What type of CPU are you running? Is it valid retail silicon? Can you add a
line before the line that is crashing:
printk("vmx_basic_msr == %08x:%08x\n", vmx_msr_high, vmx_msr_low);
BUG_ON(((vmx_msr_high >> 18) & 15) == 6);
...and then tell me what that line prints out immediately before the crash?
'cat /proc/cpuinfo' from booting Linux on that system might also be
interesting, also the CPU vendor/version string printed by Linux in dmesg as
it boots.
-- Keir
On 15/1/08 03:05, "Ralph Passgang" <xen@xxxxxxxxxxxxx> wrote:
> Hi Keir,
>
> thanks, but it seems that I have just bad news for you.
>
> now xen 3.1 (cs 15598) compiles on i386 and amd64 but at least on amd64 the
> hypervisor hangs at boot. I haven't tested i386 (besides that it compiles)
> yet. The last known working changeset for the amd64 hypervisor was 15590.
>
> The most important part of the errormessage from the hypervisor:
>
> Xen call trace:
> [<ffff828c8015f00c>] vmx_init_vmcs_config+0x1bc/0x1f0
> [<ffff828c80160430>] start_vmx+0x70/0x260
> [<ffff828c80146509>] identify_cpu+0xa8/0x200
> [<ffff828c801bbce6>] __start_xen+0x1ff6/0x24f0
> [<ffff828c801000b5>] __high_start+0xa1/0xa3
>
> Panic on CPU 0:
> Xen Bug at vmcs.c:159
>
> If more information is needed, let me know.
>
> Ralph
>
> Am Montag, 14. Januar 2008 18:19:24 schrieb Keir Fraser:
>> It was stuck in the staging tree, which I've now pushed manually.
>>
>> K.
>>
>> On 14/1/08 17:14, "Ralph Passgang" <xen@xxxxxxxxxxxxx> wrote:
>>> Hi Keir,
>>>
>>> xen-3.1-testing.hg still doesn't compile on i386 for the same reason as
>>> in my original report. I know that 3.2 is more important, but it would
>>> be nice if the 3.1 branch could get fixed.
>>>
>>> thx,
>>> Ralph
>>>
>>> Am Freitag, 11. Januar 2008 01:54:16 schrieben Sie:
>>>> We're missing xen-unstable:15526 from 3.1-testing. I'll add it
>>>> tomorrow.
>>>>
>>>> Thanks,
>>>> Keir
>>>>
>>>> On 11/1/08 00:40, "Ralph Passgang" <xen@xxxxxxxxxxxxx> wrote:
>>>>> Hi,
>>>>>
>>>>> I just found that xen 3.1-testing changeset 15577 fails to build on
>>>>> i386 on debian sid/lenny/etch. It seems to compile just fine on
>>>>> amd64/etch.
>>>>>
>>>>> The following happens compiling the 32bit version:
>>>>> [...]
>>>>> gcc -O2 -fomit-frame-pointer -m32 -march=i686 -DNDEBUG
>>>>> -fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes
>>>>> -Wno-unused-value
>>>>> -Wdeclaration-after-statement -nostdinc -fno-builtin -fno-common
>>>>> -iwithprefix include -Werror -Wno-pointer-arith -pipe
>>>>> -I/tmp/buildd/xen-3.1-3.1.3-0/debian/build/build-hypervisor_i386_i386
>>>>> /x en/incl ude
>>>>> -I/tmp/buildd/xen-3.1-3.1.3-0/debian/build/build-hypervisor_i386_i386
>>>>> /x en/incl ude/asm-x86/mach-generic
>>>>> -I/tmp/buildd/xen-3.1-3.1.3-0/debian/build/build-hypervisor_i386_i386
>>>>> /x en/incl ude/asm-x86/mach-default -msoft-float -fno-stack-protector
>>>>> -g -D__XEN__ -DACM_SECURITY -c
>>>>> vmx.c -o vmx.o
>>>>> cc1: warnings being treated as errors
>>>>> vmx.c: In function 'vmx_install_vlapic_mapping':
>>>>> vmx.c:2694: warning: right shift count >= width of type
>>>>> vmx.c:2695: warning: right shift count >= width of type
>>>>> make[8]: *** [vmx.o] Error 1
>>>>> make[8]: Leaving directory
>>>>> `/tmp/buildd/xen-3.1-3.1.3-0/debian/build/build-hypervisor_i386_i386/
>>>>> xe n/arch/ x86/hvm/vmx'
>>>>> make[7]: *** [vmx/built_in.o] Error 2
>>>>> [...]
>>>>>
>>>>> The last changeset that I tried and that compiled on i386 was 15564.
>>>>> So the last 13 changesets could be the cause. The following 4
>>>>> changesets changed the vmx.c, so most likely one of them causes this
>>>>> (but I haven't looked any further):
>>>>>
>>>>> 15565, 15567, 15571, 15575
>>>>>
>>>>> Would be great if someone could take a look...
>>>>>
>>>>> Thx,
>>>>> Ralph
>>>>>
>>>>> _______________________________________________
>>>>> Xen-devel mailing list
>>>>> Xen-devel@xxxxxxxxxxxxxxxxxxx
>>>>> http://lists.xensource.com/xen-devel
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home