drm-sync timeline signaling

drm-sync timeline signaling

[Lionel Landwerlin]

Hi all,

While picking up the IGT tests for timeline syncobj,
I noticed that although we deal with multi wait across both timeline 
(with point value > 0) and binary (point value = 0) syncobjs,
we don't seem to have a similar behavior with signaling.

Do you have any thought on this?
I'm considering writing some docs but I'm not quite sure whether this 
difference between signaling/waiting was intentional or just overlooked.

Thanks,

-Lionel
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: drm-sync timeline signaling

[Lionel Landwerlin]

Ping?

On 16/05/2019 15:49, Lionel Landwerlin wrote:
> Hi all,
>
> While picking up the IGT tests for timeline syncobj,
> I noticed that although we deal with multi wait across both timeline 
> (with point value > 0) and binary (point value = 0) syncobjs,
> we don't seem to have a similar behavior with signaling.
>
> Do you have any thought on this?
> I'm considering writing some docs but I'm not quite sure whether this 
> difference between signaling/waiting was intentional or just overlooked.
>
> Thanks,
>
> -Lionel
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel


_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: drm-sync timeline signaling

[zhoucm1]

Sorry for late response.

Although we don't expect that, drm_syncobj_timeline_signal_ioctl already 
handle this case I think. Which can handle both (point value > 0) and 
(point value = 0).


-David


On 2019年05月21日 16:44, Lionel Landwerlin wrote:
> [CAUTION: External Email]
>
> Ping?
>
> On 16/05/2019 15:49, Lionel Landwerlin wrote:
>> Hi all,
>>
>> While picking up the IGT tests for timeline syncobj,
>> I noticed that although we deal with multi wait across both timeline
>> (with point value > 0) and binary (point value = 0) syncobjs,
>> we don't seem to have a similar behavior with signaling.
>>
>> Do you have any thought on this?
>> I'm considering writing some docs but I'm not quite sure whether this
>> difference between signaling/waiting was intentional or just overlooked.
>>
>> Thanks,
>>
>> -Lionel
>> _______________________________________________
>> dri-devel mailing list
>> dri-devel@lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/dri-devel
>
>

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: drm-sync timeline signaling

[Lionel Landwerlin]

Thanks David,

Initially I thought we could run into wait issues with a dma_fence_chain 
with a value = 0, but you're right, it works fine.

We could just avoid to create a dma_fence_chain object, but maybe we 
don't care.

-Lionel

On 21/05/2019 09:51, zhoucm1 wrote:
> Sorry for late response.
>
> Although we don't expect that, drm_syncobj_timeline_signal_ioctl 
> already handle this case I think. Which can handle both (point value > 
> 0) and (point value = 0).
>
>
> -David
>
>
> On 2019年05月21日 16:44, Lionel Landwerlin wrote:
>> [CAUTION: External Email]
>>
>> Ping?
>>
>> On 16/05/2019 15:49, Lionel Landwerlin wrote:
>>> Hi all,
>>>
>>> While picking up the IGT tests for timeline syncobj,
>>> I noticed that although we deal with multi wait across both timeline
>>> (with point value > 0) and binary (point value = 0) syncobjs,
>>> we don't seem to have a similar behavior with signaling.
>>>
>>> Do you have any thought on this?
>>> I'm considering writing some docs but I'm not quite sure whether this
>>> difference between signaling/waiting was intentional or just 
>>> overlooked.
>>>
>>> Thanks,
>>>
>>> -Lionel
>>> _______________________________________________
>>> dri-devel mailing list
>>> dri-devel@lists.freedesktop.org
>>> https://lists.freedesktop.org/mailman/listinfo/dri-devel
>>
>>
>
>

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel