All of lore.kernel.org
 help / color / mirror / Atom feed
From: Chris Wilson <chris@chris-wilson.co.uk>
To: intel-gfx@lists.freedesktop.org
Subject: [PATCH 10/15] drm/i915: First try the previous execbuffer location
Date: Thu, 16 Mar 2017 13:20:01 +0000	[thread overview]
Message-ID: <20170316132006.7976-11-chris@chris-wilson.co.uk> (raw)
In-Reply-To: <20170316132006.7976-1-chris@chris-wilson.co.uk>

When choosing a slot for an execbuffer, we ideally want to use the same
address as last time (so that we don't have to rebind it) and the same
address as expected by the user (so that we don't have to fixup any
relocations pointing to it). If we first try to bind the incoming
execbuffer->offset from the user, or the currently bound offset that
should hopefully achieve the goal of avoiding the rebind cost and the
relocation penalty. However, if the object is not currently bound there
we don't want to arbitrarily unbind an object in our chosen position and
so choose to rebind/relocate the incoming object instead. After we
report the new position back to the user, on the next pass the
relocations should have settled down.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
---
 drivers/gpu/drm/i915/i915_gem_execbuffer.c | 12 ++++++++----
 drivers/gpu/drm/i915/i915_gem_gtt.c        |  6 ++++++
 drivers/gpu/drm/i915/i915_gem_gtt.h        |  1 +
 3 files changed, 15 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 5ce4f2e47eb9..0f141a5800f2 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -293,10 +293,15 @@ eb_pin_vma(struct i915_execbuffer *eb,
 {
 	u64 flags;
 
-	flags = vma->node.start;
-	flags |= PIN_USER | PIN_NONBLOCK | PIN_OFFSET_FIXED;
+	if (vma->node.size)
+		flags = vma->node.start;
+	else
+		flags = entry->offset & PIN_OFFSET_MASK;
+
+	flags |= PIN_USER | PIN_NOEVICT | PIN_OFFSET_FIXED;
 	if (unlikely(entry->flags & EXEC_OBJECT_NEEDS_GTT))
 		flags |= PIN_GLOBAL;
+
 	if (unlikely(i915_vma_pin(vma, 0, 0, flags)))
 		return;
 
@@ -407,8 +412,7 @@ eb_add_vma(struct i915_execbuffer *eb,
 		entry->flags |= eb->context_flags;
 
 	ret = 0;
-	if (vma->node.size)
-		eb_pin_vma(eb, entry, vma);
+	eb_pin_vma(eb, entry, vma);
 	if (eb_vma_misplaced(entry, vma)) {
 		eb_unreserve_vma(vma, entry);
 
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index cee9c4fec52a..9160733cbb09 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -3288,6 +3288,9 @@ int i915_gem_gtt_reserve(struct i915_address_space *vm,
 	if (err != -ENOSPC)
 		return err;
 
+	if (flags & PIN_NOEVICT)
+		return -ENOSPC;
+
 	err = i915_gem_evict_for_node(vm, node, flags);
 	if (err == 0)
 		err = drm_mm_reserve_node(&vm->mm, node);
@@ -3402,6 +3405,9 @@ int i915_gem_gtt_insert(struct i915_address_space *vm,
 	if (err != -ENOSPC)
 		return err;
 
+	if (flags & PIN_NOEVICT)
+		return -ENOSPC;
+
 	/* No free space, pick a slot at random.
 	 *
 	 * There is a pathological case here using a GTT shared between
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.h b/drivers/gpu/drm/i915/i915_gem_gtt.h
index 948b2e68131a..6be7a5c32392 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.h
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.h
@@ -589,6 +589,7 @@ int i915_gem_gtt_insert(struct i915_address_space *vm,
 #define PIN_MAPPABLE		BIT(1)
 #define PIN_ZONE_4G		BIT(2)
 #define PIN_NONFAULT		BIT(3)
+#define PIN_NOEVICT		BIT(4)
 
 #define PIN_MBZ			BIT(5) /* I915_VMA_PIN_OVERFLOW */
 #define PIN_GLOBAL		BIT(6) /* I915_VMA_GLOBAL_BIND */
-- 
2.11.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

  parent reply	other threads:[~2017-03-16 13:20 UTC|newest]

Thread overview: 39+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-03-16 13:19 Make execbuf fast and green Chris Wilson
2017-03-16 13:19 ` [PATCH 01/15] drm/i915: Copy user requested buffers into the error state Chris Wilson
2017-03-16 13:19 ` [PATCH 02/15] drm/i915: Retire an active batch pool object rather than allocate new Chris Wilson
2017-03-17  8:52   ` Joonas Lahtinen
2017-03-17  9:02     ` Chris Wilson
2017-03-16 13:19 ` [PATCH 03/15] drm/i915: Amalgamate execbuffer parameter structures Chris Wilson
2017-03-17 10:04   ` Joonas Lahtinen
2017-03-16 13:19 ` [PATCH 04/15] drm/i915: Use vma->exec_entry as our double-entry placeholder Chris Wilson
2017-03-16 13:19 ` [PATCH 05/15] drm/i915: Split vma exec_link/evict_link Chris Wilson
2017-03-16 13:19 ` [PATCH 06/15] drm/i915: Stop using obj->obj_exec_link outside of execbuf Chris Wilson
2017-03-16 13:19 ` [PATCH 07/15] drm/i915: Store a direct lookup from object handle to vma Chris Wilson
2017-03-22  9:33   ` [PATCH v2] " Chris Wilson
2017-03-22 14:22     ` Joonas Lahtinen
2017-03-23 14:23       ` Daniel Vetter
2017-03-23 14:40         ` Chris Wilson
2017-03-16 13:19 ` [PATCH 08/15] drm/i915: Pass vma to relocate entry Chris Wilson
2017-03-21 14:17   ` Joonas Lahtinen
2017-03-23 13:02     ` [PATCH] " Chris Wilson
2017-03-16 13:20 ` [PATCH 09/15] drm/i915: Eliminate lots of iterations over the execobjects array Chris Wilson
2017-03-16 13:20 ` Chris Wilson [this message]
2017-03-21 13:54   ` [PATCH 10/15] drm/i915: First try the previous execbuffer location Joonas Lahtinen
2017-03-16 13:20 ` [PATCH 11/15] drm/i915: Wait upon userptr get-user-pages within execbuffer Chris Wilson
2017-03-16 13:20 ` [PATCH 12/15] drm/i915: Remove superfluous i915_add_request_no_flush() helper Chris Wilson
2017-03-17 11:17   ` Joonas Lahtinen
2017-03-17 13:06     ` Chris Wilson
2017-03-16 13:20 ` [PATCH 13/15] drm/i915: Allow execbuffer to use the first object as the batch Chris Wilson
2017-03-17 11:15   ` Joonas Lahtinen
2017-07-07 10:17     ` Daniel Vetter
2017-07-07 11:58       ` Chris Wilson
2017-07-10  5:55         ` Daniel Vetter
2017-07-08  3:54       ` Kenneth Graunke
2017-03-16 13:20 ` [PATCH 14/15] drm/i915: Async GPU relocation processing Chris Wilson
2017-03-17 11:11   ` [PATCH v2] " Chris Wilson
2017-03-17 11:15   ` [PATCH 14/15] " Joonas Lahtinen
2017-03-16 13:20 ` [PATCH 15/15] drm/i915/scheduler: Support user-defined priorities Chris Wilson
2017-03-16 13:57 ` ✗ Fi.CI.BAT: warning for series starting with [01/15] drm/i915: Copy user requested buffers into the error state Patchwork
2017-03-17 11:28 ` ✓ Fi.CI.BAT: success for series starting with [01/15] drm/i915: Copy user requested buffers into the error state (rev2) Patchwork
2017-03-22  9:43 ` ✗ Fi.CI.BAT: failure for series starting with [01/15] drm/i915: Copy user requested buffers into the error state (rev3) Patchwork
2017-03-23 13:51 ` ✗ Fi.CI.BAT: failure for series starting with [01/15] drm/i915: Copy user requested buffers into the error state (rev4) 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=20170316132006.7976-11-chris@chris-wilson.co.uk \
    --to=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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.