From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
To: intel-gfx@lists.freedesktop.org
Subject: [Intel-gfx] [PATCH 61/61] drm/i915: Keep userpointer bindings if seqcount is unchanged
Date: Fri, 2 Oct 2020 14:59:39 +0200 [thread overview]
Message-ID: <20201002125939.50817-62-maarten.lankhorst@linux.intel.com> (raw)
In-Reply-To: <20201002125939.50817-1-maarten.lankhorst@linux.intel.com>
Instead of force unbinding and rebinding every time, we try to check
if our notifier seqcount is still correct when pages are bound. This
way we only rebind userptr when we need to, and prevent stalls.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
---
drivers/gpu/drm/i915/gem/i915_gem_userptr.c | 29 ++++++++++++++++++---
1 file changed, 26 insertions(+), 3 deletions(-)
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
index 327f01b04f21..dad68de2159f 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
@@ -268,12 +268,35 @@ int i915_gem_object_userptr_submit_init(struct drm_i915_gem_object *obj)
if (ret)
return ret;
- /* Make sure userptr is unbound for next attempt, so we don't use stale pages. */
- ret = i915_gem_object_userptr_unbind(obj, false);
+ /* optimistically try to preserve current pages while unlocked */
+ if (i915_gem_object_has_pages(obj) &&
+ !mmu_interval_check_retry(&obj->userptr.notifier,
+ obj->userptr.notifier_seq)) {
+ ret = mutex_lock_interruptible(&i915->mm.notifier_lock);
+ if (!ret) {
+ if (obj->userptr.pvec &&
+ !mmu_interval_read_retry(&obj->userptr.notifier,
+ obj->userptr.notifier_seq)) {
+ obj->userptr.page_ref++;
+ mutex_unlock(&i915->mm.notifier_lock);
+
+ /* We can keep using the current binding, this is the fastpath */
+ ret = 1;
+ }
+ }
+ }
+
+ if (!ret) {
+ /* Make sure userptr is unbound for next attempt, so we don't use stale pages. */
+ ret = i915_gem_object_userptr_unbind(obj, false);
+ }
i915_gem_object_unlock(obj);
- if (ret)
+ if (ret < 0)
return ret;
+ if (ret > 0)
+ return 0;
+
notifier_seq = mmu_interval_read_begin(&obj->userptr.notifier);
pvec = kvmalloc_array(num_pages, sizeof(struct page *), GFP_KERNEL);
--
2.28.0
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2020-10-02 13:00 UTC|newest]
Thread overview: 68+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-02 12:58 [Intel-gfx] [PATCH 00/61] drm/i915: Remove obj->mm.lock! Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 01/61] drm/i915: Move cmd parser pinning to execbuffer Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 02/61] drm/i915: Add missing -EDEADLK handling to execbuf pinning Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 03/61] drm/i915: Do not share hwsp across contexts any more, v2 Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 04/61] drm/i915: Ensure we hold the object mutex in pin correctly Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 05/61] drm/i915: Add gem object locking to madvise Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 06/61] drm/i915: Move HAS_STRUCT_PAGE to obj->flags Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 07/61] drm/i915: Rework struct phys attachment handling Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 08/61] drm/i915: Convert i915_gem_object_attach_phys() to ww locking Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 09/61] drm/i915: make lockdep slightly happier about execbuf Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 10/61] drm/i915: Disable userptr pread/pwrite support Maarten Lankhorst
2020-10-02 20:14 ` Ruhl, Michael J
2020-10-12 14:13 ` Maarten Lankhorst
2020-10-19 17:54 ` Ruhl, Michael J
2020-10-02 12:58 ` [Intel-gfx] [PATCH 11/61] drm/i915: No longer allow exporting userptr through dma-buf Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 12/61] drm/i915: Reject more ioctls for userptr Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 13/61] drm/i915: Reject UNSYNCHRONIZED " Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 14/61] drm/i915: Fix userptr so we do not have to worry about obj->mm.lock Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 15/61] drm/i915: Flatten obj->mm.lock Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 16/61] drm/i915: Pin timeline map after first timeline pin Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 17/61] drm/i915: Populate logical context during first pin Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 18/61] drm/i915: Make ring submission compatible with obj->mm.lock removal Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 19/61] drm/i915: Handle ww locking in init_status_page Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 20/61] drm/i915: Rework clflush to work correctly without obj->mm.lock Maarten Lankhorst
2020-10-02 12:58 ` [Intel-gfx] [PATCH 21/61] drm/i915: Pass ww ctx to intel_pin_to_display_plane Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 22/61] drm/i915: Add object locking to vm_fault_cpu Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 23/61] drm/i915: Move pinning to inside engine_wa_list_verify() Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 24/61] drm/i915: Take reservation lock around i915_vma_pin Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 25/61] drm/i915: Make intel_init_workaround_bb more compatible with ww locking Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 26/61] drm/i915: Make __engine_unpark() " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 27/61] drm/i915: Take obj lock around set_domain ioctl Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 28/61] drm/i915: Defer pin calls in buffer pool until first use by caller Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 29/61] drm/i915: Fix pread/pwrite to work with new locking rules Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 30/61] drm/i915: Fix workarounds selftest, part 1 Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 31/61] drm/i915: Prepare for obj->mm.lock removal Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 32/61] drm/i915: Add igt_spinner_pin() to allow for ww locking around spinner Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 33/61] drm/i915: Add ww locking around vm_access() Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 34/61] drm/i915: Increase ww locking for perf Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 35/61] drm/i915: Lock ww in ucode objects correctly Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 36/61] drm/i915: Add ww locking to dma-buf ops Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 37/61] drm/i915: Add missing ww lock in intel_dsb_prepare Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 38/61] drm/i915: Fix ww locking in shmem_create_from_object Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 39/61] drm/i915: Use a single page table lock for each gtt Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 40/61] drm/i915/selftests: Prepare huge_pages testcases for obj->mm.lock removal Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 41/61] drm/i915/selftests: Prepare client blit " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 42/61] drm/i915/selftests: Prepare coherency tests " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 43/61] drm/i915/selftests: Prepare context " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 44/61] drm/i915/selftests: Prepare dma-buf " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 45/61] drm/i915/selftests: Prepare execbuf " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 46/61] drm/i915/selftests: Prepare mman testcases " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 47/61] drm/i915/selftests: Prepare object tests " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 48/61] drm/i915/selftests: Prepare object blit " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 49/61] drm/i915/selftests: Prepare igt_gem_utils " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 50/61] drm/i915/selftests: Prepare context selftest " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 51/61] drm/i915/selftests: Prepare hangcheck " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 52/61] drm/i915/selftests: Prepare execlists " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 53/61] drm/i915/selftests: Prepare mocs tests " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 54/61] drm/i915/selftests: Prepare ring submission " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 55/61] drm/i915/selftests: Prepare timeline tests " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 56/61] drm/i915/selftests: Prepare i915_request " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 57/61] drm/i915/selftests: Prepare memory region " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 58/61] drm/i915/selftests: Prepare cs engine " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 59/61] drm/i915/selftests: Prepare gtt " Maarten Lankhorst
2020-10-02 12:59 ` [Intel-gfx] [PATCH 60/61] drm/i915: Finally remove obj->mm.lock Maarten Lankhorst
2020-10-02 12:59 ` Maarten Lankhorst [this message]
2020-10-02 13:53 ` [Intel-gfx] ✓ Fi.CI.BAT: success for drm/i915: Remove obj->mm.lock! Patchwork
2020-10-02 18:01 ` [Intel-gfx] ✗ Fi.CI.IGT: failure " Patchwork
2020-10-02 18:38 ` [Intel-gfx] [PATCH 00/61] " Chris Wilson
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=20201002125939.50817-62-maarten.lankhorst@linux.intel.com \
--to=maarten.lankhorst@linux.intel.com \
--cc=intel-gfx@lists.freedesktop.org \
/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