From 3316497bcd73dcad971d79bed32571ed785a8c01 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Tue, 1 Nov 2011 15:45:57 -0400 Subject: drm/ttm: remove userspace backed ttm object support This was never use in none of the driver, properly using userspace page for bo would need more code (vma interaction mostly). Removing this dead code in preparation of ttm_tt & backend merge. Signed-off-by: Jerome Glisse Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Thomas Hellstrom --- drivers/gpu/drm/ttm/ttm_tt.c | 105 +------------------------------------------ 1 file changed, 1 insertion(+), 104 deletions(-) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index f9cc548d6d98..c68b0e770d16 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -63,43 +63,6 @@ static void ttm_tt_free_page_directory(struct ttm_tt *ttm) ttm->dma_address = NULL; } -static void ttm_tt_free_user_pages(struct ttm_tt *ttm) -{ - int write; - int dirty; - struct page *page; - int i; - struct ttm_backend *be = ttm->be; - - BUG_ON(!(ttm->page_flags & TTM_PAGE_FLAG_USER)); - write = ((ttm->page_flags & TTM_PAGE_FLAG_WRITE) != 0); - dirty = ((ttm->page_flags & TTM_PAGE_FLAG_USER_DIRTY) != 0); - - if (be) - be->func->clear(be); - - for (i = 0; i < ttm->num_pages; ++i) { - page = ttm->pages[i]; - if (page == NULL) - continue; - - if (page == ttm->dummy_read_page) { - BUG_ON(write); - continue; - } - - if (write && dirty && !PageReserved(page)) - set_page_dirty_lock(page); - - ttm->pages[i] = NULL; - ttm_mem_global_free(ttm->glob->mem_glob, PAGE_SIZE); - put_page(page); - } - ttm->state = tt_unpopulated; - ttm->first_himem_page = ttm->num_pages; - ttm->last_lomem_page = -1; -} - static struct page *__ttm_tt_get_page(struct ttm_tt *ttm, int index) { struct page *p; @@ -326,10 +289,7 @@ void ttm_tt_destroy(struct ttm_tt *ttm) } if (likely(ttm->pages != NULL)) { - if (ttm->page_flags & TTM_PAGE_FLAG_USER) - ttm_tt_free_user_pages(ttm); - else - ttm_tt_free_alloced_pages(ttm); + ttm_tt_free_alloced_pages(ttm); ttm_tt_free_page_directory(ttm); } @@ -341,45 +301,6 @@ void ttm_tt_destroy(struct ttm_tt *ttm) kfree(ttm); } -int ttm_tt_set_user(struct ttm_tt *ttm, - struct task_struct *tsk, - unsigned long start, unsigned long num_pages) -{ - struct mm_struct *mm = tsk->mm; - int ret; - int write = (ttm->page_flags & TTM_PAGE_FLAG_WRITE) != 0; - struct ttm_mem_global *mem_glob = ttm->glob->mem_glob; - - BUG_ON(num_pages != ttm->num_pages); - BUG_ON((ttm->page_flags & TTM_PAGE_FLAG_USER) == 0); - - /** - * Account user pages as lowmem pages for now. - */ - - ret = ttm_mem_global_alloc(mem_glob, num_pages * PAGE_SIZE, - false, false); - if (unlikely(ret != 0)) - return ret; - - down_read(&mm->mmap_sem); - ret = get_user_pages(tsk, mm, start, num_pages, - write, 0, ttm->pages, NULL); - up_read(&mm->mmap_sem); - - if (ret != num_pages && write) { - ttm_tt_free_user_pages(ttm); - ttm_mem_global_free(mem_glob, num_pages * PAGE_SIZE); - return -ENOMEM; - } - - ttm->tsk = tsk; - ttm->start = start; - ttm->state = tt_unbound; - - return 0; -} - struct ttm_tt *ttm_tt_create(struct ttm_bo_device *bdev, unsigned long size, uint32_t page_flags, struct page *dummy_read_page) { @@ -453,8 +374,6 @@ int ttm_tt_bind(struct ttm_tt *ttm, struct ttm_mem_reg *bo_mem) ttm->state = tt_bound; - if (ttm->page_flags & TTM_PAGE_FLAG_USER) - ttm->page_flags |= TTM_PAGE_FLAG_USER_DIRTY; return 0; } EXPORT_SYMBOL(ttm_tt_bind); @@ -470,16 +389,6 @@ static int ttm_tt_swapin(struct ttm_tt *ttm) int i; int ret = -ENOMEM; - if (ttm->page_flags & TTM_PAGE_FLAG_USER) { - ret = ttm_tt_set_user(ttm, ttm->tsk, ttm->start, - ttm->num_pages); - if (unlikely(ret != 0)) - return ret; - - ttm->page_flags &= ~TTM_PAGE_FLAG_SWAPPED; - return 0; - } - swap_storage = ttm->swap_storage; BUG_ON(swap_storage == NULL); @@ -530,18 +439,6 @@ int ttm_tt_swapout(struct ttm_tt *ttm, struct file *persistent_swap_storage) BUG_ON(ttm->state != tt_unbound && ttm->state != tt_unpopulated); BUG_ON(ttm->caching_state != tt_cached); - /* - * For user buffers, just unpin the pages, as there should be - * vma references. - */ - - if (ttm->page_flags & TTM_PAGE_FLAG_USER) { - ttm_tt_free_user_pages(ttm); - ttm->page_flags |= TTM_PAGE_FLAG_SWAPPED; - ttm->swap_storage = NULL; - return 0; - } - if (!persistent_swap_storage) { swap_storage = shmem_file_setup("ttm swap", ttm->num_pages << PAGE_SHIFT, -- cgit v1.2.3 From 667b7a27c221acaea844ba41f1c0a713aba44d31 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Tue, 1 Nov 2011 15:57:22 -0400 Subject: drm/ttm: remove split btw highmen and lowmem page Split btw highmem and lowmem page was rendered useless by the pool code. Remove it. Note further cleanup would change the ttm page allocation helper to actualy take an array instead of relying on list this could drasticly reduce the number of function call in the common case of allocation whole buffer. Signed-off-by: Jerome Glisse Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Thomas Hellstrom --- drivers/gpu/drm/ttm/ttm_tt.c | 11 ++--------- 1 file changed, 2 insertions(+), 9 deletions(-) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index c68b0e770d16..f0c5ffd446d9 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -70,7 +70,7 @@ static struct page *__ttm_tt_get_page(struct ttm_tt *ttm, int index) struct ttm_mem_global *mem_glob = ttm->glob->mem_glob; int ret; - while (NULL == (p = ttm->pages[index])) { + if (NULL == (p = ttm->pages[index])) { INIT_LIST_HEAD(&h); @@ -86,10 +86,7 @@ static struct page *__ttm_tt_get_page(struct ttm_tt *ttm, int index) if (unlikely(ret != 0)) goto out_err; - if (PageHighMem(p)) - ttm->pages[--ttm->first_himem_page] = p; - else - ttm->pages[++ttm->last_lomem_page] = p; + ttm->pages[index] = p; } return p; out_err: @@ -271,8 +268,6 @@ static void ttm_tt_free_alloced_pages(struct ttm_tt *ttm) ttm_put_pages(&h, count, ttm->page_flags, ttm->caching_state, ttm->dma_address); ttm->state = tt_unpopulated; - ttm->first_himem_page = ttm->num_pages; - ttm->last_lomem_page = -1; } void ttm_tt_destroy(struct ttm_tt *ttm) @@ -316,8 +311,6 @@ struct ttm_tt *ttm_tt_create(struct ttm_bo_device *bdev, unsigned long size, ttm->glob = bdev->glob; ttm->num_pages = (size + PAGE_SIZE - 1) >> PAGE_SHIFT; - ttm->first_himem_page = ttm->num_pages; - ttm->last_lomem_page = -1; ttm->caching_state = tt_cached; ttm->page_flags = page_flags; -- cgit v1.2.3 From 5e2656804a34f58c2bf557465ab77f8a26a500eb Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Thu, 3 Nov 2011 01:22:39 -0400 Subject: drm/ttm: use ttm put pages function to properly restore cache attribute On failure we need to make sure the page we free has wb cache attribute. Do this pas call the proper ttm page helper function. Signed-off-by: Jerome Glisse Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Thomas Hellstrom --- drivers/gpu/drm/ttm/ttm_tt.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index f0c5ffd446d9..90527a2ce04c 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -90,7 +90,10 @@ static struct page *__ttm_tt_get_page(struct ttm_tt *ttm, int index) } return p; out_err: - put_page(p); + INIT_LIST_HEAD(&h); + list_add(&p->lru, &h); + ttm_put_pages(&h, 1, ttm->page_flags, + ttm->caching_state, &ttm->dma_address[index]); return NULL; } -- cgit v1.2.3 From f9517e63ffae6a1062a0a2ac4eea60be49b9dfd4 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Tue, 1 Nov 2011 19:07:31 -0400 Subject: drm/ttm: test for dma_address array allocation failure Signed-off-by: Jerome Glisse Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Thomas Hellstrom --- drivers/gpu/drm/ttm/ttm_tt.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 90527a2ce04c..54bbbadba93c 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -320,7 +320,7 @@ struct ttm_tt *ttm_tt_create(struct ttm_bo_device *bdev, unsigned long size, ttm->dummy_read_page = dummy_read_page; ttm_tt_alloc_page_directory(ttm); - if (!ttm->pages) { + if (!ttm->pages || !ttm->dma_address) { ttm_tt_destroy(ttm); printk(KERN_ERR TTM_PFX "Failed allocating page table\n"); return NULL; -- cgit v1.2.3 From 822c4d9ae0d55a4fcea9f0a462bc6406a06692e2 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Thu, 10 Nov 2011 18:24:09 -0500 Subject: drm/ttm: page allocation use page array instead of list Use the ttm_tt pages array for pages allocations, move the list unwinding into the page allocation functions. Signed-off-by: Jerome Glisse --- drivers/gpu/drm/ttm/ttm_tt.c | 36 +++++++----------------------------- 1 file changed, 7 insertions(+), 29 deletions(-) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 54bbbadba93c..6e079dedfc4f 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -66,22 +66,16 @@ static void ttm_tt_free_page_directory(struct ttm_tt *ttm) static struct page *__ttm_tt_get_page(struct ttm_tt *ttm, int index) { struct page *p; - struct list_head h; struct ttm_mem_global *mem_glob = ttm->glob->mem_glob; int ret; if (NULL == (p = ttm->pages[index])) { - INIT_LIST_HEAD(&h); - - ret = ttm_get_pages(&h, ttm->page_flags, ttm->caching_state, 1, + ret = ttm_get_pages(&p, ttm->page_flags, ttm->caching_state, 1, &ttm->dma_address[index]); - if (ret != 0) return NULL; - p = list_first_entry(&h, struct page, lru); - ret = ttm_mem_global_alloc_page(mem_glob, p, false, false); if (unlikely(ret != 0)) goto out_err; @@ -90,9 +84,7 @@ static struct page *__ttm_tt_get_page(struct ttm_tt *ttm, int index) } return p; out_err: - INIT_LIST_HEAD(&h); - list_add(&p->lru, &h); - ttm_put_pages(&h, 1, ttm->page_flags, + ttm_put_pages(&p, 1, ttm->page_flags, ttm->caching_state, &ttm->dma_address[index]); return NULL; } @@ -243,33 +235,19 @@ EXPORT_SYMBOL(ttm_tt_set_placement_caching); static void ttm_tt_free_alloced_pages(struct ttm_tt *ttm) { - int i; - unsigned count = 0; - struct list_head h; - struct page *cur_page; struct ttm_backend *be = ttm->be; - - INIT_LIST_HEAD(&h); + unsigned i; if (be) be->func->clear(be); for (i = 0; i < ttm->num_pages; ++i) { - - cur_page = ttm->pages[i]; - ttm->pages[i] = NULL; - if (cur_page) { - if (page_count(cur_page) != 1) - printk(KERN_ERR TTM_PFX - "Erroneous page count. " - "Leaking pages.\n"); + if (ttm->pages[i]) { ttm_mem_global_free_page(ttm->glob->mem_glob, - cur_page); - list_add(&cur_page->lru, &h); - count++; + ttm->pages[i]); + ttm_put_pages(&ttm->pages[i], 1, ttm->page_flags, + ttm->caching_state, &ttm->dma_address[i]); } } - ttm_put_pages(&h, count, ttm->page_flags, ttm->caching_state, - ttm->dma_address); ttm->state = tt_unpopulated; } -- cgit v1.2.3 From 649bf3ca77343e3be1e0af8e21356fa569b1abd9 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Tue, 1 Nov 2011 20:46:13 -0400 Subject: drm/ttm: merge ttm_backend and ttm_tt V5 ttm_backend will only exist with a ttm_tt, and ttm_tt will only be of interest when bound to a backend. Merge them to avoid code and data duplication. V2 Rebase on top of memory accounting overhaul V3 Rebase on top of more memory accounting changes V4 Rebase on top of no memory account changes (where/when is my delorean when i need it ?) V5 make sure ttm is unbound before destroying, change commit message on suggestion from Tormod Volden Signed-off-by: Jerome Glisse Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Thomas Hellstrom --- drivers/gpu/drm/ttm/ttm_tt.c | 59 +++++++++++--------------------------------- 1 file changed, 14 insertions(+), 45 deletions(-) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 6e079dedfc4f..fbc90dce1de8 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -105,7 +105,6 @@ int ttm_tt_populate(struct ttm_tt *ttm) { struct page *page; unsigned long i; - struct ttm_backend *be; int ret; if (ttm->state != tt_unpopulated) @@ -117,16 +116,11 @@ int ttm_tt_populate(struct ttm_tt *ttm) return ret; } - be = ttm->be; - for (i = 0; i < ttm->num_pages; ++i) { page = __ttm_tt_get_page(ttm, i); if (!page) return -ENOMEM; } - - be->func->populate(be, ttm->num_pages, ttm->pages, - ttm->dummy_read_page, ttm->dma_address); ttm->state = tt_unbound; return 0; } @@ -235,11 +229,8 @@ EXPORT_SYMBOL(ttm_tt_set_placement_caching); static void ttm_tt_free_alloced_pages(struct ttm_tt *ttm) { - struct ttm_backend *be = ttm->be; unsigned i; - if (be) - be->func->clear(be); for (i = 0; i < ttm->num_pages; ++i) { if (ttm->pages[i]) { ttm_mem_global_free_page(ttm->glob->mem_glob, @@ -253,20 +244,15 @@ static void ttm_tt_free_alloced_pages(struct ttm_tt *ttm) void ttm_tt_destroy(struct ttm_tt *ttm) { - struct ttm_backend *be; - if (unlikely(ttm == NULL)) return; - be = ttm->be; - if (likely(be != NULL)) { - be->func->destroy(be); - ttm->be = NULL; + if (ttm->state == tt_bound) { + ttm_tt_unbind(ttm); } if (likely(ttm->pages != NULL)) { ttm_tt_free_alloced_pages(ttm); - ttm_tt_free_page_directory(ttm); } @@ -274,52 +260,38 @@ void ttm_tt_destroy(struct ttm_tt *ttm) ttm->swap_storage) fput(ttm->swap_storage); - kfree(ttm); + ttm->swap_storage = NULL; + ttm->func->destroy(ttm); } -struct ttm_tt *ttm_tt_create(struct ttm_bo_device *bdev, unsigned long size, - uint32_t page_flags, struct page *dummy_read_page) +int ttm_tt_init(struct ttm_tt *ttm, struct ttm_bo_device *bdev, + unsigned long size, uint32_t page_flags, + struct page *dummy_read_page) { - struct ttm_bo_driver *bo_driver = bdev->driver; - struct ttm_tt *ttm; - - if (!bo_driver) - return NULL; - - ttm = kzalloc(sizeof(*ttm), GFP_KERNEL); - if (!ttm) - return NULL; - + ttm->bdev = bdev; ttm->glob = bdev->glob; ttm->num_pages = (size + PAGE_SIZE - 1) >> PAGE_SHIFT; ttm->caching_state = tt_cached; ttm->page_flags = page_flags; - ttm->dummy_read_page = dummy_read_page; + ttm->state = tt_unpopulated; ttm_tt_alloc_page_directory(ttm); if (!ttm->pages || !ttm->dma_address) { ttm_tt_destroy(ttm); printk(KERN_ERR TTM_PFX "Failed allocating page table\n"); - return NULL; - } - ttm->be = bo_driver->create_ttm_backend_entry(bdev); - if (!ttm->be) { - ttm_tt_destroy(ttm); - printk(KERN_ERR TTM_PFX "Failed creating ttm backend entry\n"); - return NULL; + return -ENOMEM; } - ttm->state = tt_unpopulated; - return ttm; + return 0; } +EXPORT_SYMBOL(ttm_tt_init); void ttm_tt_unbind(struct ttm_tt *ttm) { int ret; - struct ttm_backend *be = ttm->be; if (ttm->state == tt_bound) { - ret = be->func->unbind(be); + ret = ttm->func->unbind(ttm); BUG_ON(ret); ttm->state = tt_unbound; } @@ -328,7 +300,6 @@ void ttm_tt_unbind(struct ttm_tt *ttm) int ttm_tt_bind(struct ttm_tt *ttm, struct ttm_mem_reg *bo_mem) { int ret = 0; - struct ttm_backend *be; if (!ttm) return -EINVAL; @@ -336,13 +307,11 @@ int ttm_tt_bind(struct ttm_tt *ttm, struct ttm_mem_reg *bo_mem) if (ttm->state == tt_bound) return 0; - be = ttm->be; - ret = ttm_tt_populate(ttm); if (ret) return ret; - ret = be->func->bind(be, bo_mem); + ret = ttm->func->bind(ttm, bo_mem); if (unlikely(ret != 0)) return ret; -- cgit v1.2.3 From b1e5f172325547270f35e7d1e42416a606e1dbd2 Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Wed, 2 Nov 2011 23:59:28 -0400 Subject: drm/ttm: introduce callback for ttm_tt populate & unpopulate V4 Move the page allocation and freeing to driver callback and provide ttm code helper function for those. Most intrusive change, is the fact that we now only fully populate an object this simplify some of code designed around the page fault design. V2 Rebase on top of memory accounting overhaul V3 New rebase on top of more memory accouting changes V4 Rebase on top of no memory account changes (where/when is my delorean when i need it ?) Signed-off-by: Jerome Glisse Reviewed-by: Konrad Rzeszutek Wilk Reviewed-by: Thomas Hellstrom --- drivers/gpu/drm/ttm/ttm_tt.c | 91 +++----------------------------------------- 1 file changed, 5 insertions(+), 86 deletions(-) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index fbc90dce1de8..77f0e6f79f30 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -43,8 +43,6 @@ #include "ttm/ttm_placement.h" #include "ttm/ttm_page_alloc.h" -static int ttm_tt_swapin(struct ttm_tt *ttm); - /** * Allocates storage for pointers to the pages that back the ttm. */ @@ -63,69 +61,6 @@ static void ttm_tt_free_page_directory(struct ttm_tt *ttm) ttm->dma_address = NULL; } -static struct page *__ttm_tt_get_page(struct ttm_tt *ttm, int index) -{ - struct page *p; - struct ttm_mem_global *mem_glob = ttm->glob->mem_glob; - int ret; - - if (NULL == (p = ttm->pages[index])) { - - ret = ttm_get_pages(&p, ttm->page_flags, ttm->caching_state, 1, - &ttm->dma_address[index]); - if (ret != 0) - return NULL; - - ret = ttm_mem_global_alloc_page(mem_glob, p, false, false); - if (unlikely(ret != 0)) - goto out_err; - - ttm->pages[index] = p; - } - return p; -out_err: - ttm_put_pages(&p, 1, ttm->page_flags, - ttm->caching_state, &ttm->dma_address[index]); - return NULL; -} - -struct page *ttm_tt_get_page(struct ttm_tt *ttm, int index) -{ - int ret; - - if (unlikely(ttm->page_flags & TTM_PAGE_FLAG_SWAPPED)) { - ret = ttm_tt_swapin(ttm); - if (unlikely(ret != 0)) - return NULL; - } - return __ttm_tt_get_page(ttm, index); -} - -int ttm_tt_populate(struct ttm_tt *ttm) -{ - struct page *page; - unsigned long i; - int ret; - - if (ttm->state != tt_unpopulated) - return 0; - - if (unlikely(ttm->page_flags & TTM_PAGE_FLAG_SWAPPED)) { - ret = ttm_tt_swapin(ttm); - if (unlikely(ret != 0)) - return ret; - } - - for (i = 0; i < ttm->num_pages; ++i) { - page = __ttm_tt_get_page(ttm, i); - if (!page) - return -ENOMEM; - } - ttm->state = tt_unbound; - return 0; -} -EXPORT_SYMBOL(ttm_tt_populate); - #ifdef CONFIG_X86 static inline int ttm_tt_set_page_caching(struct page *p, enum ttm_caching_state c_old, @@ -227,21 +162,6 @@ int ttm_tt_set_placement_caching(struct ttm_tt *ttm, uint32_t placement) } EXPORT_SYMBOL(ttm_tt_set_placement_caching); -static void ttm_tt_free_alloced_pages(struct ttm_tt *ttm) -{ - unsigned i; - - for (i = 0; i < ttm->num_pages; ++i) { - if (ttm->pages[i]) { - ttm_mem_global_free_page(ttm->glob->mem_glob, - ttm->pages[i]); - ttm_put_pages(&ttm->pages[i], 1, ttm->page_flags, - ttm->caching_state, &ttm->dma_address[i]); - } - } - ttm->state = tt_unpopulated; -} - void ttm_tt_destroy(struct ttm_tt *ttm) { if (unlikely(ttm == NULL)) @@ -252,7 +172,7 @@ void ttm_tt_destroy(struct ttm_tt *ttm) } if (likely(ttm->pages != NULL)) { - ttm_tt_free_alloced_pages(ttm); + ttm->bdev->driver->ttm_tt_unpopulate(ttm); ttm_tt_free_page_directory(ttm); } @@ -307,7 +227,7 @@ int ttm_tt_bind(struct ttm_tt *ttm, struct ttm_mem_reg *bo_mem) if (ttm->state == tt_bound) return 0; - ret = ttm_tt_populate(ttm); + ret = ttm->bdev->driver->ttm_tt_populate(ttm); if (ret) return ret; @@ -321,7 +241,7 @@ int ttm_tt_bind(struct ttm_tt *ttm, struct ttm_mem_reg *bo_mem) } EXPORT_SYMBOL(ttm_tt_bind); -static int ttm_tt_swapin(struct ttm_tt *ttm) +int ttm_tt_swapin(struct ttm_tt *ttm) { struct address_space *swap_space; struct file *swap_storage; @@ -343,7 +263,7 @@ static int ttm_tt_swapin(struct ttm_tt *ttm) ret = PTR_ERR(from_page); goto out_err; } - to_page = __ttm_tt_get_page(ttm, i); + to_page = ttm->pages[i]; if (unlikely(to_page == NULL)) goto out_err; @@ -364,7 +284,6 @@ static int ttm_tt_swapin(struct ttm_tt *ttm) return 0; out_err: - ttm_tt_free_alloced_pages(ttm); return ret; } @@ -416,7 +335,7 @@ int ttm_tt_swapout(struct ttm_tt *ttm, struct file *persistent_swap_storage) page_cache_release(to_page); } - ttm_tt_free_alloced_pages(ttm); + ttm->bdev->driver->ttm_tt_unpopulate(ttm); ttm->swap_storage = swap_storage; ttm->page_flags |= TTM_PAGE_FLAG_SWAPPED; if (persistent_swap_storage) -- cgit v1.2.3 From 2334b75ffbef6b8932f09ec4418b65ddb764ae99 Mon Sep 17 00:00:00 2001 From: Konrad Rzeszutek Wilk Date: Thu, 3 Nov 2011 16:46:34 -0400 Subject: drm/ttm: provide dma aware ttm page pool code V9 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit In TTM world the pages for the graphic drivers are kept in three different pools: write combined, uncached, and cached (write-back). When the pages are used by the graphic driver the graphic adapter via its built in MMU (or AGP) programs these pages in. The programming requires the virtual address (from the graphic adapter perspective) and the physical address (either System RAM or the memory on the card) which is obtained using the pci_map_* calls (which does the virtual to physical - or bus address translation). During the graphic application's "life" those pages can be shuffled around, swapped out to disk, moved from the VRAM to System RAM or vice-versa. This all works with the existing TTM pool code - except when we want to use the software IOTLB (SWIOTLB) code to "map" the physical addresses to the graphic adapter MMU. We end up programming the bounce buffer's physical address instead of the TTM pool memory's and get a non-worky driver. There are two solutions: 1) using the DMA API to allocate pages that are screened by the DMA API, or 2) using the pci_sync_* calls to copy the pages from the bounce-buffer and back. This patch fixes the issue by allocating pages using the DMA API. The second is a viable option - but it has performance drawbacks and potential correctness issues - think of the write cache page being bounced (SWIOTLB->TTM), the WC is set on the TTM page and the copy from SWIOTLB not making it to the TTM page until the page has been recycled in the pool (and used by another application). The bounce buffer does not get activated often - only in cases where we have a 32-bit capable card and we want to use a page that is allocated above the 4GB limit. The bounce buffer offers the solution of copying the contents of that 4GB page to an location below 4GB and then back when the operation has been completed (or vice-versa). This is done by using the 'pci_sync_*' calls. Note: If you look carefully enough in the existing TTM page pool code you will notice the GFP_DMA32 flag is used - which should guarantee that the provided page is under 4GB. It certainly is the case, except this gets ignored in two cases: - If user specifies 'swiotlb=force' which bounces _every_ page. - If user is using a Xen's PV Linux guest (which uses the SWIOTLB and the underlaying PFN's aren't necessarily under 4GB). To not have this extra copying done the other option is to allocate the pages using the DMA API so that there is not need to map the page and perform the expensive 'pci_sync_*' calls. This DMA API capable TTM pool requires for this the 'struct device' to properly call the DMA API. It also has to track the virtual and bus address of the page being handed out in case it ends up being swapped out or de-allocated - to make sure it is de-allocated using the proper's 'struct device'. Implementation wise the code keeps two lists: one that is attached to the 'struct device' (via the dev->dma_pools list) and a global one to be used when the 'struct device' is unavailable (think shrinker code). The global list can iterate over all of the 'struct device' and its associated dma_pool. The list in dev->dma_pools can only iterate the device's dma_pool. /[struct device_pool]\ /---------------------------------------------------| dev | / +-------| dma_pool | /-----+------\ / \--------------------/ |struct device| /-->[struct dma_pool for WC][struct dma_pool for uncached]<-/--| dma_pool | \-----+------/ / \--------------------/ \----------------------------------------------/ [Two pools associated with the device (WC and UC), and the parallel list containing the 'struct dev' and 'struct dma_pool' entries] The maximum amount of dma pools a device can have is six: write-combined, uncached, and cached; then there are the DMA32 variants which are: write-combined dma32, uncached dma32, and cached dma32. Currently this code only gets activated when any variant of the SWIOTLB IOMMU code is running (Intel without VT-d, AMD without GART, IBM Calgary and Xen PV with PCI devices). Tested-by: Michel Dänzer [v1: Using swiotlb_nr_tbl instead of swiotlb_enabled] [v2: Major overhaul - added 'inuse_list' to seperate used from inuse and reorder the order of lists to get better performance.] [v3: Added comments/and some logic based on review, Added Jerome tag] [v4: rebase on top of ttm_tt & ttm_backend merge] [v5: rebase on top of ttm memory accounting overhaul] [v6: New rebase on top of more memory accouting changes] [v7: well rebase on top of no memory accounting changes] [v8: make sure pages list is initialized empty] [v9: calll ttm_mem_global_free_page in unpopulate for accurate accountg] Signed-off-by: Konrad Rzeszutek Wilk Reviewed-by: Jerome Glisse Acked-by: Thomas Hellstrom --- drivers/gpu/drm/ttm/ttm_tt.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 77f0e6f79f30..1625739b434b 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -196,6 +196,7 @@ int ttm_tt_init(struct ttm_tt *ttm, struct ttm_bo_device *bdev, ttm->dummy_read_page = dummy_read_page; ttm->state = tt_unpopulated; + INIT_LIST_HEAD(&ttm->alloc_list); ttm_tt_alloc_page_directory(ttm); if (!ttm->pages || !ttm->dma_address) { ttm_tt_destroy(ttm); -- cgit v1.2.3 From 8e7e70522d760c4ccd4cd370ebfa0ba69e006c6e Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Wed, 9 Nov 2011 17:15:26 -0500 Subject: drm/ttm: isolate dma data from ttm_tt V4 Move dma data to a superset ttm_dma_tt structure which herit from ttm_tt. This allow driver that don't use dma functionalities to not have to waste memory for it. V2 Rebase on top of no memory account changes (where/when is my delorean when i need it ?) V3 Make sure page list is initialized empty V4 typo/syntax fixes Signed-off-by: Jerome Glisse Reviewed-by: Thomas Hellstrom --- drivers/gpu/drm/ttm/ttm_tt.c | 60 ++++++++++++++++++++++++++++++++++++-------- 1 file changed, 49 insertions(+), 11 deletions(-) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 1625739b434b..58e1fa14fe3a 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -48,17 +48,14 @@ */ static void ttm_tt_alloc_page_directory(struct ttm_tt *ttm) { - ttm->pages = drm_calloc_large(ttm->num_pages, sizeof(*ttm->pages)); - ttm->dma_address = drm_calloc_large(ttm->num_pages, - sizeof(*ttm->dma_address)); + ttm->pages = drm_calloc_large(ttm->num_pages, sizeof(void*)); } -static void ttm_tt_free_page_directory(struct ttm_tt *ttm) +static void ttm_dma_tt_alloc_page_directory(struct ttm_dma_tt *ttm) { - drm_free_large(ttm->pages); - ttm->pages = NULL; - drm_free_large(ttm->dma_address); - ttm->dma_address = NULL; + ttm->ttm.pages = drm_calloc_large(ttm->ttm.num_pages, sizeof(void*)); + ttm->dma_address = drm_calloc_large(ttm->ttm.num_pages, + sizeof(*ttm->dma_address)); } #ifdef CONFIG_X86 @@ -173,7 +170,6 @@ void ttm_tt_destroy(struct ttm_tt *ttm) if (likely(ttm->pages != NULL)) { ttm->bdev->driver->ttm_tt_unpopulate(ttm); - ttm_tt_free_page_directory(ttm); } if (!(ttm->page_flags & TTM_PAGE_FLAG_PERSISTENT_SWAP) && @@ -196,9 +192,8 @@ int ttm_tt_init(struct ttm_tt *ttm, struct ttm_bo_device *bdev, ttm->dummy_read_page = dummy_read_page; ttm->state = tt_unpopulated; - INIT_LIST_HEAD(&ttm->alloc_list); ttm_tt_alloc_page_directory(ttm); - if (!ttm->pages || !ttm->dma_address) { + if (!ttm->pages) { ttm_tt_destroy(ttm); printk(KERN_ERR TTM_PFX "Failed allocating page table\n"); return -ENOMEM; @@ -207,6 +202,49 @@ int ttm_tt_init(struct ttm_tt *ttm, struct ttm_bo_device *bdev, } EXPORT_SYMBOL(ttm_tt_init); +void ttm_tt_fini(struct ttm_tt *ttm) +{ + drm_free_large(ttm->pages); + ttm->pages = NULL; +} +EXPORT_SYMBOL(ttm_tt_fini); + +int ttm_dma_tt_init(struct ttm_dma_tt *ttm_dma, struct ttm_bo_device *bdev, + unsigned long size, uint32_t page_flags, + struct page *dummy_read_page) +{ + struct ttm_tt *ttm = &ttm_dma->ttm; + + ttm->bdev = bdev; + ttm->glob = bdev->glob; + ttm->num_pages = (size + PAGE_SIZE - 1) >> PAGE_SHIFT; + ttm->caching_state = tt_cached; + ttm->page_flags = page_flags; + ttm->dummy_read_page = dummy_read_page; + ttm->state = tt_unpopulated; + + INIT_LIST_HEAD(&ttm_dma->pages_list); + ttm_dma_tt_alloc_page_directory(ttm_dma); + if (!ttm->pages || !ttm_dma->dma_address) { + ttm_tt_destroy(ttm); + printk(KERN_ERR TTM_PFX "Failed allocating page table\n"); + return -ENOMEM; + } + return 0; +} +EXPORT_SYMBOL(ttm_dma_tt_init); + +void ttm_dma_tt_fini(struct ttm_dma_tt *ttm_dma) +{ + struct ttm_tt *ttm = &ttm_dma->ttm; + + drm_free_large(ttm->pages); + ttm->pages = NULL; + drm_free_large(ttm_dma->dma_address); + ttm_dma->dma_address = NULL; +} +EXPORT_SYMBOL(ttm_dma_tt_fini); + void ttm_tt_unbind(struct ttm_tt *ttm) { int ret; -- cgit v1.2.3 From dea7e0ac45fd28f90bbc38ff226d36a9f788efbf Mon Sep 17 00:00:00 2001 From: Jerome Glisse Date: Tue, 3 Jan 2012 17:37:37 -0500 Subject: ttm: fix agp since ttm tt rework ttm tt rework modified the way we allocate and populate the ttm_tt structure, the AGP side was missing some bit to properly work. Fix those and fix radeon and nouveau AGP support. Tested on radeon only so far. Signed-off-by: Jerome Glisse Reviewed-by: Konrad Rzeszutek Wilk Signed-off-by: Dave Airlie --- drivers/gpu/drm/ttm/ttm_tt.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'drivers/gpu/drm/ttm/ttm_tt.c') diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 58e1fa14fe3a..2f75d203a2bf 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -191,6 +191,7 @@ int ttm_tt_init(struct ttm_tt *ttm, struct ttm_bo_device *bdev, ttm->page_flags = page_flags; ttm->dummy_read_page = dummy_read_page; ttm->state = tt_unpopulated; + ttm->swap_storage = NULL; ttm_tt_alloc_page_directory(ttm); if (!ttm->pages) { @@ -222,6 +223,7 @@ int ttm_dma_tt_init(struct ttm_dma_tt *ttm_dma, struct ttm_bo_device *bdev, ttm->page_flags = page_flags; ttm->dummy_read_page = dummy_read_page; ttm->state = tt_unpopulated; + ttm->swap_storage = NULL; INIT_LIST_HEAD(&ttm_dma->pages_list); ttm_dma_tt_alloc_page_directory(ttm_dma); -- cgit v1.2.3