Don't free snapshot objects in a critical path (under the AE lock)

Former-commit-id: d0da3d3cb74334cc8a2d14f4bdaef7935181700a

src/server.cpp

@@ -2040,6 +2040,29 @@ void flushStorageWeak()
     }
 }
 
+void freeSnapshotLazyFreesAsync()
+{
+    aeAcquireLock();
+    std::vector<robj*> vecObjs = std::move(g_pserver->vecobjLazyFree);
+    std::vector<dict*> vecDicts = std::move(g_pserver->vecdictLazyFree);
+    std::vector<std::vector<dictEntry*>> vecvecde = std::move(g_pserver->vecvecde);
+    aeReleaseLock();
+    
+    for (auto &vecdeFree : vecvecde)
+    {
+        for (auto *de : vecdeFree)
+        {
+            dbDictType.keyDestructor(nullptr, dictGetKey(de));
+            dbDictType.valDestructor(nullptr, dictGetVal(de));
+            zfree(de);
+        }
+    }
+    for (robj *o : vecObjs)
+        decrRefCount(o);
+    for (dict *d : vecDicts)
+        dictRelease(d);
+}
+
 /* This is our timer interrupt, called g_pserver->hz times per second.
  * Here is where we do a number of things that need to be done asynchronously.
  * For instance:
@@ -2313,14 +2336,18 @@ int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) {
         }
     }
 
-    bool fAnySnapshots = false;
-    for (int idb = 0; idb < cserver.dbnum && !fAnySnapshots; ++idb)
-        fAnySnapshots = fAnySnapshots || g_pserver->db[0]->FSnapshot();
-    if (fAnySnapshots)  
-    {
-        g_pserver->asyncworkqueue->AddWorkFunction([]{
-            g_pserver->db[0]->consolidate_snapshot();
-        }, true /*HiPri*/);
+    run_with_period(100) {
+        bool fAsyncFrees = g_pserver->vecobjLazyFree.size() || g_pserver->vecdictLazyFree.size() || g_pserver->vecvecde.size();
+        bool fAnySnapshots = false;
+        for (int idb = 0; idb < cserver.dbnum && !fAnySnapshots && !fAsyncFrees; ++idb)
+            fAnySnapshots = fAnySnapshots || g_pserver->db[0]->FSnapshot();
+        if (fAnySnapshots || fAsyncFrees)
+        {
+            g_pserver->asyncworkqueue->AddWorkFunction([fAsyncFrees]{
+                g_pserver->db[0]->consolidate_snapshot();
+                freeSnapshotLazyFreesAsync();
+            }, true /*HiPri*/);
+        }
     }
     
     /* Fire the cron loop modules event. */
@@ -6049,6 +6076,7 @@ int main(int argc, char **argv) {
     serverAssert(fLockAcquired);
 
     g_pserver->garbageCollector.shutdown();
+    freeSnapshotLazyFreesAsync();
     delete g_pserver->m_pstorageFactory;
 
     return 0;

src/server.h

@@ -1425,7 +1425,7 @@ protected:
     static void gcDisposeSnapshot(redisDbPersistentDataSnapshot *psnapshot);
     int snapshot_depth() const;
     void consolidate_children(redisDbPersistentData *pdbPrimary, bool fForce);
-    void freeTombstoneObjects(int depth);
+    bool freeTombstoneObjects(int depth);
 
 public:
     bool FWillFreeChildDebug() const { return m_spdbSnapshotHOLDER != nullptr; }
@@ -2452,6 +2452,10 @@ struct redisServer {
     char *aof_rewrite_cpulist; /* cpu affinity list of aof rewrite process. */
     char *bgsave_cpulist; /* cpu affinity list of bgsave process. */
 
+    std::vector<dict*> vecdictLazyFree;
+    std::vector<robj*> vecobjLazyFree;
+    std::vector<std::vector<dictEntry*>> vecvecde;
+
     bool FRdbSaveInProgress() const { return rdbThreadVars.fRdbThreadActive; }
 };
 

src/snapshot.cpp

@@ -233,17 +233,23 @@ void redisDbPersistentData::endSnapshotAsync(const redisDbPersistentDataSnapshot
     latencyAddSampleIfNeeded("end-snapshot-async-phase-2", latency);
 }
 
-void redisDbPersistentDataSnapshot::freeTombstoneObjects(int depth)
+bool redisDbPersistentDataSnapshot::freeTombstoneObjects(int depth)
 {
     if (m_pdbSnapshot == nullptr)
-        return;
+    {
+        serverAssert(dictSize(m_pdictTombstone) == 0);
+        return true;
+    }
 
-    const_cast<redisDbPersistentDataSnapshot*>(m_pdbSnapshot)->freeTombstoneObjects(depth+1);
+    bool fPrevResult = const_cast<redisDbPersistentDataSnapshot*>(m_pdbSnapshot)->freeTombstoneObjects(depth+1);
     if (m_pdbSnapshot->m_refCount != depth && (m_pdbSnapshot->m_refCount != (m_refCount+1)))
-        return;
+        return false;
     
     dictIterator *di = dictGetIterator(m_pdictTombstone);
     dictEntry *de;
+    std::vector<dictEntry*> vecdeFree;
+    vecdeFree.reserve(dictSize(m_pdictTombstone));
+    bool fAllCovered = true;
     while ((de = dictNext(di)) != nullptr)
     {
         dictEntry **dePrev = nullptr;
@@ -252,12 +258,28 @@ void redisDbPersistentDataSnapshot::freeTombstoneObjects(int depth)
         dictEntry *deObj = dictFindWithPrev(m_pdbSnapshot->m_pdict, key, (uint64_t)dictGetVal(de), &dePrev, &ht, !!sdsisshared(key));
         if (deObj != nullptr)
         {
-            decrRefCount((robj*)dictGetVal(deObj));
-            void *ptrSet = nullptr;
-            __atomic_store(&deObj->v.val, &ptrSet, __ATOMIC_RELAXED);
+            // Now unlink the DE
+            __atomic_store(dePrev, &deObj->next, __ATOMIC_RELEASE);
+            ht->used--;
+            vecdeFree.push_back(deObj);
+        }
+        else
+        {
+            fAllCovered = fPrevResult;
         }
     }
     dictReleaseIterator(di);
+
+    aeAcquireLock();
+    if (fAllCovered)
+    {
+        g_pserver->vecdictLazyFree.push_back(m_pdictTombstone);
+        m_pdictTombstone = dictCreate(&dbTombstoneDictType, nullptr);
+    }
+    g_pserver->vecvecde.emplace_back(std::move(vecdeFree));
+    aeReleaseLock();
+    
+    return fAllCovered;
 }
 
 void redisDbPersistentData::endSnapshot(const redisDbPersistentDataSnapshot *psnapshot)
@@ -308,6 +330,8 @@ void redisDbPersistentData::endSnapshot(const redisDbPersistentDataSnapshot *psn
     // Stage 1 Loop through all the tracked deletes and remove them from the snapshot DB
     dictIterator *di = dictGetIterator(m_pdictTombstone);
     dictEntry *de;
+    m_spdbSnapshotHOLDER->m_pdict->iterators++;
+    std::vector<dictEntry*> vecde;
     while ((de = dictNext(di)) != NULL)
     {
         dictEntry **dePrev;
@@ -327,15 +351,15 @@ void redisDbPersistentData::endSnapshot(const redisDbPersistentDataSnapshot *psn
         }
         
         // Delete the object from the source dict, we don't use dictDelete to avoid a second search
-        dictFreeKey(m_spdbSnapshotHOLDER->m_pdict, deSnapshot);
-        dictFreeVal(m_spdbSnapshotHOLDER->m_pdict, deSnapshot);
-        serverAssert(*dePrev == deSnapshot);
+        vecde.push_back(deSnapshot);
         *dePrev = deSnapshot->next;
-        zfree(deSnapshot);
         ht->used--;
     }
+    g_pserver->vecvecde.emplace_back(std::move(vecde));
+    m_spdbSnapshotHOLDER->m_pdict->iterators--;
     dictReleaseIterator(di);
-    dictEmpty(m_pdictTombstone, nullptr);
+    g_pserver->vecdictLazyFree.push_back(m_pdictTombstone);
+    m_pdictTombstone = dictCreate(&dbTombstoneDictType, nullptr);
 
     // Stage 2 Move all new keys to the snapshot DB
     dictMerge(m_spdbSnapshotHOLDER->m_pdict, m_pdict);