SQLite

# define UNIXFILE_DIRSYNC    0x00
#endif
#define UNIXFILE_PSOW        0x10     /* SQLITE_IOCAP_POWERSAFE_OVERWRITE */
#define UNIXFILE_DELETE      0x20     /* Delete on close */
#define UNIXFILE_URI         0x40     /* Filename might have query parameters */
#define UNIXFILE_NOLOCK      0x80     /* Do no file locking */
#define UNIXFILE_WARNED    0x0100     /* verifyDbFile() warnings issued */
#define UNIXFILE_BLOCK     0x0200     /* Next SHM lock might block */

/*
** Include code that is common to all os_*.c files
*/
#include "os_common.h"

/*

/*
** Apply posix advisory locks for all bytes from ofst through ofst+n-1.
**
** Locks block if the mask is exactly UNIX_SHM_C and are non-blocking
** otherwise.
*/
static int unixShmSystemLock(
  unixFile *pFile,       /* Open connection to the WAL file */
  int lockType,          /* F_UNLCK, F_RDLCK, or F_WRLCK */
  int ofst,              /* First byte of the locking range */
  int n                  /* Number of bytes to lock */
){
  unixShmNode *pShmNode; /* Apply locks to this open shared-memory segment */
  struct flock f;        /* The posix advisory locking structure */
  int rc = SQLITE_OK;    /* Result code form fcntl() */

  /* Access to the unixShmNode object is serialized by the caller */
  pShmNode = pFile->pInode->pShmNode;
  assert( sqlite3_mutex_held(pShmNode->mutex) || pShmNode->nRef==0 );

  /* Shared locks never span more than one byte */
  assert( n==1 || lockType!=F_RDLCK );

  /* Locks are within range */
  assert( n>=1 && n<SQLITE_SHM_NLOCK );

  if( pShmNode->h>=0 ){
    int lkType;
    /* Initialize the locking parameters */
    memset(&f, 0, sizeof(f));
    f.l_type = lockType;
    f.l_whence = SEEK_SET;
    f.l_start = ofst;
    f.l_len = n;

    lkType = (pFile->ctrlFlags & UNIXFILE_BLOCK)!=0 ? F_SETLKW : F_SETLK;
    rc = osFcntl(pShmNode->h, lkType, &f);
    rc = (rc!=(-1)) ? SQLITE_OK : SQLITE_BUSY;
    pFile->ctrlFlags &= ~UNIXFILE_BLOCK;
  }

  /* Update the global lock state and do debug tracing */
#ifdef SQLITE_DEBUG
  { u16 mask;
  OSTRACE(("SHM-LOCK "));
  mask = ofst>31 ? 0xffff : (1<<(ofst+n)) - (1<<ofst);

      */
      osFchown(pShmNode->h, sStat.st_uid, sStat.st_gid);
  
      /* Check to see if another process is holding the dead-man switch.
      ** If not, truncate the file to zero length. 
      */
      rc = SQLITE_OK;
      if( unixShmSystemLock(pDbFd, F_WRLCK, UNIX_SHM_DMS, 1)==SQLITE_OK ){
        if( robust_ftruncate(pShmNode->h, 0) ){
          rc = unixLogError(SQLITE_IOERR_SHMOPEN, "ftruncate", zShmFilename);
        }
      }
      if( rc==SQLITE_OK ){
        rc = unixShmSystemLock(pDbFd, F_RDLCK, UNIX_SHM_DMS, 1);
      }
      if( rc ) goto shm_open_err;
    }
  }

  /* Make the new connection a child of the unixShmNode */
  p->pShmNode = pShmNode;

      if( pX==p ) continue;
      assert( (pX->exclMask & (p->exclMask|p->sharedMask))==0 );
      allMask |= pX->sharedMask;
    }

    /* Unlock the system-level locks */
    if( (mask & allMask)==0 ){
      rc = unixShmSystemLock(pDbFd, F_UNLCK, ofst+UNIX_SHM_BASE, n);
    }else{
      rc = SQLITE_OK;
    }

    /* Undo the local locks */
    if( rc==SQLITE_OK ){
      p->exclMask &= ~mask;

      }
      allShared |= pX->sharedMask;
    }

    /* Get shared locks at the system level, if necessary */
    if( rc==SQLITE_OK ){
      if( (allShared & mask)==0 ){
        rc = unixShmSystemLock(pDbFd, F_RDLCK, ofst+UNIX_SHM_BASE, n);
      }else{
        rc = SQLITE_OK;
      }
    }

    /* Get the local shared locks */
    if( rc==SQLITE_OK ){

      }
    }
  
    /* Get the exclusive locks at the system level.  Then if successful
    ** also mark the local connection as being locked.
    */
    if( rc==SQLITE_OK ){
      rc = unixShmSystemLock(pDbFd, F_WRLCK, ofst+UNIX_SHM_BASE, n);
      if( rc==SQLITE_OK ){
        assert( (p->sharedMask & mask)==0 );
        p->exclMask |= mask;
      }
    }
  }
  sqlite3_mutex_leave(pShmNode->mutex);

/*
** This routine handles sqlite3_file_control() calls that are specific
** to proxy locking.
*/
static int proxyFileControl(sqlite3_file *id, int op, void *pArg){
  switch( op ){
    case SQLITE_FCNTL_WAL_BLOCK: {
      id->ctrlFlags |= UNIXFILE_BLOCK;
      return SQLITE_OK;
    }
    case SQLITE_FCNTL_GET_LOCKPROXYFILE: {
      unixFile *pFile = (unixFile*)id;
      if( pFile->pMethod == &proxyIoMethods ){
        proxyLockingContext *pCtx = (proxyLockingContext*)pFile->lockingContext;
        proxyTakeConch(pFile);
        if( pCtx->lockProxyPath ){
          *(const char **)pArg = pCtx->lockProxyPath;

**
** <li>[[SQLITE_FCNTL_WIN32_SET_HANDLE]]
** The [SQLITE_FCNTL_WIN32_SET_HANDLE] opcode is used for debugging.  This
** opcode causes the xFileControl method to swap the file handle with the one
** pointed to by the pArg argument.  This capability is used during testing
** and only needs to be supported when SQLITE_TEST is defined.
**
** <li>[[SQLITE_FCNTL_WAL_BLOCK]]
** The [SQLITE_FCNTL_WAL_BLOCK] is a single to the VFS layer that it might
** be advantageous to block on the next WAL lock if the lock is not immediately
** available.  The WAL subsystem issues this ioctl() during some rare
** circumstances in order to fix a problem with priority inversion.
** Applications should <em>not</em> use this file-control.
**
** </ul>
*/
#define SQLITE_FCNTL_LOCKSTATE               1
#define SQLITE_FCNTL_GET_LOCKPROXYFILE       2
#define SQLITE_FCNTL_SET_LOCKPROXYFILE       3
#define SQLITE_FCNTL_LAST_ERRNO              4
#define SQLITE_FCNTL_SIZE_HINT               5

#define SQLITE_FCNTL_TEMPFILENAME           16
#define SQLITE_FCNTL_MMAP_SIZE              18
#define SQLITE_FCNTL_TRACE                  19
#define SQLITE_FCNTL_HAS_MOVED              20
#define SQLITE_FCNTL_SYNC                   21
#define SQLITE_FCNTL_COMMIT_PHASETWO        22
#define SQLITE_FCNTL_WIN32_SET_HANDLE       23
#define SQLITE_FCNTL_WAL_BLOCK              24

/* deprecated names */
#define SQLITE_GET_LOCKPROXYFILE      SQLITE_FCNTL_GET_LOCKPROXYFILE
#define SQLITE_SET_LOCKPROXYFILE      SQLITE_FCNTL_SET_LOCKPROXYFILE
#define SQLITE_LAST_ERRNO             SQLITE_FCNTL_LAST_ERRNO

}
static void walUnlockShared(Wal *pWal, int lockIdx){
  if( pWal->exclusiveMode ) return;
  (void)sqlite3OsShmLock(pWal->pDbFd, lockIdx, 1,
                         SQLITE_SHM_UNLOCK | SQLITE_SHM_SHARED);
  WALTRACE(("WAL%p: release SHARED-%s\n", pWal, walLockName(lockIdx)));
}
static int walLockExclusive(Wal *pWal, int lockIdx, int n, int fBlock){
  int rc;
  if( pWal->exclusiveMode ) return SQLITE_OK;
  if( fBlock ) sqlite3OsFileControl(pWal->pDbFd, SQLITE_FCNTL_WAL_BLOCK, 0);
  rc = sqlite3OsShmLock(pWal->pDbFd, lockIdx, n,
                        SQLITE_SHM_LOCK | SQLITE_SHM_EXCLUSIVE);
  WALTRACE(("WAL%p: acquire EXCLUSIVE-%s cnt=%d %s\n", pWal,
            walLockName(lockIdx), n, rc ? "failed" : "ok"));
  VVA_ONLY( pWal->lockError = (u8)(rc!=SQLITE_OK && rc!=SQLITE_BUSY); )
  return rc;
}

  */
  assert( pWal->ckptLock==1 || pWal->ckptLock==0 );
  assert( WAL_ALL_BUT_WRITE==WAL_WRITE_LOCK+1 );
  assert( WAL_CKPT_LOCK==WAL_ALL_BUT_WRITE );
  assert( pWal->writeLock );
  iLock = WAL_ALL_BUT_WRITE + pWal->ckptLock;
  nLock = SQLITE_SHM_NLOCK - iLock;
  rc = walLockExclusive(pWal, iLock, nLock, 0);
  if( rc ){
    return rc;
  }
  WALTRACE(("WAL%p: recovery begin...\n", pWal));

  memset(&pWal->hdr, 0, sizeof(WalIndexHdr));

  int (*xBusy)(void*),            /* Function to call when busy */
  void *pBusyArg,                 /* Context argument for xBusyHandler */
  int lockIdx,                    /* Offset of first byte to lock */
  int n                           /* Number of bytes to lock */
){
  int rc;
  do {
    rc = walLockExclusive(pWal, lockIdx, n, 0);
  }while( xBusy && rc==SQLITE_BUSY && xBusy(pBusyArg) );
  return rc;
}

/*
** The cache of the wal-index header must be valid to call this function.
** Return the page-size in bytes used by the database.

  assert( badHdr==0 || pWal->writeLock==0 );
  if( badHdr ){
    if( pWal->readOnly & WAL_SHM_RDONLY ){
      if( SQLITE_OK==(rc = walLockShared(pWal, WAL_WRITE_LOCK)) ){
        walUnlockShared(pWal, WAL_WRITE_LOCK);
        rc = SQLITE_READONLY_RECOVERY;
      }
    }else if( SQLITE_OK==(rc = walLockExclusive(pWal, WAL_WRITE_LOCK, 1, 1)) ){
      pWal->writeLock = 1;
      if( SQLITE_OK==(rc = walIndexPage(pWal, 0, &page0)) ){
        badHdr = walIndexTryHdr(pWal, pChanged);
        if( badHdr ){
          /* If the wal-index header is still malformed even while holding
          ** a WRITE lock, it can only mean that the header is corrupted and
          ** needs to be reconstructed.  So run recovery to do exactly that.

  }
  /* There was once an "if" here. The extra "{" is to preserve indentation. */
  {
    if( (pWal->readOnly & WAL_SHM_RDONLY)==0
     && (mxReadMark<pWal->hdr.mxFrame || mxI==0)
    ){
      for(i=1; i<WAL_NREADER; i++){
        rc = walLockExclusive(pWal, WAL_READ_LOCK(i), 1, 0);
        if( rc==SQLITE_OK ){
          mxReadMark = pInfo->aReadMark[i] = pWal->hdr.mxFrame;
          mxI = i;
          walUnlockExclusive(pWal, WAL_READ_LOCK(i), 1);
          break;
        }else if( rc!=SQLITE_BUSY ){
          return rc;

  if( pWal->readOnly ){
    return SQLITE_READONLY;
  }

  /* Only one writer allowed at a time.  Get the write lock.  Return
  ** SQLITE_BUSY if unable.
  */
  rc = walLockExclusive(pWal, WAL_WRITE_LOCK, 1, 0);
  if( rc ){
    return rc;
  }
  pWal->writeLock = 1;

  /* If another connection has written to the database file since the
  ** time the read transaction on this connection was started, then

  if( pWal->readLock==0 ){
    volatile WalCkptInfo *pInfo = walCkptInfo(pWal);
    assert( pInfo->nBackfill==pWal->hdr.mxFrame );
    if( pInfo->nBackfill>0 ){
      u32 salt1;
      sqlite3_randomness(4, &salt1);
      rc = walLockExclusive(pWal, WAL_READ_LOCK(1), WAL_NREADER-1, 0);
      if( rc==SQLITE_OK ){
        /* If all readers are using WAL_READ_LOCK(0) (in other words if no
        ** readers are currently using the WAL), then the transactions
        ** frames will overwrite the start of the existing log. Update the
        ** wal-index header to reflect this.
        **
        ** In theory it would be Ok to update the cache of the header only

  assert( eMode!=SQLITE_CHECKPOINT_PASSIVE || xBusy==0 );

  if( pWal->readOnly ) return SQLITE_READONLY;
  WALTRACE(("WAL%p: checkpoint begins\n", pWal));

  /* IMPLEMENTATION-OF: R-62028-47212 All calls obtain an exclusive 
  ** "checkpoint" lock on the database file. */
  rc = walLockExclusive(pWal, WAL_CKPT_LOCK, 1, 0);
  if( rc ){
    /* EVIDENCE-OF: R-10421-19736 If any other process is running a
    ** checkpoint operation at the same time, the lock cannot be obtained and
    ** SQLITE_BUSY is returned.
    ** EVIDENCE-OF: R-53820-33897 Even if there is a busy-handler configured,
    ** it will not be invoked in this case.
    */