SQLite

Check-in [346388007d]
Login

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:Further comment improvements in wal.c. No code changes.
Downloads: Tarball | ZIP archive
Timelines: family | ancestors | descendants | both | readonly-wal-recovery
Files: files | file ages | folders
SHA3-256: 346388007de585083dc67ad865b91db7c7d7b78c10a06f8bb7c48767c326c47e
User & Date: drh 2017-11-11 13:30:44.595
Context
2017-11-11
20:11
In wal.c: improved comments, new assert() and testcase() macros, and replace some magic numbers with appropriate symbolic constants. (check-in: 13ec8a77a4 user: drh tags: readonly-wal-recovery)
13:30
Further comment improvements in wal.c. No code changes. (check-in: 346388007d user: drh tags: readonly-wal-recovery)
2017-11-10
20:00
Improved comments and variable names in the read-only WAL logic. (check-in: d3c25740ee user: drh tags: readonly-wal-recovery)
Changes
Unified Diff Ignore Whitespace Patch
Changes to src/wal.c.
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241













2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
  pWal->readLock = 0;

  /* Check to see if a separate writer has attached to the shared-memory area,
  ** thus making the shared-memory "reliable" again.  Do this by invoking
  ** the xShmMap() routine of the VFS and looking to see if the return
  ** is SQLITE_READONLY instead of SQLITE_READONLY_CANTINIT.
  **
  ** Once sqlite3OsShmMap() has been called for a file and has returned
  ** any SQLITE_READONLY value, it must SQLITE_READONLY or
  ** SQLITE_READONLY_CANTINIT or some error for all subsequent invocations,
  ** until sqlite3OsShmUnmap() has been called.  This is a requirement
  ** on the VFS implementation.
  **
  ** If the shared-memory is now "reliable" return WAL_RETRY, which will
  ** cause the heap-memory WAL-index to be discarded and the actual
  ** shared memory to be used in its place.













  */
  rc = sqlite3OsShmMap(pWal->pDbFd, 0, WALINDEX_PGSZ, 0, &pDummy);
  assert( rc!=SQLITE_OK ); /* SQLITE_OK not possible for read-only connection */
  if( rc!=SQLITE_READONLY_CANTINIT ){
    rc = (rc==SQLITE_READONLY ? WAL_RETRY : rc);
    goto begin_unreliable_shm_out;
  }

  /* Reach this point only if the real shared-memory is still unreliable.
  ** Assume the in-memory WAL-index substitute is correct and load it
  ** into pWal->hdr.
  */
  memcpy(&pWal->hdr, (void*)walIndexHdr(pWal), sizeof(WalIndexHdr));

  /* The WAL_READ_LOCK(0) lock held by this client prevents a checkpoint
  ** from taking place. But it does not prevent the wal from being wrapped
  ** if a checkpoint has already taken place. This means that if another
  ** client is connected at this point, it may have already checkpointed 
  ** the entire wal. In that case it would not be safe to continue with
  ** the this transaction, as the other client may overwrite wal 
  ** frames that this client is still using.
  */
  rc = sqlite3OsFileSize(pWal->pWalFd, &szWal);
  if( rc!=SQLITE_OK ){
    goto begin_unreliable_shm_out;
  }
  if( szWal<WAL_HDRSIZE ){
    /* If the wal file is too small to contain a wal-header and the







<
<
<
<
<
<



>
>
>
>
>
>
>
>
>
>
>
>
>
|







|





|
<
<
|
<
<
<







2226
2227
2228
2229
2230
2231
2232






2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263


2264



2265
2266
2267
2268
2269
2270
2271
  pWal->readLock = 0;

  /* Check to see if a separate writer has attached to the shared-memory area,
  ** thus making the shared-memory "reliable" again.  Do this by invoking
  ** the xShmMap() routine of the VFS and looking to see if the return
  ** is SQLITE_READONLY instead of SQLITE_READONLY_CANTINIT.
  **






  ** If the shared-memory is now "reliable" return WAL_RETRY, which will
  ** cause the heap-memory WAL-index to be discarded and the actual
  ** shared memory to be used in its place.
  **
  ** This step is important because, even though this connection is holding
  ** the WAL_READ_LOCK(0) which prevents a checkpoint, a writer might
  ** have already checkpointed the WAL file and, while the current
  ** is active, wrap the WAL and start overwriting frames that this
  ** process wants to use.
  **
  ** Once sqlite3OsShmMap() has been called for an sqlite3_file and has
  ** returned any SQLITE_READONLY value, it must return only SQLITE_READONLY
  ** or SQLITE_READONLY_CANTINIT or some error for all subsequent invocations,
  ** even if some external agent does a "chmod" to make the shared-memory
  ** writable by us, until sqlite3OsShmUnmap() has been called.
  ** This is a requirement on the VFS implementation.
   */
  rc = sqlite3OsShmMap(pWal->pDbFd, 0, WALINDEX_PGSZ, 0, &pDummy);
  assert( rc!=SQLITE_OK ); /* SQLITE_OK not possible for read-only connection */
  if( rc!=SQLITE_READONLY_CANTINIT ){
    rc = (rc==SQLITE_READONLY ? WAL_RETRY : rc);
    goto begin_unreliable_shm_out;
  }

  /* We reach this point only if the real shared-memory is still unreliable.
  ** Assume the in-memory WAL-index substitute is correct and load it
  ** into pWal->hdr.
  */
  memcpy(&pWal->hdr, (void*)walIndexHdr(pWal), sizeof(WalIndexHdr));

  /* Make sure some writer hasn't come in and changed the WAL file out


  ** from under us, then disconnected, while we were not looking.



  */
  rc = sqlite3OsFileSize(pWal->pWalFd, &szWal);
  if( rc!=SQLITE_OK ){
    goto begin_unreliable_shm_out;
  }
  if( szWal<WAL_HDRSIZE ){
    /* If the wal file is too small to contain a wal-header and the
2279
2280
2281
2282
2283
2284
2285



2286
2287
2288
2289
2290
2291
2292

  /* Check the salt keys at the start of the wal file still match. */
  rc = sqlite3OsRead(pWal->pWalFd, aBuf, WAL_HDRSIZE, 0);
  if( rc!=SQLITE_OK ){
    goto begin_unreliable_shm_out;
  }
  if( memcmp(&pWal->hdr.aSalt, &aBuf[16], 8) ){



    rc = WAL_RETRY;
    goto begin_unreliable_shm_out;
  }

  /* Allocate a buffer to read frames into */
  szFrame = pWal->hdr.szPage + WAL_FRAME_HDRSIZE;
  aFrame = (u8 *)sqlite3_malloc64(szFrame);







>
>
>







2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297

  /* Check the salt keys at the start of the wal file still match. */
  rc = sqlite3OsRead(pWal->pWalFd, aBuf, WAL_HDRSIZE, 0);
  if( rc!=SQLITE_OK ){
    goto begin_unreliable_shm_out;
  }
  if( memcmp(&pWal->hdr.aSalt, &aBuf[16], 8) ){
    /* Some writer has wrapped the WAL file while we were not looking.
    ** Return WAL_RETRY which will cause the in-memory WAL-index to be
    ** rebuilt. */
    rc = WAL_RETRY;
    goto begin_unreliable_shm_out;
  }

  /* Allocate a buffer to read frames into */
  szFrame = pWal->hdr.szPage + WAL_FRAME_HDRSIZE;
  aFrame = (u8 *)sqlite3_malloc64(szFrame);