[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH v14 18/22] accel: introduce AccelCPUClass extending CPUClass
From: |
Claudio Fontana |
Subject: |
Re: [PATCH v14 18/22] accel: introduce AccelCPUClass extending CPUClass |
Date: |
Mon, 1 Feb 2021 10:22:35 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.12.0 |
On 2/1/21 10:15 AM, Alex Bennée wrote:
>
> Richard Henderson <richard.henderson@linaro.org> writes:
>
>> On 1/30/21 12:53 AM, Claudio Fontana wrote:
>>> To summarize:
>>>
>>> 1) accel->cpu_realizefn extends the current cpu target-specific realize
>>> functions with accelerator-specific code,
>>> which currently does not make use of errp at all (thus, the temptation
>>> to remove errp from the interface until it is actually needed by a target).
>>
>> No, arm does use errp in realizefn already.
>> It's just that the void return value is hinky.
>
> Sounds like fixing the void return would be part of a larger qdev
> clean-up rather than this particular series. If I recall there have been
> various phases of clean-ups and refactoring of the error code paths in
> the past.
>
>>
>>
>> r~
>
>
I think you are right.
I added a patch in this series to at least not make the problem worse,
by adding a return value to accel_cpu::cpu_realizefn,
and cleaning up the only code path case that signals an error there for now in
i386.
(will add as the last patch in the series)
Thanks,
Claudio