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] drm/i915: Fix unfenced alignment on pre-G33 hardware
Date: Sat,  9 Jul 2011 09:31:25 +0100	[thread overview]
Message-ID: <1310200285-17692-1-git-send-email-chris@chris-wilson.co.uk> (raw)

Align unfenced buffers on older hardware to the power-of-two object size.
The docs suggest that it should be possible to align only to a power-of-two
tile height, but using the already computed fence size is easier and
always correct. We also have to make sure that we unbind misaligned buffers
upon tiling changes.

Reported-and-tested-by: Sitosfe Wheeler <sitsofe@yahoo.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=36326
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
---
 drivers/gpu/drm/i915/i915_drv.h        |    3 ++-
 drivers/gpu/drm/i915/i915_gem.c        |   30 +++++++++++++-----------------
 drivers/gpu/drm/i915/i915_gem_tiling.c |    3 ++-
 3 files changed, 17 insertions(+), 19 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 2798d27..a9492d9 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1216,7 +1216,8 @@ void i915_gem_free_all_phys_object(struct drm_device *dev);
 void i915_gem_release(struct drm_device *dev, struct drm_file *file);
 
 uint32_t
-i915_gem_get_unfenced_gtt_alignment(struct drm_i915_gem_object *obj);
+i915_gem_get_unfenced_gtt_alignment(struct drm_i915_gem_object *obj,
+				    int tiling_mode);
 
 int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj,
 				    enum i915_cache_level cache_level);
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index e357c73..4fc9738 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1426,36 +1426,31 @@ i915_gem_get_gtt_alignment(struct drm_i915_gem_object *obj)
  * i915_gem_get_unfenced_gtt_alignment - return required GTT alignment for an
  *					 unfenced object
  * @obj: object to check
+ * @tiling_mode: tiling mode of the object
  *
  * Return the required GTT alignment for an object, only taking into account
  * unfenced tiled surface requirements.
  */
 uint32_t
-i915_gem_get_unfenced_gtt_alignment(struct drm_i915_gem_object *obj)
+i915_gem_get_unfenced_gtt_alignment(struct drm_i915_gem_object *obj,
+				    int tiling_mode)
 {
 	struct drm_device *dev = obj->base.dev;
-	int tile_height;
+
+	if (tiling_mode == I915_TILING_NONE)
+		return 4096;
 
 	/*
 	 * Minimum alignment is 4k (GTT page size) for sane hw.
 	 */
-	if (INTEL_INFO(dev)->gen >= 4 || IS_G33(dev) ||
-	    obj->tiling_mode == I915_TILING_NONE)
+	if (INTEL_INFO(dev)->gen >= 4 || IS_G33(dev))
 		return 4096;
 
-	/*
-	 * Older chips need unfenced tiled buffers to be aligned to the left
-	 * edge of an even tile row (where tile rows are counted as if the bo is
-	 * placed in a fenced gtt region).
+	/* Previous hardware however needs to be aligned to a power-of-two
+	 * tile height. The simplest method for determining this is to reuse
+	 * the power-of-tile object size.
 	 */
-	if (IS_GEN2(dev))
-		tile_height = 16;
-	else if (obj->tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev))
-		tile_height = 32;
-	else
-		tile_height = 8;
-
-	return tile_height * obj->stride * 2;
+	return i915_gem_get_gtt_size(obj);
 }
 
 int
@@ -2781,7 +2776,8 @@ i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj,
 
 	fence_size = i915_gem_get_gtt_size(obj);
 	fence_alignment = i915_gem_get_gtt_alignment(obj);
-	unfenced_alignment = i915_gem_get_unfenced_gtt_alignment(obj);
+	unfenced_alignment =
+		i915_gem_get_unfenced_gtt_alignment(obj, obj->tiling_mode);
 
 	if (alignment == 0)
 		alignment = map_and_fenceable ? fence_alignment :
diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c
index 82d70fd..8433b97 100644
--- a/drivers/gpu/drm/i915/i915_gem_tiling.c
+++ b/drivers/gpu/drm/i915/i915_gem_tiling.c
@@ -348,7 +348,8 @@ i915_gem_set_tiling(struct drm_device *dev, void *data,
 		/* Rebind if we need a change of alignment */
 		if (!obj->map_and_fenceable) {
 			u32 unfenced_alignment =
-				i915_gem_get_unfenced_gtt_alignment(obj);
+				i915_gem_get_unfenced_gtt_alignment(obj,
+								    args->tiling_mode);
 			if (obj->gtt_offset & (unfenced_alignment - 1))
 				ret = i915_gem_object_unbind(obj);
 		}
-- 
1.7.5.4

             reply	other threads:[~2011-07-09  8:31 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-07-09  8:31 Chris Wilson [this message]
2011-07-18 15:45 ` [PATCH] drm/i915: Fix unfenced alignment on pre-G33 hardware Daniel Vetter
2011-07-18 16:17 ` Keith Packard
2011-07-18 16:25   ` Chris Wilson
2011-07-18 16:35   ` Keith Packard
2011-07-18 16:57     ` Chris Wilson
2011-07-18 20:31       ` Keith Packard
2011-07-18 22:48         ` Paul Menzel
2011-07-18 23:14           ` Keith Packard
2011-07-18 16:41   ` 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=1310200285-17692-1-git-send-email-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.