From mboxrd@z Thu Jan  1 00:00:00 1970
From: Christoph Hellwig <hch@infradead.org>
Subject: Re: [PATCH v6 4/5] dma-buf: heaps: Add CMA heap to dmabuf heaps
Date: Thu, 25 Jul 2019 05:52:06 -0700
Message-ID: <20190725125206.GE20286@infradead.org>
References: <20190624194908.121273-1-john.stultz@linaro.org>
 <20190624194908.121273-5-john.stultz@linaro.org>
 <20190718100840.GB19666@infradead.org>
 <CALAqxLWLx_tHVjZqrSNWfQ_M2RGGqh4qth3hi9GGRdSPov-gcw@mail.gmail.com>
 <20190724065958.GC16225@infradead.org>
 <CALAqxLUbsz+paJ=P2hwKecuN8DQjJt9Vj4MENCnFuEh6waxsXg@mail.gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <CALAqxLUbsz+paJ=P2hwKecuN8DQjJt9Vj4MENCnFuEh6waxsXg@mail.gmail.com>
Sender: linux-kernel-owner@vger.kernel.org
To: John Stultz <john.stultz@linaro.org>
Cc: Christoph Hellwig <hch@infradead.org>, lkml <linux-kernel@vger.kernel.org>, Laura Abbott <labbott@redhat.com>, Benjamin Gaignard <benjamin.gaignard@linaro.org>, Sumit Semwal <sumit.semwal@linaro.org>, Liam Mark <lmark@codeaurora.org>, Pratik Patel <pratikp@codeaurora.org>, Brian Starkey <Brian.Starkey@arm.com>, Vincent Donnefort <Vincent.Donnefort@arm.com>, Sudipto Paul <Sudipto.Paul@arm.com>, "Andrew F . Davis" <afd@ti.com>, Xu YiPing <xuyiping@hisilicon.com>, "Chenfeng (puck)" <puck.chen@hisilicon.com>, butao <butao@hisilicon.com>, "Xiaqing (A)" <saberlily.xia@hisilicon.com>, Yudongbin <yudongbin@hisilicon.com>, Chenbo Feng <fengc@google.com>, Alistair Strachan <astrachan@google.com>, dri-devel <dri-devel@lists.freedesktop.org>
List-Id: dri-devel@lists.freedesktop.org

On Wed, Jul 24, 2019 at 11:46:24AM -0700, John Stultz wrote:
> I'm still not understanding how this would work. Benjamin and Laura
> already commented on this point, but for a simple example, with the
> HiKey boards, the DRM driver requires contiguous memory for the
> framebuffer, but the GPU can handle non-contiguous. Thus the target
> framebuffers that we pass to the display has to be CMA allocated, but
> all the other graphics buffers that the GPU will render to and
> composite can be system.

But that just means we need a flag that memory needs to be contiguous,
which totally makes sense at the API level.  But CMA is not the only
source of contigous memory, so we should not conflate the two.

> Laura already touched on this, but similar logic can be used for
> camera buffers, which can make sure we allocate from a specifically
> reserved CMA region that is only used for the camera so we can always
> be sure to have N free buffers immediately to capture with, etc.

And for that we already have per-device CMA areas hanging off struct
device, which this should reuse instead of adding another duplicate
CMA area lookup scheme.