[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [qemu-s390x] [PATCH 1/3] qmp: expose s390-specific CPU

From: Viktor Mihajlovski
Subject: Re: [Qemu-devel] [qemu-s390x] [PATCH 1/3] qmp: expose s390-specific CPU info
Date: Tue, 13 Feb 2018 13:20:44 +0100
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.6.0

On 13.02.2018 12:16, David Hildenbrand wrote:
> On 12.02.2018 19:03, Luiz Capitulino wrote:
>> On Mon, 12 Feb 2018 13:14:30 +0100
>> Viktor Mihajlovski <address@hidden> wrote:
>>> Presently s390x is the only architecture not exposing specific
>>> CPU information via QMP query-cpus. Upstream discussion has shown
>>> that it could make sense to report the architecture specific CPU
>>> state, e.g. to detect that a CPU has been stopped.
>>> With this change the output of query-cpus will look like this on
>>> s390:
>>>    [
>>>      {"arch": "s390", "current": true,
>>>       "props": {"core-id": 0}, "cpu-state": "operating", "CPU": 0,
>>>       "qom_path": "/machine/unattached/device[0]",
>>>       "halted": false, "thread_id": 63115},
>>>      {"arch": "s390", "current": false,
>>>       "props": {"core-id": 1}, "cpu-state": "stopped", "CPU": 1,
>>>       "qom_path": "/machine/unattached/device[1]",
>>>       "halted": true, "thread_id": 63116}
>>>    ]
>> We're adding the same information to query-cpus-fast. Why do we
>> need to duplicate this into query-cpus? Do you plan to keep using
>> query-cpus? If yes, why?
> Wonder if we could simply pass a flag to query-cpus "fast=true", that
> makes it behave differently. (either not indicate the critical values or
> simply provide dummy values - e.g. simply halted=false)
That was one of the ideas in the earlier stages of this discussion (and
I was inclined to go that way initially). The major drawback of this
approach that the slow call is hard to deprecate (OK, one could change
the default to fast=true over time). It's easier to deprecate the entire
query-cpus function. The other issue, maybe not as bad, is that one has
to deal with fields that are suddenly optional or obsolete in way not
confusing every one.
Bottom line is that I'm convinced it's better to have both APIs and to
deprecate the slow one over time. But I have to confess I'm not familiar
with QAPI deprecation rules, maybe Eric can shed some light on this...
>> Libvirt for one, should move away from it. We don't want to run
>> into the risk of having the same issue we had with x86 in other
>> archs.

 Viktor Mihajlovski

reply via email to

[Prev in Thread] Current Thread [Next in Thread]