Re: [Qemu-devel] [PATCH] i386/vmmouse: Properly reset state

[Philippe Mathieu-Daudé]

Hi Jan,

On 8/27/19 9:49 PM, Eduardo Habkost wrote:
> On Sun, Aug 25, 2019 at 04:58:18PM +0200, Jan Kiszka wrote:
>> On 21.07.19 10:58, Jan Kiszka wrote:
>>> From: Jan Kiszka <address@hidden>
>>>
>>> nb_queue was not zeroed so that we no longer delivered events if a
>>> previous guest left the device in an overflow state.
>>>
>>> The state of absolute does not matter as the next vmmouse_update_handler
>>> call will align it again.
>>>
>>> Signed-off-by: Jan Kiszka <address@hidden>
>>> ---
>>>   hw/i386/vmmouse.c | 1 +
>>>   1 file changed, 1 insertion(+)
>>>
>>> diff --git a/hw/i386/vmmouse.c b/hw/i386/vmmouse.c
>>> index 5d2d278be4..e335bd07da 100644
>>> --- a/hw/i386/vmmouse.c
>>> +++ b/hw/i386/vmmouse.c
>>> @@ -257,6 +257,7 @@ static void vmmouse_reset(DeviceState *d)
>>>       VMMouseState *s = VMMOUSE(d);
>>>
>>>       s->queue_size = VMMOUSE_QUEUE_SIZE;
>>> +    s->nb_queue = 0;

Don't we also need to reset the status in case vmmouse_get_status() is
called directly after reset?

         s->status = 0;

With it:
Reviewed-by: Philippe Mathieu-Daudé <address@hidden>

>>>
>>>       vmmouse_disable(s);
>>>   }
>>> --
>>> 2.16.4
>>>
>>>
>>
>> Ping - or who is looking after this?
> 
> Despite being in hw/i386, I think we can say vmmouse.c doesn't
> have a maintainer.  Last time someone changed vmmouse.c in a
> meaningful way (not just adapting to API changes or removing
> duplicate code) was in 2012.
> 

Well it does has a few:

$ ./scripts/get_maintainer.pl -f hw/i386/vmmouse.c
"Michael S. Tsirkin" <address@hidden> (supporter:PC)
Marcel Apfelbaum <address@hidden> (supporter:PC)
Paolo Bonzini <address@hidden> (maintainer:X86 TCG CPUs)
Richard Henderson <address@hidden> (maintainer:X86 TCG CPUs)
Eduardo Habkost <address@hidden> (maintainer:X86 TCG CPUs)

However the correct section should rather be "PC Chipset".

> But the change makes sense to me, so:
> 
> Reviewed-by: Eduardo Habkost <address@hidden>
> 
> I'll queue it.
>