From: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
To: Chris Wilson <chris@chris-wilson.co.uk>, intel-gfx@lists.freedesktop.org
Subject: Re: [Intel-gfx] [PATCH 08/66] drm/i915: Make the stale cached active node available for any timeline
Date: Wed, 29 Jul 2020 16:31:40 +0100 [thread overview]
Message-ID: <63a4b94e-7730-d56d-12ba-178a07e93250@linux.intel.com> (raw)
In-Reply-To: <159603435318.25916.16947315566123739407@build.alporthouse.com>
On 29/07/2020 15:52, Chris Wilson wrote:
> Quoting Tvrtko Ursulin (2020-07-29 15:22:26)
>>
>> On 29/07/2020 14:42, Chris Wilson wrote:
>>> Quoting Tvrtko Ursulin (2020-07-29 13:40:38)
>>>>
>>>> On 28/07/2020 15:28, Chris Wilson wrote:
>>>>> Quoting Tvrtko Ursulin (2020-07-17 14:04:58)
>>>>>>
>>>>>> On 15/07/2020 12:50, Chris Wilson wrote:
>>>>>>> @@ -235,9 +239,22 @@ static struct active_node *__active_lookup(struct i915_active *ref, u64 idx)
>>>>>>> {
>>>>>>> struct active_node *it;
>>>>>>>
>>>>>>> + GEM_BUG_ON(idx == 0); /* 0 is the unordered timeline, rsvd for cache */
>>>>>>> +
>>>>>>> it = READ_ONCE(ref->cache);
>>>>>>> - if (it && it->timeline == idx)
>>>>>>> - return it;
>>>>>>> + if (it) {
>>>>>>> + u64 cached = READ_ONCE(it->timeline);
>>>>>>> +
>>>>>>> + if (cached == idx)
>>>>>>> + return it;
>>>>>>> +
>>>>>>> +#ifdef CONFIG_64BIT /* for cmpxchg(u64) */
>>>>>>> + if (!cached && !cmpxchg(&it->timeline, 0, idx)) {
>>>>>>> + GEM_BUG_ON(i915_active_fence_isset(&it->base));
>>>>>>> + return it;
>>>>>>
>>>>>> cpmxchg suggests this needs to be atomic, however above the check for
>>>>>> equality comes from a separate read.
>>>>>
>>>>> That's fine, and quite common to avoid cmpxchg if the current value
>>>>> already does not match the expected condition.
>>>>
>>>> How? What is another thread is about to install its idx into
>>>> it->timeline with cmpxchg and this thread does not see it because it
>>>> just returned on the "cached == idx" condition.
>>>
>>> Because it's nonzero.
>>>
>>> If the idx is already non-zero, it will always remain non-zero until
>>> everybody idles (and there are no more threads).
>>>
>>> If the idx is zero, it can only transition to non-zero once, atomically
>>> via cmpxchg. The first and only first cmpxchg will return that the
>>> previous value was 0, and so return with it->idx == idx.
>>
>> I think this is worthy of a comment to avoid future reader having to
>> re-figure it all out.
>
> if (it) {
> u64 cached = READ_ONCE(it->timeline);
>
> + /* Once claimed, this slot will only belong to this idx */
> if (cached == idx)
> return it;
>
> #ifdef CONFIG_64BIT /* for cmpxchg(u64) */
> + /*
> + * An unclaimed cache [.timeline=0] can only be claimed once.
> + *
> + * If the value is already non-zero, some other thread has
> + * claimed the cache and we know that is does not match our
> + * idx. If, and only if, the timeline is currently zero is it
> + * worth competing to claim it atomically for ourselves (for
> + * only the winner of that race will cmpxchg return the old
> + * value of 0).
> + */
> if (!cached && !cmpxchg(&it->timeline, 0, idx))
> return it;
> #endif
>
Sounds good. With that:
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2020-07-29 15:31 UTC|newest]
Thread overview: 154+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-15 11:50 [Intel-gfx] [PATCH 01/66] drm/i915: Reduce i915_request.lock contention for i915_request_wait Chris Wilson
2020-07-15 11:50 ` [Intel-gfx] [PATCH 02/66] drm/i915: Remove i915_request.lock requirement for execution callbacks Chris Wilson
2020-07-15 11:50 ` [Intel-gfx] [PATCH 03/66] drm/i915: Remove requirement for holding i915_request.lock for breadcrumbs Chris Wilson
2020-07-15 11:50 ` [Intel-gfx] [PATCH 04/66] drm/i915: Add a couple of missing i915_active_fini() Chris Wilson
2020-07-17 12:00 ` Tvrtko Ursulin
2020-07-21 12:23 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 05/66] drm/i915: Skip taking acquire mutex for no ref->active callback Chris Wilson
2020-07-17 12:04 ` Tvrtko Ursulin
2020-07-21 12:32 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 06/66] drm/i915: Export a preallocate variant of i915_active_acquire() Chris Wilson
2020-07-17 12:21 ` Tvrtko Ursulin
2020-07-17 12:45 ` Chris Wilson
2020-07-17 13:06 ` Tvrtko Ursulin
2020-07-21 15:33 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 07/66] drm/i915: Keep the most recently used active-fence upon discard Chris Wilson
2020-07-17 12:38 ` Tvrtko Ursulin
2020-07-28 14:22 ` Chris Wilson
2020-07-22 9:46 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 08/66] drm/i915: Make the stale cached active node available for any timeline Chris Wilson
2020-07-17 13:04 ` Tvrtko Ursulin
2020-07-28 14:28 ` Chris Wilson
2020-07-29 12:40 ` Tvrtko Ursulin
2020-07-29 13:42 ` Chris Wilson
2020-07-29 13:53 ` Chris Wilson
2020-07-29 14:22 ` Tvrtko Ursulin
2020-07-29 14:39 ` Chris Wilson
2020-07-29 14:52 ` Chris Wilson
2020-07-29 15:31 ` Tvrtko Ursulin [this message]
2020-07-22 11:19 ` Thomas Hellström (Intel)
2020-07-28 14:31 ` Chris Wilson
2020-07-15 11:50 ` [Intel-gfx] [PATCH 09/66] drm/i915: Provide a fastpath for waiting on vma bindings Chris Wilson
2020-07-17 13:23 ` Tvrtko Ursulin
2020-07-28 14:35 ` Chris Wilson
2020-07-29 12:43 ` Tvrtko Ursulin
2020-07-22 15:07 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 10/66] drm/i915: Soften the tasklet flush frequency before waits Chris Wilson
2020-07-16 14:23 ` Mika Kuoppala
2020-07-22 15:10 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 11/66] drm/i915: Preallocate stashes for vma page-directories Chris Wilson
2020-07-20 10:35 ` Matthew Auld
2020-07-23 14:33 ` Thomas Hellström (Intel)
2020-07-28 14:42 ` Chris Wilson
2020-07-31 7:43 ` Thomas Hellström (Intel)
2020-07-27 9:24 ` Thomas Hellström (Intel)
2020-07-28 14:50 ` Chris Wilson
2020-07-30 12:04 ` Thomas Hellström (Intel)
2020-07-30 12:28 ` Thomas Hellström (Intel)
2020-08-04 14:08 ` Chris Wilson
2020-08-04 16:14 ` Daniel Vetter
2020-07-15 11:50 ` [Intel-gfx] [PATCH 12/66] drm/i915: Switch to object allocations for page directories Chris Wilson
2020-07-20 10:34 ` Matthew Auld
2020-07-20 10:40 ` Chris Wilson
2020-07-15 11:50 ` [Intel-gfx] [PATCH 13/66] drm/i915/gem: Don't drop the timeline lock during execbuf Chris Wilson
2020-07-23 16:09 ` Thomas Hellström (Intel)
2020-07-28 14:46 ` Thomas Hellström (Intel)
2020-07-28 14:51 ` Chris Wilson
2020-07-31 8:09 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 14/66] drm/i915/gem: Rename execbuf.bind_link to unbound_link Chris Wilson
2020-07-31 8:11 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 15/66] drm/i915/gem: Break apart the early i915_vma_pin from execbuf object lookup Chris Wilson
2020-07-31 8:51 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 16/66] drm/i915/gem: Remove the call for no-evict i915_vma_pin Chris Wilson
2020-07-17 14:36 ` Tvrtko Ursulin
2020-07-28 15:04 ` Chris Wilson
2020-07-28 9:46 ` Thomas Hellström (Intel)
2020-07-28 15:05 ` Chris Wilson
2020-07-31 8:58 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 17/66] drm/i915: Add list_for_each_entry_safe_continue_reverse Chris Wilson
2020-07-31 8:59 ` Thomas Hellström (Intel)
2020-07-15 11:50 ` [Intel-gfx] [PATCH 18/66] drm/i915: Always defer fenced work to the worker Chris Wilson
2020-07-31 9:03 ` Thomas Hellström (Intel)
2020-07-31 13:28 ` Chris Wilson
2020-07-31 13:31 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 19/66] drm/i915/gem: Assign context id for async work Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 20/66] drm/i915/gem: Separate the ww_mutex walker into its own list Chris Wilson
2020-07-31 9:23 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 21/66] drm/i915/gem: Asynchronous GTT unbinding Chris Wilson
2020-07-31 13:09 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 22/66] drm/i915/gem: Bind the fence async for execbuf Chris Wilson
2020-07-27 18:19 ` Thomas Hellström (Intel)
2020-07-28 15:08 ` Chris Wilson
2020-07-31 13:12 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 23/66] drm/i915/gem: Include cmdparser in common execbuf pinning Chris Wilson
2020-07-31 9:43 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 24/66] drm/i915/gem: Include secure batch " Chris Wilson
2020-07-31 9:47 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 25/66] drm/i915/gem: Reintroduce multiple passes for reloc processing Chris Wilson
2020-07-31 10:05 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 26/66] drm/i915: Add an implementation for i915_gem_ww_ctx locking, v2 Chris Wilson
2020-07-31 10:07 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 27/66] drm/i915/gem: Pull execbuf dma resv under a single critical section Chris Wilson
2020-07-27 18:08 ` Thomas Hellström (Intel)
2020-07-28 15:16 ` Chris Wilson
2020-07-30 12:57 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 28/66] drm/i915/gem: Replace i915_gem_object.mm.mutex with reservation_ww_class Chris Wilson
2020-07-15 15:43 ` Maarten Lankhorst
2020-07-16 15:53 ` Tvrtko Ursulin
2020-07-28 11:17 ` Thomas Hellström (Intel)
2020-07-29 7:56 ` Thomas Hellström (Intel)
2020-07-29 12:17 ` Tvrtko Ursulin
2020-07-29 13:44 ` Thomas Hellström (Intel)
2020-08-05 12:12 ` Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 29/66] drm/i915: Hold wakeref for the duration of the vma GGTT binding Chris Wilson
2020-07-31 10:09 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 30/66] drm/i915: Specialise " Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 31/66] drm/i915/gt: Acquire backing storage for the context Chris Wilson
2020-07-31 10:27 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 32/66] drm/i915/gt: Push the wait for the context to bound to the request Chris Wilson
2020-07-31 10:48 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 33/66] drm/i915: Remove unused i915_gem_evict_vm() Chris Wilson
2020-07-31 10:51 ` Thomas Hellström (Intel)
2020-07-15 11:51 ` [Intel-gfx] [PATCH 34/66] drm/i915/gt: Decouple completed requests on unwind Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 35/66] drm/i915/gt: Check for a completed last request once Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 36/66] drm/i915/gt: Replace direct submit with direct call to tasklet Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 37/66] drm/i915/gt: Free stale request on destroying the virtual engine Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 38/66] drm/i915/gt: Use virtual_engine during execlists_dequeue Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 39/66] drm/i915/gt: Decouple inflight virtual engines Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 40/66] drm/i915/gt: Defer schedule_out until after the next dequeue Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 41/66] drm/i915/gt: Resubmit the virtual engine on schedule-out Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 42/66] drm/i915/gt: Simplify virtual engine handling for execlists_hold() Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 43/66] drm/i915/gt: ce->inflight updates are now serialised Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 44/66] drm/i915/gt: Drop atomic for engine->fw_active tracking Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 45/66] drm/i915/gt: Extract busy-stats for ring-scheduler Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 46/66] drm/i915/gt: Convert stats.active to plain unsigned int Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 47/66] drm/i915: Lift waiter/signaler iterators Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 48/66] drm/i915: Strip out internal priorities Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 49/66] drm/i915: Remove I915_USER_PRIORITY_SHIFT Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 50/66] drm/i915: Replace engine->schedule() with a known request operation Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 51/66] drm/i915/gt: Do not suspend bonded requests if one hangs Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 52/66] drm/i915: Teach the i915_dependency to use a double-lock Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 53/66] drm/i915: Restructure priority inheritance Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 54/66] drm/i915/gt: Remove timeslice suppression Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 55/66] drm/i915: Fair low-latency scheduling Chris Wilson
2020-07-15 15:33 ` [Intel-gfx] [PATCH] " Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 56/66] drm/i915/gt: Specify a deadline for the heartbeat Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 57/66] drm/i915: Replace the priority boosting for the display with a deadline Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 58/66] drm/i915: Move saturated workload detection to the GT Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 59/66] Restore "drm/i915: drop engine_pin/unpin_breadcrumbs_irq" Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 60/66] drm/i915/gt: Couple tasklet scheduling for all CS interrupts Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 61/66] drm/i915/gt: Support creation of 'internal' rings Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 62/66] drm/i915/gt: Use client timeline address for seqno writes Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 63/66] drm/i915/gt: Infrastructure for ring scheduling Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 64/66] drm/i915/gt: Implement ring scheduler for gen6/7 Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 65/66] drm/i915/gt: Enable ring scheduling " Chris Wilson
2020-07-15 11:51 ` [Intel-gfx] [PATCH 66/66] drm/i915/gem: Remove timeline nesting from snb relocs Chris Wilson
2020-07-15 13:27 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for series starting with [01/66] drm/i915: Reduce i915_request.lock contention for i915_request_wait Patchwork
2020-07-15 13:28 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2020-07-15 14:20 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork
2020-07-15 15:41 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for series starting with [01/66] drm/i915: Reduce i915_request.lock contention for i915_request_wait (rev2) Patchwork
2020-07-15 15:42 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2020-07-15 16:03 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2020-07-15 19:55 ` [Intel-gfx] ✗ Fi.CI.IGT: failure " Patchwork
2020-07-23 20:32 ` [Intel-gfx] [PATCH 01/66] drm/i915: Reduce i915_request.lock contention for i915_request_wait Dave Airlie
2020-07-27 9:35 ` Tvrtko Ursulin
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=63a4b94e-7730-d56d-12ba-178a07e93250@linux.intel.com \
--to=tvrtko.ursulin@linux.intel.com \
--cc=chris@chris-wilson.co.uk \
--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;
as well as URLs for NNTP newsgroup(s).