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: [RFC PATCH 07/42] drm/i915/region: support volatile objects
Date: Thu, 14 Feb 2019 14:57:05 +0000	[thread overview]
Message-ID: <20190214145740.14521-8-matthew.auld@intel.com> (raw)
In-Reply-To: <20190214145740.14521-1-matthew.auld@intel.com>

Volatile objects are marked as DONTNEED while pinned, therefore once
unpinned the backing store can be discarded.

Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Abdiel Janulgue <abdiel.janulgue@linux.intel.com>
---
 drivers/gpu/drm/i915/i915_gem_object.h        |  3 +-
 drivers/gpu/drm/i915/intel_memory_region.c    | 10 +++-
 .../drm/i915/selftests/intel_memory_region.c  | 54 +++++++++++++++++++
 3 files changed, 65 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem_object.h b/drivers/gpu/drm/i915/i915_gem_object.h
index efc23fc2b1df..0dfe9603e3cc 100644
--- a/drivers/gpu/drm/i915/i915_gem_object.h
+++ b/drivers/gpu/drm/i915/i915_gem_object.h
@@ -164,7 +164,8 @@ struct drm_i915_gem_object {
 	 */
 #define I915_BO_ACTIVE_REF 0
 #define I915_BO_ALLOC_CONTIGUOUS (1<<1)
-#define I915_BO_ALLOC_FLAGS (I915_BO_ALLOC_CONTIGUOUS)
+#define I915_BO_ALLOC_VOLATILE   (1<<2)
+#define I915_BO_ALLOC_FLAGS (I915_BO_ALLOC_CONTIGUOUS | I915_BO_ALLOC_VOLATILE)
 
 	/*
 	 * Is the object to be mapped as read-only to the GPU
diff --git a/drivers/gpu/drm/i915/intel_memory_region.c b/drivers/gpu/drm/i915/intel_memory_region.c
index ed37599c7784..5531ed94bdab 100644
--- a/drivers/gpu/drm/i915/intel_memory_region.c
+++ b/drivers/gpu/drm/i915/intel_memory_region.c
@@ -148,6 +148,9 @@ i915_memory_region_get_pages_buddy(struct drm_i915_gem_object *obj)
 
 	i915_sg_trim(st);
 
+	if (flags & I915_BO_ALLOC_VOLATILE)
+		obj->mm.madv = I915_MADV_DONTNEED;
+
 	__i915_gem_object_set_pages(obj, st, sg_page_sizes);
 
 	return 0;
@@ -209,7 +212,12 @@ i915_gem_object_create_region(struct intel_memory_region *mem,
 	obj->flags = flags;
 
 	mutex_lock(&mem->obj_lock);
-	list_add(&obj->region_link, &mem->objects);
+
+	if (flags & I915_BO_ALLOC_VOLATILE)
+		list_add(&obj->region_link, &mem->purgeable);
+	else
+		list_add(&obj->region_link, &mem->objects);
+
 	mutex_unlock(&mem->obj_lock);
 
 	i915_gem_object_set_cache_coherency(obj, obj->cache_level);
diff --git a/drivers/gpu/drm/i915/selftests/intel_memory_region.c b/drivers/gpu/drm/i915/selftests/intel_memory_region.c
index fbb1497d0f82..0f6432a2a4c4 100644
--- a/drivers/gpu/drm/i915/selftests/intel_memory_region.c
+++ b/drivers/gpu/drm/i915/selftests/intel_memory_region.c
@@ -309,12 +309,66 @@ static int igt_mock_continuous(void *arg)
 	return err;
 }
 
+static int igt_mock_volatile(void *arg)
+{
+	struct intel_memory_region *mem = arg;
+	struct drm_i915_gem_object *obj;
+	int err;
+
+	obj = i915_gem_object_create_region(mem, PAGE_SIZE, 0);
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
+
+	err = i915_gem_object_pin_pages(obj);
+	if (err)
+		goto err_put;
+
+	i915_gem_object_unpin_pages(obj);
+
+	err = i915_gem_shrink_memory_region(mem, PAGE_SIZE);
+	if (err != -ENOSPC) {
+		pr_err("shrink memory region\n");
+		goto err_put;
+	}
+
+	obj = i915_gem_object_create_region(mem, PAGE_SIZE, I915_BO_ALLOC_VOLATILE);
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
+
+	if (!(obj->flags & I915_BO_ALLOC_VOLATILE)) {
+		pr_err("missing flags\n");
+		goto err_put;
+	}
+
+	err = i915_gem_object_pin_pages(obj);
+	if (err)
+		goto err_put;
+
+	i915_gem_object_unpin_pages(obj);
+
+	err = i915_gem_shrink_memory_region(mem, PAGE_SIZE);
+	if (err) {
+		pr_err("failed to shrink memory\n");
+		goto err_put;
+	}
+
+	if (i915_gem_object_has_pages(obj)) {
+		pr_err("object pages not discarded\n");
+		err = -EINVAL;
+	}
+
+err_put:
+	i915_gem_object_put(obj);
+	return err;
+}
+
 int intel_memory_region_mock_selftests(void)
 {
 	static const struct i915_subtest tests[] = {
 		SUBTEST(igt_mock_fill),
 		SUBTEST(igt_mock_shrink),
 		SUBTEST(igt_mock_continuous),
+		SUBTEST(igt_mock_volatile),
 	};
 	struct intel_memory_region *mem;
 	struct drm_i915_private *i915;
-- 
2.20.1

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

  parent reply	other threads:[~2019-02-14 14:57 UTC|newest]

Thread overview: 97+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-02-14 14:56 [RFC PATCH 00/42] Introduce memory region concept (including device local memory) Matthew Auld
2019-02-14 14:56 ` [RFC PATCH 01/42] drm/i915: support 1G pages for the 48b PPGTT Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 02/42] drm/i915: enable platform support for 1G pages Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 03/42] drm/i915: buddy allocator Matthew Auld
2019-02-15 12:34   ` Jani Nikula
2019-02-15 15:03     ` Chris Wilson
2019-02-18 11:35       ` Jani Nikula
2019-02-14 14:57 ` [RFC PATCH 04/42] drm/i915: introduce intel_memory_region Matthew Auld
2019-02-14 15:16   ` Chris Wilson
2019-02-26 14:03     ` Matthew Auld
2019-02-26 14:18       ` Chris Wilson
2019-02-26 13:00   ` Tvrtko Ursulin
2019-02-26 14:20     ` Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 05/42] drm/i915/region: support basic eviction Matthew Auld
2019-02-14 15:25   ` Chris Wilson
2019-02-26 14:58     ` Matthew Auld
2019-02-26 16:49       ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 06/42] drm/i915/region: support continuous allocations Matthew Auld
2019-02-14 14:57 ` Matthew Auld [this message]
2019-02-14 14:57 ` [RFC PATCH 08/42] drm/i915: Add memory region information to device_info Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 09/42] drm/i915: support creating LMEM objects Matthew Auld
2019-02-14 15:30   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 10/42] drm/i915/lmem: add helper to get CPU visible pfn Matthew Auld
2019-02-14 15:33   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 11/42] drm/i915/selftests: exercise writes to LMEM Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 12/42] drm/i915/selftests: exercise huge-pages for LMEM Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 13/42] drm/i915: support object clearing via blitter engine Matthew Auld
2019-02-14 15:37   ` Chris Wilson
2019-02-14 15:38   ` Chris Wilson
2019-02-14 15:40   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 14/42] drm/i915: introduce kernel blitter_context Matthew Auld
2019-02-14 15:42   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 15/42] drm/i915: support copying objects via blitter engine Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 16/42] drm/i915: support basic object migration Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 17/42] drm/i915/lmem: support kernel mapping Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 18/42] drm/i915/lmem: support CPU relocations Matthew Auld
2019-02-14 15:48   ` Chris Wilson
2019-02-26 18:53     ` Matthew Auld
2019-02-26 18:58       ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 19/42] drm/i915: add vfunc for pread Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 20/42] drm/i915/lmem: support pread Matthew Auld
2019-02-14 15:50   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 21/42] drm/i915/lmem: support pwrite Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 22/42] drm/i915: define HAS_MAPPABLE_APERTURE Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 23/42] drm/i915: do not map aperture if it is not available Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 24/42] drm/i915: expose missing map_gtt support to users Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 25/42] drm/i915: set num_fence_regs to 0 if there is no aperture Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 26/42] drm/i915: error capture with no ggtt slot Matthew Auld
2019-02-14 15:56   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 27/42] drm/i915: Don't try to place HWS in non-existing mappable region Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 28/42] drm/i915: Split out GTT fault handler to make it generic Matthew Auld
2019-02-14 16:00   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 29/42] drm/i915: Set correct vmf source pages for gem objects Matthew Auld
2019-02-14 16:02   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 30/42] drm/i915: Introduce DRM_I915_GEM_MMAP_OFFSET Matthew Auld
2019-02-14 16:05   ` Chris Wilson
2019-02-26 13:34   ` Tvrtko Ursulin
2019-02-26 13:37     ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 31/42] drm/i915: cpu-map based dumb buffers Matthew Auld
2019-02-14 16:06   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 32/42] drm/i915: Add fill_pages handler for dma_buf imported objects Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 33/42] UPSTREAM: drm/i915/query: Split out query item checks Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 34/42] drm/i915/query: Expose memory regions through the query uAPI Matthew Auld
2019-02-14 16:31   ` Chris Wilson
2019-02-14 16:33     ` Chris Wilson
2019-02-14 21:12       ` Chris Wilson
2019-02-14 21:15   ` Chris Wilson
2019-02-14 21:21     ` Chris Wilson
2019-02-20 18:56     ` Jason Ekstrand
2019-02-14 14:57 ` [RFC PATCH 35/42] drm/i915: Introduce GEM_OBJECT_SETPARAM with I915_PARAM_MEMORY_REGION Matthew Auld
2019-02-14 16:20   ` Chris Wilson
2019-02-14 14:57 ` [RFC PATCH 36/42] drm/i915/lmem: include debugfs metrics Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 37/42] drm/i915: enumerate and init each supported region Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 38/42] drm/i915: treat shmem as a region Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 39/42] drm/i915: treat stolen " Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 40/42] drm/i915: setup io-mapping for LMEM Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 41/42] HAX drm/i915: add the fake lmem region Matthew Auld
2019-02-14 14:57 ` [RFC PATCH 42/42] HAX drm/i915/lmem: default userspace allocations to LMEM Matthew Auld
2019-02-14 16:13   ` Chris Wilson
2019-02-18 12:44     ` Chris Wilson
2019-02-19 17:44     ` Chris Wilson
2019-02-14 16:22   ` Chris Wilson
2019-02-14 17:58 ` ✗ Fi.CI.CHECKPATCH: warning for Introduce memory region concept (including device local memory) Patchwork
2019-02-14 18:15 ` ✗ Fi.CI.SPARSE: " Patchwork
2019-02-15  0:47 ` [RFC PATCH 00/42] " Dave Airlie
2019-02-19 13:32   ` Joonas Lahtinen
2019-02-25 20:24     ` Dave Airlie
2019-02-26  2:35       ` [Intel-gfx] " Joonas Lahtinen
2019-02-26  5:31         ` Alex Deucher
2019-02-26 10:41           ` Jani Nikula
2019-02-26 12:17           ` Joonas Lahtinen
2019-02-26 17:20             ` Alex Deucher
2019-02-26 18:52               ` Christian König
2019-02-26 19:14               ` Alex Deucher
2019-02-26 23:04                 ` Dave Airlie
2019-02-27 12:17                   ` Christian König
2019-02-27 14:40                     ` Joonas Lahtinen

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=20190214145740.14521-8-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