]> pilppa.com Git - linux-2.6-omap-h63xx.git/commitdiff
[PATCH] reiserfs: do not add save links for O_DIRECT writes
authorVladimir V. Saveliev <vs@namesys.com>
Thu, 7 Dec 2006 04:39:12 +0000 (20:39 -0800)
committerLinus Torvalds <torvalds@woody.osdl.org>
Thu, 7 Dec 2006 16:39:42 +0000 (08:39 -0800)
We add a save link for O_DIRECT writes to protect the i_size against the
crashes before we actually finish the I/O.  If we hit an -ENOSPC in
aops->prepare_write(), we would do a truncate() to release the blocks which
might have got initialized.  Now the truncate would add another save link
for the same inode causing a reiserfs panic for having multiple save links
for the same inode.

Signed-off-by: Vladimir V. Saveliev <vs@namesys.com>
Signed-off-by: Amit Arora <amitarora@in.ibm.com>
Signed-off-by: Suzuki K P <suzuki@in.ibm.com>
Cc: Jeff Mahoney <jeffm@suse.com>
Cc: Chris Mason <mason@suse.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
fs/reiserfs/file.c

index 970ecd9dbe69e225b7c4ba822cdf4d17514acc6b..373d862c3f87da6c12650a7427a64eb5740c5481 100644 (file)
@@ -406,6 +406,8 @@ static int reiserfs_allocate_blocks_for_region(struct reiserfs_transaction_handl
                                   we restart it. This will also free the path. */
                                if (journal_transaction_should_end
                                    (th, th->t_blocks_allocated)) {
+                                       inode->i_size = cpu_key_k_offset(&key) +
+                                               (to_paste << inode->i_blkbits);
                                        res =
                                            restart_transaction(th, inode,
                                                                &path);
@@ -1310,56 +1312,8 @@ static ssize_t reiserfs_file_write(struct file *file,    /* the file we are going t
                        count = MAX_NON_LFS - (unsigned long)*ppos;
        }
 
-       if (file->f_flags & O_DIRECT) { // Direct IO needs treatment
-               ssize_t result, after_file_end = 0;
-               if ((*ppos + count >= inode->i_size)
-                   || (file->f_flags & O_APPEND)) {
-                       /* If we are appending a file, we need to put this savelink in here.
-                          If we will crash while doing direct io, finish_unfinished will
-                          cut the garbage from the file end. */
-                       reiserfs_write_lock(inode->i_sb);
-                       err =
-                           journal_begin(&th, inode->i_sb,
-                                         JOURNAL_PER_BALANCE_CNT);
-                       if (err) {
-                               reiserfs_write_unlock(inode->i_sb);
-                               return err;
-                       }
-                       reiserfs_update_inode_transaction(inode);
-                       add_save_link(&th, inode, 1 /* Truncate */ );
-                       after_file_end = 1;
-                       err =
-                           journal_end(&th, inode->i_sb,
-                                       JOURNAL_PER_BALANCE_CNT);
-                       reiserfs_write_unlock(inode->i_sb);
-                       if (err)
-                               return err;
-               }
-               result = do_sync_write(file, buf, count, ppos);
-
-               if (after_file_end) {   /* Now update i_size and remove the savelink */
-                       struct reiserfs_transaction_handle th;
-                       reiserfs_write_lock(inode->i_sb);
-                       err = journal_begin(&th, inode->i_sb, 1);
-                       if (err) {
-                               reiserfs_write_unlock(inode->i_sb);
-                               return err;
-                       }
-                       reiserfs_update_inode_transaction(inode);
-                       mark_inode_dirty(inode);
-                       err = journal_end(&th, inode->i_sb, 1);
-                       if (err) {
-                               reiserfs_write_unlock(inode->i_sb);
-                               return err;
-                       }
-                       err = remove_save_link(inode, 1 /* truncate */ );
-                       reiserfs_write_unlock(inode->i_sb);
-                       if (err)
-                               return err;
-               }
-
-               return result;
-       }
+       if (file->f_flags & O_DIRECT)
+               return do_sync_write(file, buf, count, ppos);
 
        if (unlikely((ssize_t) count < 0))
                return -EINVAL;