public inbox for intel-gfx@lists.freedesktop.org
 help / color / mirror / Atom feed
From: Matthew Auld <matthew.auld@intel.com>
To: intel-gfx@lists.freedesktop.org
Subject: [PATCH 05/19] drm/i915: align the vma start to the largest gtt page size
Date: Wed, 21 Jun 2017 21:33:31 +0100	[thread overview]
Message-ID: <20170621203345.26320-6-matthew.auld@intel.com> (raw)
In-Reply-To: <20170621203345.26320-1-matthew.auld@intel.com>

For the 48b PPGTT try to align the vma start address to the required
page size boundary to guarantee we use said page size in the gtt. If we
are dealing with multiple page sizes, we can't guarantee anything and
just align to the largest. For soft pinning and objects which need to be
tightly packed into the lower 32bits we don't force any alignment.

v2: various improvements suggested by Chris

Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
---
 drivers/gpu/drm/i915/i915_vma.c | 15 +++++++++++++++
 drivers/gpu/drm/i915/i915_vma.h |  1 +
 2 files changed, 16 insertions(+)

diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
index 958be0a95960..cee1d00dc085 100644
--- a/drivers/gpu/drm/i915/i915_vma.c
+++ b/drivers/gpu/drm/i915/i915_vma.c
@@ -471,6 +471,9 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
 	if (ret)
 		return ret;
 
+	vma->page_sizes.phys = obj->mm.page_sizes.phys;
+	vma->page_sizes.sg = obj->mm.page_sizes.sg;
+
 	if (flags & PIN_OFFSET_FIXED) {
 		u64 offset = flags & PIN_OFFSET_MASK;
 		if (!IS_ALIGNED(offset, alignment) ||
@@ -485,6 +488,18 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
 		if (ret)
 			goto err_unpin;
 	} else {
+		/* We only support huge gtt pages through the 48b PPGTT,
+		 * however we also don't want to force any alignment for
+		 * objects which need to be tightly packed into the low 32bits.
+		 */
+		if (end > (1ULL << 32) &&
+		    vma->page_sizes.sg > I915_GTT_PAGE_SIZE) {
+			u64 page_alignment =
+				rounddown_pow_of_two(vma->page_sizes.sg);
+
+			alignment = max(alignment, page_alignment);
+		}
+
 		ret = i915_gem_gtt_insert(vma->vm, &vma->node,
 					  size, alignment, obj->cache_level,
 					  start, end, flags);
diff --git a/drivers/gpu/drm/i915/i915_vma.h b/drivers/gpu/drm/i915/i915_vma.h
index 4a673fc1a432..834f7ca2ada2 100644
--- a/drivers/gpu/drm/i915/i915_vma.h
+++ b/drivers/gpu/drm/i915/i915_vma.h
@@ -52,6 +52,7 @@ struct i915_vma {
 	struct drm_i915_fence_reg *fence;
 	struct reservation_object *resv; /** Alias of obj->resv */
 	struct sg_table *pages;
+	struct i915_page_sizes page_sizes;
 	void __iomem *iomap;
 	u64 size;
 	u64 display_alignment;
-- 
2.9.4

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

  parent reply	other threads:[~2017-06-21 20:33 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-06-21 20:33 [PATCH 00/19] huge gtt pages Matthew Auld
2017-06-21 20:33 ` [PATCH 01/19] drm/i915: introduce simple gemfs Matthew Auld
2017-06-21 21:19   ` Chris Wilson
2017-06-21 20:33 ` [PATCH 02/19] drm/i915/gemfs: enable THP Matthew Auld
2017-06-21 20:33 ` [PATCH 03/19] drm/i915: introduce page_size_mask to dev_info Matthew Auld
2017-06-21 20:33 ` [PATCH 04/19] drm/i915: introduce page_size members Matthew Auld
2017-06-21 21:26   ` Chris Wilson
2017-06-21 20:33 ` Matthew Auld [this message]
2017-06-21 21:35   ` [PATCH 05/19] drm/i915: align the vma start to the largest gtt page size Chris Wilson
2017-06-21 20:33 ` [PATCH 06/19] drm/i915: align 64K objects to 2M Matthew Auld
2017-06-21 21:37   ` Chris Wilson
2017-06-21 20:33 ` [PATCH 07/19] drm/i915: pass the vma to insert_entries Matthew Auld
2017-06-21 21:39   ` Chris Wilson
2017-06-21 20:33 ` [PATCH 08/19] drm/i915: enable IPS bit for 64K pages Matthew Auld
2017-06-21 20:33 ` [PATCH 09/19] drm/i915: disable GTT cache for 2M/1G pages Matthew Auld
2017-06-21 21:41   ` Chris Wilson
2017-06-21 20:33 ` [PATCH 10/19] drm/i915: support 1G pages for the 48b PPGTT Matthew Auld
2017-06-21 21:49   ` Chris Wilson
2017-06-21 22:51     ` Chris Wilson
2017-06-22 11:07       ` Matthew Auld
2017-06-22 11:38         ` Chris Wilson
2017-06-21 20:33 ` [PATCH 11/19] drm/i915: support 2M " Matthew Auld
2017-06-21 20:33 ` [PATCH 12/19] drm/i915: support 64K " Matthew Auld
2017-06-21 21:55   ` Chris Wilson
2017-06-22 11:27     ` Matthew Auld
2017-06-21 20:33 ` [PATCH 13/19] drm/i915: accurate page size tracking for the ppgtt Matthew Auld
2017-06-21 21:57   ` Chris Wilson
2017-06-21 20:33 ` [PATCH 14/19] drm/i915/debugfs: include some gtt page size metrics Matthew Auld
2017-06-21 20:33 ` [PATCH 15/19] drm/i915/selftests: basic huge page tests Matthew Auld
2017-06-22 14:17   ` Chris Wilson
2017-06-22 14:21   ` Chris Wilson
2017-06-21 20:33 ` [PATCH 16/19] drm/i915/selftests: mix huge pages Matthew Auld
2017-06-21 20:33 ` [PATCH 17/19] drm/i915: enable platform support for 64K pages Matthew Auld
2017-06-21 20:33 ` [PATCH 18/19] drm/i915: enable platform support for 2M pages Matthew Auld
2017-06-21 20:33 ` [PATCH 19/19] drm/i915: enable platform support for 1G pages Matthew Auld
2017-06-21 21:05 ` ✓ Fi.CI.BAT: success for huge gtt pages (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=20170621203345.26320-6-matthew.auld@intel.com \
    --to=matthew.auld@intel.com \
    --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