FW: SUMMARY: panic (cpu 5): Unaligned kernel space access from k ernel mode
> Thanks to :
>
> Dr. Thomas P. Blinn
> Alan Davis
> alan_at_nabeth.cxo.dec.com
> Jason Orendorf
> Kevin McDonnell
> Pat O'Brien
> Regis.Carlier_at_univ-valenciennes.fr
> Bryan Whiting
> Daniel L. Willis
>
> The fix for this problem was a patch from Compaq. The patch fixes a
> problem with assert wait in xpt_ccb_alloc where timeouts are not canceled
> on I/O requests causing the I/O block to be freed inappropriately.
>
>
> The readme for the patch is :
> ----------------------------------------------------------
> NEW PATCHID: 314.00
> PATCH ID: OSF440-281
> REQUIRED PATCHES: NONE
> CONDITIONALLY REQUIRED PATCHES: NONE
> SUPERSEDED PATCHES:
> OSF440-025 (136.00),
> OSF440-247 (283.00)
> SPECIAL INSTRUCTIONS: NONE
> FULL DESCRIPTION:
> PROBLEM: (CLD MGO104106, CLD HPAQ40QCM) (Patch ID: OSF440-025)
> ********
> Typical stack trace for the assert panic is:
> panic
> assert_wait_mesg
> xpt_ccb_alloc
> htm_immed_async
> isp_process_response_queue
>
> For the blocked thread, there will be at least one thread waiting forever
> with a stack trace of:
>
> thread_block
> mpsleep
> isp_enable_lun
> htm_enable_lun
> handle_immediate
> sim_action
> xpt_sim_func
> xpt_action
> tmv2_complete_disable
> tmv2_accept_cbf
> xpt_callback_thread
>
> PROBLEM: (CLD GOZ72114A, CLD BCGMA10Z6, CLD BCGMA0X3C, CLD BCGMA1GFC,
> CLD BCGMB06HF, CLD DSATB0SF1, CLD DSATB0X2F, CLD HPAQB0X3N,
> CLD EVT75135A) (Patch ID:OSF440-247)
> ********
>
> This patch fixes a problem on configurations having multiple Qlogic
> 1020/1040
> based SCSI controllers (such as KZPBA) and multiple CPUs. The symptom
> of the problem is a performance degradation up to having
> ss_perform_timeouts
> being logged against the KZPBA scsi devices. In some extreme cases, the
> symptoms can be so severe that it makes make it appear as if the system
> was
> hung. Tape units have had a number of problems with vdump, tar, and nsrd
> related to the command timeouts.
>
> PROBLEM: (BCGM90Z90) (Patch ID: OSF440-281)
> ********
>
> This patch fixes a problem where timeouts are not canceled on I/O requests
> causing the I/O block to be freed inappropriately.
>
>
> FILES:
> ./sys/BINARY/isp.mod
> CHECKSUM: 55763 286
> SUBSET: OSFHWBIN440
> ./kernel/io/cam/qlogic/isp1020.c,v RCS ID: 1.1.283.4
> ./usr/sys/include/io/cam/qlogic/isp1020.h
> CHECKSUM: 64702 43
> SUBSET: OSFBINCOM440
> ./kernel/io/cam/qlogic/isp1020.h,v RCS ID: 1.1.139.1
> SUPPORT NOTES: NONE
>
>
>
>
> Original Posting:
>
> Hi,
>
> My 8400 system crashed the other day. When I arrived at the console, it
> was at the P00 prompt. This is the output in the system log. I was
> wondering if any of you have seen it before and what it means.
>
>
> Jul 6 19:12:01 data-dmt1 vmunix: Unaligned kernel access
> va=0xbc0130dee0000097 pc=0xfffffc00002bd958 ra=0xfffffc00004b40cc
> inst=0xffffffff
> Jul 6 19:12:02 data-dmt1 vmunix: panic (cpu 5): Unaligned kernel space
> access from kernel mode
> Jul 6 19:12:02 data-dmt1 vmunix:
> Jul 6 19:12:02 data-dmt1 vmunix: LSM attempting to dump to SCSI device
> unit number rz0
> Jul 6 19:12:02 data-dmt1 vmunix:
> Jul 6 19:12:02 data-dmt1 vmunix: DUMP.prom: dev SCSI 0 0 0 0 0 0 0, block
> 533008
> Jul 6 19:12:02 data-dmt1 vmunix: DUMP.prom: dev SCSI 0 0 0 0 0 0 0, block
> 533008
>
>
> Thank you in advance,
> Debra Williams
>
> Debra.L.Williams_at_lmco.com
> DynCorp
> Systems Engineer
> GTN project
> Manassas, Virginia
>
>
>
Received on Tue Jul 10 2001 - 16:44:02 NZST
This archive was generated by hypermail 2.4.0
: Wed Nov 08 2023 - 11:53:42 NZDT