Bug 2941 - Boot crash w/ CONFIG_SCHED_SMT - Intel P865PERL
Summary: Boot crash w/ CONFIG_SCHED_SMT - Intel P865PERL
Status: CLOSED CODE_FIX
Alias: None
Product: ACPI
Classification: Unclassified
Component: BIOS (show other bugs)
Hardware: i386 Linux
: P2 high
Assignee: Len Brown
URL:
Keywords:
: 1908 2785 2963 3040 3102 (view as bug list)
Depends on:
Blocks:
 
Reported: 2004-06-23 08:19 UTC by Alexander Weber
Modified: 2004-11-03 17:51 UTC (History)
6 users (show)

See Also:
Kernel Version: 2.6.5-mm1, 2.6.7, 2.6.7-mm1
Subsystem:
Regression: ---
Bisected commit-id:


Attachments
.config from vanilla kernel version 2.6.7 (31.58 KB, text/plain)
2004-06-23 08:20 UTC, Alexander Weber
Details
Diff between 2.6.6 (which workd) and 2.6.7 (not working) configs (19.85 KB, text/plain)
2004-07-15 05:19 UTC, Alexander Weber
Details
2.4.27 early-init patch (4.34 KB, patch)
2004-08-13 21:54 UTC, Len Brown
Details | Diff
2.6.7 early-init patch (5.54 KB, patch)
2004-08-13 22:02 UTC, Len Brown
Details | Diff

Description Alexander Weber 2004-06-23 08:19:14 UTC
Distribution: Gentoo
Hardware Environment: 

MB: Intel P865 PERL - Bios: RL86510A.86A.P17 (latest version)

CPU:
  vendor_id       : GenuineIntel
  cpu family      : 15
  model           : 2
  model name      : Intel(R) Pentium(R) 4 CPU 3.00GHz
  stepping        : 9
  flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca
cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe cid

RAM: 2 GB Infinion DDR400

Software Environment: Kernel version 2.6.5 or above...

Problem Description: Kernel oops when trying to start ACPI - most of the time,
different messages, mainly (from a 2.6.7-mm1 kernel): 

"Unable to handle kernel NULL pointer dereference at virtual address XXXXXXX
printing eip: YYYYYYYY

with XXXXXXXX usally 00000000 or 00000002, 0000000c, ...
 and YYYYYYYY seen with f7fa0578 (XXXXXXXX was 00000002 with this), or 00000000

ACPI Subsytem Revision 20040326

One time seen:

__wake_up_common+0x35/0x55
__wake_up_coded+0x1f/0x23
__down_trylock+0x78/0x8e
__down_failed_trylock+0x7/0xc
.text.lock.osl+0x5/0x32
acpi_ut_acquire_mutex+0x5d/0x70
acpi_get_register+0x32/0x72
acpi_hw_get_mode+0x24/0x3c
acpi_hw_set_mode+0xca/0x107
acpi_enable+0x25/0x4e
acpi_enable_subsystem+0x27/0x45
acpi_bus_init+0xa2/0x11d
acpi_init+0x5e/0xd8
do_initcalls+0x28/0xae
init+0x92/0x1d9
init+0x0/0x1d9
kernel_thread_helper+0x5/0xb

code: Bad EIP value
<0> Kernel panic: Attempted to kill init

Errors seen with kernel 2.6.5-mm6:

ACPI-0183: *** ERROR: Hardware never changed modes
ACPI-0089: *** ERROR: Could not transition to APCI mode

No ACPI aswell with vanilla 2.6.7.

no Problems with kernel 2.6.4-mm1!

Steps to reproduce: Compile kernel with version => 2.6.5 and try to boot with acpi.

/proc/acpi/info from a running 2.4.25-gentoo-r2 kernel:

version:                 20040116
states:                  S0 S1 S3 S4 S5

Any help appreciated...

regards,
 Alex :)
Comment 1 Alexander Weber 2004-06-23 08:20:03 UTC
Created attachment 3237 [details]
.config from vanilla kernel version 2.6.7
Comment 2 Alexander Weber 2004-07-09 05:45:05 UTC
after some other test, i found a config that is working on the current knoppix
3.5 dvd.
Comment 3 Len Brown 2004-07-14 13:40:02 UTC
actually I think this bug is valid.
Looks exactly like the CONFIG_SCHED_SMT failure reported here:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=125841
are you sure this happened as far back as 2.6.5?
I thought it started in 2.6.7

can you diff the working and failing .config?
Comment 4 Alexander Weber 2004-07-15 05:19:19 UTC
Created attachment 3368 [details]
Diff between 2.6.6 (which workd) and 2.6.7 (not working) configs

I attached a diff between a working kernel version vanilla 2.6.6 (based on a
generic knoppix config, so there are a lot of modules used) and the nonworking
vanilla 2.6.7 (home brew).

As i've seen in the thread mentioned above, it think the CONFIG_SCHED_SMT is
the problem in the 2.6.7 kernel.

I hope i can test the 2.6.7 with the patch in the next days...
Comment 5 Alexander Weber 2004-07-19 00:05:31 UTC
The patch from the other page
(https://bugzilla.redhat.com/bugzilla/attachment.cgi?id=101570&action=view) did
the job for me. It work's :-)
Comment 6 Adrian Ulrich 2004-07-31 01:06:05 UTC
The patch also helps to solve my problem
(see Bug 2963)
Comment 7 Len Brown 2004-08-13 21:50:16 UTC
*** Bug 2963 has been marked as a duplicate of this bug. ***
Comment 8 Len Brown 2004-08-13 21:51:11 UTC
*** Bug 3040 has been marked as a duplicate of this bug. ***
Comment 9 Len Brown 2004-08-13 21:52:07 UTC
*** Bug 3102 has been marked as a duplicate of this bug. ***
Comment 10 Len Brown 2004-08-13 21:54:41 UTC
Created attachment 3507 [details]
2.4.27 early-init patch
Comment 11 Len Brown 2004-08-13 22:02:53 UTC
Created attachment 3508 [details]
2.6.7 early-init patch
Comment 12 Len Brown 2004-08-13 22:17:45 UTC
please test the 2.4 and 2.6 early init patches above. 
Comment 13 Len Brown 2004-08-24 19:52:38 UTC
*** Bug 1908 has been marked as a duplicate of this bug. ***
Comment 14 Bryan Hundven 2004-09-03 14:59:17 UTC
*** Bug 2785 has been marked as a duplicate of this bug. ***
Comment 15 Len Brown 2004-11-03 17:51:43 UTC
shipped in 2.6.9
also applied to 2.4.28
closing.

Note You need to log in before you can comment on or make changes to this bug.