CONFIG_X86_PKG_TEMP_THERMAL causes #GP fault on Core i7-740QM breaking boot

CONFIG_X86_PKG_TEMP_THERMAL causes #GP fault on Core i7-740QM breaking boot

[Mikael Pettersson]

3.11-rc1 and -rc2 refuse to boot on my Dell Latitude E6510 
(Intel Core i7-740QM processor) due to __rdmsr_on_cpu throwing
a #GP fault.

Being a laptop it doesn't have a good way to log early boot
messages, so the following was typed in by hand:

serio...
mousedev...
rtc_cmos...
general protection fault: 0000 [#1] SMP
Modules linked in:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.11.0-rc2 #1
Hardware name: Dell Inc. Latitude E6510/0N5KHN, BIOS A09 05/26/2011
task: ffffffff814d1440 ti: ffffffff814c0000 task.ti: ffffffff814c0000
RIP: 0010:[<ffffffff81206f05>]  [<ffffffff81206f05>] __rdmsr_on_cpu+0x25/0x40
RSP: 0000:ffff88012fc03f70  EFLAGS: 00010046
RAX: 0000000000000000 RBX: ffff88012b083c80 RCX: 00000000000001b1
RDX: 0000000000000000 RSI: ffff88012b083ce0 RDI: ffff88012b083cd8
RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: ffff88012fc03f78
R13: ffffffff814c1fd8 R14: 0000000000000000 R15: ffffffff814c1fd8
FS:  0000000000000000(0000) GS:ffff88012fc00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: ffff88012ffff000 CR3: 00000000014cc000 CR4: 00000000000007f0
Stack:
 ffffffff8108b5fd ffff88012fc03f78 ffff88012fc03f78 ffffffff814c1fd8
 0000000000000000 ffffffff814c1fd8 ffffffff81027272 0000000000000000
 ffffffff81382677 ffffffff814c1ea8 <EOI>  0000000000000000 0000000000000202
Call Trace:
 <IRQ>
 [<ffffffff8108b5fd>] ? generic_sml_call_function_single_interrupt+0xad/0x120
 [<ffffffff81027272>] ? smp_call_function_single_interrupt+0x22/0x40
 [<ffffffff81382677>] ? call_function_single_interrupt+0x67/0x70
 <EOI>
 [<ffffffff8100bc10>] ? idle_notifier_register+0x10/0x10
 [<ffffffff8100bc25>] ? default_idle+0x15/0xb0
 [<ffffffff8107c69d>] ? cpu_startup_entry+0x7d/0x200
 [<ffffffff8153bd80>] ? start_kernel+0x346/0x351
 [<ffffffff8153b851>] ? repair_env_strings+0x59/0x59
Code: 90 90 90 90 90 90 48 8b 47 10 48 8d 77 08 65 8b 14 25 1c b0 00 00 48 85 c0
 74 0e 48 63 d2 48 89 c6 48 03 34 d5 a0 11 52 81 8b 0f <0f> 32 48 c1 e2 20 89 c0
 48 09 c2 89 16 48 c1 ea 20 89 56 04 c3
RIP  [<ffffffff81206f05>] __rdmsr_on_cpu+0x25/0x40
 RSP <ffff88012fc03f70>
---[ end trace ... ]---
Kernel panic - not syncing: Fatal exception in interrupt

Looks like it's trying to read MSR 0x1B1, IA32_PACKAGE_THERM_STATUS,
and my i7-740QM doesn't like that.

Disabling CONFIG_X86_PKG_TEMP_THERMAL allows the kernel to boot again.

/Mikael

Re: CONFIG_X86_PKG_TEMP_THERMAL causes #GP fault on Core i7-740QM breaking boot

[Ortwin Glück]

Mikael Pettersson <mikpe@it.uu.se> >I can see that patch in git, it's NOT present in
>either
>the linux-3.11-rc2.tar.xz or the patch-3.11-rc2.xz files.  Which is
>strange
>since the patch was committed 8 days ago, and -rc2 was released 2 days
>ago.

It was merged only after rc2. Only in the tree that it was merged from it was commited before rc2. See 
gitk v3.11-rc2..

Ortwin

Re: CONFIG_X86_PKG_TEMP_THERMAL causes #GP fault on Core i7-740QM breaking boot

[Mikael Pettersson]

Ortwin Glück writes:
 > Hi,
 > 
 > I think the bug is already fixed in this commit:
 > 
 > f3ed0a17f0292300b3caca32d823ecd32554a667
 > 
 > Thermal: x86 package temp thermal crash

Thanks.  Although I can see that patch in git, it's NOT present in either
the linux-3.11-rc2.tar.xz or the patch-3.11-rc2.xz files.  Which is strange
since the patch was committed 8 days ago, and -rc2 was released 2 days ago.

I'm assuming there was a mistake when preparing this -rc2 snapshot, but if
the -rc3 snapshot also differs from git I'll alert Linus.

/Mikael

Re: CONFIG_X86_PKG_TEMP_THERMAL causes #GP fault on Core i7-740QM breaking boot

[Ortwin Glück]

Hi,

I think the bug is already fixed in this commit:

f3ed0a17f0292300b3caca32d823ecd32554a667

Thermal: x86 package temp thermal crash



Thanks,

Ortwin