Skip to contentSkip to conversation

Slab corruption with ext3-handle-cache.patch

DISCUSSION Martin Schlemmer 2 participants 7 messages Apr 5, 2003 – Apr 9, 2003

lore
Share & tools
lore
#0 lore raw

Hi

I tried this patch a few days again, but not in the mm patch set. So when I got problems with slab corruption, I figured it was maybe a patch I missed, and dropped it.

Now with 2.5.66-bk10+, it is merged, and I get it again.

----------------------------------------------------------
Slab corruption: start=cfaeecc4, expend=cfaeee77, problemat=cfaeed28
Last user: [<c01890de>](ext3_destroy_inode+0x1b/0x1f)
Data:
****************************************************************************************************3C 26 99 DF *******************************************************************************************************************************************************************************************************************************************************************************************************************************************A5
Next: 71 F0 2C .DE 90 18 C0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00
slab error in check_poison_obj(): cache `ext3_inode_cache': object was
modified after freeing

Call Trace:
 [<c0137894>] check_poison_obj+0x14f/0x18f
 [<c0138c8b>] kmem_cache_alloc+0x11c/0x141
 [<c01890aa>] ext3_alloc_inode+0x18/0x31
 [<c01890aa>] ext3_alloc_inode+0x18/0x31
 [<c0162c0c>] alloc_inode+0x1c/0x13d
 [<c0163510>] new_inode+0x13/0x7d
 [<c017fcf2>] ext3_new_inode+0x41/0x7c7
 [<c0138c8b>] kmem_cache_alloc+0x11c/0x141
 [<c018de05>] start_this_handle+0x94/0x1b9
 [<c018df46>] new_handle+0x1c/0x50
 [<c018e025>] journal_start+0xab/0xd2
 [<c0187fb4>] ext3_symlink+0xd6/0x2bd
 [<c015a4c7>] vfs_symlink+0x5c/0xa3
 [<c015a5f0>] sys_symlink+0xe2/0xf2
 [<c010a8fd>] sysenter_past_esp+0x52/0x71

-----------------------------------------------------------

If you need more info, just ask.

Regards,

Signature · 1 line
Martin Schlemmer
#1 lore raw
Quoted 7 lines · Martin Schlemmer

Hi

I tried this patch a few days again, but not in the mm patch set. So when I got problems with slab corruption, I figured it was maybe a patch I missed, and dropped it.

Quoted 7 lines

Now with 2.5.66-bk10+, it is merged, and I get it again.

----------------------------------------------------------
Slab corruption: start=cfaeecc4, expend=cfaeee77, problemat=cfaeed28
Last user: [<c01890de>](ext3_destroy_inode+0x1b/0x1f)
Data:
****************************************************************************************************3C 26 99 DF *******************************************************************************************************************************************************************************************************************************************************************************************************************************************A5

Do you have CONFIG_EXT3_FS_POSIX_ACL enabled?

#2 lore raw
Quoted 10 lines · Andrew Morton
Quoted 9 lines · Martin Schlemmer

Hi

I tried this patch a few days again, but not in the mm patch set. So when I got problems with slab corruption, I figured it was maybe a patch I missed, and dropped it.

Now with 2.5.66-bk10+, it is merged, and I get it again.

Quoted 7 lines
Quoted 5 lines

----------------------------------------------------------
Slab corruption: start=cfaeecc4, expend=cfaeee77, problemat=cfaeed28
Last user: [<c01890de>](ext3_destroy_inode+0x1b/0x1f)
Data:
****************************************************************************************************3C 26 99 DF *******************************************************************************************************************************************************************************************************************************************************************************************************************************************A5

Do you have CONFIG_EXT3_FS_POSIX_ACL enabled?

Nope.

-------------------

 # grep CONFIG_EXT3_FS_POSIX_ACL .config
# CONFIG_EXT3_FS_POSIX_ACL is not set
 # 
-------------------
Signature · 1 line
Martin Schlemmer
#3 lore raw
Quoted 4 lines · Martin Schlemmer
Quoted 2 lines

Do you have CONFIG_EXT3_FS_POSIX_ACL enabled?

Nope.

So we're leaving an inode on the orphan list. I wonder why that started happening. Thanks.

#4 lore raw
Quoted 3 lines · Andrew Morton

So we're leaving an inode on the orphan list. I wonder why that started happening. Thanks.

Well, HTREE/dir_index is used on that partition. I did mail about a weird problem some time ago where some files are deleted, but don't get removed (or something to that regards). The thread was:

Corruption problem with ext3 and htree

Sorry, cannot get onto http://www.lkml.org/ right now.

Regards,

Signature · 1 line
Martin Schlemmer
#5 lore raw
Quoted 6 lines · Martin Schlemmer
Quoted 3 lines · Andrew Morton

So we're leaving an inode on the orphan list. I wonder why that started happening. Thanks.

Quoted 8 lines

Well, HTREE/dir_index is used on that partition. I did mail about a weird problem some time ago where some files are deleted, but don't get removed (or something to that regards). The thread was:

Corruption problem with ext3 and htree

Sorry, cannot get onto http://www.lkml.org/ right now.

Err, what I tried to say here, was that I do not know if it could be related ...

Regards,

Signature · 1 line
Martin Schlemmer
#6 lore raw
In reply to Andrew Morton
Quoted 8 lines · Andrew Morton
Quoted 4 lines · Martin Schlemmer
Quoted 2 lines

Do you have CONFIG_EXT3_FS_POSIX_ACL enabled?

Nope.

So we're leaving an inode on the orphan list. I wonder why that started happening. Thanks.

I moved all my data off that partition, removed it, and recreated it without HTREE/dir_index, and moved everything back. All issues are resolved, as well as the weird corruption I had.

It seems thus that its only related to HTREE ...

Regards,

Signature · 1 line
Martin Schlemmer