SATA_SIL works with 2.6.7-bk8 seagate drive, but oops

[George Georgalis <george@galis.org>]

[-- Attachment #1: Type: text/plain, Size: 4165 bytes --]

On Thu, Jun 24, 2004 at 02:46:39PM -0400, Ricky Beam wrote:
>On Thu, 24 Jun 2004, George Georgalis wrote:
>...
>>has caused pdflush to block IO, any access to /mnt and the process
>>does not return. other than the pdflush load of ~99% the box seems to
>>function normally. 2.6.7-bk6, seagate drive
>
>-bk6 is not new enough.  bk7 has the necessary max_sectors fix.  You
>may need to add your drive model to the sil_blacklist in
>drivers/scsi/sata_sil.c.

Okay, 2.6.7-bk8 has written 8Gb to the sda4 with SATA_SIL and still
going strong! "dd if=/dev/zero of=/mnt/zero-`date +%s`"

However at about 3Gb (if that is relevant) top segfaulted with a
non critical oops. top will not restart, but the box is otherwise
functioning well considering the write load.

Is there any way to determine the drive model without first connecting
with the other sata driver (as hdc) and using hdparm?


Unable to handle kernel NULL pointer dereference at virtual address 000000b4
 printing eip:
c017c78a
*pde = 00000000
Oops: 0000 [#1]
PREEMPT 
CPU:    0
EIP:    0060:[<c017c78a>]    Not tainted
EFLAGS: 00010286   (2.6.7-sta-bk8) 
EIP is at pid_alive+0xa/0x30
eax: 000000b8   ebx: d32b0310   ecx: 00000000   edx: 00000000
esi: 00000000   edi: ef7bb7a0   ebp: d22b1b40   esp: db473e4c
ds: 007b   es: 007b   ss: 0068
Process top (pid: 489, threadinfo=db472000 task=e60ac7c0)
Stack: c017cca4 00000000 d22b1b40 db473f18 ef7bb7a0 db473ec4 c0159754 d22b1b40 
       db473f18 eaa1f006 eaa1f009 db473ec4 db473f18 c0159cc5 db473f18 db473ecc 
       db473ec4 ef7b86e0 d22b1dfc ee655240 bffff000 c0141ec8 c15cd660 c013e95c 
Call Trace:
 [<c017cca4>] pid_revalidate+0x14/0xc0
 [<c0159754>] do_lookup+0x44/0x80
 [<c0159cc5>] link_path_walk+0x535/0xa20
 [<c0141ec8>] find_extend_vma+0x18/0x70
 [<c013e95c>] follow_page+0x8c/0xb0
 [<c013ea3c>] get_user_pages+0xbc/0x3d0
 [<c015a406>] path_lookup+0x86/0x1a0
 [<c015a6a9>] __user_walk+0x39/0x70
 [<c0155a95>] vfs_stat+0x15/0x60
 [<c02445dd>] copy_to_user+0x2d/0x40
 [<c0156151>] sys_stat64+0x11/0x30
 [<c014dcbd>] __fput+0x8d/0xf0
 [<c014c6c3>] filp_close+0x43/0x70
 [<c014c744>] sys_close+0x54/0x80
 [<c0105dc7>] syscall_call+0x7/0xb




Could this be related to "Unknown HZ value! (91) Assume 100." which
started showing up with VIA motherboards on 2.5.x (I think) on top or ps
commands.  When I researched it before, It never caused ill, had been
identified as a "kernel bug" but benign. I know nothing more.

ATM, ps also seg faults, here is a corresponding oops,

 <1>Unable to handle kernel NULL pointer dereference at virtual address 000000b4
 printing eip:
c017c78a
*pde = 00000000
Oops: 0000 [#5]
PREEMPT 
CPU:    0
EIP:    0060:[<c017c78a>]    Not tainted
EFLAGS: 00010286   (2.6.7-sta-bk8) 
EIP is at pid_alive+0xa/0x30
eax: 000000b8   ebx: d32b0310   ecx: 00000000   edx: 00000000
esi: 00000000   edi: ef7bb7a0   ebp: d22b1b40   esp: ecc59e4c
ds: 007b   es: 007b   ss: 0068
Process ps (pid: 3456, threadinfo=ecc58000 task=e60ac7c0)
Stack: c017cca4 00000000 d22b1b40 ecc59f18 ef7bb7a0 ecc59ec4 c0159754 d22b1b40 
       ecc59f18 cf499006 cf499009 ecc59ec4 ecc59f18 c0159cc5 ecc59f18 ecc59ecc 
       ecc59ec4 ef7b86e0 d22b1dfc ee655240 bffff000 c0141ec8 c15cd660 c013e95c 
Call Trace:
 [<c017cca4>] pid_revalidate+0x14/0xc0
 [<c0159754>] do_lookup+0x44/0x80
 [<c0159cc5>] link_path_walk+0x535/0xa20
 [<c0141ec8>] find_extend_vma+0x18/0x70
 [<c013e95c>] follow_page+0x8c/0xb0
 [<c013ea3c>] get_user_pages+0xbc/0x3d0
 [<c015a406>] path_lookup+0x86/0x1a0
 [<c015a6a9>] __user_walk+0x39/0x70
 [<c0155a95>] vfs_stat+0x15/0x60
 [<c02445dd>] copy_to_user+0x2d/0x40
 [<c0156151>] sys_stat64+0x11/0x30
 [<c014dcbd>] __fput+0x8d/0xf0
 [<c014c6c3>] filp_close+0x43/0x70
 [<c014c744>] sys_close+0x54/0x80
 [<c0105dc7>] syscall_call+0x7/0xb
Code: 39 82 b4 00 00 00 75 07 8b 82 bc 00 00 00 c3 0f 0b 04 03 72 


config attached. I wrote 25G of zero and killed the dd process, top and
ps still segfault. Thanks all for your help!

// George



-- 
George Georgalis, Architect and administrator, Linux services. IXOYE
http://galis.org/george/  cell:646-331-2027  mailto:george@galis.org
Key fingerprint = 5415 2738 61CF 6AE1 E9A7  9EF0 0186 503B 9831 1631

[-- Attachment #2: 2.6.7-sta-bk8.config.gz --]
[-- Type: application/octet-stream, Size: 8062 bytes --]