From: Christoph Hellwig Subject: Re: [PATCH 1/2] drm: add cache support for arm64 Date: Thu, 8 Aug 2019 11:55:06 +0200 Message-ID: <20190808095506.GA32621@lst.de> References: <20190805211451.20176-1-robdclark@gmail.com> <20190806084821.GA17129@lst.de> <20190806155044.GC25050@lst.de> <20190807062545.GF6627@lst.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Daniel Vetter Cc: Christoph Hellwig , Rob Clark , Rob Clark , dri-devel , Catalin Marinas , Will Deacon , Maarten Lankhorst , Maxime Ripard , Sean Paul , David Airlie , Allison Randal , Greg Kroah-Hartman , Thomas Gleixner , Linux ARM , LKML List-Id: dri-devel@lists.freedesktop.org On Wed, Aug 07, 2019 at 10:48:56AM +0200, Daniel Vetter wrote: > > other drm drivers how do they guarantee addressability without an > > iommu?) > > We use shmem to get at swappable pages. We generally just assume that > the gpu can get at those pages, but things fall apart in fun ways: > - some setups somehow inject bounce buffers. Some drivers just give > up, others try to allocate a pool of pages with dma_alloc_coherent. > - some devices are misdesigned and can't access as much as the cpu. We > allocate using GFP_DMA32 to fix that. Well, for shmem you can't really call allocators directly, right? One thing I have in my pipeline is a dma_alloc_pages API that allocates pages that are guaranteed to be addressably by the device or otherwise fail. But that doesn't really help with the shmem fs. > Also modern gpu apis pretty much assume you can malloc() and then use > that directly with the gpu. Which is fine as long as the GPU itself supports full 64-bit addressing (or always sits behind an iommu), and the platform doesn't impose addressing limit, which unfortunately some that are shipped right now still do :( But userspace malloc really means dma_map_* anyway, so not really relevant for memory allocations.