[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Lockup with --accel tcg,thread=single
From: |
Alex Bennée |
Subject: |
Re: Lockup with --accel tcg,thread=single |
Date: |
Tue, 01 Oct 2019 16:28:03 +0100 |
User-agent: |
mu4e 1.3.4; emacs 27.0.50 |
Paolo Bonzini <address@hidden> writes:
> On 01/10/19 10:42, Alex Bennée wrote:
>>
>> Paolo Bonzini <address@hidden> writes:
>>
>>> On 30/09/19 21:20, Alex Bennée wrote:
>>>> Does seem to imply the vCPU CPUState is where the queue is. That's not
>>>> to say there shouldn't be a single work queue for thread=single.
>>>
>>> Indeed it doesn't. I confused this with commit a8efa60633 ("cpus: run
>>> work items for all vCPUs if single-threaded", 2018-11-27).
>>>
>>> Are you going to make a patch to have a single work queue, or should
>>> I?
>>
>> What's the neatest way to do it? Are you thinking just special case
>> queue_work_on_cpu to special case first_cpu when mttcg is not enabled?
>
> Yes, I cannot think of anything better.
And I am immediately stymied by the fact that cpus-common is a common
blob so can't have differentiation for SOFTMMU cases in it.
Did you have a look at:
Date: Mon, 30 Sep 2019 17:38:15 +0100
Message-ID: <address@hidden>
Is that too ugly?
>
> Paolo
--
Alex Bennée