Re: [PATCH v1 4/4] .travis.yml: reduce the load on [ppc64] GCC check-tcg

[David Gibson]

On Mon, May 04, 2020 at 08:48:46PM +0100, Alex Bennée wrote:
> 
> Richard Henderson <address@hidden> writes:
> 
> > On 5/3/20 7:10 PM, David Gibson wrote:
> >>>>>           - TEST_CMD="make check check-tcg V=1"
> >>>>> -        - CONFIG="--disable-containers 
> >>>>> --target-list=${MAIN_SOFTMMU_TARGETS},ppc64le-linux-user"
> >>>>> +        - CONFIG="--disable-containers 
> >>>>> --target-list=ppc64-softmmu,ppc64le-linux-user"
> >>>>
> >>>> Cc'ing David, since I'm not sure about this one... Maybe split as we
> >>>> did with other jobs?
> > ...
> >> Hrm.  I'd prefer not to drop this coverage if we can avoid it.  What
> >> we're not testing with the proposed patch is TCG generation for a ppc
> >> host but a non-ppc target.  e.g. if the x86 or ARM target side generates
> >> some pattern of TCG ops that's very rare for the ppc target, and is
> >> buggy in the ppc host side.
> >
> > Are we actually testing those here?  As far as I can see, we're not 
> > installing
> > any cross-compilers here, so we're not building any non-ppc binaries.  Nor 
> > are
> > we running check-acceptance which would download pre-built foreign
> > binaries.
> 
> We are testing the very minimal boot stubs that each -system binary has
> in qtest but they are hardly going to be exercising the majority of the
> TCG. Basically the $SELF-linux-user is going to be exercising more of
> the TCG than anything else.

Oh, good points.

Go ahead then.  In fact we should probably do that for all the
check-tcg builds that don't install cross compilers.

-- 
David Gibson                    | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au  | minimalist, thank you.  NOT _the_ _other_
                                | _way_ _around_!
http://www.ozlabs.org/~dgibson

signature.asc
Description: PGP signature