Re: [PATCH V3] libxl: Increase device model startup timeout to 1min.

[Ian Campbell <ian.campbell@citrix.com>]

On Tue, 2015-07-14 at 16:23 +0200, Dario Faggioli wrote:

> BTW, Ian, perhaps it's not that urgent, but how cumbersome would it be
> to setup a few ad-hoc flights, like you did here:
> 
>  http://logs.test-lab.xenproject.org/osstest/logs/58853/
> 
> when also doing what Anthony was doing for stracing QEMU,

"Fairly cumbersome", someone would have to develop and test some code in
osstest to allow the switching in of a device model wrapper and for
collecting the results afterwards.

Those previous adhoc flights were simple because they just involved
rerunning existing jobs in the same configuration and with the same
harness code.

I'm afraid that for a "to see what we see" experiment the effort I have
outweighs the predicted information gain for this.

If someone were to provide a tested and working osstest patch I could
probably throw it through a few runs with minimal effort though.

>  as he
> described in here:
> 
>  http://bugs.xenproject.org/xen/mid/%3C20150624133948.GJ3393@perard.uk.xensource.com%3E
> 
> I'm not sure what we could find, nor whether we should expect the strace
> to be similar to Anthony's. Perhaps not, as the system here is idle, and
> perhaps something would highlight or hint at where to look in this
> specific case... Or maybe none of the above, but it may be worth a try.
> 
> If there is anything I can do for helping setting this up, just let me
> know (here or on IRC).
> 
> Regards,
> Dario