From: John Stultz <john.stultz@linaro.org> To: lkml <linux-kernel@vger.kernel.org> Cc: "John Stultz" <john.stultz@linaro.org>, "Daniel Vetter" <daniel@ffwll.ch>, "Christian Koenig" <christian.koenig@amd.com>, "Sumit Semwal" <sumit.semwal@linaro.org>, "Liam Mark" <lmark@codeaurora.org>, "Chris Goldsworthy" <cgoldswo@codeaurora.org>, "Laura Abbott" <labbott@kernel.org>, "Brian Starkey" <Brian.Starkey@arm.com>, "Hridya Valsaraju" <hridya@google.com>, "Suren Baghdasaryan" <surenb@google.com>, "Sandeep Patil" <sspatil@google.com>, "Daniel Mentz" <danielmentz@google.com>, "Ørjan Eide" <orjan.eide@arm.com>, "Robin Murphy" <robin.murphy@arm.com>, "Ezequiel Garcia" <ezequiel@collabora.com>, "Simon Ser" <contact@emersion.fr>, "James Jones" <jajones@nvidia.com>, linux-media@vger.kernel.org, dri-devel@lists.freedesktop.org Subject: [PATCH v9 3/5] dma-buf: system_heap: Add drm pagepool support to system heap Date: Wed, 30 Jun 2021 01:34:19 +0000 [thread overview] Message-ID: <20210630013421.735092-4-john.stultz@linaro.org> (raw) In-Reply-To: <20210630013421.735092-1-john.stultz@linaro.org> Utilize the drm pagepool code to speed up allocation performance. This is similar to the ION pagepool usage, but tries to utilize generic code instead of a custom implementation. Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Christian Koenig <christian.koenig@amd.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Liam Mark <lmark@codeaurora.org> Cc: Chris Goldsworthy <cgoldswo@codeaurora.org> Cc: Laura Abbott <labbott@kernel.org> Cc: Brian Starkey <Brian.Starkey@arm.com> Cc: Hridya Valsaraju <hridya@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Daniel Mentz <danielmentz@google.com> Cc: Ørjan Eide <orjan.eide@arm.com> Cc: Robin Murphy <robin.murphy@arm.com> Cc: Ezequiel Garcia <ezequiel@collabora.com> Cc: Simon Ser <contact@emersion.fr> Cc: James Jones <jajones@nvidia.com> Cc: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Signed-off-by: John Stultz <john.stultz@linaro.org> --- v2: * Fix build issue caused by selecting PAGE_POOL w/o NET as Reported-by: kernel test robot <lkp@intel.com> v3: * Simplify the page zeroing logic a bit by using kmap_atomic instead of vmap as suggested by Daniel Mentz v5: * Shift away from networking page pool completely to dmabuf page pool implementation v6: * Switch again to using the drm_page_pool code shared w/ ttm_pool v7: * Slight rework for drm_page_pool changes v8: * Rework to use the rewritten drm_page_pool logic * Drop explicit buffer zeroing, as the drm_page_pool handles that v9: * Fix compiler warning Reported-by: kernel test robot <lkp@intel.com> --- drivers/dma-buf/heaps/Kconfig | 1 + drivers/dma-buf/heaps/system_heap.c | 26 +++++++++++++++++++++++--- 2 files changed, 24 insertions(+), 3 deletions(-) diff --git a/drivers/dma-buf/heaps/Kconfig b/drivers/dma-buf/heaps/Kconfig index a5eef06c4226..f19bf1f82bc2 100644 --- a/drivers/dma-buf/heaps/Kconfig +++ b/drivers/dma-buf/heaps/Kconfig @@ -1,6 +1,7 @@ config DMABUF_HEAPS_SYSTEM bool "DMA-BUF System Heap" depends on DMABUF_HEAPS + select DRM_PAGE_POOL help Choose this option to enable the system dmabuf heap. The system heap is backed by pages from the buddy allocator. If in doubt, say Y. diff --git a/drivers/dma-buf/heaps/system_heap.c b/drivers/dma-buf/heaps/system_heap.c index f57a39ddd063..85ceca2ed61d 100644 --- a/drivers/dma-buf/heaps/system_heap.c +++ b/drivers/dma-buf/heaps/system_heap.c @@ -21,6 +21,8 @@ #include <linux/slab.h> #include <linux/vmalloc.h> +#include <drm/page_pool.h> + static struct dma_heap *sys_heap; struct system_heap_buffer { @@ -54,6 +56,7 @@ static gfp_t order_flags[] = {HIGH_ORDER_GFP, MID_ORDER_GFP, LOW_ORDER_GFP}; */ static const unsigned int orders[] = {8, 4, 0}; #define NUM_ORDERS ARRAY_SIZE(orders) +struct drm_page_pool pools[NUM_ORDERS]; static struct sg_table *dup_sg_table(struct sg_table *table) { @@ -282,18 +285,27 @@ static void system_heap_vunmap(struct dma_buf *dmabuf, struct dma_buf_map *map) dma_buf_map_clear(map); } +static void system_heap_free_pages(struct drm_page_pool *pool, struct page *p) +{ + __free_pages(p, pool->order); +} + static void system_heap_dma_buf_release(struct dma_buf *dmabuf) { struct system_heap_buffer *buffer = dmabuf->priv; struct sg_table *table; struct scatterlist *sg; - int i; + int i, j; table = &buffer->sg_table; for_each_sg(table->sgl, sg, table->nents, i) { struct page *page = sg_page(sg); - __free_pages(page, compound_order(page)); + for (j = 0; j < NUM_ORDERS; j++) { + if (compound_order(page) == orders[j]) + break; + } + drm_page_pool_add(&pools[j], page); } sg_free_table(table); kfree(buffer); @@ -324,7 +336,9 @@ static struct page *alloc_largest_available(unsigned long size, if (max_order < orders[i]) continue; - page = alloc_pages(order_flags[i], orders[i]); + page = drm_page_pool_remove(&pools[i]); + if (!page) + page = alloc_pages(order_flags[i], orders[i]); if (!page) continue; return page; @@ -425,6 +439,12 @@ static const struct dma_heap_ops system_heap_ops = { static int system_heap_create(void) { struct dma_heap_export_info exp_info; + int i; + + for (i = 0; i < NUM_ORDERS; i++) { + drm_page_pool_init(&pools[i], orders[i], + system_heap_free_pages); + } exp_info.name = "system"; exp_info.ops = &system_heap_ops; -- 2.25.1
WARNING: multiple messages have this Message-ID (diff)
From: John Stultz <john.stultz@linaro.org> To: lkml <linux-kernel@vger.kernel.org> Cc: dri-devel@lists.freedesktop.org, "Sandeep Patil" <sspatil@google.com>, "Ezequiel Garcia" <ezequiel@collabora.com>, "Robin Murphy" <robin.murphy@arm.com>, "James Jones" <jajones@nvidia.com>, "Liam Mark" <lmark@codeaurora.org>, "Laura Abbott" <labbott@kernel.org>, "Chris Goldsworthy" <cgoldswo@codeaurora.org>, "Hridya Valsaraju" <hridya@google.com>, "Ørjan Eide" <orjan.eide@arm.com>, linux-media@vger.kernel.org, "Suren Baghdasaryan" <surenb@google.com>, "Christian Koenig" <christian.koenig@amd.com>, "Daniel Mentz" <danielmentz@google.com> Subject: [PATCH v9 3/5] dma-buf: system_heap: Add drm pagepool support to system heap Date: Wed, 30 Jun 2021 01:34:19 +0000 [thread overview] Message-ID: <20210630013421.735092-4-john.stultz@linaro.org> (raw) In-Reply-To: <20210630013421.735092-1-john.stultz@linaro.org> Utilize the drm pagepool code to speed up allocation performance. This is similar to the ION pagepool usage, but tries to utilize generic code instead of a custom implementation. Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Christian Koenig <christian.koenig@amd.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Liam Mark <lmark@codeaurora.org> Cc: Chris Goldsworthy <cgoldswo@codeaurora.org> Cc: Laura Abbott <labbott@kernel.org> Cc: Brian Starkey <Brian.Starkey@arm.com> Cc: Hridya Valsaraju <hridya@google.com> Cc: Suren Baghdasaryan <surenb@google.com> Cc: Sandeep Patil <sspatil@google.com> Cc: Daniel Mentz <danielmentz@google.com> Cc: Ørjan Eide <orjan.eide@arm.com> Cc: Robin Murphy <robin.murphy@arm.com> Cc: Ezequiel Garcia <ezequiel@collabora.com> Cc: Simon Ser <contact@emersion.fr> Cc: James Jones <jajones@nvidia.com> Cc: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Signed-off-by: John Stultz <john.stultz@linaro.org> --- v2: * Fix build issue caused by selecting PAGE_POOL w/o NET as Reported-by: kernel test robot <lkp@intel.com> v3: * Simplify the page zeroing logic a bit by using kmap_atomic instead of vmap as suggested by Daniel Mentz v5: * Shift away from networking page pool completely to dmabuf page pool implementation v6: * Switch again to using the drm_page_pool code shared w/ ttm_pool v7: * Slight rework for drm_page_pool changes v8: * Rework to use the rewritten drm_page_pool logic * Drop explicit buffer zeroing, as the drm_page_pool handles that v9: * Fix compiler warning Reported-by: kernel test robot <lkp@intel.com> --- drivers/dma-buf/heaps/Kconfig | 1 + drivers/dma-buf/heaps/system_heap.c | 26 +++++++++++++++++++++++--- 2 files changed, 24 insertions(+), 3 deletions(-) diff --git a/drivers/dma-buf/heaps/Kconfig b/drivers/dma-buf/heaps/Kconfig index a5eef06c4226..f19bf1f82bc2 100644 --- a/drivers/dma-buf/heaps/Kconfig +++ b/drivers/dma-buf/heaps/Kconfig @@ -1,6 +1,7 @@ config DMABUF_HEAPS_SYSTEM bool "DMA-BUF System Heap" depends on DMABUF_HEAPS + select DRM_PAGE_POOL help Choose this option to enable the system dmabuf heap. The system heap is backed by pages from the buddy allocator. If in doubt, say Y. diff --git a/drivers/dma-buf/heaps/system_heap.c b/drivers/dma-buf/heaps/system_heap.c index f57a39ddd063..85ceca2ed61d 100644 --- a/drivers/dma-buf/heaps/system_heap.c +++ b/drivers/dma-buf/heaps/system_heap.c @@ -21,6 +21,8 @@ #include <linux/slab.h> #include <linux/vmalloc.h> +#include <drm/page_pool.h> + static struct dma_heap *sys_heap; struct system_heap_buffer { @@ -54,6 +56,7 @@ static gfp_t order_flags[] = {HIGH_ORDER_GFP, MID_ORDER_GFP, LOW_ORDER_GFP}; */ static const unsigned int orders[] = {8, 4, 0}; #define NUM_ORDERS ARRAY_SIZE(orders) +struct drm_page_pool pools[NUM_ORDERS]; static struct sg_table *dup_sg_table(struct sg_table *table) { @@ -282,18 +285,27 @@ static void system_heap_vunmap(struct dma_buf *dmabuf, struct dma_buf_map *map) dma_buf_map_clear(map); } +static void system_heap_free_pages(struct drm_page_pool *pool, struct page *p) +{ + __free_pages(p, pool->order); +} + static void system_heap_dma_buf_release(struct dma_buf *dmabuf) { struct system_heap_buffer *buffer = dmabuf->priv; struct sg_table *table; struct scatterlist *sg; - int i; + int i, j; table = &buffer->sg_table; for_each_sg(table->sgl, sg, table->nents, i) { struct page *page = sg_page(sg); - __free_pages(page, compound_order(page)); + for (j = 0; j < NUM_ORDERS; j++) { + if (compound_order(page) == orders[j]) + break; + } + drm_page_pool_add(&pools[j], page); } sg_free_table(table); kfree(buffer); @@ -324,7 +336,9 @@ static struct page *alloc_largest_available(unsigned long size, if (max_order < orders[i]) continue; - page = alloc_pages(order_flags[i], orders[i]); + page = drm_page_pool_remove(&pools[i]); + if (!page) + page = alloc_pages(order_flags[i], orders[i]); if (!page) continue; return page; @@ -425,6 +439,12 @@ static const struct dma_heap_ops system_heap_ops = { static int system_heap_create(void) { struct dma_heap_export_info exp_info; + int i; + + for (i = 0; i < NUM_ORDERS; i++) { + drm_page_pool_init(&pools[i], orders[i], + system_heap_free_pages); + } exp_info.name = "system"; exp_info.ops = &system_heap_ops; -- 2.25.1
next prev parent reply other threads:[~2021-06-30 1:34 UTC|newest] Thread overview: 49+ messages / expand[flat|nested] mbox.gz Atom feed top 2021-06-30 1:34 [PATCH v9 0/5] Generic page pool & deferred freeing for system dmabuf hea John Stultz 2021-06-30 1:34 ` John Stultz 2021-06-30 1:34 ` [PATCH v9 1/5] drm: Add a sharable drm page-pool implementation John Stultz 2021-06-30 1:34 ` John Stultz 2021-06-30 9:10 ` Christian König 2021-06-30 9:10 ` Christian König 2021-06-30 22:24 ` John Stultz 2021-06-30 22:24 ` John Stultz 2021-07-01 6:52 ` Christian König 2021-07-01 6:52 ` Christian König 2021-07-06 21:03 ` John Stultz 2021-07-06 21:03 ` John Stultz 2021-07-06 21:15 ` Daniel Vetter 2021-07-06 21:15 ` Daniel Vetter 2021-07-06 21:19 ` John Stultz 2021-07-06 21:19 ` John Stultz 2021-07-07 6:52 ` Christian König 2021-07-07 6:52 ` Christian König 2021-07-07 6:38 ` page pools, was " Christoph Hellwig 2021-07-07 7:10 ` Christian König 2021-07-07 7:10 ` Christian König 2021-07-07 7:14 ` Christoph Hellwig 2021-07-07 9:32 ` Christian König 2021-07-07 9:32 ` Christian König 2021-07-07 19:42 ` John Stultz 2021-07-07 19:42 ` John Stultz 2021-07-07 19:42 ` John Stultz 2021-07-07 19:35 ` John Stultz 2021-07-07 19:35 ` John Stultz 2021-07-07 19:35 ` John Stultz 2021-07-08 4:20 ` Christoph Hellwig 2021-07-08 7:37 ` Christian König 2021-07-08 7:37 ` Christian König 2021-06-30 1:34 ` [PATCH v9 2/5] drm: ttm_pool: Rework ttm_pool to use drm_page_pool John Stultz 2021-06-30 1:34 ` John Stultz 2021-06-30 5:11 ` kernel test robot 2021-06-30 5:11 ` kernel test robot 2021-06-30 1:34 ` John Stultz [this message] 2021-06-30 1:34 ` [PATCH v9 3/5] dma-buf: system_heap: Add drm pagepool support to system heap John Stultz 2021-06-30 4:34 ` kernel test robot 2021-06-30 4:34 ` kernel test robot 2021-06-30 5:25 ` kernel test robot 2021-06-30 5:25 ` kernel test robot 2021-06-30 1:34 ` [PATCH v9 4/5] dma-buf: heaps: Add deferred-free-helper library code John Stultz 2021-06-30 1:34 ` John Stultz 2021-06-30 1:34 ` [PATCH v9 5/5] dma-buf: system_heap: Add deferred freeing to the system heap John Stultz 2021-06-30 1:34 ` John Stultz 2021-06-30 9:13 ` [PATCH v9 0/5] Generic page pool & deferred freeing for system dmabuf hea Christian König 2021-06-30 9:13 ` Christian König
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=20210630013421.735092-4-john.stultz@linaro.org \ --to=john.stultz@linaro.org \ --cc=Brian.Starkey@arm.com \ --cc=cgoldswo@codeaurora.org \ --cc=christian.koenig@amd.com \ --cc=contact@emersion.fr \ --cc=daniel@ffwll.ch \ --cc=danielmentz@google.com \ --cc=dri-devel@lists.freedesktop.org \ --cc=ezequiel@collabora.com \ --cc=hridya@google.com \ --cc=jajones@nvidia.com \ --cc=labbott@kernel.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-media@vger.kernel.org \ --cc=lmark@codeaurora.org \ --cc=orjan.eide@arm.com \ --cc=robin.murphy@arm.com \ --cc=sspatil@google.com \ --cc=sumit.semwal@linaro.org \ --cc=surenb@google.com \ /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: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.