Re: [Oops] 2.6.10: PREEMPT SMP

From: Andrew Taylor
Date: Wed Mar 23 2005 - 19:52:40 EST

Hello,

On Mon, Mar 21, 2005 at 03:35:37PM -0800, Andrew Morton wrote:
> Andrew Taylor <taylor@xxxxxxxx> wrote:
> >
> > Hello,
> >
> > We have been experiencing a very similar oops on a Dell poweredge 2600, running Fedora Core 3 (FC3). This host is mainly used as a samba fileserver. The problem has been seen on 2 different machines (both Dell-pe-2600s).
>
> Could you pleae confirm that 2.6.12-rc1 fixes this?

Would be glad to however, between 2.6.11-5 (which appears to runs fine) and 2.6.12-rc1 the mptfusion driver got broken. Attempts to isolate the mptfusion changes have not been successful. Will look at this issue again on tuesday.

Cheers,

Andrew

>
> Thanks.
>
> > The problem is only ever seen when running the SMP kernel in conjunction with samba.
> >
> > We have been experiencing problems since early January when the samba/nfs/io load was increased considerably. Since then a number of 2.4 kernels have been used:
> > 2.4.22.2115nptlsmp (FC1)
> > 2.4.22-1.2199smp (FC1)
> > 2.4.29 (vanilla kernel)
> > 2.6.10-1.741_FC3smp (FC3)
> > 2.6.10-1.766_FC3smp (FC3)
> >
> > All 2.4 kernels produced the same result: full system lock-up, with no messages on the console or the log files.
> >
> > The 2.6.10 kernel is giving an oops which has been seen 3 time in the last 3 days. A very similar one was seen pointing to eip: c01b2d98 on the 2.6.10-1.741 kernel.
> >
> > I have included the kallsyms and System.map relevant to the instruction pointed to by the EIP. I can include the full files if required.
> >
> > printing eip:
> > c01b387c
> > *pde = 2fcdd001
> > Oops: 0000 [#1]
> > SMP
> > Modules linked in: nfsd exportfs md5 ipv6 parport_pc lp parport autofs4 nfs lockd sunrpc video button battery ac uhci_hcd hw_random e1000 floppy sg dm_snapshot dm_zero dm_mirror ext3 jbd dm_mod mptscsih mptbase sd_mod scsi_mod
> > CPU: 1
> > EIP: 0060:[<c01b387c>] Not tainted VLI
> > EFLAGS: 00010203 (2.6.10-1.766_FC3smp)
> > EIP is at __rb_rotate_left+0x8/0x36
> > eax: f57be640 ebx: c041e5e4 ecx: f57be640 edx: 00000000
> > esi: f57be640 edi: f5f39c80 ebp: c041e5e4 esp: efa2eed0
> > ds: 007b es: 007b ss: 0068
> > Process smbd (pid: 8536, threadinfo=efa2e000 task=f7612a60)
> > Stack: f5ac1980 c01b3988 f5f39c80 f5f39c80 f5ac1988 00000267 c01965d7 f5ac1980
> > 0000000d efa2ef54 efa2ef61 ffffffea c019668b 00000015 00000000 00000267
> > c03191e0 efa2ef54 00000000 f599ed80 00000267 c01977df ffffffff 001f0000
> > Call Trace:
> > [<c01b3988>] rb_insert_color+0xa8/0xc1
> > [<c01965d7>] key_user_lookup+0xcf/0xfc
> > [<c019668b>] key_alloc+0x53/0x2b6
> > [<c01977df>] keyring_alloc+0x1a/0x48
> > [<c0198d43>] alloc_uid_keyring+0x2b/0x7c
> > [<c0125d2e>] alloc_uid+0xae/0x133
> > [<c01291d4>] set_user+0xb/0x8b
> > [<c01295f2>] sys_setresuid+0x11a/0x1b9
> > [<c0103c97>] syscall_call+0x7/0xb
> > Code: 82 04 01 00 00 00 75 ea 41 83 f9 01 76 ed 31 c0 5b c3 57 b9 45 00 00 00 89 c7 31 c0 f3 ab 5f c3 90 90 90 53 89 d3 8b 50 08 89 c1 <8b> 42 0c 85 c0 89 41 08 74 02 89 08 89 4a 0c 8b 01 85 c0 89 02
> >
> >
> > /proc/kallsyms grepped on c01b3.
> >
> > c01b3024 t newary
> > c01b313c T sys_semget
> > c01b325a t try_atomic_semop
> > c01b336b t update_queue
> > c01b33de t count_semncnt
> > c01b342c t count_semzcnt
> > c01b347a t freeary
> > c01b34f6 t copy_semid_to_user
> > c01b354b t semctl_nolock
> > c01b3731 t semctl_main
> > c01b3ad6 t semctl_down
> > c01b3c39 T sys_semctl
> > c01b3ce9 t lookup_undo
> > c01b3d19 t find_undo
> >
> >
> > root@msh2-c boot]# grep -i c01b3 /boot/System.map-2.6.10-1.766_FC3smp
> > c01b304b T match_int
> > c01b3052 T match_octal
> > c01b305c T match_hex
> > c01b3066 T match_strcpy
> > c01b3098 T match_strdup
> > c01b30c0 t radix_tree_node_alloc
> > c01b3109 T radix_tree_preload
> > c01b317d t radix_tree_extend
> > c01b321c T radix_tree_insert
> > c01b3303 T radix_tree_lookup
> > c01b3349 T radix_tree_tag_set
> > c01b33b8 T radix_tree_tag_clear
> > c01b347a t __lookup
> > c01b3547 T radix_tree_gang_lookup
> > c01b3595 t __lookup_tag
> > c01b369d T radix_tree_gang_lookup_tag
> > c01b36f0 T radix_tree_delete
> > c01b3835 T radix_tree_tagged
> > c01b3863 t radix_tree_node_ctor
> > c01b3874 t __rb_rotate_left
> > c01b38aa t __rb_rotate_right
> > c01b38e0 T rb_insert_color
> > c01b39a1 t __rb_erase_color
> > c01b3b0c T rb_erase
> > c01b3bcd T rb_first
> > c01b3be5 T rb_last
> > c01b3bfd T rb_next
> > c01b3c26 T rb_prev
> > c01b3c4f T rb_replace_node
> > c01b3c94 T rwsem_wake
> > c01b3d89 T rwsem_downgrade_wake
> > c01b3e14 T strnicmp
> > c01b3e75 T strcpy
> > c01b3e89 T strncpy
> > c01b3ea8 T strlcpy
> > c01b3eeb T strcat
> > c01b3f0a T strncat
> > c01b3f35 T strlcat
> > c01b3f97 T strcmp
> > c01b3fae T strncmp
> > c01b3fe0 T strchr
> > c01b3ff0 T strrchr
> >
> > Cheers,
> >
> > Andrew Taylor
> >
> > Systems Engineer
> > Array Systems Computing Inc.
> > email: taylor@xxxxxxxx
> > http: www.array.ca
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at http://www.tux.org/lkml/
> Andrew Taylor <taylor@xxxxxxxx> wrote:
> >
> > Hello,
> >
> > We have been experiencing a very similar oops on a Dell poweredge 2600, running Fedora Core 3 (FC3). This host is mainly used as a samba fileserver. The problem has been seen on 2 different machines (both Dell-pe-2600s).
>
> Could you pleae confirm that 2.6.12-rc1 fixes this?
>
> Thanks.
>
> > The problem is only ever seen when running the SMP kernel in conjunction with samba.
> >
> > We have been experiencing problems since early January when the samba/nfs/io load was increased considerably. Since then a number of 2.4 kernels have been used:
> > 2.4.22.2115nptlsmp (FC1)
> > 2.4.22-1.2199smp (FC1)
> > 2.4.29 (vanilla kernel)
> > 2.6.10-1.741_FC3smp (FC3)
> > 2.6.10-1.766_FC3smp (FC3)
> >
> > All 2.4 kernels produced the same result: full system lock-up, with no messages on the console or the log files.
> >
> > The 2.6.10 kernel is giving an oops which has been seen 3 time in the last 3 days. A very similar one was seen pointing to eip: c01b2d98 on the 2.6.10-1.741 kernel.
> >
> > I have included the kallsyms and System.map relevant to the instruction pointed to by the EIP. I can include the full files if required.
> >
> > printing eip:
> > c01b387c
> > *pde = 2fcdd001
> > Oops: 0000 [#1]
> > SMP
> > Modules linked in: nfsd exportfs md5 ipv6 parport_pc lp parport autofs4 nfs lockd sunrpc video button battery ac uhci_hcd hw_random e1000 floppy sg dm_snapshot dm_zero dm_mirror ext3 jbd dm_mod mptscsih mptbase sd_mod scsi_mod
> > CPU: 1
> > EIP: 0060:[<c01b387c>] Not tainted VLI
> > EFLAGS: 00010203 (2.6.10-1.766_FC3smp)
> > EIP is at __rb_rotate_left+0x8/0x36
> > eax: f57be640 ebx: c041e5e4 ecx: f57be640 edx: 00000000
> > esi: f57be640 edi: f5f39c80 ebp: c041e5e4 esp: efa2eed0
> > ds: 007b es: 007b ss: 0068
> > Process smbd (pid: 8536, threadinfo=efa2e000 task=f7612a60)
> > Stack: f5ac1980 c01b3988 f5f39c80 f5f39c80 f5ac1988 00000267 c01965d7 f5ac1980
> > 0000000d efa2ef54 efa2ef61 ffffffea c019668b 00000015 00000000 00000267
> > c03191e0 efa2ef54 00000000 f599ed80 00000267 c01977df ffffffff 001f0000
> > Call Trace:
> > [<c01b3988>] rb_insert_color+0xa8/0xc1
> > [<c01965d7>] key_user_lookup+0xcf/0xfc
> > [<c019668b>] key_alloc+0x53/0x2b6
> > [<c01977df>] keyring_alloc+0x1a/0x48
> > [<c0198d43>] alloc_uid_keyring+0x2b/0x7c
> > [<c0125d2e>] alloc_uid+0xae/0x133
> > [<c01291d4>] set_user+0xb/0x8b
> > [<c01295f2>] sys_setresuid+0x11a/0x1b9
> > [<c0103c97>] syscall_call+0x7/0xb
> > Code: 82 04 01 00 00 00 75 ea 41 83 f9 01 76 ed 31 c0 5b c3 57 b9 45 00 00 00 89 c7 31 c0 f3 ab 5f c3 90 90 90 53 89 d3 8b 50 08 89 c1 <8b> 42 0c 85 c0 89 41 08 74 02 89 08 89 4a 0c 8b 01 85 c0 89 02
> >
> >
> > /proc/kallsyms grepped on c01b3.
> >
> > c01b3024 t newary
> > c01b313c T sys_semget
> > c01b325a t try_atomic_semop
> > c01b336b t update_queue
> > c01b33de t count_semncnt
> > c01b342c t count_semzcnt
> > c01b347a t freeary
> > c01b34f6 t copy_semid_to_user
> > c01b354b t semctl_nolock
> > c01b3731 t semctl_main
> > c01b3ad6 t semctl_down
> > c01b3c39 T sys_semctl
> > c01b3ce9 t lookup_undo
> > c01b3d19 t find_undo
> >
> >
> > root@msh2-c boot]# grep -i c01b3 /boot/System.map-2.6.10-1.766_FC3smp
> > c01b304b T match_int
> > c01b3052 T match_octal
> > c01b305c T match_hex
> > c01b3066 T match_strcpy
> > c01b3098 T match_strdup
> > c01b30c0 t radix_tree_node_alloc
> > c01b3109 T radix_tree_preload
> > c01b317d t radix_tree_extend
> > c01b321c T radix_tree_insert
> > c01b3303 T radix_tree_lookup
> > c01b3349 T radix_tree_tag_set
> > c01b33b8 T radix_tree_tag_clear
> > c01b347a t __lookup
> > c01b3547 T radix_tree_gang_lookup
> > c01b3595 t __lookup_tag
> > c01b369d T radix_tree_gang_lookup_tag
> > c01b36f0 T radix_tree_delete
> > c01b3835 T radix_tree_tagged
> > c01b3863 t radix_tree_node_ctor
> > c01b3874 t __rb_rotate_left
> > c01b38aa t __rb_rotate_right
> > c01b38e0 T rb_insert_color
> > c01b39a1 t __rb_erase_color
> > c01b3b0c T rb_erase
> > c01b3bcd T rb_first
> > c01b3be5 T rb_last
> > c01b3bfd T rb_next
> > c01b3c26 T rb_prev
> > c01b3c4f T rb_replace_node
> > c01b3c94 T rwsem_wake
> > c01b3d89 T rwsem_downgrade_wake
> > c01b3e14 T strnicmp
> > c01b3e75 T strcpy
> > c01b3e89 T strncpy
> > c01b3ea8 T strlcpy
> > c01b3eeb T strcat
> > c01b3f0a T strncat
> > c01b3f35 T strlcat
> > c01b3f97 T strcmp
> > c01b3fae T strncmp
> > c01b3fe0 T strchr
> > c01b3ff0 T strrchr
> >
> > Cheers,
> >
> > Andrew Taylor
> >
> > Systems Engineer
> > Array Systems Computing Inc.
> > email: taylor@xxxxxxxx
> > http: www.array.ca
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at http://www.tux.org/lkml/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/