https://bugzilla.novell.com/show_bug.cgi?id=332048#c1 --- Comment #1 from joerg reisenweber 2007-10-08 21:29:54 MST ---

uname -a Linux andromeda 2.6.22.5-31-default #1 SMP 2007/09/21 22:29:00 UTC i686 i686 i386 GNU/Linux

-- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=332048#c3 --- Comment #3 from joerg reisenweber 2007-10-16 02:12:32 MST --- (In reply to comment #2 from Jeff Mahoney)

The next time you reproduce this, could you provide the lines below blocking_notifier_call_chain? There has to be something actually calling it.

Sorry, i do not think i truncated the "copy" at bottom. Why should i do this? Anyway i'll reproduce and verify this, after sending hwinfo -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=332048#c5 --- Comment #5 from joerg reisenweber 2007-10-16 05:45:38 MST --- now i reproduced 2 times: first was so fast i still was on X console. Second i had to wait 1/2h and then got kernel: PANIC - not syncing: for safety as the *only* msg at console10, besides the blinking LEDs. now starting a third try... ~30min: --------------------------------------------------------------------------- [[[ some SFW2-INext-drop-dflt standard firewall trace log. Next first line: ]]] qinpos = 65, SCB index = 3 Kernel panic - not syncing: Loop 1 BUG: scheduling while atomic: beagled/ox10010000/4560 [<c02c362e>] __sched_text_start+0x5e/0x7c5 [<c011f2fc>] __cond_resched+0x18/0x29 [<c02c3dfb>] cond_resched+0x26/0x31 [<c0136c19>] down_read+0x08/0x17 [<c012eb07>] __blocking_notifier_call_chain+0x1q/0x47 [<c01ebc38>] bit_cursor+0x0/0x498 [<c012eb4b>] blocking_notifier_call_chain+0x17/0x1a [<c01eac73>] fbcon_blank+0x17/0x1e7 [<c022f2df>] lf+0x20/0x3d [<c0232fcb>] vt_console_print+0x0/0x226 [<c0123d45>] __call_console_drivers+0x4f/0x5b [<c012408e>] release_console_sem+0x186/0x19f [<c0123f06>] wake_up_klogd+0x33/0x35 [<c0158b7f>] fblockable_page_cache_readahead+0x4x/0x9f [<c012b904>] lock_timer_base+0x19/0x35 [<c012ba18>] __mod_timer+0x9a/0xa4 [<c0230b47>] do_unblank_screen+0xca/0x111 [<c01d17d0>] ust_spinlocks+0x14/0x24 [<c0123c0c>] panic+0x48/0xf8 [<dcf4cac1>] ahc_search_qinfifo+0x2b0/0x775 [aic7xxx] [<dcf4d98e>] ahc_freezedevq+0x7f/0x8f [aic7xxx] [<dcf4fbe5>] ahc_handle_seqint+0x1a1/0x23e0 [aic7xxx] [<c016fc81>] do_sync_read+0xc7/0x10a [<dcf0689c>] scsi_end_request+0xa1/0xab [scsi_mod] [<c01341b9>] autoremove_wake_function+0x0/0x33 [<dcf5aff0>] ahc_linux_isr+0x184/0x1b0 [aic7xxx] [<c014f2c7>] handle_IRQ_event+0x23/0x51 [<c01506c9>] handle_level_irq+0xb9/0xc8 [<c0107479>] do_IRQ+0x79/0x90 [<c0105e73>] common_interrupt+0x23/0x30 [<c02c0000>] unix_ioctl+0x75/0x95 ======================= Oct 16 12:38:40 andromeda kernel: qinpos = 65, SCB index = 3 Oct 16 12:38:40 andromeda kernel: Oct 16 12:38:40 andromeda kernel: BUG: scheduling while atomic: beagled/ox10010000/4560 Oct 16 12:38:40 andromeda kernel: [<c02c362e>] __sched_text_start+0x5e/0x7c5 Oct 16 12:38:40 andromeda kernel: [<c011f2fc>] __cond_resched+0x18/0x29 Oct 16 12:38:40 andromeda kernel: [<c02c3dfb>] cond_resched+0x26/0x31 Oct 16 12:38:40 andromeda kernel: [<c0136c19>] down_read+0x08/0x17 Oct 16 12:38:40 andromeda kernel: [<c012eb07>] __blocking_notifier_call_chain+0x1q/0x47 Oct 16 12:38:40 andromeda kernel: [<c01ebc38>] bit_cursor+0x0/0x498 Oct 16 12:38:40 andromeda kernel: [<c012eb4b>] blocking_notifier_call_chain+0x17/0x1a [[[ stalls here. Above is last line on bottom of screen]]] ------------------------------------------------------------------ obviously this is quite a different kernel panic log than the one i had before. Hope it will help to trace the BUG anyway. ps: (dunno if this is of any interest. 2 blocked tasks:) SysReq: Show Blocked State task syslog-ng call-trace: mod_timer .... *) smartd call-trace: scsi_request_fn blk_plug_device cfg_set_request wait_for_completion default_wake_funktion .... *)give me a means to copy&paste the screen, if you like more of this console info (i might use a serial console and minicom, if you tell me to do so, and how to do so). tired of copying... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=332048#c7 --- Comment #7 from joerg reisenweber 2007-10-16 07:00:57 MST --- (In reply to comment #3 from joerg reisenweber)

(In reply to comment #2 from Jeff Mahoney)

...

The next time you reproduce this, could you provide the lines below blocking_notifier_call_chain? There has to be something actually calling it.

Sorry, i do not think i truncated the "copy" at bottom. Why should i do this? Anyway i'll reproduce and verify this, after sending hwinfo

Please note: in 2. copy of panic-screen, the syslog lines _after_ kernel-bug-msg also stall at blocking_notifier_call_chain. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=332048#c9 Jeff Mahoney changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|reisenweber@web.de | --- Comment #9 from Jeff Mahoney 2007-10-16 12:04:53 MST --- Actually, nevermind. I see it in the output now: qinpos = 65, SCB index = 3 Kernel panic - not syncing: Loop 1 That's printed at ./drivers/scsi/aic7xxx/aic7xxx_core.c:5312: while (qinpos != qintail) { scb = ahc_lookup_scb(ahc, ahc->qinfifo[qinpos]); if (scb == NULL) { printf("qinpos = %d, SCB index = %d\n", qinpos, ahc->qinfifo[qinpos]); panic("Loop 1\n"); } -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=332048#c11 Hannes Reinecke changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED --- Comment #11 from Hannes Reinecke 2007-10-17 00:56:33 MST --- Ah-ha. Suspending aic7xxx. Tricky that. And unimplemented. We'll see what we can do. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=332048#c13 --- Comment #13 from joerg reisenweber 2007-10-17 01:20:26 MST --- (In reply to comment #12 from Hannes Reinecke)

Tricky _and_ complicated.

As far as _I_ am concerned, it's fine to simply block s2disk, if aic7xxx module is loaded, to avoid kernel freeze. I really accidentally hit the "sleep" button on keyboard. And i easily could kill the scsi-swap partition and load aic7xxx on demand (what is really rare).

From my point of view, AHA-29160 controller is a server hardware, and who would ever want to s2disk a server?!

-- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=332048#c15 --- Comment #15 from joerg reisenweber 2007-10-17 02:41:13 MST --- So feel free to contact me, if there is need to test any patched modules on my hw. Please binaries only, no src-patches (you see the severe lack of raw computing power in this hw, which makes kernel-building an annoying task). Btw(sorry a *little* OT): is there a way meanwhile to handle spindown (hdparm -S) for scsi-drives? Or at least to send a scsi-cmd to cause drive to spindown immediately? In ancient times (last millenium?) there was a way to do this via win98 tool, so it should be feasible. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=332048#c17 Jeff Mahoney changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|reisenweber@web.de | --- Comment #17 from Jeff Mahoney 2007-10-23 11:37:39 MST --- I set the other bug to NEEDINFO for this issue, setting this backed to assigned. Sorry for the noise. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

https://bugzilla.novell.com/show_bug.cgi?id=332048#c19 --- Comment #19 from Hannes Reinecke 2007-10-24 00:53:33 MST --- As it turned out it wasn't too hard; some bits and pieces were already in the driver. So with the above patch suspend/resume actually works for aic7xxx and aic79xx. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.