during the load run (see definition above). Any attempt to use it outside of the load run will result in an error. The parameters pass to the API are: * function_name - Function name (must be a Lua string) * callback - Lua function object that will be called when the function is invokes using fcall/fcall_ro * description - Function description, optional (must be a Lua string). ### Example The following example creates a library called `lib` with 2 functions, `f1` and `f1`, returns 1 and 2 respectively: ``` local function f1(keys, args)     return 1 end local function f2(keys, args)     return 2 end redis.register_function('f1', f1) redis.register_function('f2', f2) ``` Notice: Unlike `eval`, functions inside a library get the KEYS and ARGV as arguments to the functions and not as global. ### Technical Details On the load run we only want the user to be able to call a white list on API's. This way, in the future, if new API's will be added, the new API's will not be available to the load run unless specifically added to this white list. We put the while list on the `library` object and make sure the `library` object is only available to the load run by using [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) API. This API allows us to set the `globals` of a function (and all the function it creates). Before starting the load run we create a new fresh Lua table (call it `g`) that only contains the `library` API (we make sure to set global protection on this table just like the general global protection already exists today), then we use [lua_setfenv](https://www.lua.org/manual/5.1/manual.html#lua_setfenv) to set `g` as the global table of the load run. After the load run finished we update `g` metatable and set `__index` and `__newindex` functions to be `_G` (Lua default globals), we also pop out the `library` object as we do not need it anymore. This way, any function that was created on the load run (and will be invoke using `fcall`) will see the default globals as it expected to see them and will not have the `library` API anymore. An important outcome of this new approach is that now we can achieve a distinct global table for each library (it is not yet like that but it is very easy to achieve it now). In the future we can decide to remove global protection because global on different libraries will not collide or we can chose to give different API to different libraries base on some configuration or input. Notice that this technique was meant to prevent errors and was not meant to prevent malicious user from exploit it. For example, the load run can still save the `library` object on some local variable and then using in `fcall` context. To prevent such a malicious use, the C code also make sure it is running in the right context and if not raise an error.
2022-01-06 13:39:38 +02:00
* already defined, we can call it.
* On eval mode, we have zero arguments and expect a single return value.
* In addition the error handler is located on position -2 on the Lua stack.
* On function mode, we pass 2 arguments (the keys and args tables),
* and the error handler is located on position -4 (stack: error_handler, callback, keys, args) */
int err;
if (run_ctx->flags & SCRIPT_EVAL_MODE) {
err = lua_pcall(lua, 0, 1, -2);
} else {
err = lua_pcall(lua, 2, 1, -4);
}
/* Call the Lua garbage collector from time to time to avoid a
* full cycle performed by Lua, which adds too latency.
*
* The call is performed every LUA_GC_CYCLE_PERIOD executed commands
* (and for LUA_GC_CYCLE_PERIOD collection steps) because calling it
* for every command uses too much CPU. */
#define LUA_GC_CYCLE_PERIOD 50
{
static long gc_count = 0;
gc_count++;
if (gc_count == LUA_GC_CYCLE_PERIOD) {
lua_gc(lua, LUA_GCSTEP, LUA_GC_CYCLE_PERIOD);
gc_count = 0;
}
}
if (err) {
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
/* Error object is a table of the following format:
* {err='<error msg>', source='<source file>', line=<line>}
* We can construct the error message from this information */
if (!lua_istable(lua, -1)) {
const char *msg = "execution failure";
if (lua_isstring(lua, -1)) {
msg = lua_tostring(lua, -1);
}
addReplyErrorFormat(c, "Error running script %s, %.100s\n", run_ctx->funcname, msg);
Sort out the mess around Lua error messages and error stats (#10329) This PR fix 2 issues on Lua scripting: * Server error reply statistics (some errors were counted twice). * Error code and error strings returning from scripts (error code was missing / misplaced). ## Statistics a Lua script user is considered part of the user application, a sophisticated transaction, so we want to count an error even if handled silently by the script, but when it is propagated outwards from the script we don't wanna count it twice. on the other hand, if the script decides to throw an error on its own (using `redis.error_reply`), we wanna count that too. Besides, we do count the `calls` in command statistics for the commands the script calls, we we should certainly also count `failed_calls`. So when a simple `eval "return redis.call('set','x','y')" 0` fails, it should count the failed call to both SET and EVAL, but the `errorstats` and `total_error_replies` should be counted only once. The PR changes the error object that is raised on errors. Instead of raising a simple Lua string, Redis will raise a Lua table in the following format: ``` { err='<error message (including error code)>', source='<User source file name>', line='<line where the error happned>', ignore_error_stats_update=true/false, } ``` The `luaPushError` function was modified to construct the new error table as describe above. The `luaRaiseError` was renamed to `luaError` and is now simply called `lua_error` to raise the table on the top of the Lua stack as the error object. The reason is that since its functionality is changed, in case some Redis branch / fork uses it, it's better to have a compilation error than a bug. The `source` and `line` fields are enriched by the error handler (if possible) and the `ignore_error_stats_update` is optional and if its not present then the default value is `false`. If `ignore_error_stats_update` is true, the error will not be counted on the error stats. When parsing Redis call reply, each error is translated to a Lua table on the format describe above and the `ignore_error_stats_update` field is set to `true` so we will not count errors twice (we counted this error when we invoke the command). The changes in this PR might have been considered as a breaking change for users that used Lua `pcall` function. Before, the error was a string and now its a table. To keep backward comparability the PR override the `pcall` implementation and extract the error message from the error table and return it. Example of the error stats update: ``` 127.0.0.1:6379> lpush l 1 (integer) 2 127.0.0.1:6379> eval "return redis.call('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value. script: e471b73f1ef44774987ab00bdf51f21fd9f7974a, on @user_script:1. 127.0.0.1:6379> info Errorstats # Errorstats errorstat_WRONGTYPE:count=1 127.0.0.1:6379> info commandstats # Commandstats cmdstat_eval:calls=1,usec=341,usec_per_call=341.00,rejected_calls=0,failed_calls=1 cmdstat_info:calls=1,usec=35,usec_per_call=35.00,rejected_calls=0,failed_calls=0 cmdstat_lpush:calls=1,usec=14,usec_per_call=14.00,rejected_calls=0,failed_calls=0 cmdstat_get:calls=1,usec=10,usec_per_call=10.00,rejected_calls=0,failed_calls=1 ``` ## error message We can now construct the error message (sent as a reply to the user) from the error table, so this solves issues where the error message was malformed and the error code appeared in the middle of the error message: ```diff 127.0.0.1:6379> eval "return redis.call('set','x','y')" 0 -(error) ERR Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479): @user_script:1: OOM command not allowed when used memory > 'maxmemory'. +(error) OOM command not allowed when used memory > 'maxmemory' @user_script:1. Error running script (call to 71e6319f97b0fe8bdfa1c5df3ce4489946dda479) ``` ```diff 127.0.0.1:6379> eval "redis.call('get', 'l')" 0 -(error) ERR Error running script (call to f_8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1): @user_script:1: WRONGTYPE Operation against a key holding the wrong kind of value +(error) WRONGTYPE Operation against a key holding the wrong kind of value script: 8a705cfb9fb09515bfe57ca2bd84a5caee2cbbd1, on @user_script:1. ``` Notica that `redis.pcall` was not change: ``` 127.0.0.1:6379> eval "return redis.pcall('get', 'l')" 0 (error) WRONGTYPE Operation against a key holding the wrong kind of value ``` ## other notes Notice that Some commands (like GEOADD) changes the cmd variable on the client stats so we can not count on it to update the command stats. In order to be able to update those stats correctly we needed to promote `realcmd` variable to be located on the client struct. Tests was added and modified to verify the changes. Related PR's: #10279, #10218, #10278, #10309 Co-authored-by: Oran Agra <oran@redislabs.com>
2022-02-27 13:40:57 +02:00
} else {
errorInfo err_info = {0};
sds final_msg = sdsempty();
luaExtractErrorInformation(lua, &err_info);
final_msg = sdscatfmt(final_msg, "-%s", err_info.msg);
if (err_info.line && err_info.source) {
final_msg =
sdscatfmt(final_msg, " script: %s, on %s:%s.", run_ctx->funcname, err_info.source, err_info.line);
}
addReplyErrorSdsEx(c, final_msg, err_info.ignore_err_stats_update ? ERR_REPLY_FLAG_NO_STATS_UPDATE : 0);
luaErrorInformationDiscard(&err_info);
}
lua_pop(lua, 1); /* Consume the Lua error */
} else {
/* On success convert the Lua return value into RESP, and
* send it to * the client. */
luaReplyToServerReply(c, run_ctx->c, lua); /* Convert and consume the reply. */
}
/* Perform some cleanup that we need to do both on error and success. */
if (delhook) lua_sethook(lua, NULL, 0, 0); /* Disable hook */
/* remove run_ctx from registry, its only applicable for the current script. */
luaSaveOnRegistry(lua, REGISTRY_RUN_CTX_NAME, NULL);
}
unsigned long luaMemory(lua_State *lua) {
return lua_gc(lua, LUA_GCCOUNT, 0) * 1024LL;
}