3.14.19: Oops on boot in rapl_cpu_prepare()

3.14.19: Oops on boot in rapl_cpu_prepare()

[Thomas Jarosch]

Hi Artem,

I just upgraded from kernel 3.4.101 to 3.14.19. 
Now the system fails to boot using qemu (kernel Oops):

------------------------------------------------------
...
UDP hash table entries: 512 (order: 2, 16384 bytes)
UDP-Lite hash table entries: 512 (order: 2, 16384 bytes)
NET: Registered protocol family 1             
pci 0000:00:00.0: Limiting direct PCI/PCI transfers   
pci 0000:00:01.0: PIIX3: Enabling Passive Release     
pci 0000:00:01.0: Activating ISA DMA hang workarounds         
Trying to unpack rootfs image as initramfs...
Freeing initrd memory: 16688K (f67b2000 - f77fe000)
general protection fault: 0000 [#1] SMP
Modules linked in:
CPU: 0 PID: 1 Comm: swapper/0 Not tainted 3.14.19-1.i2n.i686 #1
Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
task: f58779a0 ti: f5878000 task.ti: f5878000  
EIP: 0060:[<c101f9b2>] EFLAGS: 00000282 CPU: 0
EIP is at rapl_cpu_prepare+0x62/0xf0   
EAX: 00000606 EBX: c14ef3e4 ECX: 00000606 EDX: 00000000
ESI: 00000000 EDI: f5918280 EBP: f5879ec4 ESP: f5879eb4
    DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
CR0: 80050033 CR2: ffe16000 CR3: 014fb000 CR4: 000406d0
Stack:                            
    00000006 c1491784 00000000 00000000 f5879efc c149e670 c14ee5c0 c146a000
    00000008 f5879ee0 c103b45d f5879eec c132100e c1491784 f5879efc 00000006
    0000009e 00000000 f5879f70 c1000366 00000000 f58ab300 f58ab080 0000000f
Call Trace:
    [<c149e670>] rapl_pmu_init+0x88/0x1af
    [<c103b45d>] ? cpu_maps_update_done+0xd/0x10
    [<c132100e>] ? register_cpu_notifier+0x1e/0x30    
    [<c1000366>] do_one_initcall+0x36/0x150
    [<c149e5e8>] ? intel_uncore_init+0x332/0x332
    [<c1495600>] ? kernel_init_freeable+0x143/0x18c
    [<c1051f23>] ? parse_args+0x1a3/0x330
    [<c106aba0>] ? __wake_up+0x40/0x50
    [<c14955a4>] kernel_init_freeable+0xe7/0x18c
    [<c1495649>] ? kernel_init_freeable+0x18c/0x18c
    [<c132087b>] kernel_init+0xb/0xe0
    [<c13279b7>] ret_from_kernel_thread+0x1b/0x28          
    [<c1320870>] ? rest_init+0x60/0x60
Code: c1 ba d0 80 00 00 e8 8e 5b 0b 00 89 c7 b8 ff ff ff ff 85 ff 74 d9 8d 47 0c
    89 47 0c 89 47 10 b8 06 06 00 00 66 c7 07 00 00 89 c1 <0f> 32 c1 f8 08 66 b9 1f
    00 83 e0 1f 31 d2 29 c1 89 47 04 b8 05
EIP: [<c101f9b2>] rapl_cpu_prepare+0x62/0xf0 SS:ESP 0068:f5879eb4
---[ end trace 52657a58a39d7b04 ]---
------------------------------------------------------

The kernel is 32bit with SMP + PAE, qemu is invoked like this:

/usr/bin/qemu-kvm \
    -name 'vm1' \
    -M pc  \
    -nodefaults  \
    -vga std  \
    -display curses \
    -curses \
    -drive id=drive_image1,if=none,cache=none,file=/mnt/local/vm1/vm1.qcow2 \
    -device virtio-blk-pci,id=image1,drive=drive_image1,bus=pci.0,addr=03 \
    -m 1024  \
    -smp 2,cores=1,threads=1,sockets=2  \
    -cpu 'SandyBridge' \
    -drive media=cdrom,file=/mnt/local/isos/latest.iso,bus=0,unit=0  \
    -rtc base=utc,clock=host,driftfix=none  \
    -boot order=cdn,once=dcn,menu=off \
    -enable-kvm


qemu version is 1.6.2 from Fedora 20.

I tried reverting commit 825600c0f20e595daaa7a6dd8970f84fa2a2ee57
but the issue remained. So it doesn't seem to be related to your patch :)

A workaround is to change the guest CPU type
from 'SandyBridge' to 'qemu32'.

Any idea what could be going on?
I can run debug output patches if needed.


The problem is similar to this previous report from 2014-07-31:
https://groups.google.com/forum/#!topic/fa.linux.kernel/sizx3cmxj0k
"RE: [x86] BUG: unable to handle kernel paging request at ffff880012770000"

Cheers,
Thomas

Re: 3.14.19: Oops on boot in rapl_cpu_prepare()

[Thomas Jarosch]

Hi again,

On Wednesday, 24. September 2014 17:09:01 Thomas Jarosch wrote:
> ...
> general protection fault: 0000 [#1] SMP
> Modules linked in:
> CPU: 0 PID: 1 Comm: swapper/0 Not tainted 3.14.19-1.i2n.i686 #1
> Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
> task: f58779a0 ti: f5878000 task.ti: f5878000
> EIP: 0060:[<c101f9b2>] EFLAGS: 00000282 CPU: 0
> EIP is at rapl_cpu_prepare+0x62/0xf0
> EAX: 00000606 EBX: c14ef3e4 ECX: 00000606 EDX: 00000000
> ESI: 00000000 EDI: f5918280 EBP: f5879ec4 ESP: f5879eb4
>     DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
> CR0: 80050033 CR2: ffe16000 CR3: 014fb000 CR4: 000406d0
> Stack:
>     00000006 c1491784 00000000 00000000 f5879efc c149e670 c14ee5c0
> c146a000 00000008 f5879ee0 c103b45d f5879eec c132100e c1491784 f5879efc
> 00000006 0000009e 00000000 f5879f70 c1000366 00000000 f58ab300 f58ab080
> 0000000f Call Trace:
>     [<c149e670>] rapl_pmu_init+0x88/0x1af
>     [<c103b45d>] ? cpu_maps_update_done+0xd/0x10
>     [<c132100e>] ? register_cpu_notifier+0x1e/0x30
>     [<c1000366>] do_one_initcall+0x36/0x150
>     [<c149e5e8>] ? intel_uncore_init+0x332/0x332
>     [<c1495600>] ? kernel_init_freeable+0x143/0x18c
>     [<c1051f23>] ? parse_args+0x1a3/0x330
>     [<c106aba0>] ? __wake_up+0x40/0x50
>     [<c14955a4>] kernel_init_freeable+0xe7/0x18c
>     [<c1495649>] ? kernel_init_freeable+0x18c/0x18c
>     [<c132087b>] kernel_init+0xb/0xe0
>     [<c13279b7>] ret_from_kernel_thread+0x1b/0x28
>     [<c1320870>] ? rest_init+0x60/0x60
> ...

the issue is fixed by upgrading to kernel 3.14.20.

Probably it's this patch:

----------------------------------------
commit 2968314094d8db0af32de20ad51349a30ea54a01
Author: Venkatesh Srinivas <venkateshs@google.com>
Date:   Thu Mar 13 12:36:26 2014 -0700

    perf/x86/intel: Use rdmsrl_safe() when initializing RAPL PMU
    
    commit 24223657806a0ebd0ae5c9caaf7b021091889cf2 upstream.
    
    CPUs which should support the RAPL counters according to
    Family/Model/Stepping may still issue #GP when attempting to access
    the RAPL MSRs. This may happen when Linux is running under KVM and
    we are passing-through host F/M/S data, for example. Use rdmsrl_safe
    to first access the RAPL_POWER_UNIT MSR; if this fails, do not
    attempt to use this PMU.
----------------------------------------

Thanks for the backport to 3.14.

Cheers,
Thomas