[Ocfs2-users] ocfs2_delete_inode kernel BUG
Sunil Mushran
Sunil.Mushran at oracle.com
Wed May 31 15:38:29 CDT 2006
ahh.. you built 1.2.1 on 2.6.15/16. Then it's explainable.
Remove the following from fs/ocfs2/Makefile and rebuild.
EXTRA_CFLAGS += -DOCFS2_DELETE_INODE_WORKAROUND
It has to do with a change in generic_delete_inode() introduced in 2.6.14.
1.2.1 has this #define because it supports older kernels too.
Sjon Stigter wrote:
> Here is some more debug information with inode debugging turned on in OCFS2:
>
> <6>(3826,1):ocfs2_populate_inode:282 blkno = 262971, ino = 262971, create_ino = true
> (3826,1):ocfs2_extend_file:802 extend inode 262971, new_i_size = 2, i_size = 0, fe->i_clusters = 0, clusters_to_add = 1
> (3826,1):ocfs2_extend_allocation:691 Allocating 1 clusters at block 32306 for inode 262971
> (3826,1):ocfs2_extend_file:944 fe: i_clusters = 1, i_size=2
> (3826,1):ocfs2_extend_file:947 inode: ip_clusters=1, i_size=0
> (3826,1):ocfs2_file_finish_extension:165 inode 262971, newsize = 2, direct_extend = 0
> (8255,3):ocfs2_drop_inode:989 Drop inode 262971, nlink = 0, ip_flags = 0x80
> (8255,3):ocfs2_query_inode_wipe:776 Inode 262971 is ok to wipe from orphan dir 0
> (8255,3):ocfs2_cleanup_delete_inode:791 Cleanup inode 262971, sync = 0
> ------------[ cut here ]------------
> kernel BUG at <bad filename>:33675!
> invalid operand: 0000 [#3]
> PREEMPT SMP
> Modules linked in: ocfs2 ipv6 ocfs2_dlmfs ocfs2_dlm ocfs2_nodemanager configfs iscsi_tcp scsi_transport_iscsi bonding floppy pcspkr generic hw_random shpchp pci_hotplug ehci_hcd uhci_hcd usbcore piix tsdev mousedev e1000 ip_vs psmouse ide_disk ide_cd cdrom ide_core genrtc unix
> CPU: 3
> EIP: 0060:[<c0162a71>] Not tainted VLI
> EFLAGS: 00010202 (2.6.15)
> EIP is at clear_inode+0x16/0xd2
> eax: f2eeb624 ebx: f2eeb624 ecx: ffffffff edx: f2eeb624
> esi: 00000000 edi: f2eeb624 ebp: f2eeb624 esp: f2b4dee4
> ds: 007b es: 007b ss: 0068
> Process rm (pid: 8255, threadinfo=f2b4c000 task=f586b030)
> Stack: f2eeb624 00000000 f9c388ba f2eeb624 00000001 f2ea6bcc 00000000 00000000
> ffffffff ffffffff f2eeb624 f9c3839e f2eeb400 f2eeb624 c01637a9 f2eeb624
> f2b4c000 00000003 f9c399c2 f2eeb624 f2eeb624 f2f32000 f56407d8 f2eeb624
> Call Trace:
> [<f9c388ba>] ocfs2_delete_inode+0x51c/0x58b [ocfs2]
> [<f9c3839e>] ocfs2_delete_inode+0x0/0x58b [ocfs2]
> [<c01637a9>] generic_delete_inode+0xa1/0x10d
> [<f9c399c2>] ocfs2_drop_inode+0x158/0x1c4 [ocfs2]
> [<c01639b6>] iput+0x5d/0x60
> [<c015aff3>] sys_unlink+0xb3/0xf8
> [<c0102951>] syscall_call+0x7/0xb
> Code: c7 42 04 24 f1 31 c0 89 15 24 f1 31 c0 ff 0d 84 72 3d c0 5b c3 56 53 8b 5c 24 0c 53 e8 7f bf fe ff 83 bb f0 00 00 00 00 58 74 02 <0f> 0b 8b 83 64 01 00 00 a8 10 75 02 0f 0b a8 20 74 02 0f 0b 8b
>
>
> This is with version 1.2.1 of OCFS2 and the 1.2.1 tools version on a clean install of Debian Sarge 3.1r2, problem occurs with both kernel version 2.6.15 as with 2.6.16.15 from kernel.org.
>
>
> -----Original Message-----
> From: Sunil Mushran [mailto:Sunil.Mushran at oracle.com]
> Sent: woensdag 31 mei 2006 20:18
> To: Sjon Stigter
> Cc: ocfs2-users at oss.oracle.com
> Subject: Re: [Ocfs2-users] ocfs2_delete_inode kernel BUG
>
> Looks like a build issue in more ways than one.
>
> kernel BUG at <bad filename>:33675! ????
>
>
> I suspect the location in clear_inode is:
>
> if (inode->i_data.nrpages)
> BUG();
>
> That means this is related to the
> truncate_inode_pages()/generic_delete_inode()
> change introduced in 2.6.14.
>
> Joel?
>
> Sjon Stigter wrote:
>
>> When removing data from our OCFS2 filesystem we run into a kernel bug
>> on ocfs2_delete_inode, we have this problem on both kernel 2.6.16.15
>> and 2.6.15 on Debian Sarge. Anyone have any suggestions on how to fix
>> this?
>>
>>
>>
>> ------------[ cut here ]------------
>>
>> kernel BUG at <bad filename>:33675!
>>
>> invalid operand: 0000 [#1]
>>
>> PREEMPT SMP
>>
>> Modules linked in: ocfs2 ipv6 ocfs2_dlmfs ocfs2_dlm ocfs2_nodemanager
>> configfs iscsi_tcp scsi_transport_iscsi bonding floppy pcspkr generic
>> hw_random shpchp pci_hotplug ehci_hcd uhci_hcd usbcore piix tsdev
>> mousedev e1000 ip_vs psmouse ide_disk ide_cd cdrom ide_core genrtc
>> unix
>>
>> CPU: 0
>>
>> EIP: 0060:[<c0162a71>] Not tainted VLI
>>
>> EFLAGS: 00010202 (2.6.15)
>>
>> EIP is at clear_inode+0x16/0xd2
>>
>> eax: f2eebe24 ebx: f2eebe24 ecx: ffffffff edx: f2eebe24
>>
>> esi: 00000000 edi: f2eebe24 ebp: f2eebe24 esp: f2ed7ee4
>>
>> ds: 007b es: 007b ss: 0068
>>
>> Process mysqld (pid: 4112, threadinfo=f2ed6000 task=f5e3a550)
>>
>> Stack: f2eebe24 00000000 f9c388ba f2eebe24 00000001 f2ea6bcc 00087007
>> 00000000
>>
>> ffffffff ffffffff f2eebe24 f9c3839e f2eebc00 f2eebe24 c01637a9
>> f2eebe24
>>
>> f2eebe24 f2f11000 f9c399c2 f2eebe24 f2eebe24 f2f11000 f5640358
>> f2eebe24
>>
>> Call Trace:
>>
>> [<f9c388ba>] ocfs2_delete_inode+0x51c/0x58b [ocfs2]
>>
>> [<f9c3839e>] ocfs2_delete_inode+0x0/0x58b [ocfs2]
>>
>> [<c01637a9>] generic_delete_inode+0xa1/0x10d
>>
>> [<f9c399c2>] ocfs2_drop_inode+0x158/0x1c4 [ocfs2]
>>
>> [<c01639b6>] iput+0x5d/0x60
>>
>> [<c015aff3>] sys_unlink+0xb3/0xf8
>>
>> [<c014c4cf>] filp_close+0x4c/0x55
>>
>> [<c014c53c>] sys_close+0x64/0x7d
>>
>> [<c0102951>] syscall_call+0x7/0xb
>>
>> Code: c7 42 04 24 f1 31 c0 89 15 24 f1 31 c0 ff 0d 84 72 3d c0 5b c3
>> 56 53 8b 5c 24 0c 53 e8 7f bf fe ff 83 bb f0 00 00 00 00 58 74 02 <0f>
>> 0b 8b 83 64 01 00 00 a8 10 75 02 0f 0b a8 20 74 02 0f 0b 8b
>>
>>
>>
>>
>>
>>
>> *HCN Logo*
>>
>> HCN, Hét Callcenter Netwerk B.V
>>
>> Stationsstraat 19
>>
>> Postbus 680, 5000 AR Tilburg
>>
>> Sjon Stigter* * * *
>> System Engineer
>>
>> Tel: 013 - 464 99 99
>>
>> Mobiel: 06 53 86 16 87
>>
>> Fax: 013 - 464 99 90
>>
>> Email: s.stigter at hcn.nl
>>
>> Internet: http://www.hcn.nl
>>
>>
>> /Dit bericht kan vertrouwelijke informatie bevatten. Indien u niet de
>> geadresseerde van dit bericht bent, wordt u verzocht dit bericht te
>> vernietigen zonder van de inhoud kennis te nemen en de inhoud ervan
>> niet te gebruiken, niet te kopiëren en niet onder derden te verspreiden.
>>
>> This message may contain information which is privileged or
>> confidential. If you are not the named addressee of this message
>> please destroy it without reading, using, copying or disclosing its
>> contents to any other person. /
>>
>> ----------------------------------------------------------------------
>> --
>>
>> _______________________________________________
>> Ocfs2-users mailing list
>> Ocfs2-users at oss.oracle.com
>> http://oss.oracle.com/mailman/listinfo/ocfs2-users
>>
>>
>
> _______________________________________________
> Ocfs2-users mailing list
> Ocfs2-users at oss.oracle.com
> http://oss.oracle.com/mailman/listinfo/ocfs2-users
>
More information about the Ocfs2-users
mailing list