linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] [RESEND] ReiserFS _get_block_create_0 wrong behavior when I/O fails
@ 2005-06-22 20:19 fs
  2005-06-23  6:19 ` Hans Reiser
  0 siblings, 1 reply; 2+ messages in thread
From: fs @ 2005-06-22 20:19 UTC (permalink / raw)
  To: reiserfs-list; +Cc: Vladimir Saveliev, Hans Reiser, linux-fsdevel

[-- Attachment #1: Type: text/plain, Size: 1520 bytes --]

Seems my domain is filtered , so I resend the modified version.

Related FS:
    ReiserFS

Related Files:
    fs/reiserfs/inode.c

Bug description:
    Make a ReiserFS partition in USB storage HDD, create a test file
with
enough size.
    Write a program, do: open(O_RDONLY) - read - close. After each
operation, pause for a while, such as 3s. Between open and read, unlug
the
USB wire. open returns zero-filled buffer, no error returns.

Bug analysis:
    do_mpage_readpage will call FS-specific get_block to get buffer
mapped
from disk. reiserfs_get_block doesn't return non-zero when I/O failure
occurs.
    reiserfs_get_block -> _get_block_create_0 ->
search_by_position_by_key
search_by_position_by_key returns IO_ERROR, but the original code just
simply
returns 0

research:
    if (search_for_position_by_key (inode->i_sb, &key, &path) !=
POSITION_FOUND) {
        pathrelse (&path);
        if (p)
            kunmap(bh_result->b_page) ;
        // We do not return -ENOENT if there is a hole but page is
uptodate, because it means
        // That there is some MMAPED data associated with it that is yet
to be written to disk.
        if ((args & GET_BLOCK_NO_HOLE) &&
!PageUptodate(bh_result->b_page) ) {
            return -ENOENT ;
        }
        return 0 ; <- 0 retuns for IO_ERROR
    }

Way around:
    test result of search_for_position_by_key

Signed-off-by: Qu Fuping<fs@ercist.iscas.ac.cn>

Patch:
diff -uNp /tmp/linux-2.6.12-rc6/fs/reiserfs/inode.c
/tmp/linux-2.6.12-rc6.new/fs/reiserfs/inode.c



[-- Attachment #2: reiserfs_read.diff --]
[-- Type: text/x-patch, Size: 1751 bytes --]

--- linux-2.6.12-rc6.old/fs/reiserfs/inode.c	2005-06-06 11:22:29.000000000 -0400
+++ linux-2.6.12-rc6.new/fs/reiserfs/inode.c	2005-06-17 16:12:18.000000000 -0400
@@ -254,6 +254,7 @@ static int _get_block_create_0 (struct i
     char * p = NULL;
     int chars;
     int ret ;
+    int	result ;
     int done = 0 ;
     unsigned long offset ;
 
@@ -262,7 +263,8 @@ static int _get_block_create_0 (struct i
 		  (loff_t)block * inode->i_sb->s_blocksize + 1, TYPE_ANY, 3);
 
 research:
-    if (search_for_position_by_key (inode->i_sb, &key, &path) != POSITION_FOUND) {
+    result = search_for_position_by_key (inode->i_sb, &key, &path) ;
+    if (result != POSITION_FOUND) {
 	pathrelse (&path);
         if (p)
             kunmap(bh_result->b_page) ;
@@ -270,7 +272,8 @@ research:
 	// That there is some MMAPED data associated with it that is yet to be written to disk.
 	if ((args & GET_BLOCK_NO_HOLE) && !PageUptodate(bh_result->b_page) ) {
 	    return -ENOENT ;
-	}
+	}else if(result == IO_ERROR)
+		return -EIO;
         return 0 ;
     }
     
@@ -382,7 +385,8 @@ research:
 
 	// update key to look for the next piece
 	set_cpu_key_k_offset (&key, cpu_key_k_offset (&key) + chars);
-	if (search_for_position_by_key (inode->i_sb, &key, &path) != POSITION_FOUND)
+	result = search_for_position_by_key (inode->i_sb, &key, &path);
+	if (result != POSITION_FOUND)
 	    // we read something from tail, even if now we got IO_ERROR
 	    break;
 	bh = get_last_bh (&path);
@@ -394,6 +398,10 @@ research:
 
 finished:
     pathrelse (&path);
+
+    if(result == IO_ERROR)
+	    return -EIO;
+    
     /* this buffer has valid data, but isn't valid for io.  mapping it to
      * block #0 tells the rest of reiserfs it just has a tail in it
      */

^ permalink raw reply	[flat|nested] 2+ messages in thread
end of thread, other threads:[~2005-06-23  6:19 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2005-06-22 20:19 [PATCH] [RESEND] ReiserFS _get_block_create_0 wrong behavior when I/O fails fs
2005-06-23  6:19 ` Hans Reiser

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).