[igt-dev] [PATCH i-g-t v2 08/10] tests/i915/exec_fence: add timeline fence tests

[Lionel Landwerlin]

On 22/11/2019 17:23, Chris Wilson wrote:
> Quoting Lionel Landwerlin (2019-11-22 15:17:02)
>> On 22/11/2019 17:13, Chris Wilson wrote:
>>> Quoting Lionel Landwerlin (2019-11-22 13:02:58)
>>>> We can now give a tuple (handle, point_value) for timeline semaphores.
>>>>
>>>> v2: Fix up syncobj-timeline-repeat test (Lionel)
>>>>       Update following kernel rename
>>>>
>>>> v3: Add more ordering tests for syncobj/buffer handle state (Chris)
>>> One thing I did not spot was using the syncobj to schedule different
>>> contexts. Is that interesting?
>>> -Chris
>> The chaining tests are using different context & different engines for
>> each point of the timeline.
>>
>> Is that what you're looking for?
> Patch 10? Yeah, that probably does more than what I was thinking, which
> is just that contexts cannot be rescheduled past their syncpts.


One reason I kind of somewhat confident is that you cannot build 
circular dependencies.

When we create a new node in the timeline, we make it out of existing 
fences and the dma-fence-chain structure does not allow putting new 
nodes anywhere other than the front of the timeline.


-Lionel


>
> /me goes back to trying to find gaps
>
> Where's gcov when you need it?
> -Chris