From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?UTF-8?B?6rmA7Iq57Jqw?= Subject: Re: [RFC][PATCH 0/2] dma-buf: add importer private data for reimporting Date: Tue, 04 Jun 2013 19:42:22 +0900 Message-ID: <51ADC48E.80907@samsung.com> References: <1369990487-23510-1-git-send-email-sw0312.kim@samsung.com> <51A879E0.3080106@samsung.com> <20130531152956.GX15743@phenom.ffwll.local> Reply-To: sw0312.kim@samsung.com Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: In-reply-to: <20130531152956.GX15743@phenom.ffwll.local> Sender: linux-kernel-owner@vger.kernel.org To: dri-devel , "linux-media@vger.kernel.org" , "linaro-mm-sig@lists.linaro.org" , Sumit Semwal , Dave Airlie , Linux Kernel Mailing List , Inki Dae , Kyungmin Park , Seung-Woo Kim List-Id: dri-devel@lists.freedesktop.org On 2013=EB=85=84 06=EC=9B=94 01=EC=9D=BC 00:29, Daniel Vetter wrote: > On Fri, May 31, 2013 at 07:22:24PM +0900, =EA=B9=80=EC=8A=B9=EC=9A=B0= wrote: >> Hello Daniel, >> >> Thanks for your comment. >> >> On 2013=EB=85=84 05=EC=9B=94 31=EC=9D=BC 18:14, Daniel Vetter wrote: >>> On Fri, May 31, 2013 at 10:54 AM, Seung-Woo Kim wrote: >>>> importer private data in dma-buf attachment can be used by importe= r to >>>> reimport same dma-buf. >>>> >>>> Seung-Woo Kim (2): >>>> dma-buf: add importer private data to attachment >>>> drm/prime: find gem object from the reimported dma-buf >>> >>> Self-import should already work (at least with the latest refcount >>> fixes merged). At least the tests to check both re-import on the sa= me >>> drm fd and on a different all work as expected now. >> >> Currently, prime works well for all case including self-importing, >> importing, and reimporting as you describe. Just, importing dma-buf = from >> other driver twice with different drm_fd, each import create its own= gem >> object even two import is done for same buffer because prime_priv is= in >> struct drm_file. This means mapping to the device is done also twice= =2E >> IMHO, these duplicated creations and maps are not necessary if drm c= an >> find previous import in different prime_priv. >=20 > Well, that's imo a bug with the other driver. If it doesn't export > something really simple (e.g. contiguous memory which doesn't require= any > mmio resources at all) it should have a cache of exported dma_buf fds= so > that it hands out the same dma_buf every time. Hm, all existing dma-buf exporter including i915 driver implements its map_dma_buf callback as allocating scatter-gather table with pages in its buffer and calling dma_map_sg() with the sgt. With different drm_fds, importing one dma-buf *twice*, then importer calls dma_buf_attach() and dma_buf_map_attachment() twice at least in drm importer because re-importing case can only checked with prime_priv in drm_file as I described. >=20 > Or it needs to be more clever in it's dma_buf_attachment_map function= s and > lookup up a pre-existing iommu mapping. >=20 > But dealing with this in the importer is just broken. >=20 >>> Second, the dma_buf_attachment is _definitely_ the wrong place to d= o >>> this. If you need iommu mapping caching, that should happen at a lo= wer >>> level (i.e. in the map_attachment callback somewhere of the exporte= r, >>> that's what the priv field in the attachment is for). Snatching awa= y >>> the attachement from some random other import is certainly not the = way >>> to go - attachements are _not_ refcounted! >> >> Yes, attachments do not have refcount, so importer should handle and= drm >> case in my patch, importer private data is gem object and it has, of >> course, refcount. >> >> And at current, exporter can not classify map_dma_buf requests of sa= me >> importer to same buffer with different attachment because dma_buf_at= tach >> always makes new attachments. To resolve this exporter should search= all >> different attachment from same importer of dma-buf and it seems more >> complex than importer private data to me. >> >> If I misunderstood something, please let me know. >=20 > Like I've said above, just fix this in the exporter. If an importer s= ees > two different dma_bufs it can very well presume that it those two ind= eed > point to different backing storage. Yes, my patch does not break this concept. I just fixed case importing _one_ dma-buf twice with different drm_fds. >=20 > This will be even more important if we attach fences two dma_bufs. If= your > broken exporter creates multiple dma_bufs each one of them will have = their > own fences attached, leading to a complete disasters. Ok, strictly > speaking if you keep the same reservation pointer for each dma_buf it= 'll > work, but that's just a detail of how you solve this in the exporter. I can not understand about broken exporter you addressed. I don't mean exporter makes dma-bufs from one backing storage. While, my patch prevents not to create drm gem objects from one back storage by importing one dma-buf with different drm-fds. I do not believe the fix of importer is the best way, but at this moment, I have no idea how I can fix the exporter for this issue. Best Regards, - Seung-Woo Kim >=20 > Cheers, Daniel >=20 --=20 Seung-Woo Kim Samsung Software R&D Center --