Allow a wal mode recovery to proceed even if there are readers.

FossilOrigin-Name: 74374aebf9abf3d6b6a3920967a079ceaa4c6276dc6c177682742c2be405b7b7

diff --git a/manifest b/manifest
index 80b02e2e4b1d2a586d3e5a52ef4348694e43bbc3..78768b9243d72e362f04f98555108b47134b4bd2 100644 (file)
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Remove\sa\ssurplus\sspace\sfrom\sa\scomment
-D 2020-07-24T11:01:29.118
+C Allow\sa\swal\smode\srecovery\sto\sproceed\seven\sif\sthere\sare\sreaders.
+D 2020-07-25T20:16:27.306
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F LICENSE.md df5091916dbb40e6e9686186587125e1b2ff51f022cc334e886c19a0e9982724
@@ -619,7 +619,7 @@ F src/vdbetrace.c fa3bf238002f0bbbdfb66cc8afb0cea284ff9f148d6439bc1f6f2b4c3b7143
 F src/vdbevtab.c ee5b4c902fdda2230f9503ac7b84c6d614c91e8f6f4dc1633e2e8dfef8ffb144
 F src/vtab.c 5f5fc793092f53bbdfde296c50f563fb7bda58cf48e9cf6a8bdfbc5abd409845
 F src/vxworks.h d2988f4e5a61a4dfe82c6524dd3d6e4f2ce3cdb9
-F src/wal.c 231044ecf7d5d78bc705af9dcec6c10ec59e891366362b6be54bb6a0bc7c17db
+F src/wal.c 477491b1996746fb197ac44cb7e5e2bdc14645e0663c6422445534254bee37fe
 F src/wal.h c3aa7825bfa2fe0d85bef2db94655f99870a285778baa36307c0a16da32b226a
 F src/walker.c 3df26a33dc4f54e8771600fb7fdebe1ece0896c2ad68c30ab40b017aa4395049
 F src/where.c 2ea911238674e9baaeddf105dddabed92692a01996073c4d4983f9a7efe481f9
@@ -1878,7 +1878,10 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P 270ac1a0f232d75537be40abae559004e950b992cb2c7e94cd6de66e96ae17bd
-R 0212ebf857b7a36fab6e1708725a20d6
-U drh
-Z 7c8911d20a18c22808c4eb3363f6ff12
+P 73fecc688ab5c459245c9bb89432139a083ef378404b4dae3397a7f4e253f72a
+R 7245f53be8944a8b99a3f69058418f3e
+T *branch * unlocked-recovery
+T *sym-unlocked-recovery *
+T -sym-trunk *
+U dan
+Z c489f8aaf2222395682cc1ddd5e61e86

diff --git a/manifest.uuid b/manifest.uuid
index ef6483f2d60cc1f1587b013bc23ec10d9e47eb1b..6e841e436f625a6f436441e6cf4bb9c7593cf712 100644 (file)
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-73fecc688ab5c459245c9bb89432139a083ef378404b4dae3397a7f4e253f72a
\ No newline at end of file
+74374aebf9abf3d6b6a3920967a079ceaa4c6276dc6c177682742c2be405b7b7
\ No newline at end of file

diff --git a/src/wal.c b/src/wal.c
index c6d4476e7b3b5d65765a816060139b69bd5e4b5c..9ad0b496e299bfeca6790adba0357e17d6aac9e5 100644 (file)
--- a/src/wal.c
+++ b/src/wal.c
@@ -1161,12 +1161,6 @@ static int walIndexRecover(Wal *pWal){
   assert( pWal->writeLock );
   iLock = WAL_ALL_BUT_WRITE + pWal->ckptLock;
   rc = walLockExclusive(pWal, iLock, WAL_READ_LOCK(0)-iLock);
-  if( rc==SQLITE_OK ){
-    rc = walLockExclusive(pWal, WAL_READ_LOCK(1), WAL_NREADER-1);
-    if( rc!=SQLITE_OK ){
-      walUnlockExclusive(pWal, iLock, WAL_READ_LOCK(0)-iLock);
-    }
-  }
   if( rc ){
     return rc;
   }
@@ -1182,15 +1176,16 @@ static int walIndexRecover(Wal *pWal){
 
   if( nSize>WAL_HDRSIZE ){
     u8 aBuf[WAL_HDRSIZE];         /* Buffer to load WAL header into */
+    u32 *aPrivate = 0;            /* Heap copy of *-shm hash being populated */
     u8 *aFrame = 0;               /* Malloc'd buffer to load entire frame */
     int szFrame;                  /* Number of bytes in buffer aFrame[] */
     u8 *aData;                    /* Pointer to data part of aFrame buffer */
-    int iFrame;                   /* Index of last frame read */
-    i64 iOffset;                  /* Next offset to read from log file */
     int szPage;                   /* Page size according to the log */
     u32 magic;                    /* Magic value read from WAL header */
     u32 version;                  /* Magic value read from WAL header */
     int isValid;                  /* True if this frame is valid */
+    int iPg;                      /* Current 32KB wal-index page */
+    int iLastFrame;               /* Last frame in wal, based on nSize alone */
 
     /* Read in the WAL header. */
     rc = sqlite3OsRead(pWal->pWalFd, aBuf, WAL_HDRSIZE, 0);
@@ -1237,38 +1232,59 @@ static int walIndexRecover(Wal *pWal){
 
     /* Malloc a buffer to read frames into. */
     szFrame = szPage + WAL_FRAME_HDRSIZE;
-    aFrame = (u8 *)sqlite3_malloc64(szFrame);
+    aFrame = (u8 *)sqlite3_malloc64(szFrame + WALINDEX_PGSZ);
     if( !aFrame ){
       rc = SQLITE_NOMEM_BKPT;
       goto recovery_error;
     }
     aData = &aFrame[WAL_FRAME_HDRSIZE];
+    aPrivate = (u32*)&aData[szPage];
 
     /* Read all frames from the log file. */
-    iFrame = 0;
-    for(iOffset=WAL_HDRSIZE; (iOffset+szFrame)<=nSize; iOffset+=szFrame){
-      u32 pgno;                   /* Database page number for frame */
-      u32 nTruncate;              /* dbsize field from frame header */
-
-      /* Read and decode the next log frame. */
-      iFrame++;
-      rc = sqlite3OsRead(pWal->pWalFd, aFrame, szFrame, iOffset);
-      if( rc!=SQLITE_OK ) break;
-      isValid = walDecodeFrame(pWal, &pgno, &nTruncate, aData, aFrame);
-      if( !isValid ) break;
-      rc = walIndexAppend(pWal, iFrame, pgno);
-      if( rc!=SQLITE_OK ) break;
-
-      /* If nTruncate is non-zero, this is a commit record. */
-      if( nTruncate ){
-        pWal->hdr.mxFrame = iFrame;
-        pWal->hdr.nPage = nTruncate;
-        pWal->hdr.szPage = (u16)((szPage&0xff00) | (szPage>>16));
-        testcase( szPage<=32768 );
-        testcase( szPage>=65536 );
-        aFrameCksum[0] = pWal->hdr.aFrameCksum[0];
-        aFrameCksum[1] = pWal->hdr.aFrameCksum[1];
+    iLastFrame = (nSize - WAL_HDRSIZE) / szFrame;
+    for(iPg=0; iPg<=walFramePage(iLastFrame); iPg++){
+      u32 *aShare;
+      int iFrame;                 /* Index of last frame read */
+      int iLast = MIN(iLastFrame, HASHTABLE_NPAGE_ONE+iPg*HASHTABLE_NPAGE);
+      int iFirst = 1 + (iPg==0?0:HASHTABLE_NPAGE_ONE+(iPg-1)*HASHTABLE_NPAGE);
+      rc = walIndexPage(pWal, iPg, (volatile u32**)&aShare);
+      if( rc ) break;
+      pWal->apWiData[iPg] = aPrivate;
+      
+      for(iFrame=iFirst; iFrame<=iLast; iFrame++){
+        i64 iOffset = walFrameOffset(iFrame, szPage);
+        u32 pgno;                 /* Database page number for frame */
+        u32 nTruncate;            /* dbsize field from frame header */
+
+        /* Read and decode the next log frame. */
+        rc = sqlite3OsRead(pWal->pWalFd, aFrame, szFrame, iOffset);
+        if( rc!=SQLITE_OK ) break;
+        isValid = walDecodeFrame(pWal, &pgno, &nTruncate, aData, aFrame);
+        if( !isValid ) break;
+        rc = walIndexAppend(pWal, iFrame, pgno);
+        if( rc!=SQLITE_OK ) break;
+
+        /* If nTruncate is non-zero, this is a commit record. */
+        if( nTruncate ){
+          pWal->hdr.mxFrame = iFrame;
+          pWal->hdr.nPage = nTruncate;
+          pWal->hdr.szPage = (u16)((szPage&0xff00) | (szPage>>16));
+          testcase( szPage<=32768 );
+          testcase( szPage>=65536 );
+          aFrameCksum[0] = pWal->hdr.aFrameCksum[0];
+          aFrameCksum[1] = pWal->hdr.aFrameCksum[1];
+        }
       }
+      pWal->apWiData[iPg] = aShare;
+
+      {
+        int nHdr = (iPg==0 ? WALINDEX_HDR_SIZE : 0);
+        int nHdr32 = nHdr / sizeof(u32);
+        if( memcpy(&aShare[nHdr32], &aPrivate[nHdr32], WALINDEX_PGSZ-nHdr) ){
+          memcpy(&aShare[nHdr32], &aPrivate[nHdr32], WALINDEX_PGSZ-nHdr);
+        }
+      }
+      if( iFrame<=iLast ) break;
     }
 
     sqlite3_free(aFrame);
@@ -1283,15 +1299,24 @@ finished:
     walIndexWriteHdr(pWal);
 
     /* Reset the checkpoint-header. This is safe because this thread is 
-    ** currently holding locks that exclude all other readers, writers and
-    ** checkpointers.
+    ** currently holding locks that exclude all other writers and 
+    ** checkpointers. Then set the values of read-mark slots 1 through N.
     */
     pInfo = walCkptInfo(pWal);
     pInfo->nBackfill = 0;
     pInfo->nBackfillAttempted = pWal->hdr.mxFrame;
     pInfo->aReadMark[0] = 0;
-    for(i=1; i<WAL_NREADER; i++) pInfo->aReadMark[i] = READMARK_NOT_USED;
-    if( pWal->hdr.mxFrame ) pInfo->aReadMark[1] = pWal->hdr.mxFrame;
+    for(i=1; i<WAL_NREADER; i++){
+      rc = walLockExclusive(pWal, WAL_READ_LOCK(i), 1);
+      if( rc==SQLITE_OK ){
+        if( i==1 && pWal->hdr.mxFrame ){
+          pInfo->aReadMark[i] = pWal->hdr.mxFrame;
+        }else{
+          pInfo->aReadMark[i] = READMARK_NOT_USED;
+        }
+        walUnlockExclusive(pWal, WAL_READ_LOCK(i), 1);
+      }
+    }
 
     /* If more than one frame was recovered from the log file, report an
     ** event via sqlite3_log(). This is to help with identifying performance
@@ -1309,7 +1334,6 @@ finished:
 recovery_error:
   WALTRACE(("WAL%p: recovery %s\n", pWal, rc ? "failed" : "ok"));
   walUnlockExclusive(pWal, iLock, WAL_READ_LOCK(0)-iLock);
-  walUnlockExclusive(pWal, WAL_READ_LOCK(1), WAL_NREADER-1);
   return rc;
 }