[Intel-gfx] [PATCH 3/4] dma-fence: Refactor signaling for manual invocation

[Koenig, Christian]

Am 13.08.19 um 10:25 schrieb Chris Wilson:
> Quoting Koenig, Christian (2019-08-13 07:59:28)
>> Am 12.08.19 um 16:53 schrieb Chris Wilson:
>>> Quoting Koenig, Christian (2019-08-12 15:50:59)
>>>> Am 12.08.19 um 16:43 schrieb Chris Wilson:
>>>>> Quoting Koenig, Christian (2019-08-12 15:34:32)
>>>>>> Am 10.08.19 um 17:34 schrieb Chris Wilson:
>>>>>>> Move the duplicated code within dma-fence.c into the header for wider
>>>>>>> reuse. In the process apply a small micro-optimisation to only prune the
>>>>>>> fence->cb_list once rather than use list_del on every entry.
>>>>>>>
>>>>>>> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
>>>>>>> Cc: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
>>>>>>> ---
>>>>>>>      drivers/dma-buf/Makefile                    |  10 +-
>>>>>>>      drivers/dma-buf/dma-fence-trace.c           |  28 +++
>>>>>>>      drivers/dma-buf/dma-fence.c                 |  33 +--
>>>>>>>      drivers/gpu/drm/i915/gt/intel_breadcrumbs.c |  32 +--
>>>>>>>      include/linux/dma-fence-impl.h              |  83 +++++++
>>>>>>>      include/linux/dma-fence-types.h             | 258 ++++++++++++++++++++
>>>>>>>      include/linux/dma-fence.h                   | 228 +----------------
>>>>>> Mhm, I don't really see the value in creating more header files.
>>>>>>
>>>>>> Especially I'm pretty sure that the types should stay in dma-fence.h
>>>>> iirc, when I included the trace.h from dma-fence.h or dma-fence-impl.h
>>>>> without separating the types, amdgpu failed to compile (which is more
>>>>> than likely to be simply due to be first drm in the list to compile).
>>>> Ah, but why do you want to include trace.h in a header in the first place?
>>>>
>>>> That's usually not something I would recommend either.
>>> The problem is that we do emit a tracepoint as part of the sequence I
>>> want to put into the reusable chunk of code.
>> Ok, we are going in circles here. Why do you want to reuse the code then?
> I am reusing the code to avoid fun and games with signal-vs-free.

Yeah, but that doesn't seems to be valid.

See the dma_fence should more or less be a contained object and you now 
expose quite a bit of the internal functionality inside a headers.

And creating headers which when included make other drivers fail to 
compile sounds like a rather bad idea to me.

Please explain the background a bit more.

Thanks,
Christian.

> -Chris