Re: Seabios incompatible with Linux 2.6.26 host?

[Jan Kiszka <jan.kiszka@siemens.com>]

Pierre Riteau wrote:
> On 5 févr. 2010, at 16:07, Jan Kiszka wrote:
> 
>> Pierre Riteau wrote:
>>> On 5 févr. 2010, at 15:46, Jan Kiszka wrote:
>>>
>>>> Pierre Riteau wrote:
>>>>> On 4 févr. 2010, at 15:34, Pierre Riteau wrote:
>>>>>
>>>>>> Hello,
>>>>>> I'm having trouble running the latest qemu-kvm code on Debian Lenny (Linux 2.6.26).
>>>>>> qemu-kvm dies with an error like this one:
>>>>>> exception 13 (0)
>>>>>> rax 0000000000000010 rbx 0000000000008c00 rcx 0000000000006ebe rdx 00000000000c8c00
>>>>>> rsi 00000000e2010000 rdi 00000000000c0000 rsp 0000000000006eb4 rbp 00000000e2010000
>>>>>> r8  0000000000000000 r9  0000000000000000 r10 0000000000000000 r11 0000000000000000
>>>>>> r12 0000000000000000 r13 0000000000000000 r14 0000000000000000 r15 0000000000000000
>>>>>> rip 00000000000fdeb0 rflags 00033002
>>>>>> cs 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0)
>>>>>> ds 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0)
>>>>>> es 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0)
>>>>>> ss 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0)
>>>>>> fs 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0)
>>>>>> gs 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0)
>>>>>> tr 0000 (feffd000/00002088 p 1 dpl 0 db 0 s 0 type b l 0 g 0 avl 0)
>>>>>> ldt 0000 (00000000/0000ffff p 1 dpl 0 db 0 s 0 type 2 l 0 g 0 avl 0)
>>>>>> gdt f7a20/37
>>>>>> idt f8aa0/0
>>>>>> cr0 10 cr2 0 cr3 0 cr4 0 cr8 0 efer 0
>>>>>> code: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 --> 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
>>>>>>
>>>>>> I think I traced back the issue to the switch from Bochs BIOS to Seabios. By forcing the usage of Bochs BIOS 5f08bb45861f54be478b25075b90d2406a0f8bb3 works, while it dies without the -bios override.
>>>>>> Unfortunately, newer versions don't seem to work with Bochs BIOS.
>>>>>>
>>>>>> Upgrading the host kernel to 2.6.32 (Debian Squeeze) solves the issue. No problem on Fedora 12 as well.
>>>>> Am I the only one to see this issue?
>>>>>
>>>> The oldest kernels I've running here are 2.6.27. They all run kvm-kmod,
>>>> and they are all happy (mostly).
>>>>
>>>> Are you using the KVM support 2.6.26 provided or are you on some recent
>>>> kvm-kmod? The latter is highly recommended for tons of reasons.
>>>>
>>>> Jan
>>>
>>> I use the KVM modules provided by the distribution.
>>>
>> Then it makes some sense to try with kvm-kmod first. I do not know what
>> KVM patches debian people back-ported to their kernel, maybe you run
>> into long-fixed bugs.
> 
> 
> I confirm that this exception does not happen with kvm-kmod-2.6.33-rc6.
> 
> However, after I run and stop a guest VM just executing the bios sequence (no disk/CD image attached), I get this kind of errors on the host when untar a small archive (less than 1MB, about 300 files):
> 
> Feb  8 12:00:32 paraquad-7 kernel: [  200.073865] Bad page state in process 'gzip'
> Feb  8 12:00:32 paraquad-7 kernel: [  200.073867] page:ffffe200040c4638 flags:0x020000000000001c mapping:0000000000000000 mapcount:0 count:0
> Feb  8 12:00:32 paraquad-7 kernel: [  200.102991] Trying to fix it up, but a reboot is needed
> Feb  8 12:00:32 paraquad-7 kernel: [  200.102992] Backtrace:
> Feb  8 12:00:32 paraquad-7 kernel: [  200.129821] Pid: 3559, comm: gzip Tainted: P          2.6.26-2-amd64 #1
> Feb  8 12:00:32 paraquad-7 kernel: [  200.143267] 
> Feb  8 12:00:32 paraquad-7 kernel: [  200.143268] Call Trace:
> Feb  8 12:00:32 paraquad-7 kernel: [  200.156024]  [<ffffffff8022f0d2>] hrtick_set+0x88/0xf7
> Feb  8 12:00:32 paraquad-7 kernel: [  200.167250]  [<ffffffff8027504d>] bad_page+0x6b/0x95
> Feb  8 12:00:32 paraquad-7 kernel: [  200.179879]  [<ffffffff80276482>] get_page_from_freelist+0x3e1/0x606
> Feb  8 12:00:32 paraquad-7 kernel: [  200.195345]  [<ffffffff80276929>] __alloc_pages_internal+0xd6/0x3bf
> Feb  8 12:00:32 paraquad-7 kernel: [  200.210580]  [<ffffffff80281707>] handle_mm_fault+0x244/0x867
> Feb  8 12:00:32 paraquad-7 kernel: [  200.223694]  [<ffffffff8027313a>] filemap_fault+0x1c2/0x33c
> Feb  8 12:00:32 paraquad-7 kernel: [  200.237948]  [<ffffffff80221fbc>] do_page_fault+0x5d8/0x9c8
> Feb  8 12:00:32 paraquad-7 kernel: [  200.251793]  [<ffffffff8042a599>] error_exit+0x0/0x60
> Feb  8 12:00:32 paraquad-7 kernel: [  200.265754]  [<ffffffff80270df1>] file_read_actor+0x52/0x125
> Feb  8 12:00:32 paraquad-7 kernel: [  200.278595]  [<ffffffff802764fb>] get_page_from_freelist+0x45a/0x606
> Feb  8 12:00:32 paraquad-7 kernel: [  200.294022]  [<ffffffff80272c03>] generic_file_aio_read+0x29e/0x4ae
> Feb  8 12:00:32 paraquad-7 kernel: [  200.309319]  [<ffffffff8029ae47>] do_sync_read+0xc9/0x10c
> Feb  8 12:00:32 paraquad-7 kernel: [  200.322816]  [<ffffffff80246221>] autoremove_wake_function+0x0/0x2e
> Feb  8 12:00:32 paraquad-7 kernel: [  200.338051]  [<ffffffff80221fbc>] do_page_fault+0x5d8/0x9c8
> Feb  8 12:00:32 paraquad-7 kernel: [  200.351919]  [<ffffffff8029b638>] vfs_read+0xaa/0x152
> Feb  8 12:00:32 paraquad-7 kernel: [  200.366108]  [<ffffffff8029ba19>] sys_read+0x45/0x6e
> Feb  8 12:00:32 paraquad-7 kernel: [  200.378750]  [<ffffffff8020beca>] system_call_after_swapgs+0x8a/0x8f
> 

Haven't seen this before. Can you reproduce it with

a) kvm-kmod-2.6.32.7?
b) a host kernel >= 2.6.27?

Thanks,
Jan

-- 
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux