From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
To: Chris Wilson <chris@chris-wilson.co.uk>, intel-gfx@lists.freedesktop.org
Subject: Re: [PATCH 06/16] drm/i915: Move dev_priv->mm.[un]bound_list to its own lock
Date: Thu, 10 Aug 2017 19:22:49 +0300 [thread overview]
Message-ID: <1502382169.4310.50.camel@linux.intel.com> (raw)
In-Reply-To: <20170726132610.3336-7-chris@chris-wilson.co.uk>
On ke, 2017-07-26 at 14:26 +0100, Chris Wilson wrote:
> Remove the struct_mutex requirement around dev_priv->mm.bound_list and
> dev_priv->mm.unbound_list by giving it its own spinlock. This reduces
> one more requirement for struct_mutex and in the process gives us
> slightly more accurate unbound_list tracking, which should improve the
> shrinker.
Please mention the global_link -> mm.link rename too.
>
> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
<SNIP>
> +++ b/drivers/gpu/drm/i915/i915_drv.h
> @@ -1443,6 +1443,8 @@ struct i915_gem_mm {
> * always the inner lock when overlapping with struct_mutex. */
> struct mutex stolen_lock;
Please comment what this lock achieves, here too. Maybe cross-link with
the other doc?
> + spinlock_t obj_lock;
> +
> /** List of all objects in gtt_space. Used to restore gtt
> * mappings on resume */
<SNIP>
> +++ b/drivers/gpu/drm/i915/i915_gem_shrinker.c
> @@ -92,9 +92,6 @@ static bool swap_available(void)
>
> static bool can_release_pages(struct drm_i915_gem_object *obj)
> {
> - if (!i915_gem_object_has_pages(obj))
> - return false;
So you think the inaccuracies we get for being lockless don't matter
compared to better forward progress?
Maybe the cons should be documented in the commit message to help in
bisecting.
> @@ -149,8 +149,6 @@ static int igt_overcommit(void *arg)
> goto cleanup;
> }
>
> - list_move(&obj->global_link, &i915->mm.unbound_list);
Accidentally dropped line?
> -
> vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, 0);
> if (!IS_ERR(vma) || PTR_ERR(vma) != -ENOSPC) {
> pr_err("Failed to evict+insert, i915_gem_object_ggtt_pin returned err=%d\n", (int)PTR_ERR(vma));
--
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:[~2017-08-10 16:22 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-07-26 13:25 More tweaks Chris Wilson
2017-07-26 13:25 ` [PATCH 01/16] drm/i915: Keep a small stash of preallocated WC pages Chris Wilson
2017-08-07 20:15 ` Matthew Auld
2017-08-11 7:34 ` Joonas Lahtinen
2017-08-11 11:55 ` Chris Wilson
2017-08-11 9:51 ` Jani Nikula
2017-07-26 13:25 ` [PATCH 02/16] drm/i915: Pin fence for iomap Chris Wilson
2017-07-26 13:25 ` [PATCH 03/16] drm/i915: Consolidate get_fence with pin_fence Chris Wilson
2017-07-26 13:25 ` [PATCH 04/16] drm/i915: Emit pipelined fence changes Chris Wilson
2017-07-26 13:25 ` [PATCH 05/16] drm/i915: Refactor testing obj->mm.pages Chris Wilson
2017-08-10 16:26 ` Joonas Lahtinen
2017-07-26 13:26 ` [PATCH 06/16] drm/i915: Move dev_priv->mm.[un]bound_list to its own lock Chris Wilson
2017-08-10 16:22 ` Joonas Lahtinen [this message]
2017-08-12 10:26 ` Chris Wilson
2017-08-12 11:34 ` Chris Wilson
2017-08-12 11:41 ` Chris Wilson
2017-07-26 13:26 ` [PATCH 07/16] drm/i915: Trim struct_mutex hold duration for i915_gem_free_objects Chris Wilson
2017-08-11 7:36 ` Joonas Lahtinen
2017-07-26 13:26 ` [PATCH 08/16] drm/i915: Only free the oldest stale object before a fresh allocation Chris Wilson
2017-07-26 13:26 ` [PATCH 09/16] drm/i915: Track user GTT faulting per-vma Chris Wilson
2017-07-26 13:26 ` [PATCH 10/16] drm/i915: Try a minimal attempt to insert the whole object for relocations Chris Wilson
2017-07-26 13:26 ` [PATCH 11/16] drm/i915: Check context status before looking up our obj/vma Chris Wilson
2017-07-26 13:26 ` [PATCH 12/16] drm/i915: Convert execbuf to use struct-of-array packing for critical fields Chris Wilson
2017-07-26 13:26 ` [PATCH 13/16] drm/i915: Simplify eb_lookup_vmas() Chris Wilson
2017-07-26 13:26 ` [PATCH 14/16] drm/i915: Replace execbuf vma ht with an idr Chris Wilson
2017-07-26 13:26 ` [PATCH 15/16] drm/i915: Trivial grammar fix s/opt of/opt out of/ in comment Chris Wilson
2017-07-26 13:26 ` [PATCH 16/16] drm/i915: Mark the GT as busy before idling the previous request Chris Wilson
2017-07-26 14:13 ` ✓ Fi.CI.BAT: success for series starting with [01/16] drm/i915: Keep a small stash of preallocated WC pages 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=1502382169.4310.50.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