Re: [Qemu-devel] travis failures

[Emilio G. Cota]

On Fri, Dec 21, 2018 at 11:26:34 +0100, Greg Kurz wrote:
> On Fri, 21 Dec 2018 09:36:23 +0100
> Greg Kurz <address@hidden> wrote:
> 
> > On Fri, 21 Dec 2018 07:47:07 +0100
> > Thomas Huth <address@hidden> wrote:
> > 
> > > On 2018-12-21 04:06, Alexey Kardashevskiy wrote:  
> > > > Hi
> > > > 
> > > > I am trying https://travis-ci.org/aik/qemu/ and that thing fails every
> > > > time I am not so sure why.
> > > > 
> > > > One example:
> > > > https://travis-ci.org/aik/qemu/jobs/470796318
> > > > 
> > > > The errors are like this:
> > > > 
> > > >   GTESTER check-qtest-unicore32
> > > >   GTESTER check-qtest-x86_64
> > > > Could not access KVM kernel module: No such file or directory
> > > > qemu-system-x86_64: failed to initialize KVM: No such file or directory
> > > > qemu-system-x86_64: Back to tcg accelerator
> > > > 
> > > > 
> > > > Does anyone else see those? How do we fix them? Thanks.    
> > > 
> > > Some test are explicitly requesting "-M accel=kvm:tcg" and this is
> > > causing this message if KVM is not available. We could maybe silence
> > > them if qtest_enabled() ?
> > > 
> > >  Thomas
> > > 
> > >   
> > 
> > I've also experienced a bunch of similar failures with travis, like:
> > 
> > https://travis-ci.org/gkurz/qemu/jobs/466994463
> > 
> 
> Today with tests/test-qht-par again:
> 
> https://travis-ci.org/gkurz/qemu/jobs/470871251#L4455
> 
> Cc'ing Emilio for suggestions on how to go forward with this.

Yes, this was being discussed in this thread:

  https://lists.gnu.org/archive/html/qemu-devel/2018-12/msg01638.html

I haven't heard from Eduardo in a while, I guess he's off for the
holidays.

My latest suggestion was to try a branch where test-qht-par is
standalone, instead of using system(3) to call qht-bench--see
this message:
  https://lists.gnu.org/archive/html/qemu-devel/2018-12/msg02056.html

Can anyone please try that? Or at least let me know how to run
a test Travis build so that I can debug this faster.

Thanks,

                Emilio