From 677e8622a9ae8cd9d351f98ecf120fb1c83b59d1 Mon Sep 17 00:00:00 2001 From: Nayan Deshmukh Date: Fri, 25 May 2018 10:15:48 +0530 Subject: drm/doc: add a chapter for gpu scheduler MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Nayan Deshmukh Reviewed-by: Christian König Acked-by: Daniel Vetter Signed-off-by: Alex Deucher --- Documentation/gpu/drm-mm.rst | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/drm-mm.rst b/Documentation/gpu/drm-mm.rst index b08e9dcd9177..96ebcc2a7b41 100644 --- a/Documentation/gpu/drm-mm.rst +++ b/Documentation/gpu/drm-mm.rst @@ -496,3 +496,21 @@ DRM Sync Objects .. kernel-doc:: drivers/gpu/drm/drm_syncobj.c :export: + +GPU Scheduler +============= + +Overview +-------- + +.. kernel-doc:: drivers/gpu/drm/scheduler/gpu_scheduler.c + :doc: Overview + +Scheduler Function References +----------------------------- + +.. kernel-doc:: include/drm/gpu_scheduler.h + :internal: + +.. kernel-doc:: drivers/gpu/drm/scheduler/gpu_scheduler.c + :export: -- cgit v1.2.3 From 41f507080e45010f916698165420640422942f7c Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Wed, 30 May 2018 13:00:46 +0200 Subject: drm/doc: Add a label for the PRIME Buffer Sharing chapter MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit So that it can be referenced from e.g. DOC comments. Reviewed-by: Alex Deucher Signed-off-by: Michel Dänzer Signed-off-by: Alex Deucher --- Documentation/gpu/drm-mm.rst | 2 ++ 1 file changed, 2 insertions(+) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/drm-mm.rst b/Documentation/gpu/drm-mm.rst index 96ebcc2a7b41..21b6b72a9ba8 100644 --- a/Documentation/gpu/drm-mm.rst +++ b/Documentation/gpu/drm-mm.rst @@ -395,6 +395,8 @@ VMA Offset Manager .. kernel-doc:: drivers/gpu/drm/drm_vma_manager.c :export: +.. _prime_buffer_sharing: + PRIME Buffer Sharing ==================== -- cgit v1.2.3 From 99fa7ce3a82bf8047b6104682da6ae276beb22d8 Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Tue, 29 May 2018 18:39:04 +0200 Subject: drm/doc: Add initial amdgpu driver documentation MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reviewed-by: Alex Deucher Signed-off-by: Michel Dänzer Signed-off-by: Alex Deucher --- Documentation/gpu/amdgpu.rst | 6 ++++++ Documentation/gpu/drivers.rst | 1 + 2 files changed, 7 insertions(+) create mode 100644 Documentation/gpu/amdgpu.rst (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst new file mode 100644 index 000000000000..41a14e4aa4ac --- /dev/null +++ b/Documentation/gpu/amdgpu.rst @@ -0,0 +1,6 @@ +========================= + drm/amdgpu AMDgpu driver +========================= + +The drm/amdgpu driver supports all AMD Radeon GPUs based on the Graphics Core +Next (GCN) architecture. diff --git a/Documentation/gpu/drivers.rst b/Documentation/gpu/drivers.rst index f982558fc25d..65be325bf282 100644 --- a/Documentation/gpu/drivers.rst +++ b/Documentation/gpu/drivers.rst @@ -4,6 +4,7 @@ GPU Driver Documentation .. toctree:: + amdgpu i915 meson pl111 -- cgit v1.2.3 From baca30fabdc9ba2712481e1b9e14ded1de6cfa3a Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Tue, 29 May 2018 18:33:41 +0200 Subject: drm/amdgpu: Add documentation for PRIME related code MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reviewed-by: Alex Deucher Signed-off-by: Michel Dänzer Signed-off-by: Alex Deucher --- Documentation/gpu/amdgpu.rst | 14 ++++ drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c | 119 ++++++++++++++++++++++++++++++ 2 files changed, 133 insertions(+) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst index 41a14e4aa4ac..f557866f6788 100644 --- a/Documentation/gpu/amdgpu.rst +++ b/Documentation/gpu/amdgpu.rst @@ -4,3 +4,17 @@ The drm/amdgpu driver supports all AMD Radeon GPUs based on the Graphics Core Next (GCN) architecture. + +Core Driver Infrastructure +========================== + +This section covers core driver infrastructure. + +PRIME Buffer Sharing +-------------------- + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c + :doc: PRIME Buffer Sharing + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c + :internal: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c index 4683626b065f..d1f05489595b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c @@ -23,6 +23,14 @@ * * Authors: Alex Deucher */ + +/** + * DOC: PRIME Buffer Sharing + * + * The following callback implementations are used for :ref:`sharing GEM buffer + * objects between different devices via PRIME `. + */ + #include #include "amdgpu.h" @@ -32,6 +40,14 @@ static const struct dma_buf_ops amdgpu_dmabuf_ops; +/** + * amdgpu_gem_prime_get_sg_table - &drm_driver.gem_prime_get_sg_table + * implementation + * @obj: GEM buffer object + * + * Returns: + * A scatter/gather table for the pinned pages of the buffer object's memory. + */ struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj) { struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); @@ -40,6 +56,15 @@ struct sg_table *amdgpu_gem_prime_get_sg_table(struct drm_gem_object *obj) return drm_prime_pages_to_sg(bo->tbo.ttm->pages, npages); } +/** + * amdgpu_gem_prime_vmap - &dma_buf_ops.vmap implementation + * @obj: GEM buffer object + * + * Sets up an in-kernel virtual mapping of the buffer object's memory. + * + * Returns: + * The virtual address of the mapping or an error pointer. + */ void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj) { struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); @@ -53,6 +78,13 @@ void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj) return bo->dma_buf_vmap.virtual; } +/** + * amdgpu_gem_prime_vunmap - &dma_buf_ops.vunmap implementation + * @obj: GEM buffer object + * @vaddr: virtual address (unused) + * + * Tears down the in-kernel virtual mapping of the buffer object's memory. + */ void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr) { struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); @@ -60,6 +92,17 @@ void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr) ttm_bo_kunmap(&bo->dma_buf_vmap); } +/** + * amdgpu_gem_prime_mmap - &drm_driver.gem_prime_mmap implementation + * @obj: GEM buffer object + * @vma: virtual memory area + * + * Sets up a userspace mapping of the buffer object's memory in the given + * virtual memory area. + * + * Returns: + * 0 on success or negative error code. + */ int amdgpu_gem_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma) { struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); @@ -94,6 +137,19 @@ int amdgpu_gem_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma return ret; } +/** + * amdgpu_gem_prime_import_sg_table - &drm_driver.gem_prime_import_sg_table + * implementation + * @dev: DRM device + * @attach: DMA-buf attachment + * @sg: Scatter/gather table + * + * Import shared DMA buffer memory exported by another device. + * + * Returns: + * A new GEM buffer object of the given DRM device, representing the memory + * described by the given DMA-buf attachment and scatter/gather table. + */ struct drm_gem_object * amdgpu_gem_prime_import_sg_table(struct drm_device *dev, struct dma_buf_attachment *attach, @@ -132,6 +188,19 @@ error: return ERR_PTR(ret); } +/** + * amdgpu_gem_map_attach - &dma_buf_ops.attach implementation + * @dma_buf: shared DMA buffer + * @target_dev: target device + * @attach: DMA-buf attachment + * + * Makes sure that the shared DMA buffer can be accessed by the target device. + * For now, simply pins it to the GTT domain, where it should be accessible by + * all DMA devices. + * + * Returns: + * 0 on success or negative error code. + */ static int amdgpu_gem_map_attach(struct dma_buf *dma_buf, struct device *target_dev, struct dma_buf_attachment *attach) @@ -181,6 +250,14 @@ error_detach: return r; } +/** + * amdgpu_gem_map_detach - &dma_buf_ops.detach implementation + * @dma_buf: shared DMA buffer + * @attach: DMA-buf attachment + * + * This is called when a shared DMA buffer no longer needs to be accessible by + * the other device. For now, simply unpins the buffer from GTT. + */ static void amdgpu_gem_map_detach(struct dma_buf *dma_buf, struct dma_buf_attachment *attach) { @@ -202,6 +279,13 @@ error: drm_gem_map_detach(dma_buf, attach); } +/** + * amdgpu_gem_prime_res_obj - &drm_driver.gem_prime_res_obj implementation + * @obj: GEM buffer object + * + * Returns: + * The buffer object's reservation object. + */ struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *obj) { struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); @@ -209,6 +293,18 @@ struct reservation_object *amdgpu_gem_prime_res_obj(struct drm_gem_object *obj) return bo->tbo.resv; } +/** + * amdgpu_gem_begin_cpu_access - &dma_buf_ops.begin_cpu_access implementation + * @dma_buf: shared DMA buffer + * @direction: direction of DMA transfer + * + * This is called before CPU access to the shared DMA buffer's memory. If it's + * a read access, the buffer is moved to the GTT domain if possible, for optimal + * CPU read performance. + * + * Returns: + * 0 on success or negative error code. + */ static int amdgpu_gem_begin_cpu_access(struct dma_buf *dma_buf, enum dma_data_direction direction) { @@ -253,6 +349,18 @@ static const struct dma_buf_ops amdgpu_dmabuf_ops = { .vunmap = drm_gem_dmabuf_vunmap, }; +/** + * amdgpu_gem_prime_export - &drm_driver.gem_prime_export implementation + * @dev: DRM device + * @gobj: GEM buffer object + * @flags: flags like DRM_CLOEXEC and DRM_RDWR + * + * The main work is done by the &drm_gem_prime_export helper, which in turn + * uses &amdgpu_gem_prime_res_obj. + * + * Returns: + * Shared DMA buffer representing the GEM buffer object from the given device. + */ struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev, struct drm_gem_object *gobj, int flags) @@ -273,6 +381,17 @@ struct dma_buf *amdgpu_gem_prime_export(struct drm_device *dev, return buf; } +/** + * amdgpu_gem_prime_import - &drm_driver.gem_prime_import implementation + * @dev: DRM device + * @dma_buf: Shared DMA buffer + * + * The main work is done by the &drm_gem_prime_import helper, which in turn + * uses &amdgpu_gem_prime_import_sg_table. + * + * Returns: + * GEM buffer object representing the shared DMA buffer for the given device. + */ struct drm_gem_object *amdgpu_gem_prime_import(struct drm_device *dev, struct dma_buf *dma_buf) { -- cgit v1.2.3 From 0152ac16522b6c6dd9319e005bf2945f585328d4 Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Fri, 1 Jun 2018 12:10:02 +0200 Subject: drm/amdgpu: Hook up documentation about memory domains MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reviewed-by: Alex Deucher Signed-off-by: Michel Dänzer Signed-off-by: Alex Deucher --- Documentation/gpu/amdgpu.rst | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst index f557866f6788..ad3711fd2a28 100644 --- a/Documentation/gpu/amdgpu.rst +++ b/Documentation/gpu/amdgpu.rst @@ -10,6 +10,14 @@ Core Driver Infrastructure This section covers core driver infrastructure. +.. _amdgpu_memory_domains: + +Memory Domains +-------------- + +.. kernel-doc:: include/uapi/drm/amdgpu_drm.h + :doc: memory domains + PRIME Buffer Sharing -------------------- -- cgit v1.2.3 From a2d19d66b1454b88bfcf1929837b329378d3377f Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Fri, 1 Jun 2018 12:30:44 +0200 Subject: drm/amdgpu: Hook up amdgpu_object.c documentation MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Reviewed-by: Alex Deucher Signed-off-by: Michel Dänzer Signed-off-by: Alex Deucher --- Documentation/gpu/amdgpu.rst | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst index ad3711fd2a28..1fbf3876a3d8 100644 --- a/Documentation/gpu/amdgpu.rst +++ b/Documentation/gpu/amdgpu.rst @@ -18,6 +18,15 @@ Memory Domains .. kernel-doc:: include/uapi/drm/amdgpu_drm.h :doc: memory domains +Buffer Objects +-------------- + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_object.c + :doc: amdgpu_object + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_object.c + :internal: + PRIME Buffer Sharing -------------------- -- cgit v1.2.3 From ad7f0b6334fe3cf52f2d79345791a4ef4547353f Mon Sep 17 00:00:00 2001 From: Christian König Date: Tue, 5 Jun 2018 11:47:43 +0200 Subject: drm/amdgpu: fix documentation of amdgpu_mn.c v2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit And wire it up as well. v2: improve the wording, fix label mismatch Signed-off-by: Christian König Reviewed-by: Michel Dänzer Signed-off-by: Alex Deucher --- Documentation/gpu/amdgpu.rst | 9 +++++ drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c | 74 ++++++++++++++++++++++++++-------- 2 files changed, 67 insertions(+), 16 deletions(-) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst index 1fbf3876a3d8..a4852f9a6141 100644 --- a/Documentation/gpu/amdgpu.rst +++ b/Documentation/gpu/amdgpu.rst @@ -35,3 +35,12 @@ PRIME Buffer Sharing .. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c :internal: + +MMU Notifier +------------ + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c + :doc: MMU Notifier + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c + :internal: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c index 37570a1c6db8..40fcb2af2914 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c @@ -28,6 +28,21 @@ * Christian König */ +/** + * DOC: MMU Notifier + * + * For coherent userptr handling registers an MMU notifier to inform the driver + * about updates on the page tables of a process. + * + * When somebody tries to invalidate the page tables we block the update until + * all operations on the pages in question are completed, then those pages are + * marked as accessed and also dirty if it wasn't a read only access. + * + * New command submissions using the userptrs in question are delayed until all + * page table invalidation are completed and we once more see a coherent process + * address space. + */ + #include #include #include @@ -38,6 +53,21 @@ #include "amdgpu.h" #include "amdgpu_amdkfd.h" +/** + * struct amdgpu_mn + * + * @adev: amdgpu device pointer + * @mm: process address space + * @mn: MMU notifier structur + * @work: destrution work item + * @node: hash table node to find structure by adev and mn + * @lock: rw semaphore protecting the notifier nodes + * @objects: interval tree containing amdgpu_mn_nodes + * @read_lock: mutex for recursive locking of @lock + * @recursion: depth of recursion + * + * Data for each amdgpu device and process address space. + */ struct amdgpu_mn { /* constant after initialisation */ struct amdgpu_device *adev; @@ -58,13 +88,21 @@ struct amdgpu_mn { atomic_t recursion; }; +/** + * struct amdgpu_mn_node + * + * @it: interval node defining start-last of the affected address range + * @bos: list of all BOs in the affected address range + * + * Manages all BOs which are affected of a certain range of address space. + */ struct amdgpu_mn_node { struct interval_tree_node it; struct list_head bos; }; /** - * amdgpu_mn_destroy - destroy the amn + * amdgpu_mn_destroy - destroy the MMU notifier * * @work: previously sheduled work item * @@ -98,7 +136,7 @@ static void amdgpu_mn_destroy(struct work_struct *work) * amdgpu_mn_release - callback to notify about mm destruction * * @mn: our notifier - * @mn: the mm this callback is about + * @mm: the mm this callback is about * * Shedule a work item to lazy destroy our notifier. */ @@ -106,13 +144,16 @@ static void amdgpu_mn_release(struct mmu_notifier *mn, struct mm_struct *mm) { struct amdgpu_mn *amn = container_of(mn, struct amdgpu_mn, mn); + INIT_WORK(&amn->work, amdgpu_mn_destroy); schedule_work(&amn->work); } /** - * amdgpu_mn_lock - take the write side lock for this mn + * amdgpu_mn_lock - take the write side lock for this notifier + * + * @mn: our notifier */ void amdgpu_mn_lock(struct amdgpu_mn *mn) { @@ -121,7 +162,9 @@ void amdgpu_mn_lock(struct amdgpu_mn *mn) } /** - * amdgpu_mn_unlock - drop the write side lock for this mn + * amdgpu_mn_unlock - drop the write side lock for this notifier + * + * @mn: our notifier */ void amdgpu_mn_unlock(struct amdgpu_mn *mn) { @@ -130,11 +173,9 @@ void amdgpu_mn_unlock(struct amdgpu_mn *mn) } /** - * amdgpu_mn_read_lock - take the amn read lock + * amdgpu_mn_read_lock - take the read side lock for this notifier * * @amn: our notifier - * - * Take the amn read side lock. */ static void amdgpu_mn_read_lock(struct amdgpu_mn *amn) { @@ -145,11 +186,9 @@ static void amdgpu_mn_read_lock(struct amdgpu_mn *amn) } /** - * amdgpu_mn_read_unlock - drop the amn read lock + * amdgpu_mn_read_unlock - drop the read side lock for this notifier * * @amn: our notifier - * - * Drop the amn read side lock. */ static void amdgpu_mn_read_unlock(struct amdgpu_mn *amn) { @@ -161,9 +200,11 @@ static void amdgpu_mn_read_unlock(struct amdgpu_mn *amn) * amdgpu_mn_invalidate_node - unmap all BOs of a node * * @node: the node with the BOs to unmap + * @start: start of address range affected + * @end: end of address range affected * - * We block for all BOs and unmap them by move them - * into system domain again. + * Block for operations on BOs to finish and mark pages as accessed and + * potentially dirty. */ static void amdgpu_mn_invalidate_node(struct amdgpu_mn_node *node, unsigned long start, @@ -190,12 +231,12 @@ static void amdgpu_mn_invalidate_node(struct amdgpu_mn_node *node, * amdgpu_mn_invalidate_range_start_gfx - callback to notify about mm change * * @mn: our notifier - * @mn: the mm this callback is about + * @mm: the mm this callback is about * @start: start of updated range * @end: end of updated range * - * We block for all BOs between start and end to be idle and - * unmap them by move them into system domain again. + * Block for operations on BOs to finish and mark pages as accessed and + * potentially dirty. */ static void amdgpu_mn_invalidate_range_start_gfx(struct mmu_notifier *mn, struct mm_struct *mm, @@ -268,7 +309,7 @@ static void amdgpu_mn_invalidate_range_start_hsa(struct mmu_notifier *mn, * amdgpu_mn_invalidate_range_end - callback to notify about mm change * * @mn: our notifier - * @mn: the mm this callback is about + * @mm: the mm this callback is about * @start: start of updated range * @end: end of updated range * @@ -456,6 +497,7 @@ void amdgpu_mn_unregister(struct amdgpu_bo *bo) if (list_empty(head)) { struct amdgpu_mn_node *node; + node = container_of(head, struct amdgpu_mn_node, bos); interval_tree_remove(&node->it, &amn->objects); kfree(node); -- cgit v1.2.3 From 7fc48e5912795caf78b1e2fb86d5f6ee02fbdb8c Mon Sep 17 00:00:00 2001 From: Andrey Grodzovsky Date: Mon, 11 Jun 2018 11:11:24 -0400 Subject: drm/amdgpu: Update function level documentation for GPUVM v3 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add/update function level documentation and add reference to amdgpu_vm.c in amdgpu.rst v2: Fix reference in rst file. Fix compilation warnings. Add space between function names and params list where it's missing. v3: Fix some funtion comments. Add formatted documentation to structs. Signed-off-by: Andrey Grodzovsky Reviewed-by: Christian König Signed-off-by: Alex Deucher --- Documentation/gpu/amdgpu.rst | 9 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 253 ++++++++++++++++++++++++++++----- 2 files changed, 224 insertions(+), 38 deletions(-) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst index a4852f9a6141..3ffb2a226fae 100644 --- a/Documentation/gpu/amdgpu.rst +++ b/Documentation/gpu/amdgpu.rst @@ -44,3 +44,12 @@ MMU Notifier .. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c :internal: + +AMDGPU Virtual Memory +--------------------- + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c + :doc: GPUVM + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c + :internal: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index d88687b617e2..7f03f8c38708 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -34,8 +34,9 @@ #include "amdgpu_trace.h" #include "amdgpu_amdkfd.h" -/* - * GPUVM +/** + * DOC: GPUVM + * * GPUVM is similar to the legacy gart on older asics, however * rather than there being a single global gart table * for the entire GPU, there are multiple VM page tables active @@ -63,37 +64,84 @@ INTERVAL_TREE_DEFINE(struct amdgpu_bo_va_mapping, rb, uint64_t, __subtree_last, #undef START #undef LAST -/* Local structure. Encapsulate some VM table update parameters to reduce +/** + * struct amdgpu_pte_update_params - Local structure + * + * Encapsulate some VM table update parameters to reduce * the number of function parameters + * */ struct amdgpu_pte_update_params { - /* amdgpu device we do this update for */ + + /** + * @adev: amdgpu device we do this update for + */ struct amdgpu_device *adev; - /* optional amdgpu_vm we do this update for */ + + /** + * @vm: optional amdgpu_vm we do this update for + */ struct amdgpu_vm *vm; - /* address where to copy page table entries from */ + + /** + * @src: address where to copy page table entries from + */ uint64_t src; - /* indirect buffer to fill with commands */ + + /** + * @ib: indirect buffer to fill with commands + */ struct amdgpu_ib *ib; - /* Function which actually does the update */ + + /** + * @func: Function which actually does the update + */ void (*func)(struct amdgpu_pte_update_params *params, struct amdgpu_bo *bo, uint64_t pe, uint64_t addr, unsigned count, uint32_t incr, uint64_t flags); - /* The next two are used during VM update by CPU - * DMA addresses to use for mapping - * Kernel pointer of PD/PT BO that needs to be updated + /** + * @pages_addr: + * + * DMA addresses to use for mapping, used during VM update by CPU */ dma_addr_t *pages_addr; + + /** + * @kptr: + * + * Kernel pointer of PD/PT BO that needs to be updated, + * used during VM update by CPU + */ void *kptr; }; -/* Helper to disable partial resident texture feature from a fence callback */ +/** + * struct amdgpu_prt_cb - Helper to disable partial resident texture feature from a fence callback + */ struct amdgpu_prt_cb { + + /** + * @adev: amdgpu device + */ struct amdgpu_device *adev; + + /** + * @cb: callback + */ struct dma_fence_cb cb; }; +/** + * amdgpu_vm_bo_base_init - Adds bo to the list of bos associated with the vm + * + * @base: base structure for tracking BO usage in a VM + * @vm: vm to which bo is to be added + * @bo: amdgpu buffer object + * + * Initialize a bo_va_base structure and add it to the appropriate lists + * + */ static void amdgpu_vm_bo_base_init(struct amdgpu_vm_bo_base *base, struct amdgpu_vm *vm, struct amdgpu_bo *bo) @@ -126,8 +174,10 @@ static void amdgpu_vm_bo_base_init(struct amdgpu_vm_bo_base *base, * amdgpu_vm_level_shift - return the addr shift for each level * * @adev: amdgpu_device pointer + * @level: VMPT level * - * Returns the number of bits the pfn needs to be right shifted for a level. + * Returns: + * The number of bits the pfn needs to be right shifted for a level. */ static unsigned amdgpu_vm_level_shift(struct amdgpu_device *adev, unsigned level) @@ -155,8 +205,10 @@ static unsigned amdgpu_vm_level_shift(struct amdgpu_device *adev, * amdgpu_vm_num_entries - return the number of entries in a PD/PT * * @adev: amdgpu_device pointer + * @level: VMPT level * - * Calculate the number of entries in a page directory or page table. + * Returns: + * The number of entries in a page directory or page table. */ static unsigned amdgpu_vm_num_entries(struct amdgpu_device *adev, unsigned level) @@ -179,8 +231,10 @@ static unsigned amdgpu_vm_num_entries(struct amdgpu_device *adev, * amdgpu_vm_bo_size - returns the size of the BOs in bytes * * @adev: amdgpu_device pointer + * @level: VMPT level * - * Calculate the size of the BO for a page directory or page table in bytes. + * Returns: + * The size of the BO for a page directory or page table in bytes. */ static unsigned amdgpu_vm_bo_size(struct amdgpu_device *adev, unsigned level) { @@ -218,6 +272,9 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm, * @param: parameter for the validation callback * * Validate the page table BOs on command submission if neccessary. + * + * Returns: + * Validation result. */ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm, int (*validate)(void *p, struct amdgpu_bo *bo), @@ -273,6 +330,9 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm, * @vm: VM to check * * Check if all VM PDs/PTs are ready for updates + * + * Returns: + * True if eviction list is empty. */ bool amdgpu_vm_ready(struct amdgpu_vm *vm) { @@ -283,10 +343,14 @@ bool amdgpu_vm_ready(struct amdgpu_vm *vm) * amdgpu_vm_clear_bo - initially clear the PDs/PTs * * @adev: amdgpu_device pointer + * @vm: VM to clear BO from * @bo: BO to clear * @level: level this BO is at * * Root PD needs to be reserved when calling this. + * + * Returns: + * 0 on success, errno otherwise. */ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev, struct amdgpu_vm *vm, struct amdgpu_bo *bo, @@ -382,10 +446,16 @@ error: * * @adev: amdgpu_device pointer * @vm: requested vm + * @parent: parent PT * @saddr: start of the address range * @eaddr: end of the address range + * @level: VMPT level + * @ats: indicate ATS support from PTE * * Make sure the page directories and page tables are allocated + * + * Returns: + * 0 on success, errno otherwise. */ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev, struct amdgpu_vm *vm, @@ -494,6 +564,9 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev, * @size: Size from start address we need. * * Make sure the page tables are allocated. + * + * Returns: + * 0 on success, errno otherwise. */ int amdgpu_vm_alloc_pts(struct amdgpu_device *adev, struct amdgpu_vm *vm, @@ -559,6 +632,15 @@ void amdgpu_vm_check_compute_bug(struct amdgpu_device *adev) } } +/** + * amdgpu_vm_need_pipeline_sync - Check if pipe sync is needed for job. + * + * @ring: ring on which the job will be submitted + * @job: job to submit + * + * Returns: + * True if sync is needed. + */ bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring, struct amdgpu_job *job) { @@ -586,6 +668,14 @@ bool amdgpu_vm_need_pipeline_sync(struct amdgpu_ring *ring, return vm_flush_needed || gds_switch_needed; } +/** + * amdgpu_vm_is_large_bar - Check if BAR is large enough + * + * @adev: amdgpu_device pointer + * + * Returns: + * True if BAR is large enough. + */ static bool amdgpu_vm_is_large_bar(struct amdgpu_device *adev) { return (adev->gmc.real_vram_size == adev->gmc.visible_vram_size); @@ -595,10 +685,12 @@ static bool amdgpu_vm_is_large_bar(struct amdgpu_device *adev) * amdgpu_vm_flush - hardware flush the vm * * @ring: ring to use for flush - * @vmid: vmid number to use - * @pd_addr: address of the page directory + * @need_pipe_sync: is pipe sync needed * * Emit a VM flush when it is necessary. + * + * Returns: + * 0 on success, errno otherwise. */ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_pipe_sync) { @@ -706,6 +798,9 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_ * Returns the found bo_va or NULL if none is found * * Object has to be reserved! + * + * Returns: + * Found bo_va or NULL. */ struct amdgpu_bo_va *amdgpu_vm_bo_find(struct amdgpu_vm *vm, struct amdgpu_bo *bo) @@ -787,7 +882,10 @@ static void amdgpu_vm_do_copy_ptes(struct amdgpu_pte_update_params *params, * @addr: the unmapped addr * * Look up the physical address of the page that the pte resolves - * to and return the pointer for the page table entry. + * to. + * + * Returns: + * The pointer for the page table entry. */ static uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr) { @@ -840,6 +938,17 @@ static void amdgpu_vm_cpu_set_ptes(struct amdgpu_pte_update_params *params, } } + +/** + * amdgpu_vm_wait_pd - Wait for PT BOs to be free. + * + * @adev: amdgpu_device pointer + * @vm: related vm + * @owner: fence owner + * + * Returns: + * 0 on success, errno otherwise. + */ static int amdgpu_vm_wait_pd(struct amdgpu_device *adev, struct amdgpu_vm *vm, void *owner) { @@ -893,7 +1002,10 @@ static void amdgpu_vm_update_pde(struct amdgpu_pte_update_params *params, /* * amdgpu_vm_invalidate_level - mark all PD levels as invalid * + * @adev: amdgpu_device pointer + * @vm: related vm * @parent: parent PD + * @level: VMPT level * * Mark all PD level as invalid after an error. */ @@ -928,7 +1040,9 @@ static void amdgpu_vm_invalidate_level(struct amdgpu_device *adev, * @vm: requested vm * * Makes sure all directories are up to date. - * Returns 0 for success, error for failure. + * + * Returns: + * 0 for success, error for failure. */ int amdgpu_vm_update_directories(struct amdgpu_device *adev, struct amdgpu_vm *vm) @@ -1115,14 +1229,15 @@ static void amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p, * amdgpu_vm_update_ptes - make sure that page tables are valid * * @params: see amdgpu_pte_update_params definition - * @vm: requested vm * @start: start of GPU address range * @end: end of GPU address range * @dst: destination address to map to, the next dst inside the function * @flags: mapping flags * * Update the page tables in the range @start - @end. - * Returns 0 for success, -EINVAL for failure. + * + * Returns: + * 0 for success, -EINVAL for failure. */ static int amdgpu_vm_update_ptes(struct amdgpu_pte_update_params *params, uint64_t start, uint64_t end, @@ -1176,7 +1291,9 @@ static int amdgpu_vm_update_ptes(struct amdgpu_pte_update_params *params, * @end: last PTE to handle * @dst: addr those PTEs should point to * @flags: hw mapping flags - * Returns 0 for success, -EINVAL for failure. + * + * Returns: + * 0 for success, -EINVAL for failure. */ static int amdgpu_vm_frag_ptes(struct amdgpu_pte_update_params *params, uint64_t start, uint64_t end, @@ -1248,7 +1365,9 @@ static int amdgpu_vm_frag_ptes(struct amdgpu_pte_update_params *params, * @fence: optional resulting fence * * Fill in the page table entries between @start and @last. - * Returns 0 for success, -EINVAL for failure. + * + * Returns: + * 0 for success, -EINVAL for failure. */ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev, struct dma_fence *exclusive, @@ -1403,7 +1522,9 @@ error_free: * * Split the mapping into smaller chunks so that each update fits * into a SDMA IB. - * Returns 0 for success, -EINVAL for failure. + * + * Returns: + * 0 for success, -EINVAL for failure. */ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev, struct dma_fence *exclusive, @@ -1514,7 +1635,9 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev, * @clear: if true clear the entries * * Fill in the page table entries for @bo_va. - * Returns 0 for success, -EINVAL for failure. + * + * Returns: + * 0 for success, -EINVAL for failure. */ int amdgpu_vm_bo_update(struct amdgpu_device *adev, struct amdgpu_bo_va *bo_va, @@ -1609,6 +1732,8 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev, /** * amdgpu_vm_update_prt_state - update the global PRT state + * + * @adev: amdgpu_device pointer */ static void amdgpu_vm_update_prt_state(struct amdgpu_device *adev) { @@ -1623,6 +1748,8 @@ static void amdgpu_vm_update_prt_state(struct amdgpu_device *adev) /** * amdgpu_vm_prt_get - add a PRT user + * + * @adev: amdgpu_device pointer */ static void amdgpu_vm_prt_get(struct amdgpu_device *adev) { @@ -1635,6 +1762,8 @@ static void amdgpu_vm_prt_get(struct amdgpu_device *adev) /** * amdgpu_vm_prt_put - drop a PRT user + * + * @adev: amdgpu_device pointer */ static void amdgpu_vm_prt_put(struct amdgpu_device *adev) { @@ -1644,6 +1773,8 @@ static void amdgpu_vm_prt_put(struct amdgpu_device *adev) /** * amdgpu_vm_prt_cb - callback for updating the PRT status + * + * @fence: fence for the callback */ static void amdgpu_vm_prt_cb(struct dma_fence *fence, struct dma_fence_cb *_cb) { @@ -1655,6 +1786,9 @@ static void amdgpu_vm_prt_cb(struct dma_fence *fence, struct dma_fence_cb *_cb) /** * amdgpu_vm_add_prt_cb - add callback for updating the PRT status + * + * @adev: amdgpu_device pointer + * @fence: fence for the callback */ static void amdgpu_vm_add_prt_cb(struct amdgpu_device *adev, struct dma_fence *fence) @@ -1746,9 +1880,11 @@ static void amdgpu_vm_prt_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm) * or if an error occurred) * * Make sure all freed BOs are cleared in the PT. - * Returns 0 for success. - * * PTs have to be reserved and mutex must be locked! + * + * Returns: + * 0 for success. + * */ int amdgpu_vm_clear_freed(struct amdgpu_device *adev, struct amdgpu_vm *vm, @@ -1793,10 +1929,11 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev, * * @adev: amdgpu_device pointer * @vm: requested vm - * @sync: sync object to add fences to * * Make sure all BOs which are moved are updated in the PTs. - * Returns 0 for success. + * + * Returns: + * 0 for success. * * PTs have to be reserved! */ @@ -1851,7 +1988,9 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev, * * Add @bo into the requested vm. * Add @bo to the list of bos associated with the vm - * Returns newly added bo_va or NULL for failure + * + * Returns: + * Newly added bo_va or NULL for failure * * Object has to be reserved! */ @@ -1917,7 +2056,9 @@ static void amdgpu_vm_bo_insert_map(struct amdgpu_device *adev, * @flags: attributes of pages (read/write/valid/etc.) * * Add a mapping of the BO at the specefied addr into the VM. - * Returns 0 for success, error for failure. + * + * Returns: + * 0 for success, error for failure. * * Object has to be reserved and unreserved outside! */ @@ -1979,7 +2120,9 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev, * * Add a mapping of the BO at the specefied addr into the VM. Replace existing * mappings as we do so. - * Returns 0 for success, error for failure. + * + * Returns: + * 0 for success, error for failure. * * Object has to be reserved and unreserved outside! */ @@ -2036,7 +2179,9 @@ int amdgpu_vm_bo_replace_map(struct amdgpu_device *adev, * @saddr: where to the BO is mapped * * Remove a mapping of the BO at the specefied addr from the VM. - * Returns 0 for success, error for failure. + * + * Returns: + * 0 for success, error for failure. * * Object has to be reserved and unreserved outside! */ @@ -2090,7 +2235,9 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev, * @size: size of the range * * Remove all mappings in a range, split them as appropriate. - * Returns 0 for success, error for failure. + * + * Returns: + * 0 for success, error for failure. */ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev, struct amdgpu_vm *vm, @@ -2189,6 +2336,10 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev, * @vm: the requested VM * * Find a mapping by it's address. + * + * Returns: + * The amdgpu_bo_va_mapping matching for addr or NULL + * */ struct amdgpu_bo_va_mapping *amdgpu_vm_bo_lookup_mapping(struct amdgpu_vm *vm, uint64_t addr) @@ -2240,7 +2391,6 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev, * amdgpu_vm_bo_invalidate - mark the bo as invalid * * @adev: amdgpu_device pointer - * @vm: requested vm * @bo: amdgpu buffer object * * Mark @bo as invalid. @@ -2281,6 +2431,14 @@ void amdgpu_vm_bo_invalidate(struct amdgpu_device *adev, } } +/** + * amdgpu_vm_get_block_size - calculate VM page table size as power of two + * + * @vm_size: VM size + * + * Returns: + * VM page table as power of two + */ static uint32_t amdgpu_vm_get_block_size(uint64_t vm_size) { /* Total bits covered by PD + PTs */ @@ -2368,6 +2526,9 @@ void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t vm_size, * @vm_context: Indicates if it GFX or Compute context * * Init @vm fields. + * + * Returns: + * 0 for success, error for failure. */ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, int vm_context, unsigned int pasid) @@ -2488,6 +2649,9 @@ error_free_sched_entity: /** * amdgpu_vm_make_compute - Turn a GFX VM into a compute VM * + * @adev: amdgpu_device pointer + * @vm: requested vm + * * This only works on GFX VMs that don't have any BOs added and no * page tables allocated yet. * @@ -2500,7 +2664,8 @@ error_free_sched_entity: * setting. May leave behind an unused shadow BO for the page * directory when switching from SDMA updates to CPU updates. * - * Returns 0 for success, -errno for errors. + * Returns: + * 0 for success, -errno for errors. */ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm) { @@ -2655,8 +2820,10 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm) * @adev: amdgpu_device pointer * @pasid: PASID do identify the VM * - * This function is expected to be called in interrupt context. Returns - * true if there was fault credit, false otherwise + * This function is expected to be called in interrupt context. + * + * Returns: + * True if there was fault credit, false otherwise */ bool amdgpu_vm_pasid_fault_credit(struct amdgpu_device *adev, unsigned int pasid) @@ -2740,6 +2907,16 @@ void amdgpu_vm_manager_fini(struct amdgpu_device *adev) amdgpu_vmid_mgr_fini(adev); } +/** + * amdgpu_vm_ioctl - Manages VMID reservation for vm hubs. + * + * @dev: drm device pointer + * @data: drm_amdgpu_vm + * @filp: drm file pointer + * + * Returns: + * 0 for success, -errno for errors. + */ int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) { union drm_amdgpu_vm *args = data; -- cgit v1.2.3 From 923ff76eb035b0de3725edd7c13cda3b068ba197 Mon Sep 17 00:00:00 2001 From: Slava Abramov Date: Thu, 7 Jun 2018 17:27:07 -0400 Subject: drm/amdgpu: update documentation for amdgpu_irq.c v3 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add/update function level documentation and add reference to amdgpu_irq.c in amdgpu.rst v2: Added DOC comment Added more explanations for amdgpu_hotplug_work_func Properly formatted unused parameters Properly formatted return values Fixed usage of acronyms More consistent styling v3: Removed duplicate "not" Using '&' to refer to functions and types Signed-off-by: Slava Abramov Acked-by: Christian König Reviewed-by: Michel Dänzer Signed-off-by: Alex Deucher --- Documentation/gpu/amdgpu.rst | 8 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c | 202 ++++++++++++++++++++++---------- 2 files changed, 147 insertions(+), 63 deletions(-) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst index 3ffb2a226fae..abf87360ab15 100644 --- a/Documentation/gpu/amdgpu.rst +++ b/Documentation/gpu/amdgpu.rst @@ -52,4 +52,12 @@ AMDGPU Virtual Memory :doc: GPUVM .. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c + +Interrupt Handling +------------------ + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c + :doc: Interrupt Handling + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c :internal: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index 3a5ca462abf0..1abf5b5bac9e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -25,6 +25,23 @@ * Alex Deucher * Jerome Glisse */ + +/** + * DOC: Interrupt Handling + * + * Interrupts generated within GPU hardware raise interrupt requests that are + * passed to amdgpu IRQ handler which is responsible for detecting source and + * type of the interrupt and dispatching matching handlers. If handling an + * interrupt requires calling kernel functions that may sleep processing is + * dispatched to work handlers. + * + * If MSI functionality is not disabled by module parameter then MSI + * support will be enabled. + * + * For GPU interrupt sources that may be driven by another driver, IRQ domain + * support is used (with mapping between virtual and hardware IRQs). + */ + #include #include #include @@ -43,19 +60,21 @@ #define AMDGPU_WAIT_IDLE_TIMEOUT 200 -/* - * Handle hotplug events outside the interrupt handler proper. - */ /** - * amdgpu_hotplug_work_func - display hotplug work handler + * amdgpu_hotplug_work_func - work handler for display hotplug event * - * @work: work struct + * @work: work struct pointer * - * This is the hot plug event work handler (all asics). - * The work gets scheduled from the irq handler if there - * was a hot plug interrupt. It walks the connector table - * and calls the hotplug handler for each one, then sends - * a drm hotplug event to alert userspace. + * This is the hotplug event work handler (all ASICs). + * The work gets scheduled from the IRQ handler if there + * was a hotplug interrupt. It walks through the connector table + * and calls hotplug handler for each connector. After this, it sends + * a DRM hotplug event to alert userspace. + * + * This design approach is required in order to defer hotplug event handling + * from the IRQ handler to a work handler because hotplug handler has to use + * mutexes which cannot be locked in an IRQ handler (since &mutex_lock may + * sleep). */ static void amdgpu_hotplug_work_func(struct work_struct *work) { @@ -74,13 +93,12 @@ static void amdgpu_hotplug_work_func(struct work_struct *work) } /** - * amdgpu_irq_reset_work_func - execute gpu reset + * amdgpu_irq_reset_work_func - execute GPU reset * - * @work: work struct + * @work: work struct pointer * - * Execute scheduled gpu reset (cayman+). - * This function is called when the irq handler - * thinks we need a gpu reset. + * Execute scheduled GPU reset (Cayman+). + * This function is called when the IRQ handler thinks we need a GPU reset. */ static void amdgpu_irq_reset_work_func(struct work_struct *work) { @@ -91,7 +109,13 @@ static void amdgpu_irq_reset_work_func(struct work_struct *work) amdgpu_device_gpu_recover(adev, NULL, false); } -/* Disable *all* interrupts */ +/** + * amdgpu_irq_disable_all - disable *all* interrupts + * + * @adev: amdgpu device pointer + * + * Disable all types of interrupts from all sources. + */ void amdgpu_irq_disable_all(struct amdgpu_device *adev) { unsigned long irqflags; @@ -123,11 +147,15 @@ void amdgpu_irq_disable_all(struct amdgpu_device *adev) } /** - * amdgpu_irq_handler - irq handler + * amdgpu_irq_handler - IRQ handler + * + * @irq: IRQ number (unused) + * @arg: pointer to DRM device * - * @int irq, void *arg: args + * IRQ handler for amdgpu driver (all ASICs). * - * This is the irq handler for the amdgpu driver (all asics). + * Returns: + * result of handling the IRQ, as defined by &irqreturn_t */ irqreturn_t amdgpu_irq_handler(int irq, void *arg) { @@ -142,18 +170,18 @@ irqreturn_t amdgpu_irq_handler(int irq, void *arg) } /** - * amdgpu_msi_ok - asic specific msi checks + * amdgpu_msi_ok - check whether MSI functionality is enabled * - * @adev: amdgpu device pointer + * @adev: amdgpu device pointer (unused) + * + * Checks whether MSI functionality has been disabled via module parameter + * (all ASICs). * - * Handles asic specific MSI checks to determine if - * MSIs should be enabled on a particular chip (all asics). - * Returns true if MSIs should be enabled, false if MSIs - * should not be enabled. + * Returns: + * *true* if MSIs are allowed to be enabled or *false* otherwise */ static bool amdgpu_msi_ok(struct amdgpu_device *adev) { - /* force MSI on */ if (amdgpu_msi == 1) return true; else if (amdgpu_msi == 0) @@ -163,12 +191,15 @@ static bool amdgpu_msi_ok(struct amdgpu_device *adev) } /** - * amdgpu_irq_init - init driver interrupt info + * amdgpu_irq_init - initialize interrupt handling * * @adev: amdgpu device pointer * - * Sets up the work irq handlers, vblank init, MSIs, etc. (all asics). - * Returns 0 for success, error for failure. + * Sets up work functions for hotplug and reset interrupts, enables MSI + * functionality, initializes vblank, hotplug and reset interrupt handling. + * + * Returns: + * 0 on success or error code on failure */ int amdgpu_irq_init(struct amdgpu_device *adev) { @@ -176,7 +207,7 @@ int amdgpu_irq_init(struct amdgpu_device *adev) spin_lock_init(&adev->irq.lock); - /* enable msi */ + /* Enable MSI if not disabled by module parameter */ adev->irq.msi_enabled = false; if (amdgpu_msi_ok(adev)) { @@ -189,7 +220,7 @@ int amdgpu_irq_init(struct amdgpu_device *adev) if (!amdgpu_device_has_dc_support(adev)) { if (!adev->enable_virtual_display) - /* Disable vblank irqs aggressively for power-saving */ + /* Disable vblank IRQs aggressively for power-saving */ /* XXX: can this be enabled for DC? */ adev->ddev->vblank_disable_immediate = true; @@ -197,7 +228,7 @@ int amdgpu_irq_init(struct amdgpu_device *adev) if (r) return r; - /* pre DCE11 */ + /* Pre-DCE11 */ INIT_WORK(&adev->hotplug_work, amdgpu_hotplug_work_func); } @@ -220,11 +251,13 @@ int amdgpu_irq_init(struct amdgpu_device *adev) } /** - * amdgpu_irq_fini - tear down driver interrupt info + * amdgpu_irq_fini - shut down interrupt handling * * @adev: amdgpu device pointer * - * Tears down the work irq handlers, vblank handlers, MSIs, etc. (all asics). + * Tears down work functions for hotplug and reset interrupts, disables MSI + * functionality, shuts down vblank, hotplug and reset interrupt handling, + * turns off interrupts from all sources (all ASICs). */ void amdgpu_irq_fini(struct amdgpu_device *adev) { @@ -264,12 +297,17 @@ void amdgpu_irq_fini(struct amdgpu_device *adev) } /** - * amdgpu_irq_add_id - register irq source + * amdgpu_irq_add_id - register IRQ source * * @adev: amdgpu device pointer - * @src_id: source id for this source - * @source: irq source + * @client_id: client id + * @src_id: source id + * @source: IRQ source pointer + * + * Registers IRQ source on a client. * + * Returns: + * 0 on success or error code otherwise */ int amdgpu_irq_add_id(struct amdgpu_device *adev, unsigned client_id, unsigned src_id, @@ -312,12 +350,12 @@ int amdgpu_irq_add_id(struct amdgpu_device *adev, } /** - * amdgpu_irq_dispatch - dispatch irq to IP blocks + * amdgpu_irq_dispatch - dispatch IRQ to IP blocks * * @adev: amdgpu device pointer - * @entry: interrupt vector + * @entry: interrupt vector pointer * - * Dispatches the irq to the different IP blocks + * Dispatches IRQ to IP blocks. */ void amdgpu_irq_dispatch(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) @@ -361,13 +399,13 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev, } /** - * amdgpu_irq_update - update hw interrupt state + * amdgpu_irq_update - update hardware interrupt state * * @adev: amdgpu device pointer - * @src: interrupt src you want to enable - * @type: type of interrupt you want to update + * @src: interrupt source pointer + * @type: type of interrupt * - * Updates the interrupt state for a specific src (all asics). + * Updates interrupt state for the specific source (all ASICs). */ int amdgpu_irq_update(struct amdgpu_device *adev, struct amdgpu_irq_src *src, unsigned type) @@ -378,7 +416,7 @@ int amdgpu_irq_update(struct amdgpu_device *adev, spin_lock_irqsave(&adev->irq.lock, irqflags); - /* we need to determine after taking the lock, otherwise + /* We need to determine after taking the lock, otherwise we might disable just enabled interrupts again */ if (amdgpu_irq_enabled(adev, src, type)) state = AMDGPU_IRQ_STATE_ENABLE; @@ -390,6 +428,14 @@ int amdgpu_irq_update(struct amdgpu_device *adev, return r; } +/** + * amdgpu_irq_gpu_reset_resume_helper - update interrupt states on all sources + * + * @adev: amdgpu device pointer + * + * Updates state of all types of interrupts on all sources on resume after + * reset. + */ void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev) { int i, j, k; @@ -413,10 +459,13 @@ void amdgpu_irq_gpu_reset_resume_helper(struct amdgpu_device *adev) * amdgpu_irq_get - enable interrupt * * @adev: amdgpu device pointer - * @src: interrupt src you want to enable - * @type: type of interrupt you want to enable + * @src: interrupt source pointer + * @type: type of interrupt * - * Enables the interrupt type for a specific src (all asics). + * Enables specified type of interrupt on the specified source (all ASICs). + * + * Returns: + * 0 on success or error code otherwise */ int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src, unsigned type) @@ -440,10 +489,13 @@ int amdgpu_irq_get(struct amdgpu_device *adev, struct amdgpu_irq_src *src, * amdgpu_irq_put - disable interrupt * * @adev: amdgpu device pointer - * @src: interrupt src you want to disable - * @type: type of interrupt you want to disable + * @src: interrupt source pointer + * @type: type of interrupt + * + * Enables specified type of interrupt on the specified source (all ASICs). * - * Disables the interrupt type for a specific src (all asics). + * Returns: + * 0 on success or error code otherwise */ int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src, unsigned type) @@ -464,12 +516,17 @@ int amdgpu_irq_put(struct amdgpu_device *adev, struct amdgpu_irq_src *src, } /** - * amdgpu_irq_enabled - test if irq is enabled or not + * amdgpu_irq_enabled - check whether interrupt is enabled or not * * @adev: amdgpu device pointer - * @idx: interrupt src you want to test + * @src: interrupt source pointer + * @type: type of interrupt * - * Tests if the given interrupt source is enabled or not + * Checks whether the given type of interrupt is enabled on the given source. + * + * Returns: + * *true* if interrupt is enabled, *false* if interrupt is disabled or on + * invalid parameters */ bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src, unsigned type) @@ -486,7 +543,7 @@ bool amdgpu_irq_enabled(struct amdgpu_device *adev, struct amdgpu_irq_src *src, return !!atomic_read(&src->enabled_types[type]); } -/* gen irq */ +/* XXX: Generic IRQ handling */ static void amdgpu_irq_mask(struct irq_data *irqd) { /* XXX */ @@ -497,12 +554,26 @@ static void amdgpu_irq_unmask(struct irq_data *irqd) /* XXX */ } +/* amdgpu hardware interrupt chip descriptor */ static struct irq_chip amdgpu_irq_chip = { .name = "amdgpu-ih", .irq_mask = amdgpu_irq_mask, .irq_unmask = amdgpu_irq_unmask, }; +/** + * amdgpu_irqdomain_map - create mapping between virtual and hardware IRQ numbers + * + * @d: amdgpu IRQ domain pointer (unused) + * @irq: virtual IRQ number + * @hwirq: hardware irq number + * + * Current implementation assigns simple interrupt handler to the given virtual + * IRQ. + * + * Returns: + * 0 on success or error code otherwise + */ static int amdgpu_irqdomain_map(struct irq_domain *d, unsigned int irq, irq_hw_number_t hwirq) { @@ -514,17 +585,21 @@ static int amdgpu_irqdomain_map(struct irq_domain *d, return 0; } +/* Implementation of methods for amdgpu IRQ domain */ static const struct irq_domain_ops amdgpu_hw_irqdomain_ops = { .map = amdgpu_irqdomain_map, }; /** - * amdgpu_irq_add_domain - create a linear irq domain + * amdgpu_irq_add_domain - create a linear IRQ domain * * @adev: amdgpu device pointer * - * Create an irq domain for GPU interrupt sources + * Creates an IRQ domain for GPU interrupt sources * that may be driven by another driver (e.g., ACP). + * + * Returns: + * 0 on success or error code otherwise */ int amdgpu_irq_add_domain(struct amdgpu_device *adev) { @@ -539,11 +614,11 @@ int amdgpu_irq_add_domain(struct amdgpu_device *adev) } /** - * amdgpu_irq_remove_domain - remove the irq domain + * amdgpu_irq_remove_domain - remove the IRQ domain * * @adev: amdgpu device pointer * - * Remove the irq domain for GPU interrupt sources + * Removes the IRQ domain for GPU interrupt sources * that may be driven by another driver (e.g., ACP). */ void amdgpu_irq_remove_domain(struct amdgpu_device *adev) @@ -555,16 +630,17 @@ void amdgpu_irq_remove_domain(struct amdgpu_device *adev) } /** - * amdgpu_irq_create_mapping - create a mapping between a domain irq and a - * Linux irq + * amdgpu_irq_create_mapping - create mapping between domain Linux IRQs * * @adev: amdgpu device pointer * @src_id: IH source id * - * Create a mapping between a domain irq (GPU IH src id) and a Linux irq + * Creates mapping between a domain IRQ (GPU IH src id) and a Linux IRQ * Use this for components that generate a GPU interrupt, but are driven * by a different driver (e.g., ACP). - * Returns the Linux irq. + * + * Returns: + * Linux IRQ */ unsigned amdgpu_irq_create_mapping(struct amdgpu_device *adev, unsigned src_id) { -- cgit v1.2.3 From dc85db256d47062955145264df5923f03b85b0c6 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Fri, 1 Jun 2018 12:28:14 -0500 Subject: drm/doc: Add amdgpu hwmon/power documentation (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Document the hwmon and power control interfaces exposed by the amdgpu driver. v2: use section rather than chapter for now Reviewed-by: Michel Dänzer Signed-off-by: Alex Deucher --- Documentation/gpu/amdgpu.rst | 53 ++++++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c | 45 ++++++++++++++++++++++++----- 2 files changed, 90 insertions(+), 8 deletions(-) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst index abf87360ab15..e1656fe8f247 100644 --- a/Documentation/gpu/amdgpu.rst +++ b/Documentation/gpu/amdgpu.rst @@ -61,3 +61,56 @@ Interrupt Handling .. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c :internal: + +GPU Power/Thermal Controls and Monitoring +========================================= + +This section covers hwmon and power/thermal controls. + +HWMON Interfaces +---------------- + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c + :doc: hwmon + +GPU sysfs Power State Interfaces +-------------------------------- + +GPU power controls are exposed via sysfs files. + +power_dpm_state +~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c + :doc: power_dpm_state + +power_dpm_force_performance_level +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c + :doc: power_dpm_force_performance_level + +pp_table +~~~~~~~~ + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c + :doc: pp_table + +pp_od_clk_voltage +~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c + :doc: pp_od_clk_voltage + +pp_dpm_sclk pp_dpm_mclk pp_dpm_pcie +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c + :doc: pp_dpm_sclk pp_dpm_mclk pp_dpm_pcie + +pp_power_profile_mode +~~~~~~~~~~~~~~~~~~~~~ + +.. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c + :doc: pp_power_profile_mode + diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c index b455da487782..f667cb9eb614 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c @@ -80,12 +80,15 @@ void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev) /** * DOC: power_dpm_state * - * This is a legacy interface and is only provided for backwards compatibility. - * The amdgpu driver provides a sysfs API for adjusting certain power - * related parameters. The file power_dpm_state is used for this. + * The power_dpm_state file is a legacy interface and is only provided for + * backwards compatibility. The amdgpu driver provides a sysfs API for adjusting + * certain power related parameters. The file power_dpm_state is used for this. * It accepts the following arguments: + * * - battery + * * - balanced + * * - performance * * battery @@ -169,14 +172,21 @@ fail: * The amdgpu driver provides a sysfs API for adjusting certain power * related parameters. The file power_dpm_force_performance_level is * used for this. It accepts the following arguments: + * * - auto + * * - low + * * - high + * * - manual - * - GPU fan + * * - profile_standard + * * - profile_min_sclk + * * - profile_min_mclk + * * - profile_peak * * auto @@ -463,8 +473,11 @@ static ssize_t amdgpu_set_pp_table(struct device *dev, * this. * * Reading the file will display: + * * - a list of engine clock levels and voltages labeled OD_SCLK + * * - a list of memory clock levels and voltages labeled OD_MCLK + * * - a list of valid ranges for sclk, mclk, and voltage labeled OD_RANGE * * To manually adjust these settings, first select manual using @@ -1285,35 +1298,51 @@ static ssize_t amdgpu_hwmon_set_power_cap(struct device *dev, * DOC: hwmon * * The amdgpu driver exposes the following sensor interfaces: + * * - GPU temperature (via the on-die sensor) + * * - GPU voltage + * * - Northbridge voltage (APUs only) + * * - GPU power + * * - GPU fan * * hwmon interfaces for GPU temperature: + * * - temp1_input: the on die GPU temperature in millidegrees Celsius + * * - temp1_crit: temperature critical max value in millidegrees Celsius + * * - temp1_crit_hyst: temperature hysteresis for critical limit in millidegrees Celsius * * hwmon interfaces for GPU voltage: + * * - in0_input: the voltage on the GPU in millivolts + * * - in1_input: the voltage on the Northbridge in millivolts * * hwmon interfaces for GPU power: + * * - power1_average: average power used by the GPU in microWatts + * * - power1_cap_min: minimum cap supported in microWatts + * * - power1_cap_max: maximum cap supported in microWatts + * * - power1_cap: selected power cap in microWatts * * hwmon interfaces for GPU fan: + * * - pwm1: pulse width modulation fan level (0-255) - * - pwm1_enable: pulse width modulation fan control method - * 0: no fan speed control - * 1: manual fan speed control using pwm interface - * 2: automatic fan speed control + * + * - pwm1_enable: pulse width modulation fan control method (0: no fan speed control, 1: manual fan speed control using pwm interface, 2: automatic fan speed control) + * * - pwm1_min: pulse width modulation fan control minimum level (0) + * * - pwm1_max: pulse width modulation fan control maximum level (255) + * * - fan1_input: fan speed in RPM * * You can use hwmon tools like sensors to view this information on your system. -- cgit v1.2.3 From ab3ec8a1a9f1c785103f395e9502ab5804705134 Mon Sep 17 00:00:00 2001 From: Michel Dänzer Date: Thu, 14 Jun 2018 11:59:20 +0200 Subject: drm/amdgpu: Restore :internal: for amdgpu_vm.c documentation MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This was accidentally dropped by the "drm/amdgpu: update documentation for amdgpu_irq.c v3" change, resulting in the GPUVM documentation body being included twice in the generated documentation. Trivial. Signed-off-by: Michel Dänzer Signed-off-by: Alex Deucher --- Documentation/gpu/amdgpu.rst | 1 + 1 file changed, 1 insertion(+) (limited to 'Documentation/gpu') diff --git a/Documentation/gpu/amdgpu.rst b/Documentation/gpu/amdgpu.rst index e1656fe8f247..e52d0ce186fe 100644 --- a/Documentation/gpu/amdgpu.rst +++ b/Documentation/gpu/amdgpu.rst @@ -52,6 +52,7 @@ AMDGPU Virtual Memory :doc: GPUVM .. kernel-doc:: drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c + :internal: Interrupt Handling ------------------ -- cgit v1.2.3