[Qemu-devel] Re: [PATCH v3 uq/master 00/22] Win32 iothread support

[Jan Kiszka]

On 2011-03-03 09:48, Avi Kivity wrote:
> On 03/02/2011 09:42 PM, Marcelo Tosatti wrote:
>> On Wed, Mar 02, 2011 at 04:36:34PM -0300, Marcelo Tosatti wrote:
>>>  On Wed, Mar 02, 2011 at 08:03:42PM +0100, Jan Kiszka wrote:
>>>  >  On 2011-03-02 19:43, Marcelo Tosatti wrote:
>>>  >  >  On Tue, Mar 01, 2011 at 02:35:56PM +0200, Avi Kivity wrote:
>>>  >  >>  On 02/28/2011 04:05 PM, Paolo Bonzini wrote:
>>>  >  >>>  On 02/28/2011 01:13 PM, Avi Kivity wrote:
>>>  >  >>>>>
>>>  >  >>>>
>>>  >  >>>>  If there's a git tree of this I'll be happy to do an autotest run.
>>>  >  >>>
>>>  >  >>>  Sure, it's branch iothread-win32 of 
>>> git://github.com/bonzini/qemu.git
>>>  >  >>
>>>  >  >>  Fails on Fedora 9 i386 install, hangs right after "Performing post
>>>  >  >>  install configuration...".  The guest is processing interrupts but
>>>  >  >>  the mouse won't move, and it doesn't make progress.
>>>  >  >>
>>>  >  >>  Configured with --enable-io-thread.  Perhaps the problem exists even
>>>  >  >>  before the patchset.
>>>  >  >
>>>  >  >  Probably unrelated, looks similar to the regression seen with 
>>> qemu-kvm.
>>>  >  >
>>>  >
>>>  >  Do these patches change some behavior or not?
>>>
>>>  Yes, they change some behaviour. Autotest fails.
>>
>> Sorry, i misunderstood. I don't think these patches change any
>> behaviour.
>>
>> Avi's failure case is similar to what i've seen earlier upon
>> qemu->qemu-kvm merge. Conclusion is there is no new regression
>> introduced by these patches.
>>
> 
> Well that's strange, since qemu-kvm now passes autotest (and this was 
> qemu.git + patchset, not qemu-kvm).
> 

Does it fail every time? Have you tested if qemu.git also fails without
the patches? Then we may have a chance to bisect.

Jan

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