[Intel-gfx] [PATCH 181/190] drm/i915: Introduce an internal allocator for disposable private objects
Chris Wilson
chris at chris-wilson.co.uk
Mon Jan 11 03:01:22 PST 2016
Quite a few of our objects used for internal hardware programming do not
benefit from being swappable or from being zero initialised. As such
they do not benefit from using a shmemfs backing storage and since they
are internal and never directly exposed to the user, we do not need to
worry about providing a filp. For these we can use an
drm_i915_gem_object wrapper around a sg_table of plain struct page. They
are not swapped backed and not automatically pinned. If they are reaped
by the shrinker, the pages are released and the contents discarded. For
the internal use case, this is fine as for example, ringbuffers are
pinned from being written by a request to be read by the hardware. Once
they are idle, they can be discarded entirely. As such they are a good
match for execlist ringbuffers and a small varierty of other internal
objects.
In the first iteration, this is limited to the scratch batch buffers we
use (for command parsing and state initialisation).
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
---
drivers/gpu/drm/i915/Makefile | 1 +
drivers/gpu/drm/i915/i915_drv.h | 8 ++
drivers/gpu/drm/i915/i915_gem.c | 9 +-
drivers/gpu/drm/i915/i915_gem_batch_pool.c | 27 ++---
drivers/gpu/drm/i915/i915_gem_internal.c | 157 +++++++++++++++++++++++++++
drivers/gpu/drm/i915/i915_gem_render_state.c | 2 +-
6 files changed, 180 insertions(+), 24 deletions(-)
create mode 100644 drivers/gpu/drm/i915/i915_gem_internal.c
diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile
index a362425ef862..8d0fae65a5bd 100644
--- a/drivers/gpu/drm/i915/Makefile
+++ b/drivers/gpu/drm/i915/Makefile
@@ -28,6 +28,7 @@ i915-y += i915_cmd_parser.o \
i915_gem_execbuffer.o \
i915_gem_fence.o \
i915_gem_gtt.o \
+ i915_gem_internal.o \
i915_gem.o \
i915_gem_render_state.o \
i915_gem_request.o \
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 2f8b5e7f9320..e3c77d245a6b 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1989,6 +1989,9 @@ enum hdmi_force_audio {
#define I915_GTT_OFFSET_NONE ((u32)-1)
struct drm_i915_gem_object_ops {
+ unsigned int flags;
+#define I915_GEM_OBJECT_HAS_STRUCT_PAGE 0x1
+
/* Interface between the GEM object and its backing storage.
* get_pages() is called once prior to the use of the associated set
* of pages before to binding them into the GTT, and put_pages() is
@@ -3117,6 +3120,11 @@ i915_gem_object_create_stolen_for_preallocated(struct drm_device *dev,
u32 gtt_offset,
u32 size);
+/* i915_gem_internal.c */
+struct drm_i915_gem_object *
+i915_gem_object_create_internal(struct drm_device *dev,
+ unsigned size);
+
/* i915_gem_shrinker.c */
unsigned long i915_gem_shrink(struct drm_i915_private *dev_priv,
unsigned long target,
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 9cd161645041..cca45f60d0bd 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -497,7 +497,7 @@ int i915_gem_obj_prepare_shmem_read(struct drm_i915_gem_object *obj,
int ret;
*needs_clflush = 0;
- if (!obj->base.filp)
+ if ((obj->ops->flags & I915_GEM_OBJECT_HAS_STRUCT_PAGE) == 0)
return -EINVAL;
ret = i915_gem_object_get_pages(obj);
@@ -547,7 +547,7 @@ int i915_gem_obj_prepare_shmem_write(struct drm_i915_gem_object *obj,
int ret;
*needs_clflush = 0;
- if (!obj->base.filp)
+ if ((obj->ops->flags & I915_GEM_OBJECT_HAS_STRUCT_PAGE) == 0)
return -EINVAL;
ret = i915_gem_object_get_pages(obj);
@@ -800,7 +800,7 @@ i915_gem_pread_ioctl(struct drm_device *dev, void *data,
/* prime objects have no backing filp to GEM pread/pwrite
* pages from.
*/
- if (!obj->base.filp) {
+ if ((obj->ops->flags & I915_GEM_OBJECT_HAS_STRUCT_PAGE) == 0) {
ret = -EINVAL;
goto out;
}
@@ -1131,7 +1131,7 @@ i915_gem_pwrite_ioctl(struct drm_device *dev, void *data,
/* prime objects have no backing filp to GEM pread/pwrite
* pages from.
*/
- if (!obj->base.filp) {
+ if ((obj->ops->flags & I915_GEM_OBJECT_HAS_STRUCT_PAGE) == 0) {
ret = -EINVAL;
goto out;
}
@@ -3750,6 +3750,7 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,
}
static const struct drm_i915_gem_object_ops i915_gem_object_ops = {
+ .flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE,
.get_pages = i915_gem_object_get_pages_gtt,
.put_pages = i915_gem_object_put_pages_gtt,
};
diff --git a/drivers/gpu/drm/i915/i915_gem_batch_pool.c b/drivers/gpu/drm/i915/i915_gem_batch_pool.c
index d46012234db1..b876e334da58 100644
--- a/drivers/gpu/drm/i915/i915_gem_batch_pool.c
+++ b/drivers/gpu/drm/i915/i915_gem_batch_pool.c
@@ -98,9 +98,9 @@ i915_gem_batch_pool_get(struct i915_gem_batch_pool *pool,
size_t size)
{
struct drm_i915_gem_object *obj = NULL;
- struct drm_i915_gem_object *tmp, *next;
+ struct drm_i915_gem_object *tmp;
struct list_head *list;
- int n;
+ int n, ret;
lockdep_assert_held(&pool->engine->dev->struct_mutex);
@@ -113,7 +113,7 @@ i915_gem_batch_pool_get(struct i915_gem_batch_pool *pool,
n = ARRAY_SIZE(pool->cache_list) - 1;
list = &pool->cache_list[n];
- list_for_each_entry_safe(tmp, next, list, batch_pool_link) {
+ list_for_each_entry(tmp, list, batch_pool_link) {
/* The batches are strictly LRU ordered */
if (i915_gem_object_is_active(tmp)) {
struct drm_i915_gem_request *rq;
@@ -126,13 +126,6 @@ i915_gem_batch_pool_get(struct i915_gem_batch_pool *pool,
GEM_BUG_ON(tmp->last_write.request);
}
- /* While we're looping, do some clean up */
- if (tmp->madv == __I915_MADV_PURGED) {
- list_del(&tmp->batch_pool_link);
- drm_gem_object_unreference(&tmp->base);
- continue;
- }
-
if (tmp->base.size >= size) {
obj = tmp;
break;
@@ -140,19 +133,15 @@ i915_gem_batch_pool_get(struct i915_gem_batch_pool *pool,
}
if (obj == NULL) {
- int ret;
-
- obj = i915_gem_alloc_object(pool->engine->dev, size);
+ obj = i915_gem_object_create_internal(pool->engine->dev, size);
if (obj == NULL)
return ERR_PTR(-ENOMEM);
-
- ret = i915_gem_object_get_pages(obj);
- if (ret)
- return ERR_PTR(ret);
-
- obj->madv = I915_MADV_DONTNEED;
}
+ ret = i915_gem_object_get_pages(obj);
+ if (ret)
+ return ERR_PTR(ret);
+
list_move_tail(&obj->batch_pool_link, list);
i915_gem_object_pin_pages(obj);
return obj;
diff --git a/drivers/gpu/drm/i915/i915_gem_internal.c b/drivers/gpu/drm/i915/i915_gem_internal.c
new file mode 100644
index 000000000000..0685a304f3ab
--- /dev/null
+++ b/drivers/gpu/drm/i915/i915_gem_internal.c
@@ -0,0 +1,157 @@
+/*
+ * Copyright © 2015 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#include <drm/drmP.h>
+#include <drm/i915_drm.h>
+#include "i915_drv.h"
+
+static void __i915_gem_object_free_pages(struct sg_table *st)
+{
+ struct sg_page_iter sg_iter;
+
+ for_each_sg_page(st->sgl, &sg_iter, st->nents, 0)
+ page_cache_release(sg_page_iter_page(&sg_iter));
+
+ sg_free_table(st);
+ kfree(st);
+}
+
+static int i915_gem_object_get_pages_internal(struct drm_i915_gem_object *obj)
+{
+ const unsigned npages = obj->base.size / PAGE_SIZE;
+ struct sg_table *st;
+ struct scatterlist *sg;
+ unsigned long last_pfn = 0; /* suppress gcc warning */
+ gfp_t gfp;
+ int i;
+
+ st = kmalloc(sizeof(*st), GFP_KERNEL);
+ if (st == NULL)
+ return -ENOMEM;
+
+ if (sg_alloc_table(st, npages, GFP_KERNEL)) {
+ kfree(st);
+ return -ENOMEM;
+ }
+
+ sg = st->sgl;
+ st->nents = 0;
+
+ gfp = GFP_KERNEL | __GFP_HIGHMEM;
+ gfp |= __GFP_NORETRY | __GFP_NOWARN;
+ gfp &= ~(__GFP_IO | __GFP_RECLAIM);
+ for (i = 0; i < npages; i++) {
+ struct page *page;
+
+ page = alloc_page(gfp);
+ if (page == NULL) {
+ i915_gem_shrink_all(to_i915(obj->base.dev));
+ page = alloc_page(GFP_KERNEL | __GFP_HIGHMEM);
+ if (page == NULL)
+ goto err;
+ }
+
+#ifdef CONFIG_SWIOTLB
+ if (swiotlb_nr_tbl()) {
+ st->nents++;
+ sg_set_page(sg, page, PAGE_SIZE, 0);
+ sg = sg_next(sg);
+ continue;
+ }
+#endif
+ if (!i || page_to_pfn(page) != last_pfn + 1) {
+ if (i)
+ sg = sg_next(sg);
+ st->nents++;
+ sg_set_page(sg, page, PAGE_SIZE, 0);
+ } else {
+ sg->length += PAGE_SIZE;
+ }
+ last_pfn = page_to_pfn(page);
+ }
+#ifdef CONFIG_SWIOTLB
+ if (!swiotlb_nr_tbl())
+#endif
+ sg_mark_end(sg);
+ obj->pages = st;
+
+ if (i915_gem_gtt_prepare_object(obj)) {
+ obj->pages = NULL;
+ goto err;
+ }
+
+ obj->madv = I915_MADV_DONTNEED;
+ i915_gem_object_set_dirty(obj);
+ return 0;
+
+err:
+ sg_mark_end(sg);
+ __i915_gem_object_free_pages(st);
+ return -ENOMEM;
+}
+
+static void i915_gem_object_put_pages_internal(struct drm_i915_gem_object *obj)
+{
+ __i915_gem_object_free_pages(obj->pages);
+
+ i915_gem_object_unset_dirty(obj);
+ obj->madv = I915_MADV_WILLNEED;
+}
+
+static const struct drm_i915_gem_object_ops i915_gem_object_internal_ops = {
+ .flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE,
+ .get_pages = i915_gem_object_get_pages_internal,
+ .put_pages = i915_gem_object_put_pages_internal,
+};
+
+/**
+ * Creates a new object that wraps some internal memory for private use.
+ * This object is not backed by swappable storage, and as such its contents
+ * are volatile and only valid whilst pinned. If the object is reaped by the
+ * shrinker, its pages and data will be discarded. Equally, it is not a full
+ * GEM object and so not valid for access from userspace. This makes it useful
+ * for hardware interfaces like ringbuffers (which are pinned from the time
+ * the request is written to the time the hardware stops accessing it), but
+ * not for contexts (which need to be preserved when not active for later
+ * reuse).
+ */
+struct drm_i915_gem_object *
+i915_gem_object_create_internal(struct drm_device *dev,
+ unsigned size)
+{
+ struct drm_i915_gem_object *obj;
+
+ obj = i915_gem_object_alloc(dev);
+ if (obj == NULL)
+ return NULL;
+
+ drm_gem_private_object_init(dev, &obj->base, size);
+ i915_gem_object_init(obj, &i915_gem_object_internal_ops);
+
+ obj->base.write_domain = I915_GEM_DOMAIN_CPU;
+ obj->base.read_domains = I915_GEM_DOMAIN_CPU;
+ obj->cache_level = HAS_LLC(dev) ? I915_CACHE_LLC: I915_CACHE_NONE;
+
+ return obj;
+}
diff --git a/drivers/gpu/drm/i915/i915_gem_render_state.c b/drivers/gpu/drm/i915/i915_gem_render_state.c
index 2fac95b0ba44..436f097d1907 100644
--- a/drivers/gpu/drm/i915/i915_gem_render_state.c
+++ b/drivers/gpu/drm/i915/i915_gem_render_state.c
@@ -66,7 +66,7 @@ static int render_state_init(struct render_state *so, struct drm_device *dev)
if (so->rodata->batch_items * 4 > 4096)
return -EINVAL;
- so->obj = i915_gem_alloc_object(dev, 4096);
+ so->obj = i915_gem_object_create_internal(dev, 4096);
if (so->obj == NULL)
return -ENOMEM;
--
2.7.0.rc3
More information about the Intel-gfx
mailing list