From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> To: Jason Ekstrand <jason@jlekstrand.net>, dri-devel@lists.freedesktop.org, intel-gfx@lists.freedesktop.org Cc: "Daniel Vetter" <daniel.vetter@ffwll.ch>, "Christian König" <christian.koenig@amd.com>, "Matthew Auld" <matthew.auld@intel.com>, "Dave Airlie" <airlied@redhat.com> Subject: Re: [Intel-gfx] [PATCH 0/5] dma-fence, i915: Stop allowing SLAB_TYPESAFE_BY_RCU for dma_fence Date: Thu, 10 Jun 2021 10:29:12 +0100 [thread overview] Message-ID: <ae289388-911f-2928-d695-050863ee1007@linux.intel.com> (raw) In-Reply-To: <20210609212959.471209-1-jason@jlekstrand.net> On 09/06/2021 22:29, Jason Ekstrand wrote: > Ever since 0eafec6d3244 ("drm/i915: Enable lockless lookup of request > tracking via RCU"), the i915 driver has used SLAB_TYPESAFE_BY_RCU (it > was called SLAB_DESTROY_BY_RCU at the time) in order to allow RCU on > i915_request. As nifty as SLAB_TYPESAFE_BY_RCU may be, it comes with > some serious disclaimers. In particular, objects can get recycled while > RCU readers are still in-flight. This can be ok if everyone who touches > these objects knows about the disclaimers and is careful. However, > because we've chosen to use SLAB_TYPESAFE_BY_RCU for i915_request and > because i915_request contains a dma_fence, we've leaked > SLAB_TYPESAFE_BY_RCU and its whole pile of disclaimers to every driver > in the kernel which may consume a dma_fence. I don't think the part about leaking is true... > We've tried to keep it somewhat contained by doing most of the hard work > to prevent access of recycled objects via dma_fence_get_rcu_safe(). > However, a quick grep of kernel sources says that, of the 30 instances > of dma_fence_get_rcu*, only 11 of them use dma_fence_get_rcu_safe(). > It's likely there bear traps in DRM and related subsystems just waiting > for someone to accidentally step in them. ...because dma_fence_get_rcu_safe apears to be about whether the *pointer* to the fence itself is rcu protected, not about the fence object itself. If one has a stable pointer to a fence dma_fence_get_rcu is I think enough to deal with SLAB_TYPESAFE_BY_RCU used by i915_request (as dma fence is a base object there). Unless you found a bug in rq field recycling. But access to the dma fence is all tightly controlled so I don't get what leaks. > This patch series stops us using SLAB_TYPESAFE_BY_RCU for i915_request > and, instead, does an RCU-safe slab free via rcu_call(). This should > let us keep most of the perf benefits of slab allocation while avoiding > the bear traps inherent in SLAB_TYPESAFE_BY_RCU. It then removes support > for SLAB_TYPESAFE_BY_RCU from dma_fence entirely. According to the rationale behind SLAB_TYPESAFE_BY_RCU traditional RCU freeing can be a lot more costly so I think we need a clear justification on why this change is being considered. Regards, Tvrtko > > Note: The last patch is labled DONOTMERGE. This was at Daniel Vetter's > request as we may want to let this bake for a couple releases before we > rip out dma_fence_get_rcu_safe entirely. > > Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> > Cc: Jon Bloomfield <jon.bloomfield@intel.com> > Cc: Daniel Vetter <daniel.vetter@ffwll.ch> > Cc: Christian König <christian.koenig@amd.com> > Cc: Dave Airlie <airlied@redhat.com> > Cc: Matthew Auld <matthew.auld@intel.com> > Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> > > Jason Ekstrand (5): > drm/i915: Move intel_engine_free_request_pool to i915_request.c > drm/i915: Use a simpler scheme for caching i915_request > drm/i915: Stop using SLAB_TYPESAFE_BY_RCU for i915_request > dma-buf: Stop using SLAB_TYPESAFE_BY_RCU in selftests > DONOTMERGE: dma-buf: Get rid of dma_fence_get_rcu_safe > > drivers/dma-buf/dma-fence-chain.c | 8 +- > drivers/dma-buf/dma-resv.c | 4 +- > drivers/dma-buf/st-dma-fence-chain.c | 24 +--- > drivers/dma-buf/st-dma-fence.c | 27 +--- > drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c | 4 +- > drivers/gpu/drm/i915/gt/intel_engine_cs.c | 8 -- > drivers/gpu/drm/i915/i915_active.h | 4 +- > drivers/gpu/drm/i915/i915_request.c | 147 ++++++++++++---------- > drivers/gpu/drm/i915/i915_request.h | 2 - > drivers/gpu/drm/i915/i915_vma.c | 4 +- > include/drm/drm_syncobj.h | 4 +- > include/linux/dma-fence.h | 50 -------- > include/linux/dma-resv.h | 4 +- > 13 files changed, 110 insertions(+), 180 deletions(-) >
WARNING: multiple messages have this Message-ID (diff)
From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> To: Jason Ekstrand <jason@jlekstrand.net>, dri-devel@lists.freedesktop.org, intel-gfx@lists.freedesktop.org Cc: "Daniel Vetter" <daniel.vetter@ffwll.ch>, "Christian König" <christian.koenig@amd.com>, "Matthew Auld" <matthew.auld@intel.com>, "Dave Airlie" <airlied@redhat.com> Subject: Re: [Intel-gfx] [PATCH 0/5] dma-fence, i915: Stop allowing SLAB_TYPESAFE_BY_RCU for dma_fence Date: Thu, 10 Jun 2021 10:29:12 +0100 [thread overview] Message-ID: <ae289388-911f-2928-d695-050863ee1007@linux.intel.com> (raw) In-Reply-To: <20210609212959.471209-1-jason@jlekstrand.net> On 09/06/2021 22:29, Jason Ekstrand wrote: > Ever since 0eafec6d3244 ("drm/i915: Enable lockless lookup of request > tracking via RCU"), the i915 driver has used SLAB_TYPESAFE_BY_RCU (it > was called SLAB_DESTROY_BY_RCU at the time) in order to allow RCU on > i915_request. As nifty as SLAB_TYPESAFE_BY_RCU may be, it comes with > some serious disclaimers. In particular, objects can get recycled while > RCU readers are still in-flight. This can be ok if everyone who touches > these objects knows about the disclaimers and is careful. However, > because we've chosen to use SLAB_TYPESAFE_BY_RCU for i915_request and > because i915_request contains a dma_fence, we've leaked > SLAB_TYPESAFE_BY_RCU and its whole pile of disclaimers to every driver > in the kernel which may consume a dma_fence. I don't think the part about leaking is true... > We've tried to keep it somewhat contained by doing most of the hard work > to prevent access of recycled objects via dma_fence_get_rcu_safe(). > However, a quick grep of kernel sources says that, of the 30 instances > of dma_fence_get_rcu*, only 11 of them use dma_fence_get_rcu_safe(). > It's likely there bear traps in DRM and related subsystems just waiting > for someone to accidentally step in them. ...because dma_fence_get_rcu_safe apears to be about whether the *pointer* to the fence itself is rcu protected, not about the fence object itself. If one has a stable pointer to a fence dma_fence_get_rcu is I think enough to deal with SLAB_TYPESAFE_BY_RCU used by i915_request (as dma fence is a base object there). Unless you found a bug in rq field recycling. But access to the dma fence is all tightly controlled so I don't get what leaks. > This patch series stops us using SLAB_TYPESAFE_BY_RCU for i915_request > and, instead, does an RCU-safe slab free via rcu_call(). This should > let us keep most of the perf benefits of slab allocation while avoiding > the bear traps inherent in SLAB_TYPESAFE_BY_RCU. It then removes support > for SLAB_TYPESAFE_BY_RCU from dma_fence entirely. According to the rationale behind SLAB_TYPESAFE_BY_RCU traditional RCU freeing can be a lot more costly so I think we need a clear justification on why this change is being considered. Regards, Tvrtko > > Note: The last patch is labled DONOTMERGE. This was at Daniel Vetter's > request as we may want to let this bake for a couple releases before we > rip out dma_fence_get_rcu_safe entirely. > > Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> > Cc: Jon Bloomfield <jon.bloomfield@intel.com> > Cc: Daniel Vetter <daniel.vetter@ffwll.ch> > Cc: Christian König <christian.koenig@amd.com> > Cc: Dave Airlie <airlied@redhat.com> > Cc: Matthew Auld <matthew.auld@intel.com> > Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> > > Jason Ekstrand (5): > drm/i915: Move intel_engine_free_request_pool to i915_request.c > drm/i915: Use a simpler scheme for caching i915_request > drm/i915: Stop using SLAB_TYPESAFE_BY_RCU for i915_request > dma-buf: Stop using SLAB_TYPESAFE_BY_RCU in selftests > DONOTMERGE: dma-buf: Get rid of dma_fence_get_rcu_safe > > drivers/dma-buf/dma-fence-chain.c | 8 +- > drivers/dma-buf/dma-resv.c | 4 +- > drivers/dma-buf/st-dma-fence-chain.c | 24 +--- > drivers/dma-buf/st-dma-fence.c | 27 +--- > drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c | 4 +- > drivers/gpu/drm/i915/gt/intel_engine_cs.c | 8 -- > drivers/gpu/drm/i915/i915_active.h | 4 +- > drivers/gpu/drm/i915/i915_request.c | 147 ++++++++++++---------- > drivers/gpu/drm/i915/i915_request.h | 2 - > drivers/gpu/drm/i915/i915_vma.c | 4 +- > include/drm/drm_syncobj.h | 4 +- > include/linux/dma-fence.h | 50 -------- > include/linux/dma-resv.h | 4 +- > 13 files changed, 110 insertions(+), 180 deletions(-) > _______________________________________________ Intel-gfx mailing list Intel-gfx@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2021-06-10 9:29 UTC|newest] Thread overview: 82+ messages / expand[flat|nested] mbox.gz Atom feed top 2021-06-09 21:29 [PATCH 0/5] dma-fence, i915: Stop allowing SLAB_TYPESAFE_BY_RCU for dma_fence Jason Ekstrand 2021-06-09 21:29 ` [Intel-gfx] " Jason Ekstrand 2021-06-09 21:29 ` [PATCH 1/5] drm/i915: Move intel_engine_free_request_pool to i915_request.c Jason Ekstrand 2021-06-09 21:29 ` [Intel-gfx] " Jason Ekstrand 2021-06-10 10:03 ` Tvrtko Ursulin 2021-06-10 10:03 ` Tvrtko Ursulin 2021-06-10 13:57 ` Jason Ekstrand 2021-06-10 13:57 ` Jason Ekstrand 2021-06-10 15:07 ` Tvrtko Ursulin 2021-06-10 15:07 ` Tvrtko Ursulin 2021-06-10 16:32 ` Jason Ekstrand 2021-06-10 16:32 ` Jason Ekstrand 2021-06-09 21:29 ` [PATCH 2/5] drm/i915: Use a simpler scheme for caching i915_request Jason Ekstrand 2021-06-09 21:29 ` [Intel-gfx] " Jason Ekstrand 2021-06-10 10:08 ` Tvrtko Ursulin 2021-06-10 10:08 ` Tvrtko Ursulin 2021-06-10 13:50 ` Jason Ekstrand 2021-06-10 13:50 ` Jason Ekstrand 2021-06-09 21:29 ` [PATCH 3/5] drm/i915: Stop using SLAB_TYPESAFE_BY_RCU for i915_request Jason Ekstrand 2021-06-09 21:29 ` [Intel-gfx] " Jason Ekstrand 2021-06-09 21:29 ` [PATCH 4/5] dma-buf: Stop using SLAB_TYPESAFE_BY_RCU in selftests Jason Ekstrand 2021-06-09 21:29 ` [Intel-gfx] " Jason Ekstrand 2021-06-16 12:47 ` kernel test robot 2021-06-16 12:47 ` kernel test robot 2021-06-16 12:47 ` kernel test robot 2021-06-09 21:29 ` [PATCH 5/5] DONOTMERGE: dma-buf: Get rid of dma_fence_get_rcu_safe Jason Ekstrand 2021-06-09 21:29 ` [Intel-gfx] " Jason Ekstrand 2021-06-10 6:51 ` Christian König 2021-06-10 6:51 ` [Intel-gfx] " Christian König 2021-06-10 13:59 ` Jason Ekstrand 2021-06-10 13:59 ` [Intel-gfx] " Jason Ekstrand 2021-06-10 15:13 ` Daniel Vetter 2021-06-10 15:13 ` [Intel-gfx] " Daniel Vetter 2021-06-10 16:24 ` Jason Ekstrand 2021-06-10 16:24 ` [Intel-gfx] " Jason Ekstrand 2021-06-10 16:37 ` Daniel Vetter 2021-06-10 16:37 ` [Intel-gfx] " Daniel Vetter 2021-06-10 16:52 ` Jason Ekstrand 2021-06-10 16:52 ` [Intel-gfx] " Jason Ekstrand 2021-06-10 17:06 ` Daniel Vetter 2021-06-10 17:06 ` [Intel-gfx] " Daniel Vetter 2021-06-10 16:54 ` Christian König 2021-06-10 16:54 ` [Intel-gfx] " Christian König 2021-06-10 17:11 ` Daniel Vetter 2021-06-10 17:11 ` [Intel-gfx] " Daniel Vetter 2021-06-10 18:12 ` Christian König 2021-06-10 18:12 ` [Intel-gfx] " Christian König 2021-06-16 16:38 ` kernel test robot 2021-06-16 16:38 ` kernel test robot 2021-06-16 16:38 ` kernel test robot 2021-06-09 21:52 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for dma-fence, i915: Stop allowing SLAB_TYPESAFE_BY_RCU for dma_fence Patchwork 2021-06-09 21:54 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork 2021-06-09 22:22 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork 2021-06-09 22:22 ` [Intel-gfx] ✗ Fi.CI.BUILD: warning " Patchwork 2021-06-10 9:29 ` Tvrtko Ursulin [this message] 2021-06-10 9:29 ` [Intel-gfx] [PATCH 0/5] " Tvrtko Ursulin 2021-06-10 9:39 ` Christian König 2021-06-10 9:39 ` Christian König 2021-06-10 11:29 ` Daniel Vetter 2021-06-10 11:29 ` Daniel Vetter 2021-06-10 11:53 ` Daniel Vetter 2021-06-10 11:53 ` Daniel Vetter 2021-06-10 13:07 ` Tvrtko Ursulin 2021-06-10 13:07 ` Tvrtko Ursulin 2021-06-10 13:35 ` Jason Ekstrand 2021-06-10 13:35 ` Jason Ekstrand 2021-06-10 20:09 ` Jason Ekstrand 2021-06-10 20:09 ` Jason Ekstrand 2021-06-10 20:42 ` Daniel Vetter 2021-06-10 20:42 ` Daniel Vetter 2021-06-11 6:55 ` Christian König 2021-06-11 6:55 ` Christian König 2021-06-11 7:20 ` Daniel Vetter 2021-06-11 7:20 ` Daniel Vetter 2021-06-11 7:42 ` Christian König 2021-06-11 7:42 ` Christian König 2021-06-11 9:33 ` Daniel Vetter 2021-06-11 9:33 ` Daniel Vetter 2021-06-11 10:03 ` Christian König 2021-06-11 10:03 ` Christian König 2021-06-11 15:08 ` Daniel Vetter 2021-06-11 15:08 ` Daniel Vetter
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=ae289388-911f-2928-d695-050863ee1007@linux.intel.com \ --to=tvrtko.ursulin@linux.intel.com \ --cc=airlied@redhat.com \ --cc=christian.koenig@amd.com \ --cc=daniel.vetter@ffwll.ch \ --cc=dri-devel@lists.freedesktop.org \ --cc=intel-gfx@lists.freedesktop.org \ --cc=jason@jlekstrand.net \ --cc=matthew.auld@intel.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.