[Exim] Kernel Panics with exim4. Any clue?

Etusivu
Poista viesti
Vastaa
Lähettäjä: Sander Smeenk
Päiväys:  
Vastaanottaja: Exim Users
Aihe: [Exim] Kernel Panics with exim4. Any clue?
Hi,

I'm having trouble with my 2.5.53 kernel (and 2.5.59 too) and exim4.

Now I know it's an unstable untested kernel, and an unstable untested
exim4, so do not look at this message as a 'Complaint of It Doesn't Work',
but rather see it as informative and maybe helpful, please.

The exim4 binary came from a Debian package of exim4 which i'm helping
developing for Debian.

My guess is that on some occasions the daemon tends to trigger a
kernelpanic when attempting to bind to port 25. Sometimes it does hang
the kernel, sometimes it doesn't.

This is a message I sent to the LKM today. I hope it might help exim
upstream too? Please let me know if there is anything I can test / try
for you. I'm willing to "crash" my PC for you ;)

| From: Sander Smeenk <ssmeenk@???>
| To: linux-kernel@???
| Subject: KernelPanic with 2.5.53 & 59 in tcp_v4_get_port?

|
| Hi,

|
| I run exim 4.12 on my system, and while doing a restart of that service
| on a system running 2.5.53 at that moment...

|
| : Unable to handle kernel paging request at virtual address 5a5a5aa3
| :  printing eip:
| : c02df929
| : *pde = 00000000
| : Oops: 0000
| : CPU:    0
| : EIP:    0060:[tcp_v4_get_port+377/720]    Not tainted
| : EFLAGS: 00010246
| : EIP is at tcp_v4_get_port+0x179/0x2d0
| : eax: 5a5a5a5a   ebx: 00000019   ecx: deb6fb00   edx: dadba844
| : esi: df71b2cc   edi: 00000001   ebp: deb6fc4c   esp: db64feb0
| : ds: 0068   es: 0068   ss: 0068
| : Process exim4 (pid: 11613, threadinfo=db64e000 task=c525c180)
| : Stack: db64e000 deb6fb00 deb6fc4c ffffffea 00000000 c01ddd4b 00000001 dfc000c8
| :        c02ef3ee deb6fb00 00000019 c916d2f0 db64ff14 00000010 bffff1c8 00000003
| :        00190010 c02b55b4 c916d2f0 db64ff14 00000010 00000001 bffff190 00000010
| : Call Trace:
| :  [capable+27/64] capable+0x1b/0x40
| :  [inet_bind+398/720] inet_bind+0x18e/0x2d0
| :  [sys_bind+84/128] sys_bind+0x54/0x80
| :  [sock_setsockopt+48/1488] sock_setsockopt+0x30/0x5d0
| :  [inet_setsockopt+42/64] inet_setsockopt+0x2a/0x40
| :  [sys_setsockopt+97/128] sys_setsockopt+0x61/0x80
| :  [copy_from_user+51/64] copy_from_user+0x33/0x40
| :  [sys_socketcall+124/512] sys_socketcall+0x7c/0x200
| :  [syscall_call+7/11] syscall_call+0x7/0xb
| :
| : Code: f6 40 49 20 75 31 8b 4c 24 24 8b 42 04 39 41 04 75 25 85 ff
| :  <0>Kernel panic: Aiee, killing interrupt handler!
| : In interrupt handler - not syncing

|
| Does anyone have any idea why this happened? It looks like there's
| something in exim that binds to port 25 but also makes the kernel panic.

|
| But that's not all...

|
| I try to "save what's left to save" by doing MagicSysRQ to unmount my
| disks and that resulted in this huge loop-ish kernel-oops. I have pasted
| only one section of them, the complete log (1068 lines!) is at:
| http://www.hoho.nl/~ssmeenk/kernelpanic.txt

|
| : <6>SysRq : Emergency Remount R/O
| : Remounting device ide0(3,2) ... <3>bad: scheduling while atomic!
| : Call Trace:
| : [schedule+61/704] schedule+0x3d/0x2c0
| : [sleep_on+91/144] sleep_on+0x5b/0x90
| : [default_wake_function+0/64] default_wake_function+0x0/0x40
| : [log_wait_commit+193/272] log_wait_commit+0xc1/0x110
| : [ext3_sync_fs+76/144] ext3_sync_fs+0x4c/0x90
| : [fsync_super+81/128] fsync_super+0x51/0x80
| : [fsync_bdev+27/64] fsync_bdev+0x1b/0x40
| : [go_sync+204/352] go_sync+0xcc/0x160
| : [do_emergency_sync+94/256] do_emergency_sync+0x5e/0x100
| : [panic+222/224] panic+0xde/0xe0
| : [do_exit+32/928] do_exit+0x20/0x3a0
| : [die+119/128] die+0x77/0x80
| : [do_page_fault+743/1044] do_page_fault+0x2e7/0x414
| : [do_page_fault+0/1044] do_page_fault+0x0/0x414
| : [do_generic_mapping_read+854/880] do_generic_mapping_read+0x356/0x370
| : [__generic_file_aio_read+445/480] __generic_file_aio_read+0x1bd/0x1e0
| : [error_code+45/56] error_code+0x2d/0x38
| : [tcp_v4_get_port+377/720] tcp_v4_get_port+0x179/0x2d0
| : [capable+27/64] capable+0x1b/0x40
| : [inet_bind+398/720] inet_bind+0x18e/0x2d0
| : [sys_bind+84/128] sys_bind+0x54/0x80
| : [sock_setsockopt+48/1488] sock_setsockopt+0x30/0x5d0
| : [inet_setsockopt+42/64] inet_setsockopt+0x2a/0x40
| : [sys_setsockopt+97/128] sys_setsockopt+0x61/0x80
| : [copy_from_user+51/64] copy_from_user+0x33/0x40
| : [sys_socketcall+124/512] sys_socketcall+0x7c/0x200
| : [syscall_call+7/11] syscall_call+0x7/0xb

|
| < .. and more .. >

|
| Nothing to do now but reboot Ctrl-Alt-SysRQ-B ...

|
| I hope this helps and that someone will be able to spot a possible
| cause? I only experience this behaviour with exim4 ...

|
| i'm running lvm2, with ext3, i'm telling you because I see stuff about
| device mapper, ext3 and syncing.

|
| With regards,
| Sander.


--
| Waarvoor zijn de witte halve maantjes op je nagels ?
| 1024D/08CEC94D - 34B3 3314 B146 E13C 70C8 9BDB D463 7E41 08CE C94D