[Intel-gfx] [PATCH 1/3] drm: use common drm_gem_dmabuf_release in i915/exynos drivers

[Daniel Vetter]

On Wed, Aug 7, 2013 at 2:01 PM, Inki Dae <inki.dae at samsung.com> wrote:
>
>
> 2013/8/7 Daniel Vetter <daniel at ffwll.ch>
>>
>> On Wed, Aug 07, 2013 at 07:18:45PM +0900, Joonyoung Shim wrote:
>> > On 08/07/2013 06:55 PM, Daniel Vetter wrote:
>> > >On Wed, Aug 7, 2013 at 11:40 AM, Inki Dae <inki.dae at samsung.com> wrote:
>> > >>>-----Original Message-----
>> > >>>From: Daniel Vetter [mailto:daniel.vetter at ffwll.ch]
>> > >>>Sent: Wednesday, August 07, 2013 6:15 PM
>> > >>>To: DRI Development
>> > >>>Cc: Intel Graphics Development; Daniel Vetter; Inki Dae
>> > >>>Subject: [PATCH 1/3] drm: use common drm_gem_dmabuf_release in
>> > >>> i915/exynos
>> > >>>drivers
>> > >>>
>> > >>>Note that this is slightly tricky since both drivers store their
>> > >>>native objects in dma_buf->priv. But both also embed the base
>> > >>>drm_gem_object at the first position, so the implicit cast is ok.
>> > >>>
>> > >>>To use the release helper we need to export it, too.
>> > >>Yeah, may I repost this patch with additional work?  We also need to
>> > >> export
>> > >>with a gem object instead of specific one like you did.
>> >
>> > I think dmabuf stuff of exynos can be replaced to common drm_gem_dmabuf.
>> > Already dmabuf stuff of drm_gem_cma_helper.c was substituted to common
>> > drm_gem_dmabuf with low-level hook functions to use prime helpers.
>>
>> Ah, but that can easily be done on top of this, right?
>
>
> Daniel, could you remove exynos related codes from your patch set? Your
> patch set would make exynos broken because you didn't consider exporting
> with a gem object for exynos like [PATCH 3/3] drm/i915: explicit store base
> gem object in dma_buf->priv. So I think your patch set is not complete set,
> and That is why exynos needs the additional work I mentioned above. So I
> just wanted to repost your patch set + new one.

Nope, my patch should not break exynos since the base gem_object is
the first member of the exynos object, so we don't have any issues
with upcasting in exynos dma-buf code. The same applies to i915
dma-buf code, my follow-up patch just makes the code a bit safer.

> However, I think not only exynos could go to common drm_gem_dmabuf directly
> but also it would make your patch set to be complete set if you remove
> exynos related codes from your patch set. Otherwise, we have to work twice.
> one is the additional work for resolving exynos broken issue by your patch
> set, and other is to replace existing dmabuf stuff of exynos to common
> drm_gem_dmabuf.

Yeah np, I'll drop exynos then.
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch