xf86-video-intel: 2 commits - src/sna/kgem.c src/sna/kgem.h
Chris Wilson
ickle at kemper.freedesktop.org
Tue Sep 20 00:32:41 PDT 2011
src/sna/kgem.c | 88 +++++++++++++++++++++++++++++----------------------------
src/sna/kgem.h | 2 -
2 files changed, 46 insertions(+), 44 deletions(-)
New commits:
commit 902a0234191d77816259affe331339caaa9131d6
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date: Mon Sep 19 20:34:15 2011 +0100
sna: Use pot buckets for active cache
On x11perf one of the major hotspots is the search through the active
list for an object large enough to reuse as the target surface. We can
eliminate that overhead by keeping those active objects in pre-sorted
lists by size.
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
diff --git a/src/sna/kgem.c b/src/sna/kgem.c
index 1e7065d..4a73971 100644
--- a/src/sna/kgem.c
+++ b/src/sna/kgem.c
@@ -302,6 +302,15 @@ static struct list *inactive(struct kgem *kgem,
return &kgem->inactive[order];
}
+static struct list *active(struct kgem *kgem,
+ int size)
+{
+ uint32_t order = __fls(size / PAGE_SIZE);
+ if (order >= ARRAY_SIZE(kgem->inactive))
+ order = ARRAY_SIZE(kgem->inactive)-1;
+ return &kgem->active[order];
+}
+
static size_t
agp_aperture_size(struct pci_device *dev, int gen)
{
@@ -324,10 +333,11 @@ void kgem_init(struct kgem *kgem, int fd, struct pci_device *dev, int gen)
list_init(&kgem->partial);
list_init(&kgem->requests);
- list_init(&kgem->active);
list_init(&kgem->flushing);
for (i = 0; i < ARRAY_SIZE(kgem->inactive); i++)
list_init(&kgem->inactive[i]);
+ for (i = 0; i < ARRAY_SIZE(kgem->active); i++)
+ list_init(&kgem->active[i]);
kgem->next_request = __kgem_request_alloc();
@@ -546,11 +556,11 @@ static void __kgem_bo_destroy(struct kgem *kgem, struct kgem_bo *bo)
kgem->need_expire = true;
if (bo->rq) {
- list_move(&bo->list, &kgem->active);
+ list_move(&bo->list, active(kgem, bo->size));
} else if (bo->needs_flush) {
assert(list_is_empty(&bo->request));
list_add(&bo->request, &kgem->flushing);
- list_move(&bo->list, &kgem->active);
+ list_move(&bo->list, active(kgem, bo->size));
} else {
assert(!kgem_busy(kgem, bo->handle));
list_move(&bo->list, inactive(kgem, bo->size));
@@ -1155,17 +1165,17 @@ void kgem_cleanup_cache(struct kgem *kgem)
}
static struct kgem_bo *
-search_linear_cache(struct kgem *kgem, unsigned int size, bool active)
+search_linear_cache(struct kgem *kgem, unsigned int size, bool use_active)
{
struct kgem_bo *bo;
struct list *cache;
- cache = active ? &kgem->active : inactive(kgem, size);
+ cache = use_active ? active(kgem, size): inactive(kgem, size);
list_for_each_entry(bo, cache, list) {
if (size > bo->size)
continue;
- if (active && bo->tiling != I915_TILING_NONE)
+ if (use_active && bo->tiling != I915_TILING_NONE)
continue;
if (bo->deleted) {
@@ -1192,10 +1202,10 @@ search_linear_cache(struct kgem *kgem, unsigned int size, bool active)
bo->delta = 0;
DBG((" %s: found handle=%d (size=%d) in linear %s cache\n",
__FUNCTION__, bo->handle, bo->size,
- active ? "active" : "inactive"));
+ use_active ? "active" : "inactive"));
assert(bo->refcnt == 0);
assert(bo->reusable);
- assert(active || !kgem_busy(kgem, bo->handle));
+ assert(use_active || !kgem_busy(kgem, bo->handle));
return bo;
}
@@ -1383,8 +1393,8 @@ struct kgem_bo *kgem_create_2d(struct kgem *kgem,
uint32_t flags)
{
struct list *cache;
- struct kgem_bo *bo, *next, *best;
- uint32_t pitch, tiled_height[3], size, best_size;
+ struct kgem_bo *bo, *next;
+ uint32_t pitch, tiled_height[3], size;
uint32_t handle;
int exact = flags & CREATE_EXACT;
int i;
@@ -1405,14 +1415,9 @@ struct kgem_bo *kgem_create_2d(struct kgem *kgem,
tiled_height[i] = kgem_aligned_height(kgem, height, i);
search_active: /* Best active match first */
- best = NULL;
- best_size = -1;
- list_for_each_entry_safe(bo, next, &kgem->active, list) {
+ list_for_each_entry(bo, active(kgem, size), list) {
uint32_t s;
- if (bo->size > best_size)
- continue;
-
if (exact) {
if (bo->tiling != tiling)
continue;
@@ -1433,36 +1438,32 @@ search_active: /* Best active match first */
s = bo->pitch * tiled_height[bo->tiling];
if (s <= bo->size) {
- best_size = s;
- best = bo;
- }
- }
- if (best) {
- list_del(&best->list);
- if (best->rq == NULL)
- list_del(&best->request);
+ list_del(&bo->list);
+ if (bo->rq == NULL)
+ list_del(&bo->request);
- if (best->deleted) {
- if (!gem_madvise(kgem->fd, best->handle,
- I915_MADV_WILLNEED)) {
- kgem->need_purge |= best->gpu;
- gem_close(kgem->fd, best->handle);
- list_del(&best->request);
- free(best);
- best = NULL;
- goto search_active;
+ if (bo->deleted) {
+ if (!gem_madvise(kgem->fd, bo->handle,
+ I915_MADV_WILLNEED)) {
+ kgem->need_purge |= bo->gpu;
+ gem_close(kgem->fd, bo->handle);
+ list_del(&bo->request);
+ free(bo);
+ bo = NULL;
+ goto search_active;
+ }
+
+ bo->deleted = 0;
}
- best->deleted = 0;
+ bo->unique_id = kgem_get_unique_id(kgem);
+ bo->delta = 0;
+ DBG((" from active: pitch=%d, tiling=%d, handle=%d, id=%d\n",
+ bo->pitch, bo->tiling, bo->handle, bo->unique_id));
+ assert(bo->refcnt == 0);
+ assert(bo->reusable);
+ return kgem_bo_reference(bo);
}
-
- best->unique_id = kgem_get_unique_id(kgem);
- best->delta = 0;
- DBG((" from active: pitch=%d, tiling=%d, handle=%d, id=%d\n",
- best->pitch, best->tiling, best->handle, best->unique_id));
- assert(best->refcnt == 0);
- assert(best->reusable);
- return kgem_bo_reference(best);
}
skip_active_search:
diff --git a/src/sna/kgem.h b/src/sna/kgem.h
index 43211f3..072280d 100644
--- a/src/sna/kgem.h
+++ b/src/sna/kgem.h
@@ -85,7 +85,7 @@ struct kgem {
KGEM_BLT,
} mode, ring;
- struct list active, flushing, inactive[16];
+ struct list flushing, active[16], inactive[16];
struct list partial;
struct list requests;
struct kgem_request *next_request;
commit 7bb2c5c4d55560f7901fd8036cab164460f81edc
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date: Mon Sep 19 19:12:46 2011 +0100
sna: Move an assertion to the point of import
All the guesswork is so that when we require an inactive bo, we do
actually get a buffer that is not currently on a GPU active list. For
some unresolved reason, this assertion was firing when putting the
buffer onto the inactive list - so just workaround the worrisome issue
by delaying the check until use.
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
diff --git a/src/sna/kgem.c b/src/sna/kgem.c
index 7df019a..1e7065d 100644
--- a/src/sna/kgem.c
+++ b/src/sna/kgem.c
@@ -613,7 +613,6 @@ void kgem_retire(struct kgem *kgem)
if (bo->needs_flush) {
list_add(&bo->request, &kgem->flushing);
} else if (bo->reusable) {
- assert(!kgem_busy(kgem, bo->handle));
list_move(&bo->list,
inactive(kgem, bo->size));
} else {
@@ -1507,6 +1506,8 @@ skip_active_search:
bo->pitch, bo->tiling, bo->handle, bo->unique_id));
assert(bo->refcnt == 0);
assert(bo->reusable);
+ assert((flags & CREATE_INACTIVE) == 0 ||
+ !kgem_busy(kgem, bo->handle));
return kgem_bo_reference(bo);
next_bo:
More information about the xorg-commit
mailing list