IDE + SMP lockups, further findings, 2.2.13pre14

thx@rivalnet.de
Mon, 4 Oct 1999 19:34:15 +0200


Unfortunately, 2 hours after posting the IDE+SMP success message, the kernel exploded.

It gave me the same oops others have noticed (kernel NULL pointer dereference, see below).

I then also tested the original 2.2.13pre14, without unified IDE patch.

my findings until now:

with unified IDE patch (for 13pre12, 990925):
- kernel locks immediately if two controllers on the same irq are accessed (promise controller)
- kernel locks after a few minutes if the two controllers are not on the same irq
- with Tom´s patch: the kernel survived many hours stresstesting (simultaneous access to five drives on three controllers), but then oopsed with kernel NULL pointer dereference

without unified IDE patch (I need to cheat the kernel into believing the promise66 board is a promise33):
- exploded after 30 mins
- with Tom´s patch: exploded after 2.5h

ongoing/further tests:
- non SMP kernel (to ensure that the problems only occur with SMP)
- SMP with Andre´s newest unified IDE patch (2.2.12 or 13, depending on the patch)

the kernel was 2.2.13pre14 with
- newest raid patch (for 2.2.11)
- experimental large disk patch (for 2.3.X, needed for the IBM 37GB drives)
- with and without unified IDE patch
- with and without Tom´s patch
- tasks.h: increased to 4000 tasks

the oops (seems to appear with unified IDE patch only):

Oct 3 03:55:37 s_all@grs1 kernel: Unable to handle kernel NULL pointer dereference at virtual address 00000004
Oct 3 03:55:37 s_all@grs1 kernel: current->tss.cr3 = 00101000, %cr3 = 00101000
Oct 3 03:55:37 s_all@grs1 kernel: *pde = 00000000
Oct 3 03:55:37 s_all@grs1 kernel: Oops: 0000
Oct 3 03:55:37 s_all@grs1 kernel: CPU: 0
Oct 3 03:55:37 s_all@grs1 kernel: EIP: 0010:[ide_do_request+814/1416]
Oct 3 03:55:37 s_all@grs1 kernel: EFLAGS: 00010282
Oct 3 03:55:37 s_all@grs1 kernel: eax: 00000000 ebx: 00001260 ecx: ffffd000 edx: c0290000
Oct 3 03:55:37 s_all@grs1 kernel: esi: c02d297c edi: c02d2940 ebp: 00000000 esp: dabbbe5c
Oct 3 03:55:37 s_all@grs1 kernel: ds: 0018 es: 0018 ss: 0018
Oct 3 03:55:37 s_all@grs1 kernel: Process raid5d (pid: 245, process nr: 48, stackpage=dabbb000)
Oct 3 03:55:37 s_all@grs1 kernel: Stack: c02ce1c0 c02d297c 00016b10 0000000c c01b50c0 c02d297c 00000008 c02d2940
Oct 3 03:55:37 s_all@grs1 kernel: 00000000 00000046 c02d2940 c01a0a79 effda8e0 dabbbe9c 00000000 c02ce1c0
Oct 3 03:55:37 s_all@grs1 kernel: 00000282 c01a0b40 effda8e0 c019e7b1 c02ba464 00000038 da77d340 dabbbf10
Oct 3 03:55:37 s_all@grs1 kernel: Call Trace: [do_rw_disk+0/696] [do_hwgroup_request+73/92] [do_ide4_request+20/40] [add_request+657/680] [make_request+1487/1520] [handle_stripe+1945/3268] [handle_IRQ_event+90/140]
Oct 3 03:55:37 s_all@grs1 kernel: [raid5d+55/268] [raid5d+183/268] [md_thread+216/428] [kernel_thread+35/48]
Oct 3 03:55:37 s_all@grs1 kernel: Code: 0f b6 68 04 89 6c 24 18 c1 6c 24 18 06 8b 56 48 89 54 24 24

--
the online community service for gamers & friends -  http://www.rivalnet.com
* unterstützt über 50 PC-Spiele im Multiplayer-Modus
* Dateien senden & empfangen bis 500 MB am Stück
* Newsgroups, Mail, Chat & mehr

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/