From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
To: Chris Wilson <chris@chris-wilson.co.uk>, intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 01/17] drm/i915: Skip holding an object reference for execbuf preparation
Date: Mon, 22 Aug 2016 14:21:16 +0300 [thread overview]
Message-ID: <1471864876.3601.10.camel@linux.intel.com> (raw)
In-Reply-To: <20160822080350.4964-2-chris@chris-wilson.co.uk>
On ma, 2016-08-22 at 09:03 +0100, Chris Wilson wrote:
> This is a golden oldie! We can shave a couple of locked instructions for
> about 10% of the per-object overhead by not taking an extra kref whilst
> reserving objects for an execbuf. Due to lock management this is safe,
> as we cannot lose the original object reference without the lock.
> Equally, because this relies on the heavy BKL^W struct_mutex, it is also
> likely to be only a temporary optimisation until we have fine grained
> locking. (That's what we said 5 years ago, so there's probably another
> 10 years before we get around to finer grained locking!)
>
Should we sprinkle a couple of lockdep_assert_held for documentation?
Regards, Joonas
--
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
next prev parent reply other threads:[~2016-08-22 11:21 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-08-22 8:03 Execbuf fixes and major tuning Chris Wilson
2016-08-22 8:03 ` [PATCH 01/17] drm/i915: Skip holding an object reference for execbuf preparation Chris Wilson
2016-08-22 11:21 ` Joonas Lahtinen [this message]
2016-08-22 11:56 ` Chris Wilson
2016-08-22 8:03 ` [PATCH 02/17] drm/i915: Defer active reference until required Chris Wilson
2016-08-22 8:03 ` [PATCH 03/17] drm/i915: Allow the user to pass a context to any ring Chris Wilson
2016-08-22 11:23 ` Joonas Lahtinen
2016-08-22 12:23 ` Chris Wilson
2016-08-23 13:28 ` John Harrison
2016-08-23 13:33 ` John Harrison
2016-08-25 15:28 ` John Harrison
2016-08-29 12:21 ` Joonas Lahtinen
2016-08-22 8:03 ` [PATCH 04/17] drm/i915: Fix i915_gem_evict_for_vma (soft-pinning) Chris Wilson
2016-08-22 8:03 ` [PATCH 05/17] drm/i915: Pin the pages whilst operating on them Chris Wilson
2016-08-23 11:54 ` Joonas Lahtinen
2016-08-22 8:03 ` [PATCH 06/17] drm/i915: Move obj->dirty:1 to obj->flags Chris Wilson
2016-08-23 12:01 ` Joonas Lahtinen
2016-09-06 11:37 ` Dave Gordon
2016-09-06 13:16 ` Chris Wilson
2016-08-22 8:03 ` [PATCH 07/17] drm/i915: Use the precomputed value for whether to enable command parsing Chris Wilson
2016-08-24 14:33 ` John Harrison
2016-08-27 9:12 ` Chris Wilson
2016-08-22 8:03 ` [PATCH 08/17] drm/i915: Drop spinlocks around adding to the client request list Chris Wilson
2016-08-24 13:16 ` Mika Kuoppala
2016-08-24 13:25 ` Chris Wilson
2016-08-26 9:13 ` Mika Kuoppala
2016-09-02 10:30 ` John Harrison
2016-09-02 10:59 ` Chris Wilson
2016-09-02 11:02 ` Chris Wilson
2016-09-02 13:20 ` John Harrison
2016-09-02 13:38 ` Chris Wilson
2016-08-22 8:03 ` [PATCH 09/17] drm/i915: Amalgamate execbuffer parameter structures Chris Wilson
2016-08-24 13:20 ` John Harrison
2016-08-22 8:03 ` [PATCH 10/17] drm/i915: Use vma->exec_entry as our double-entry placeholder Chris Wilson
2016-08-22 8:03 ` [PATCH 11/17] drm/i915: Store a direct lookup from object handle to vma Chris Wilson
2016-08-22 8:03 ` [PATCH 12/17] drm/i915: Pass vma to relocate entry Chris Wilson
2016-08-22 8:03 ` [PATCH 13/17] drm/i915: Eliminate lots of iterations over the execobjects array Chris Wilson
2016-08-25 6:03 ` [PATCH v2] " Chris Wilson
2016-08-22 8:03 ` [PATCH 14/17] drm/i915: First try the previous execbuffer location Chris Wilson
2016-08-22 8:03 ` [PATCH 15/17] drm/i915: Wait upon userptr get-user-pages within execbuffer Chris Wilson
2016-08-23 10:53 ` Joonas Lahtinen
2016-08-23 11:14 ` Chris Wilson
2016-08-22 8:03 ` [PATCH 16/17] drm/i915: Remove superfluous i915_add_request_no_flush() helper Chris Wilson
2016-08-23 10:21 ` Joonas Lahtinen
2016-08-22 8:03 ` [PATCH 17/17] drm/i915: Use the MRU stack search after evicting Chris Wilson
2016-08-23 13:52 ` ✗ Fi.CI.BAT: failure for series starting with [01/17] drm/i915: Skip holding an object reference for execbuf preparation Patchwork
2016-08-25 6:51 ` ✗ Fi.CI.BAT: warning for series starting with [01/17] drm/i915: Skip holding an object reference for execbuf preparation (rev2) 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=1471864876.3601.10.camel@linux.intel.com \
--to=joonas.lahtinen@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