From: "Thomas Hellström" <thomas.hellstrom@linux.intel.com> To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org Cc: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>, matthew.auld@intel.com Subject: [PATCH v6 1/4] drm/i915: Avoid allocating a page array for the gpu coredump Date: Mon, 8 Nov 2021 18:45:44 +0100 [thread overview] Message-ID: <20211108174547.979714-2-thomas.hellstrom@linux.intel.com> (raw) In-Reply-To: <20211108174547.979714-1-thomas.hellstrom@linux.intel.com> The gpu coredump typically takes place in a dma_fence signalling critical path, and hence can't use GFP_KERNEL allocations, as that means we might hit deadlocks under memory pressure. However changing to __GFP_KSWAPD_RECLAIM which will be done in an upcoming patch will instead mean a lower chance of the allocation succeeding. In particular large contigous allocations like the coredump page vector. Remove the page vector in favor of a linked list of single pages. Use the page lru list head as the list link, as the page owner is allowed to do that. Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com> --- drivers/gpu/drm/i915/i915_gpu_error.c | 50 +++++++++++++++------------ drivers/gpu/drm/i915/i915_gpu_error.h | 4 +-- 2 files changed, 28 insertions(+), 26 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c index 2a2d7643b551..14de64282697 100644 --- a/drivers/gpu/drm/i915/i915_gpu_error.c +++ b/drivers/gpu/drm/i915/i915_gpu_error.c @@ -275,16 +275,16 @@ static bool compress_start(struct i915_vma_compress *c) static void *compress_next_page(struct i915_vma_compress *c, struct i915_vma_coredump *dst) { - void *page; + void *page_addr; + struct page *page; - if (dst->page_count >= dst->num_pages) - return ERR_PTR(-ENOSPC); - - page = pool_alloc(&c->pool, ALLOW_FAIL); - if (!page) + page_addr = pool_alloc(&c->pool, ALLOW_FAIL); + if (!page_addr) return ERR_PTR(-ENOMEM); - return dst->pages[dst->page_count++] = page; + page = virt_to_page(page_addr); + list_add_tail(&page->lru, &dst->page_list); + return page_addr; } static int compress_page(struct i915_vma_compress *c, @@ -397,7 +397,7 @@ static int compress_page(struct i915_vma_compress *c, if (!(wc && i915_memcpy_from_wc(ptr, src, PAGE_SIZE))) memcpy(ptr, src, PAGE_SIZE); - dst->pages[dst->page_count++] = ptr; + list_add_tail(&virt_to_page(ptr)->lru, &dst->page_list); cond_resched(); return 0; @@ -614,7 +614,7 @@ static void print_error_vma(struct drm_i915_error_state_buf *m, const struct i915_vma_coredump *vma) { char out[ASCII85_BUFSZ]; - int page; + struct page *page; if (!vma) return; @@ -628,16 +628,17 @@ static void print_error_vma(struct drm_i915_error_state_buf *m, err_printf(m, "gtt_page_sizes = 0x%08x\n", vma->gtt_page_sizes); err_compression_marker(m); - for (page = 0; page < vma->page_count; page++) { + list_for_each_entry(page, &vma->page_list, lru) { int i, len; + const u32 *addr = page_address(page); len = PAGE_SIZE; - if (page == vma->page_count - 1) + if (page == list_last_entry(&vma->page_list, typeof(*page), lru)) len -= vma->unused; len = ascii85_encode_len(len); for (i = 0; i < len; i++) - err_puts(m, ascii85_encode(vma->pages[page][i], out)); + err_puts(m, ascii85_encode(addr[i], out)); } err_puts(m, "\n"); } @@ -946,10 +947,12 @@ static void i915_vma_coredump_free(struct i915_vma_coredump *vma) { while (vma) { struct i915_vma_coredump *next = vma->next; - int page; + struct page *page, *n; - for (page = 0; page < vma->page_count; page++) - free_page((unsigned long)vma->pages[page]); + list_for_each_entry_safe(page, n, &vma->page_list, lru) { + list_del_init(&page->lru); + __free_page(page); + } kfree(vma); vma = next; @@ -1016,7 +1019,6 @@ i915_vma_coredump_create(const struct intel_gt *gt, struct i915_ggtt *ggtt = gt->ggtt; const u64 slot = ggtt->error_capture.start; struct i915_vma_coredump *dst; - unsigned long num_pages; struct sgt_iter iter; int ret; @@ -1025,9 +1027,7 @@ i915_vma_coredump_create(const struct intel_gt *gt, if (!vma || !vma->pages || !compress) return NULL; - num_pages = min_t(u64, vma->size, vma->obj->base.size) >> PAGE_SHIFT; - num_pages = DIV_ROUND_UP(10 * num_pages, 8); /* worstcase zlib growth */ - dst = kmalloc(sizeof(*dst) + num_pages * sizeof(u32 *), ALLOW_FAIL); + dst = kmalloc(sizeof(*dst), ALLOW_FAIL); if (!dst) return NULL; @@ -1036,14 +1036,13 @@ i915_vma_coredump_create(const struct intel_gt *gt, return NULL; } + INIT_LIST_HEAD(&dst->page_list); strcpy(dst->name, name); dst->next = NULL; dst->gtt_offset = vma->node.start; dst->gtt_size = vma->node.size; dst->gtt_page_sizes = vma->page_sizes.gtt; - dst->num_pages = num_pages; - dst->page_count = 0; dst->unused = 0; ret = -EINVAL; @@ -1106,8 +1105,13 @@ i915_vma_coredump_create(const struct intel_gt *gt, } if (ret || compress_flush(compress, dst)) { - while (dst->page_count--) - pool_free(&compress->pool, dst->pages[dst->page_count]); + struct page *page, *n; + + list_for_each_entry_safe_reverse(page, n, &dst->page_list, lru) { + list_del_init(&page->lru); + pool_free(&compress->pool, page_address(page)); + } + kfree(dst); dst = NULL; } diff --git a/drivers/gpu/drm/i915/i915_gpu_error.h b/drivers/gpu/drm/i915/i915_gpu_error.h index b98d8cdbe4f2..5aedf5129814 100644 --- a/drivers/gpu/drm/i915/i915_gpu_error.h +++ b/drivers/gpu/drm/i915/i915_gpu_error.h @@ -39,10 +39,8 @@ struct i915_vma_coredump { u64 gtt_size; u32 gtt_page_sizes; - int num_pages; - int page_count; int unused; - u32 *pages[]; + struct list_head page_list; }; struct i915_request_coredump { -- 2.31.1
WARNING: multiple messages have this Message-ID (diff)
From: "Thomas Hellström" <thomas.hellstrom@linux.intel.com> To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org Cc: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>, matthew.auld@intel.com Subject: [Intel-gfx] [PATCH v6 1/4] drm/i915: Avoid allocating a page array for the gpu coredump Date: Mon, 8 Nov 2021 18:45:44 +0100 [thread overview] Message-ID: <20211108174547.979714-2-thomas.hellstrom@linux.intel.com> (raw) In-Reply-To: <20211108174547.979714-1-thomas.hellstrom@linux.intel.com> The gpu coredump typically takes place in a dma_fence signalling critical path, and hence can't use GFP_KERNEL allocations, as that means we might hit deadlocks under memory pressure. However changing to __GFP_KSWAPD_RECLAIM which will be done in an upcoming patch will instead mean a lower chance of the allocation succeeding. In particular large contigous allocations like the coredump page vector. Remove the page vector in favor of a linked list of single pages. Use the page lru list head as the list link, as the page owner is allowed to do that. Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com> --- drivers/gpu/drm/i915/i915_gpu_error.c | 50 +++++++++++++++------------ drivers/gpu/drm/i915/i915_gpu_error.h | 4 +-- 2 files changed, 28 insertions(+), 26 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c index 2a2d7643b551..14de64282697 100644 --- a/drivers/gpu/drm/i915/i915_gpu_error.c +++ b/drivers/gpu/drm/i915/i915_gpu_error.c @@ -275,16 +275,16 @@ static bool compress_start(struct i915_vma_compress *c) static void *compress_next_page(struct i915_vma_compress *c, struct i915_vma_coredump *dst) { - void *page; + void *page_addr; + struct page *page; - if (dst->page_count >= dst->num_pages) - return ERR_PTR(-ENOSPC); - - page = pool_alloc(&c->pool, ALLOW_FAIL); - if (!page) + page_addr = pool_alloc(&c->pool, ALLOW_FAIL); + if (!page_addr) return ERR_PTR(-ENOMEM); - return dst->pages[dst->page_count++] = page; + page = virt_to_page(page_addr); + list_add_tail(&page->lru, &dst->page_list); + return page_addr; } static int compress_page(struct i915_vma_compress *c, @@ -397,7 +397,7 @@ static int compress_page(struct i915_vma_compress *c, if (!(wc && i915_memcpy_from_wc(ptr, src, PAGE_SIZE))) memcpy(ptr, src, PAGE_SIZE); - dst->pages[dst->page_count++] = ptr; + list_add_tail(&virt_to_page(ptr)->lru, &dst->page_list); cond_resched(); return 0; @@ -614,7 +614,7 @@ static void print_error_vma(struct drm_i915_error_state_buf *m, const struct i915_vma_coredump *vma) { char out[ASCII85_BUFSZ]; - int page; + struct page *page; if (!vma) return; @@ -628,16 +628,17 @@ static void print_error_vma(struct drm_i915_error_state_buf *m, err_printf(m, "gtt_page_sizes = 0x%08x\n", vma->gtt_page_sizes); err_compression_marker(m); - for (page = 0; page < vma->page_count; page++) { + list_for_each_entry(page, &vma->page_list, lru) { int i, len; + const u32 *addr = page_address(page); len = PAGE_SIZE; - if (page == vma->page_count - 1) + if (page == list_last_entry(&vma->page_list, typeof(*page), lru)) len -= vma->unused; len = ascii85_encode_len(len); for (i = 0; i < len; i++) - err_puts(m, ascii85_encode(vma->pages[page][i], out)); + err_puts(m, ascii85_encode(addr[i], out)); } err_puts(m, "\n"); } @@ -946,10 +947,12 @@ static void i915_vma_coredump_free(struct i915_vma_coredump *vma) { while (vma) { struct i915_vma_coredump *next = vma->next; - int page; + struct page *page, *n; - for (page = 0; page < vma->page_count; page++) - free_page((unsigned long)vma->pages[page]); + list_for_each_entry_safe(page, n, &vma->page_list, lru) { + list_del_init(&page->lru); + __free_page(page); + } kfree(vma); vma = next; @@ -1016,7 +1019,6 @@ i915_vma_coredump_create(const struct intel_gt *gt, struct i915_ggtt *ggtt = gt->ggtt; const u64 slot = ggtt->error_capture.start; struct i915_vma_coredump *dst; - unsigned long num_pages; struct sgt_iter iter; int ret; @@ -1025,9 +1027,7 @@ i915_vma_coredump_create(const struct intel_gt *gt, if (!vma || !vma->pages || !compress) return NULL; - num_pages = min_t(u64, vma->size, vma->obj->base.size) >> PAGE_SHIFT; - num_pages = DIV_ROUND_UP(10 * num_pages, 8); /* worstcase zlib growth */ - dst = kmalloc(sizeof(*dst) + num_pages * sizeof(u32 *), ALLOW_FAIL); + dst = kmalloc(sizeof(*dst), ALLOW_FAIL); if (!dst) return NULL; @@ -1036,14 +1036,13 @@ i915_vma_coredump_create(const struct intel_gt *gt, return NULL; } + INIT_LIST_HEAD(&dst->page_list); strcpy(dst->name, name); dst->next = NULL; dst->gtt_offset = vma->node.start; dst->gtt_size = vma->node.size; dst->gtt_page_sizes = vma->page_sizes.gtt; - dst->num_pages = num_pages; - dst->page_count = 0; dst->unused = 0; ret = -EINVAL; @@ -1106,8 +1105,13 @@ i915_vma_coredump_create(const struct intel_gt *gt, } if (ret || compress_flush(compress, dst)) { - while (dst->page_count--) - pool_free(&compress->pool, dst->pages[dst->page_count]); + struct page *page, *n; + + list_for_each_entry_safe_reverse(page, n, &dst->page_list, lru) { + list_del_init(&page->lru); + pool_free(&compress->pool, page_address(page)); + } + kfree(dst); dst = NULL; } diff --git a/drivers/gpu/drm/i915/i915_gpu_error.h b/drivers/gpu/drm/i915/i915_gpu_error.h index b98d8cdbe4f2..5aedf5129814 100644 --- a/drivers/gpu/drm/i915/i915_gpu_error.h +++ b/drivers/gpu/drm/i915/i915_gpu_error.h @@ -39,10 +39,8 @@ struct i915_vma_coredump { u64 gtt_size; u32 gtt_page_sizes; - int num_pages; - int page_count; int unused; - u32 *pages[]; + struct list_head page_list; }; struct i915_request_coredump { -- 2.31.1
next prev parent reply other threads:[~2021-11-08 18:01 UTC|newest] Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top 2021-11-08 17:45 [PATCH v6 0/4] drm/i915: Prepare error capture for asynchronous migration Thomas Hellström 2021-11-08 17:45 ` [Intel-gfx] " Thomas Hellström 2021-11-08 17:45 ` Thomas Hellström [this message] 2021-11-08 17:45 ` [Intel-gfx] [PATCH v6 1/4] drm/i915: Avoid allocating a page array for the gpu coredump Thomas Hellström 2021-11-22 15:08 ` Ramalingam C 2021-11-22 15:08 ` [Intel-gfx] " Ramalingam C 2021-11-08 17:45 ` [PATCH v6 2/4] drm/i915: Use __GFP_KSWAPD_RECLAIM in the capture code Thomas Hellström 2021-11-08 17:45 ` [Intel-gfx] " Thomas Hellström 2021-11-22 16:50 ` Ramalingam C 2021-11-08 17:45 ` [PATCH v6 3/4] drm/i915: Update error capture code to avoid using the current vma state Thomas Hellström 2021-11-08 17:45 ` [Intel-gfx] " Thomas Hellström 2021-11-29 9:13 ` Ramalingam C 2021-11-29 9:13 ` [Intel-gfx] " Ramalingam C 2021-11-08 17:45 ` [PATCH v6 4/4] drm/i915: Initial introduction of vma resources Thomas Hellström 2021-11-08 17:45 ` [Intel-gfx] " Thomas Hellström 2021-11-08 18:29 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915: Prepare error capture for asynchronous migration (rev2) Patchwork 2021-11-08 19:02 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork 2021-11-08 20:21 ` [Intel-gfx] ✓ Fi.CI.IGT: " Patchwork
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=20211108174547.979714-2-thomas.hellstrom@linux.intel.com \ --to=thomas.hellstrom@linux.intel.com \ --cc=dri-devel@lists.freedesktop.org \ --cc=intel-gfx@lists.freedesktop.org \ --cc=matthew.auld@intel.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.