Re: [PATCH v2 2/4] tests: Add capabilities for QEMU 9.2.0 on ppc64

[Narayana Murty N <nnmlinux@xxxxxxxxxxxxx>]

On 04/03/25 10:00 PM, Peter Krempa wrote:

> On Tue, Mar 04, 2025 at 10:40:22 -0500, Narayana Murty N wrote:
> > Add the qemu test capabilities xml and reply files for
> > QEMU 9.2.0 on ppc64. A QEMU v9.2.0 was used for generating
> > this data.
> >       
> That's what would be expected based on the summary, although it isn't
> true; see below.
>

Sorry, I didn't realize that the tree was dirty, I will regenerate it correctly in version 3.

> > Signed-off-by: Narayana Murty N <nnmlinux@xxxxxxxxxxxxx>
> > ---
> >       
> [...]
>
> > diff --git a/tests/qemucapabilitiesdata/caps_9.2.0_ppc64.replies b/tests/qemucapabilitiesdata/caps_9.2.0_ppc64.replies
> > new file mode 100644
> > index 0000000000..feae6134fc
> > --- /dev/null
> > +++ b/tests/qemucapabilitiesdata/caps_9.2.0_ppc64.replies
> > @@ -0,0 +1,39302 @@
> > +{
> > +  "execute": "qmp_capabilities",
> > +  "id": "libvirt-1"
> > +}
> > +
> > +{
> > +  "return": {},
> > +  "id": "libvirt-1"
> > +}
> > +
> > +{
> > +  "execute": "query-version",
> > +  "id": "libvirt-2"
> > +}
> > +
> > +{
> > +  "return": {
> > +    "qemu": {
> > +      "micro": 1,
> > +      "minor": 2,
> > +      "major": 9
> > +    },
> > +    "package": "v9.2.1-dirty"
> >       
> This is qemu v9.2.1 and not 9.2.0 as you claim in the commit message.
>
> Your tree also wasn't clean when building qemu. That can happen if you have
> non-commited code or didn't update all submodules.
>     
    Yes
> > +  },
> >       
> [...]
>
> > diff --git a/tests/qemuhotplugtestcpus/ppc64-modern-bulk-result-conf.xml b/tests/qemuhotplugtestcpus/ppc64-modern-bulk-result-conf.xml
> > index 1a0d28257e..d6ae9995c3 100644
> > --- a/tests/qemuhotplugtestcpus/ppc64-modern-bulk-result-conf.xml
> > +++ b/tests/qemuhotplugtestcpus/ppc64-modern-bulk-result-conf.xml
> > @@ -43,7 +43,7 @@
> >      <boot dev='network'/>
> >    </os>
> >    <cpu mode='custom' match='exact' check='none'>
> > -    <model fallback='forbid'>POWER9</model>
> > +    <model fallback='forbid'>POWER10</model>
> >      <topology sockets='1' dies='1' clusters='1' cores='4' threads='8'/>
> >    </cpu>
> >    <clock offset='utc'/>
> > diff --git a/tests/qemuhotplugtestcpus/ppc64-modern-bulk-result-live.xml b/tests/qemuhotplugtestcpus/ppc64-modern-bulk-result-live.xml
> > index b127883b36..c02d2bb10f 100644
> > --- a/tests/qemuhotplugtestcpus/ppc64-modern-bulk-result-live.xml
> > +++ b/tests/qemuhotplugtestcpus/ppc64-modern-bulk-result-live.xml
> > @@ -43,7 +43,7 @@
> >      <boot dev='network'/>
> >    </os>
> >    <cpu mode='custom' match='exact' check='none'>
> > -    <model fallback='forbid'>POWER9</model>
> > +    <model fallback='forbid'>POWER10</model>
> >       
> Please also explain in the commit message why the CPU has changed.
>

This is because of

commit 140ff3c514159ba58d04e0ca43aa1b9a3b3fcd2d Author: Peter Krempa <pkrempa@xxxxxxxxxx> Date: Mon Aug 28 14:24:52 2023 +0200 tests: qemuhotplugtest: Fix arch-specific parts of 'ppc64' test XMLs The tests were using a copy of a x86_64 based XML and thus 'qemuhotplugtest' was selecting wrong capabilities to use for that specific test. Signed-off-by: Peter Krempa <pkrempa@xxxxxxxxxx>

which fixated on power9 for most of the tests skipping this specific test.

I will fix it separately in my next version.

Narayana

> >      <topology sockets='1' dies='1' clusters='1' cores='4' threads='8'/>
> >    </cpu>
> >    <clock offset='utc'/>
> >       
>