Re: [Qemu-devel] [PULL 0/7] ppc-for-2.10 queue 20170822

[Laurent Vivier]

On 22/08/2017 11:53, Peter Maydell wrote:
> On 22 August 2017 at 10:43, Laurent Vivier <address@hidden> wrote:
>> On 22/08/2017 11:34, Peter Maydell wrote:
>>> On 22 August 2017 at 05:24, David Gibson <address@hidden> wrote:
>>>> The following changes since commit 
>>>> 1f296733876434118fd766cfef5eb6f29ecab6a8:
>>>>
>>>>   Update version for v2.10.0-rc3 release (2017-08-15 18:53:31 +0100)
>>>>
>>>> are available in the git repository at:
>>>>
>>>>   git://github.com/dgibson/qemu.git tags/ppc-for-2.10-20170822
>>>>
>>>> for you to fetch changes up to d3234e2851f1630c695c681beac1e87ac0881260:
>>>>
>>>>   hw/ppc/spapr_iommu: Fix crash when removing the "spapr-tce-table" device 
>>>> (2017-08-22 11:11:30 +1000)
>>>>
>>>> ----------------------------------------------------------------
>>>> ppc patch queue 2017-08-22
>>>>
>>>> Last minute ppc related fixes for qemu-2.10.  I'm not sure if these
>>>> are critical enough to prompt another rc, but I'm submitting them for
>>>> consideration.
>>>>
>>>> First, is Cornelia's fix for 480bc11e6 which meant "make check" would
>>>> always fail on a ppc host.  Tracking that down delayed submission of
>>>> the rest of these patches, sorry.
>>>>
>>>> The rest are all fairly important bugfixes for qemu crashes or guest
>>>> behaviour regression on ppc.  Patches 2-4 specifically are fixes for
>>>> regressions from qemu-2.9, caused by the compatibility mode and
>>>> hotplug handling cleanups for the pseries machine type.
>>>>
>>>> ----------------------------------------------------------------
>>>
>>> I get a make check failure on ppc64 Linux:
>>>
>>> TEST: tests/postcopy-test... (pid=12468)
>>>   /ppc64/postcopy:
>>> Broken pipe
>>> qemu-system-ppc64: RP: Received invalid message 0x0000 length 0x0000
>>> FAIL
>>> GTester: last random seed: R02Se5468e06f561627824306d95b0566d2b
>>> (pid=13011)
>>> FAIL: tests/postcopy-test

The problem is in:

bool kvmppc_pvr_workaround_required(PowerPCCPU *cpu)
{
    CPUState *cs = CPU(cpu);

    if (cap_ppc_pvr_compat) {
        return false;
    }

    return !kvmppc_is_pr(cs->kvm_state);
}

It guesses !kvm pr means kvm_hv. That is not true, it can be TCG.

This fixes the problem for me:
--- a/target/ppc/kvm.c
+++ b/target/ppc/kvm.c
@@ -2817,5 +2817,5 @@ bool kvmppc_pvr_workaround_required(PowerPCCPU *cpu)
         return false;
     }

-    return !kvmppc_is_pr(cs->kvm_state);
+    return kvm_enabled() && !kvmppc_is_pr(cs->kvm_state);
 }

[CC' Daniel]
Thanks,
Laurent