On 22.03.2018 10:41, Marc-André Lureau wrote: > Hi > > On Thu, Mar 22, 2018 at 6:42 AM, Thomas Huth <th...@redhat.com> wrote: >> On 21.03.2018 12:52, Marc-André Lureau wrote: [...] >>> diff --git a/qmp.c b/qmp.c >>> index d8f80cb04e..14972b78df 100644 >>> --- a/qmp.c >>> +++ b/qmp.c >>> @@ -601,20 +601,6 @@ CpuModelExpansionInfo >>> *qmp_query_cpu_model_expansion(CpuModelExpansionType type, >>> return arch_query_cpu_model_expansion(type, model, errp); >>> } >>> >>> -CpuModelCompareInfo *qmp_query_cpu_model_comparison(CpuModelInfo *modela, >>> - CpuModelInfo *modelb, >>> - Error **errp) >>> -{ >>> - return arch_query_cpu_model_comparison(modela, modelb, errp); >>> -} >>> - >>> -CpuModelBaselineInfo *qmp_query_cpu_model_baseline(CpuModelInfo *modela, >>> - CpuModelInfo *modelb, >>> - Error **errp) >>> -{ >>> - return arch_query_cpu_model_baseline(modela, modelb, errp); >>> -} >> >> Not sure, but couldn't these two commands be implemented on other >> architectures in the long run, too? So removing them now here seems >> somewhat counterproductive? > > They would have modify the qapi ifdef and implement the qmp handler in > their target, similar to what would be done by implementing arch_query > handlers. How counterproductive is that? The benefit is that we don't > have to have stubs and "de-register" the commands.
Yes, thinking about that again, I guess you're right, this should be fine. I first thought that there would likely be some common code between the targets finally, but it's more likely that this CPU stuff is completely different everywhere. So never mind! Thomas