From: "Thomas Hellström (Intel)" <thomas_os@shipmail.org>
To: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>,
intel-gfx@lists.freedesktop.org
Subject: Re: [Intel-gfx] [PATCH v6 16/64] drm/i915: Fix userptr so we do not have to worry about obj->mm.lock, v5.
Date: Mon, 18 Jan 2021 12:30:09 +0100 [thread overview]
Message-ID: <5b38d4b5-ded7-9f4d-118e-74f706c6a8ad@shipmail.org> (raw)
In-Reply-To: <20210105153558.134272-17-maarten.lankhorst@linux.intel.com>
Hi,
On 1/5/21 4:35 PM, Maarten Lankhorst wrote:
> Instead of doing what we do currently, which will never work with
> PROVE_LOCKING, do the same as AMD does, and something similar to
> relocation slowpath. When all locks are dropped, we acquire the
> pages for pinning. When the locks are taken, we transfer those
> pages in .get_pages() to the bo. As a final check before installing
> the fences, we ensure that the mmu notifier was not called; if it is,
> we return -EAGAIN to userspace to signal it has to start over.
>
> Changes since v1:
> - Unbinding is done in submit_init only. submit_begin() removed.
> - MMU_NOTFIER -> MMU_NOTIFIER
> Changes since v2:
> - Make i915->mm.notifier a spinlock.
> Changes since v3:
> - Add WARN_ON if there are any page references left, should have been 0.
> - Return 0 on success in submit_init(), bug from spinlock conversion.
> - Release pvec outside of notifier_lock (Thomas).
> Changes since v4:
> - Mention why we're clearing eb->[i + 1].vma in the code. (Thomas)
> - Actually check all invalidations in eb_move_to_gpu. (Thomas)
> - Do not wait when process is exiting to fix gem_ctx_persistence.userptr.
>
> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
...
>
> -static int
> -userptr_mn_invalidate_range_start(struct mmu_notifier *_mn,
> - const struct mmu_notifier_range *range)
> -{
> - struct i915_mmu_notifier *mn =
> - container_of(_mn, struct i915_mmu_notifier, mn);
> - struct interval_tree_node *it;
> - unsigned long end;
> - int ret = 0;
> -
> - if (RB_EMPTY_ROOT(&mn->objects.rb_root))
> - return 0;
> -
> - /* interval ranges are inclusive, but invalidate range is exclusive */
> - end = range->end - 1;
> -
> - spin_lock(&mn->lock);
> - it = interval_tree_iter_first(&mn->objects, range->start, end);
> - while (it) {
> - struct drm_i915_gem_object *obj;
> -
> - if (!mmu_notifier_range_blockable(range)) {
> - ret = -EAGAIN;
> - break;
> - }
> + spin_lock(&i915->mm.notifier_lock);
>
> - /*
> - * The mmu_object is released late when destroying the
> - * GEM object so it is entirely possible to gain a
> - * reference on an object in the process of being freed
> - * since our serialisation is via the spinlock and not
> - * the struct_mutex - and consequently use it after it
> - * is freed and then double free it. To prevent that
> - * use-after-free we only acquire a reference on the
> - * object if it is not in the process of being destroyed.
> - */
> - obj = container_of(it, struct i915_mmu_object, it)->obj;
> - if (!kref_get_unless_zero(&obj->base.refcount)) {
> - it = interval_tree_iter_next(it, range->start, end);
> - continue;
> - }
> - spin_unlock(&mn->lock);
> + mmu_interval_set_seq(mni, cur_seq);
>
> - ret = i915_gem_object_unbind(obj,
> - I915_GEM_OBJECT_UNBIND_ACTIVE |
> - I915_GEM_OBJECT_UNBIND_BARRIER);
> - if (ret == 0)
> - ret = __i915_gem_object_put_pages(obj);
> - i915_gem_object_put(obj);
> - if (ret)
> - return ret;
> + spin_unlock(&i915->mm.notifier_lock);
>
> - spin_lock(&mn->lock);
> + /* During exit there's no need to wait */
> + if (current->flags & PF_EXITING)
> + return true;
Did we ever find out why this is needed, that is why the old userptr
invalidation called doesn't hang here in a similar way?
/Thomas
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2021-01-18 11:30 UTC|newest]
Thread overview: 92+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-05 15:34 [Intel-gfx] [PATCH v6 00/64] drm/i915: Remove obj->mm.lock! Maarten Lankhorst
2021-01-05 15:34 ` [Intel-gfx] [PATCH v6 01/64] drm/i915: Do not share hwsp across contexts any more, v6 Maarten Lankhorst
2021-01-14 12:06 ` Tvrtko Ursulin
2021-01-18 16:09 ` Maarten Lankhorst
2021-01-05 15:34 ` [Intel-gfx] [PATCH v6 02/64] drm/i915: Pin timeline map after first timeline pin, v3 Maarten Lankhorst
2021-01-18 10:28 ` Thomas Hellström (Intel)
2021-01-05 15:34 ` [Intel-gfx] [PATCH v6 03/64] drm/i915: Move cmd parser pinning to execbuffer Maarten Lankhorst
2021-01-05 15:34 ` [Intel-gfx] [PATCH v6 04/64] drm/i915: Add missing -EDEADLK handling to execbuf pinning, v2 Maarten Lankhorst
2021-01-05 15:34 ` [Intel-gfx] [PATCH v6 05/64] drm/i915: Ensure we hold the object mutex in pin correctly Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 06/64] drm/i915: Add gem object locking to madvise Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 07/64] drm/i915: Move HAS_STRUCT_PAGE to obj->flags Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 08/64] drm/i915: Rework struct phys attachment handling Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 09/64] drm/i915: Convert i915_gem_object_attach_phys() to ww locking, v2 Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 10/64] drm/i915: make lockdep slightly happier about execbuf Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 11/64] drm/i915: Disable userptr pread/pwrite support Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 12/64] drm/i915: No longer allow exporting userptr through dma-buf Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 13/64] drm/i915: Reject more ioctls for userptr Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 14/64] drm/i915: Reject UNSYNCHRONIZED for userptr, v2 Maarten Lankhorst
2021-01-11 20:50 ` Dave Airlie
2021-01-18 10:34 ` Thomas Hellström
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 15/64] drm/i915: Make compilation of userptr code depend on MMU_NOTIFIER Maarten Lankhorst
2021-01-18 11:17 ` Thomas Hellström
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 16/64] drm/i915: Fix userptr so we do not have to worry about obj->mm.lock, v5 Maarten Lankhorst
2021-01-11 20:51 ` Dave Airlie
2021-01-18 11:30 ` Thomas Hellström (Intel) [this message]
2021-01-18 12:43 ` Maarten Lankhorst
2021-01-18 12:55 ` Thomas Hellström (Intel)
2021-01-18 14:43 ` Maarten Lankhorst
2021-01-20 13:32 ` Thomas Hellström (Intel)
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 17/64] drm/i915: Flatten obj->mm.lock Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 18/64] drm/i915: Populate logical context during first pin Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 19/64] drm/i915: Make ring submission compatible with obj->mm.lock removal, v2 Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 20/64] drm/i915: Handle ww locking in init_status_page Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 21/64] drm/i915: Rework clflush to work correctly without obj->mm.lock Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 22/64] drm/i915: Pass ww ctx to intel_pin_to_display_plane Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 23/64] drm/i915: Add object locking to vm_fault_cpu Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 24/64] drm/i915: Move pinning to inside engine_wa_list_verify() Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 25/64] drm/i915: Take reservation lock around i915_vma_pin Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 26/64] drm/i915: Make lrc_init_wa_ctx compatible with ww locking Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 27/64] drm/i915: Make __engine_unpark() " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 28/64] drm/i915: Take obj lock around set_domain ioctl Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 29/64] drm/i915: Defer pin calls in buffer pool until first use by caller Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 30/64] drm/i915: Fix pread/pwrite to work with new locking rules Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 31/64] drm/i915: Fix workarounds selftest, part 1 Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 32/64] drm/i915: Prepare for obj->mm.lock removal Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 33/64] drm/i915: Add igt_spinner_pin() to allow for ww locking around spinner Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 34/64] drm/i915: Add ww locking around vm_access() Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 35/64] drm/i915: Increase ww locking for perf Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 36/64] drm/i915: Lock ww in ucode objects correctly Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 37/64] drm/i915: Add ww locking to dma-buf ops Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 38/64] drm/i915: Add missing ww lock in intel_dsb_prepare Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 39/64] drm/i915: Fix ww locking in shmem_create_from_object Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 40/64] drm/i915: Use a single page table lock for each gtt Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 41/64] drm/i915/selftests: Prepare huge_pages testcases for obj->mm.lock removal Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 42/64] drm/i915/selftests: Prepare client blit " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 43/64] drm/i915/selftests: Prepare coherency tests " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 44/64] drm/i915/selftests: Prepare context " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 45/64] drm/i915/selftests: Prepare dma-buf " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 46/64] drm/i915/selftests: Prepare execbuf " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 47/64] drm/i915/selftests: Prepare mman testcases " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 48/64] drm/i915/selftests: Prepare object tests " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 49/64] drm/i915/selftests: Prepare object blit " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 50/64] drm/i915/selftests: Prepare igt_gem_utils " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 51/64] drm/i915/selftests: Prepare context selftest " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 52/64] drm/i915/selftests: Prepare hangcheck " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 53/64] drm/i915/selftests: Prepare execlists and lrc selftests " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 54/64] drm/i915/selftests: Prepare mocs tests " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 55/64] drm/i915/selftests: Prepare ring submission " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 56/64] drm/i915/selftests: Prepare timeline tests " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 57/64] drm/i915/selftests: Prepare i915_request " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 58/64] drm/i915/selftests: Prepare memory region " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 59/64] drm/i915/selftests: Prepare cs engine " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 60/64] drm/i915/selftests: Prepare gtt " Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 61/64] drm/i915: Finally remove obj->mm.lock Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 62/64] drm/i915: Keep userpointer bindings if seqcount is unchanged, v2 Maarten Lankhorst
2021-01-18 10:58 ` Thomas Hellström
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 63/64] drm/i915: Move gt_revoke() slightly Maarten Lankhorst
2021-01-18 11:11 ` Thomas Hellström
2021-01-18 12:01 ` Maarten Lankhorst
2021-01-18 13:22 ` Thomas Hellström
2021-01-18 13:28 ` Thomas Hellström
2021-01-18 14:46 ` Maarten Lankhorst
2021-01-18 15:05 ` Thomas Hellström
2021-01-18 15:32 ` Maarten Lankhorst
2021-01-05 15:35 ` [Intel-gfx] [PATCH v6 64/64] drm/i915: Avoid some false positives in assert_object_held() Maarten Lankhorst
2021-01-05 17:24 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915: Remove obj->mm.lock! (rev12) Patchwork
2021-01-05 17:26 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2021-01-05 17:29 ` [Intel-gfx] ✗ Fi.CI.DOCS: " Patchwork
2021-01-07 15:42 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915: Remove obj->mm.lock! (rev13) Patchwork
2021-01-07 15:44 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2021-01-07 15:47 ` [Intel-gfx] ✗ Fi.CI.DOCS: " Patchwork
2021-01-07 16:11 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork
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=5b38d4b5-ded7-9f4d-118e-74f706c6a8ad@shipmail.org \
--to=thomas_os@shipmail.org \
--cc=intel-gfx@lists.freedesktop.org \
--cc=maarten.lankhorst@linux.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: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).