Re: [Qemu-devel] [PATCH for-4.0?] arm: Allow system registers for KVM guests to be changed by QEMU code

[gengdongjiu]

On 2019/3/18 20:49, Peter Maydell wrote:
> On Mon, 18 Mar 2019 at 12:34, gengdongjiu <address@hidden> wrote:
>>
>>
>>
>> On 2019/3/16 4:11, Philippe Mathieu-Daudé wrote:
>>>> Signed-off-by: Peter Maydell <address@hidden>
>>>> ---
>>>> Should we try to put this in for rc1? Not sure... Testing
>>>> definitely appreciated.
>>> You might include it for rc1 and we still have rc2/rc3 to revert it.
>>
>> why we still have rc2/rc3 to revert it?
> 
> I think Philippe's point is that it's reasonably safe to
> apply this patch in rc1 (ie now), because if we do do that
> and then discover that we have some other bug in it, we
> still have plenty of time to take the patch out again
> before release.
> 
> (If we did discover another bug in this patch in future, I
> would favour reverting it for the 4.0 release rather than
> trying to fix whatever that bug was, because two unexpected
> bugs in the patch means I clearly didn't understand
> the code well enough to produce a reliable patch. The cases
> that this patch fixes are pretty rare -- it does fix a bug
> but only in handling of some cases of debugging a KVM guest;
> but the patch potentially affects the behaviour of any
> KVM guest, even if the user isn't trying to debug. So it's
> riskier than some other kinds of change, and on balance
> that means that when we're making a decision at rc2 or rc3
> I'm more in favour of just reverting it rather than
> applying what we hope is a fix.)

Got it, understand it now, thanks very much for Peter's detailed explanation.

> 
> thanks
> -- PMM
> 
> .
>