summaryrefslogtreecommitdiffstats
path: root/freedreno/freedreno_bo_cache.c
diff options
context:
space:
mode:
Diffstat (limited to '')
-rw-r--r--freedreno/freedreno_bo_cache.c220
1 files changed, 220 insertions, 0 deletions
diff --git a/freedreno/freedreno_bo_cache.c b/freedreno/freedreno_bo_cache.c
new file mode 100644
index 0000000..bb0605a
--- /dev/null
+++ b/freedreno/freedreno_bo_cache.c
@@ -0,0 +1,220 @@
+/* -*- mode: C; c-file-style: "k&r"; tab-width 4; indent-tabs-mode: t; -*- */
+
+/*
+ * Copyright (C) 2016 Rob Clark <robclark@freedesktop.org>
+ *
+ * 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.
+ *
+ * Authors:
+ * Rob Clark <robclark@freedesktop.org>
+ */
+
+#include "freedreno_drmif.h"
+#include "freedreno_priv.h"
+
+drm_private void bo_del(struct fd_bo *bo);
+drm_private extern pthread_mutex_t table_lock;
+
+static void
+add_bucket(struct fd_bo_cache *cache, int size)
+{
+ unsigned int i = cache->num_buckets;
+
+ assert(i < ARRAY_SIZE(cache->cache_bucket));
+
+ list_inithead(&cache->cache_bucket[i].list);
+ cache->cache_bucket[i].size = size;
+ cache->num_buckets++;
+}
+
+/**
+ * @coarse: if true, only power-of-two bucket sizes, otherwise
+ * fill in for a bit smoother size curve..
+ */
+drm_private void
+fd_bo_cache_init(struct fd_bo_cache *cache, int coarse)
+{
+ unsigned long size, cache_max_size = 64 * 1024 * 1024;
+
+ /* OK, so power of two buckets was too wasteful of memory.
+ * Give 3 other sizes between each power of two, to hopefully
+ * cover things accurately enough. (The alternative is
+ * probably to just go for exact matching of sizes, and assume
+ * that for things like composited window resize the tiled
+ * width/height alignment and rounding of sizes to pages will
+ * get us useful cache hit rates anyway)
+ */
+ add_bucket(cache, 4096);
+ add_bucket(cache, 4096 * 2);
+ if (!coarse)
+ add_bucket(cache, 4096 * 3);
+
+ /* Initialize the linked lists for BO reuse cache. */
+ for (size = 4 * 4096; size <= cache_max_size; size *= 2) {
+ add_bucket(cache, size);
+ if (!coarse) {
+ add_bucket(cache, size + size * 1 / 4);
+ add_bucket(cache, size + size * 2 / 4);
+ add_bucket(cache, size + size * 3 / 4);
+ }
+ }
+}
+
+/* Frees older cached buffers. Called under table_lock */
+drm_private void
+fd_bo_cache_cleanup(struct fd_bo_cache *cache, time_t time)
+{
+ int i;
+
+ if (cache->time == time)
+ return;
+
+ for (i = 0; i < cache->num_buckets; i++) {
+ struct fd_bo_bucket *bucket = &cache->cache_bucket[i];
+ struct fd_bo *bo;
+
+ while (!LIST_IS_EMPTY(&bucket->list)) {
+ bo = LIST_ENTRY(struct fd_bo, bucket->list.next, list);
+
+ /* keep things in cache for at least 1 second: */
+ if (time && ((time - bo->free_time) <= 1))
+ break;
+
+ VG_BO_OBTAIN(bo);
+ list_del(&bo->list);
+ bo_del(bo);
+ }
+ }
+
+ cache->time = time;
+}
+
+static struct fd_bo_bucket * get_bucket(struct fd_bo_cache *cache, uint32_t size)
+{
+ int i;
+
+ /* hmm, this is what intel does, but I suppose we could calculate our
+ * way to the correct bucket size rather than looping..
+ */
+ for (i = 0; i < cache->num_buckets; i++) {
+ struct fd_bo_bucket *bucket = &cache->cache_bucket[i];
+ if (bucket->size >= size) {
+ return bucket;
+ }
+ }
+
+ return NULL;
+}
+
+static int is_idle(struct fd_bo *bo)
+{
+ return fd_bo_cpu_prep(bo, NULL,
+ DRM_FREEDRENO_PREP_READ |
+ DRM_FREEDRENO_PREP_WRITE |
+ DRM_FREEDRENO_PREP_NOSYNC) == 0;
+}
+
+static struct fd_bo *find_in_bucket(struct fd_bo_bucket *bucket, uint32_t flags)
+{
+ struct fd_bo *bo = NULL;
+
+ /* TODO .. if we had an ALLOC_FOR_RENDER flag like intel, we could
+ * skip the busy check.. if it is only going to be a render target
+ * then we probably don't need to stall..
+ *
+ * NOTE that intel takes ALLOC_FOR_RENDER bo's from the list tail
+ * (MRU, since likely to be in GPU cache), rather than head (LRU)..
+ */
+ pthread_mutex_lock(&table_lock);
+ if (!LIST_IS_EMPTY(&bucket->list)) {
+ bo = LIST_ENTRY(struct fd_bo, bucket->list.next, list);
+ /* TODO check for compatible flags? */
+ if (is_idle(bo)) {
+ list_del(&bo->list);
+ } else {
+ bo = NULL;
+ }
+ }
+ pthread_mutex_unlock(&table_lock);
+
+ return bo;
+}
+
+/* NOTE: size is potentially rounded up to bucket size: */
+drm_private struct fd_bo *
+fd_bo_cache_alloc(struct fd_bo_cache *cache, uint32_t *size, uint32_t flags)
+{
+ struct fd_bo *bo = NULL;
+ struct fd_bo_bucket *bucket;
+
+ *size = ALIGN(*size, 4096);
+ bucket = get_bucket(cache, *size);
+
+ /* see if we can be green and recycle: */
+retry:
+ if (bucket) {
+ *size = bucket->size;
+ bo = find_in_bucket(bucket, flags);
+ if (bo) {
+ VG_BO_OBTAIN(bo);
+ if (bo->funcs->madvise(bo, TRUE) <= 0) {
+ /* we've lost the backing pages, delete and try again: */
+ pthread_mutex_lock(&table_lock);
+ bo_del(bo);
+ pthread_mutex_unlock(&table_lock);
+ goto retry;
+ }
+ atomic_set(&bo->refcnt, 1);
+ fd_device_ref(bo->dev);
+ return bo;
+ }
+ }
+
+ return NULL;
+}
+
+drm_private int
+fd_bo_cache_free(struct fd_bo_cache *cache, struct fd_bo *bo)
+{
+ struct fd_bo_bucket *bucket = get_bucket(cache, bo->size);
+
+ /* see if we can be green and recycle: */
+ if (bucket) {
+ struct timespec time;
+
+ bo->funcs->madvise(bo, FALSE);
+
+ clock_gettime(CLOCK_MONOTONIC, &time);
+
+ bo->free_time = time.tv_sec;
+ VG_BO_RELEASE(bo);
+ list_addtail(&bo->list, &bucket->list);
+ fd_bo_cache_cleanup(cache, time.tv_sec);
+
+ /* bo's in the bucket cache don't have a ref and
+ * don't hold a ref to the dev:
+ */
+ fd_device_del_locked(bo->dev);
+
+ return 0;
+ }
+
+ return -1;
+}