From mboxrd@z Thu Jan 1 00:00:00 1970 From: Chris Wilson Subject: Re: [Intel-gfx] [PATCH 01/13] drm/i915: fall through pwrite_gtt_slow to the shmem slow path Date: Mon, 21 Nov 2011 10:20:15 +0000 Message-ID: References: <1320606840-21132-1-git-send-email-daniel.vetter@ffwll.ch> <1320606840-21132-2-git-send-email-daniel.vetter@ffwll.ch> <20111120190918.2b138476@bwidawsk.net> Return-path: In-Reply-To: <20111120190918.2b138476@bwidawsk.net> Sender: linux-kernel-owner@vger.kernel.org To: Ben Widawsky , Daniel Vetter Cc: intel-gfx , linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org List-Id: intel-gfx@lists.freedesktop.org On Sun, 20 Nov 2011 19:09:18 -0800, Ben Widawsky wrote: > On Sun, 6 Nov 2011 20:13:48 +0100 > Daniel Vetter wrote: > > > The gtt_pwrite slowpath grabs the userspace memory with > > get_user_pages. This will not work for non-page backed memory, like a > > gtt mmapped gem object. Hence fall throuh to the shmem paths if we hit > > -EFAULT in the gtt paths. > > > > Now the shmem paths have exactly the same problem, but this way we > > only need to rearrange the code in one write path. > > > > v2: v1 accidentaly falls back to shmem pwrite for phys objects. Fixed. > > > > Signed-Off-by: Daniel Vetter > > It would be nice if there was some way to notify users that pwriting a > gtt mmapped address can be really damn slow. That's also the one > behavior change this patch introduces. It's possible that some SW was > expecting to get a, "fast path" and would deal with the -EFAULT if it > didn't get it. The behaviour change is intentional. Before this patch we would deadlock... -Chris -- Chris Wilson, Intel Open Source Technology Centre