Re: Reiserfs 3.6 + quota enabled, crash on delete

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 

Hello

Guillaume Pelat wrote:

Hi,
I'm having a crash with reiserfs 3.6 + user quota enabled, on 2.6.11.10 kernel (no smp), apparently when deleting files (or maybe during a truncate operation). The problem seems to happen under high load. When the error occurs, all the processes accessing the reiserfs partition seems to hang. This problem happened several times on different servers (having the same hardware configuration) during last weeks. 

You can find the error logs below :

======================================================================
ReiserFS: sda3: warning: vs-15011: reiserfs_release_objectid: tried to free free object id (96557091) ReiserFS: sda3: warning: PAP-5660: reiserfs_do_truncate: wrong result -1 of search for [847141 185744 0xfffffffffffffff DIRECT] 
ReiserFS: sda3: warning: clm-2100: nesting info a different FS


The attached patch should help to get rid of clm-2100 and to avoid crash.

Also, I think you should reiserfsck sda3.


ReiserFS: sda3: warning: clm-2100: nesting info a different FS
ReiserFS: sda3: warning: clm-2100: nesting info a different FS
ReiserFS: sda3: warning: clm-2100: nesting info a different FS
REISERFS: panic (device sda3): journal-1577: handle trans id 1122409068 != current trans id 3947596 
------------[ cut here ]------------
kernel BUG at fs/reiserfs/prints.c:362!
invalid operand: 0000 [#1]
CPU:    0
EIP:    0060:[<c019785f>]    Not tainted VLI
EFLAGS: 00010282   (2.6.11.6)
EIP is at reiserfs_panic+0x4f/0x80
eax: 00000068   ebx: c02be9bf   ecx: c038e8f8   edx: c02eecac
esi: f7eea000   edi: f7eea140   ebp: f3391944   esp: f339192c
ds: 007b   es: 007b   ss: 0068
Process pure-ftpd (pid: 27841, threadinfo=f3390000 task=d3053530)
Stack: c02c3ee0 f7eea140 c03b3580 f7eea000 00000000 f895d000 f3391984 c01a6d01 f7eea000 c02c6440 42e69a6c 003c3c4c 00000002 f3391970 cdc49b40 00000002 cdc49b40 f3391984 00000000 052f8dee 00000000 f2bbfe60 f3391a08 c019ebba 
Call Trace:
 [<c010282f>] show_stack+0x7f/0xa0
 [<c01029d1>] show_registers+0x151/0x1c0
 [<c0102bc8>] die+0xc8/0x150
 [<c010307c>] do_invalid_op+0xbc/0xd0
 [<c01024bb>] error_code+0x2b/0x30
 [<c01a6d01>] journal_mark_dirty+0x271/0x2a0
 [<c019ebba>] prepare_for_delete_or_cut+0x54a/0x720
 [<c019fdaa>] reiserfs_cut_from_item+0xca/0x5f0
 [<c01a0668>] reiserfs_do_truncate+0x2e8/0x610
 [<c019f7ef>] reiserfs_delete_object+0x3f/0x80
 [<c018636c>] reiserfs_delete_inode+0x8c/0x110
 [<c015ed85>] generic_delete_inode+0x95/0x130
 [<c015efc6>] iput+0x56/0x80
 [<c018995a>] reiserfs_new_inode+0x13a/0x740
 [<c01847b7>] reiserfs_create+0x97/0x1b0
 [<c0153e0f>] vfs_create+0x9f/0x120
 [<c01546f9>] open_namei+0x5d9/0x630
 [<c0144bbc>] filp_open+0x3c/0x60
 [<c0144ed6>] sys_open+0x46/0x90
 [<c0102313>] syscall_call+0x7/0xb
Code: 01 00 00 89 04 24 e8 31 fd ff ff c7 04 24 e0 3e 2c c0 85 f6 89 d8 0f 45 c7 ba 80 35 3b c0 89 54 24 08 89 44 24 04 e8 c1 a9 f7ff <0f> 0b 6a 01 02 ef 2b c0 c7 04 24 20 3f 2c c0 85 f6 b9 80 35 3b 
======================================================================


This patch adds missing calls to journal_end on error handling code paths.


 fs/reiserfs/inode.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletion(-)

diff -puN fs/reiserfs/inode.c~reiserfs-add-missing-journal_end fs/reiserfs/inode.c
--- linux-2.6.11.10/fs/reiserfs/inode.c~reiserfs-add-missing-journal_end	2005-07-27 18:03:06.197544430 +0400
+++ linux-2.6.11.10-vs/fs/reiserfs/inode.c	2005-07-27 18:04:07.363047159 +0400
@@ -46,6 +46,7 @@ void reiserfs_delete_inode (struct inode
 	reiserfs_update_inode_transaction(inode) ;
 
 	if (reiserfs_delete_object (&th, inode)) {
+	    journal_end(&th, inode->i_sb, jbegin_count);
 	    up (&inode->i_sem);
 	    goto out;
 	}
@@ -2015,8 +2016,10 @@ int reiserfs_truncate_file(struct inode 
 	       either appears truncated properly or not truncated at all */
 	add_save_link (&th, p_s_inode, 1);
     error = reiserfs_do_truncate (&th, p_s_inode, page, update_timestamps) ;
-    if (error)
+    if (error) {
+	journal_end (&th, p_s_inode->i_sb, JOURNAL_PER_BALANCE_CNT * 2 + 1);
         goto out;
+    }
     error = journal_end (&th, p_s_inode->i_sb, JOURNAL_PER_BALANCE_CNT * 2 + 1);
     if (error)
         goto out;

_
[Index of Archives]     [Kernel Newbies]     [Netfilter]     [Bugtraq]     [Photo]     [Gimp]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Video 4 Linux]     [Linux for the blind]
  Powered by Linux