drm fence fixes and cross-dev sync.
Subsystems (10)
cover drm fence fixes and cross-dev sync.
This has a nouveau fix for suspend (no interruptible wait during suspend) and bumps patchlevel to abi 1.2.1 for telling shared fences are in use.
Maarten Lankhorst (8):
drm/nouveau: bump driver patchlevel to 1.2.1 drm/nouveau: specify if interruptible wait is desired in
nouveau_fence_sync
drm: Pass dma-buf as argument to gem_prime_import_sg_table drm/ttm: add reservation_object as argument to ttm_bo_init drm/radeon: cope with foreign fences inside display drm/radeon: cope with foreign fences inside the reservation object drm/radeon: export reservation_object from dmabuf to ttm drm/nouveau: export reservation_object from dmabuf to ttm
drivers/gpu/drm/ast/ast_ttm.c | 2 +- drivers/gpu/drm/bochs/bochs_mm.c | 2 +- drivers/gpu/drm/cirrus/cirrus_ttm.c | 2 +- drivers/gpu/drm/drm_gem_cma_helper.c | 5 +++-- drivers/gpu/drm/drm_prime.c | 2 +- drivers/gpu/drm/mgag200/mgag200_ttm.c | 2 +- drivers/gpu/drm/msm/msm_drv.h | 2 +- drivers/gpu/drm/msm/msm_gem_prime.c | 4 ++-- drivers/gpu/drm/nouveau/dispnv04/crtc.c | 2 +- drivers/gpu/drm/nouveau/nouveau_bo.c | 6 +++--- drivers/gpu/drm/nouveau/nouveau_bo.h | 1 + drivers/gpu/drm/nouveau/nouveau_chan.c | 2 +- drivers/gpu/drm/nouveau/nouveau_display.c | 4 ++-- drivers/gpu/drm/nouveau/nouveau_drm.h | 4 +++- drivers/gpu/drm/nouveau/nouveau_fence.c | 20 +++++++++++--------- drivers/gpu/drm/nouveau/nouveau_fence.h | 2 +- drivers/gpu/drm/nouveau/nouveau_gem.c | 4 ++-- drivers/gpu/drm/nouveau/nouveau_gem.h | 2 +- drivers/gpu/drm/nouveau/nouveau_prime.c | 10 +++++++--- drivers/gpu/drm/nouveau/nv17_fence.c | 2 +- drivers/gpu/drm/nouveau/nv50_display.c | 6 +++--- drivers/gpu/drm/nouveau/nv50_fence.c | 2 +- drivers/gpu/drm/nouveau/nv84_fence.c | 4 ++-- drivers/gpu/drm/qxl/qxl_drv.h | 2 +- drivers/gpu/drm/qxl/qxl_object.c | 2 +- drivers/gpu/drm/qxl/qxl_prime.c | 2 +- drivers/gpu/drm/radeon/cik.c | 6 +++--- drivers/gpu/drm/radeon/cik_sdma.c | 2 +- drivers/gpu/drm/radeon/evergreen.c | 6 +++--- drivers/gpu/drm/radeon/evergreen_dma.c | 2 +- drivers/gpu/drm/radeon/r600.c | 6 +++--- drivers/gpu/drm/radeon/r600_dma.c | 2 +- drivers/gpu/drm/radeon/radeon.h | 9 +++++---- drivers/gpu/drm/radeon/radeon_benchmark.c | 4 ++-- drivers/gpu/drm/radeon/radeon_cs.c | 27 ++++++++++++++++++++------- drivers/gpu/drm/radeon/radeon_device.c | 2 +- drivers/gpu/drm/radeon/radeon_display.c | 30 +++++++++++++++++++----------- drivers/gpu/drm/radeon/radeon_drv.c | 2 +- drivers/gpu/drm/radeon/radeon_fence.c | 3 +++ drivers/gpu/drm/radeon/radeon_gart.c | 2 +- drivers/gpu/drm/radeon/radeon_gem.c | 2 +- drivers/gpu/drm/radeon/radeon_object.c | 8 +++++--- drivers/gpu/drm/radeon/radeon_object.h | 1 + drivers/gpu/drm/radeon/radeon_prime.c | 10 +++++++--- drivers/gpu/drm/radeon/radeon_ring.c | 2 +- drivers/gpu/drm/radeon/radeon_sa.c | 2 +- drivers/gpu/drm/radeon/radeon_semaphore.c | 24 ++++++++++++++++++------ drivers/gpu/drm/radeon/radeon_test.c | 5 +++-- drivers/gpu/drm/radeon/radeon_ttm.c | 2 +- drivers/gpu/drm/radeon/radeon_uvd.c | 3 ++- drivers/gpu/drm/radeon/radeon_vce.c | 3 ++- drivers/gpu/drm/radeon/radeon_vm.c | 9 +++++---- drivers/gpu/drm/radeon/rv770_dma.c | 2 +- drivers/gpu/drm/radeon/si_dma.c | 2 +- drivers/gpu/drm/ttm/ttm_bo.c | 24 ++++++++++++++++++------ drivers/gpu/drm/vmwgfx/vmwgfx_resource.c | 2 +- include/drm/drmP.h | 3 ++- include/drm/drm_gem_cma_helper.h | 3 ++- include/drm/ttm/ttm_bo_api.h | 2 ++ 59 files changed, 192 insertions(+), 118 deletions(-)
Signature · 1 line
2.0.4
PATCH 1/8 drm/nouveau: bump driver patchlevel to 1.2.1
Allows userspace to detect shared fences are supported.
--- drivers/gpu/drm/nouveau/nouveau_drm.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nouveau_drm.h +3 -1
@@ -10,7 +10,7 @@#define DRIVER_MAJOR 1#define DRIVER_MINOR 2 -#define DRIVER_PATCHLEVEL 0 +#define DRIVER_PATCHLEVEL 1/* * 1.1.1:
@@ -26,6 +26,8 @@* 1.2.0: * - object api exposed to userspace * - fermi,kepler,maxwell zbc +* 1.2.1: +* - allow concurrent access to bo's mapped read/write. */ #include <nvif/client.h>
Signature · 1 line
2.0.4
PATCH 2/8 drm/nouveau: specify if interruptible wait is desired in nouveau_fence_sync
In reply to Maarten Lankhorst--- drivers/gpu/drm/nouveau/nouveau_bo.c | 2 +- drivers/gpu/drm/nouveau/nouveau_display.c | 4 ++-- drivers/gpu/drm/nouveau/nouveau_fence.c | 14 ++++++-------- drivers/gpu/drm/nouveau/nouveau_fence.h | 2 +- drivers/gpu/drm/nouveau/nouveau_gem.c | 2 +- 5 files changed, 11 insertions(+), 13 deletions(-)
drivers/gpu/drm/nouveau/nouveau_bo.c +1 -1
@@ -970,7 +970,7 @@ nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, bool intr, }mutex_lock_nested (&cli->mutex,SINGLE_DEPTH_NESTING ); - ret =nouveau_fence_sync (nouveau_bo (bo), chan, true); + ret =nouveau_fence_sync (nouveau_bo (bo), chan, true, intr);if (ret ==0 ) { ret = drm->ttm.move (chan, bo, &bo->mem, new_mem);if (ret ==0 ) {
drivers/gpu/drm/nouveau/nouveau_display.c +2 -2
@@ -658,7 +658,7 @@ nouveau_page_flip_emit(struct nouveau_channel *chan,spin_unlock_irqrestore (&dev->event_lock, flags);/* Synchronize with the old framebuffer */ - ret =nouveau_fence_sync (old_bo, chan, false); + ret =nouveau_fence_sync (old_bo, chan, false, false);if (ret) goto fail;
@@ -722,7 +722,7 @@ nouveau_crtc_page_flip(struct drm_crtc *crtc, struct drm_framebuffer *fb, goto fail_unpin;/* synchronise rendering channel with the kernel's channel */ - ret =nouveau_fence_sync (new_bo, chan, false); + ret =nouveau_fence_sync (new_bo, chan, false, true);if (ret) {ttm_bo_unreserve (&new_bo->bo); goto fail_unpin;
drivers/gpu/drm/nouveau/nouveau_fence.c +6 -8
@@ -226,12 +226,10 @@ nouveau_fence_emit(struct nouveau_fence *fence, struct nouveau_channel *chan)if (priv->uevent)fence_init (&fence->base, &nouveau_fence_ops_uevent, - &fctx->lock, - priv->context_base + chan->chid, ++fctx->sequence); + &fctx->lock, fctx->context, ++fctx->sequence);else fence_init (&fence->base, &nouveau_fence_ops_legacy, - &fctx->lock, - priv->context_base + chan->chid, ++fctx->sequence); + &fctx->lock, fctx->context, ++fctx->sequence);trace_fence_emit (&fence->base); ret = fctx->emit (fence);
@@ -342,7 +340,7 @@ nouveau_fence_wait(struct nouveau_fence *fence, bool lazy, bool intr) }int -nouveau_fence_sync (struct nouveau_bo *nvbo,struct nouveau_channel *chan,bool exclusive) +nouveau_fence_sync (struct nouveau_bo *nvbo,struct nouveau_channel *chan,bool exclusive,bool intr) {struct nouveau_fence_chan *fctx = chan->fence;struct fence *fence;
@@ -369,7 +367,7 @@ nouveau_fence_sync(struct nouveau_bo *nvbo, struct nouveau_channel *chan, bool e prev = f->channel;if (!prev || (prev != chan && (ret = fctx->sync (f, prev, chan)))) - ret =fence_wait (fence, true); + ret =fence_wait (fence, intr);return ret; }
@@ -387,8 +385,8 @@ nouveau_fence_sync(struct nouveau_bo *nvbo, struct nouveau_channel *chan, bool eif (f) prev = f->channel; -if (!prev || (ret = fctx->sync (f, prev, chan))) - ret =fence_wait (fence, true); +if (!prev || (prev != chan && (ret = fctx->sync (f, prev, chan)))) + ret =fence_wait (fence, intr);if (ret)break ;
drivers/gpu/drm/nouveau/nouveau_fence.h +1 -1
@@ -26,7 +26,7 @@ int nouveau_fence_emit(struct nouveau_fence *, struct nouveau_channel *);bool nouveau_fence_done (struct nouveau_fence *);void nouveau_fence_work (struct fence *,void (*)(void *),void *);int nouveau_fence_wait (struct nouveau_fence *,bool lazy,bool intr); -int nouveau_fence_sync (struct nouveau_bo *,struct nouveau_channel *,bool exclusive); +int nouveau_fence_sync (struct nouveau_bo *,struct nouveau_channel *,bool exclusive,bool intr);struct nouveau_fence_chan {spinlock_t lock;
drivers/gpu/drm/nouveau/nouveau_gem.c +1 -1
@@ -459,7 +459,7 @@ validate_list(struct nouveau_channel *chan, struct nouveau_cli *cli,return ret; } - ret =nouveau_fence_sync (nvbo, chan, !!b->write_domains); + ret =nouveau_fence_sync (nvbo, chan, !!b->write_domains, true);if (unlikely (ret)) {if (ret != -ERESTARTSYS )NV_PRINTK (error, cli,"fail post-validate sync\n" );
Signature · 1 line
2.0.4
PATCH 3/8 drm: Pass dma-buf as argument to gem_prime_import_sg_table
In reply to Maarten LankhorstAllows importing dma_reservation_objects from a dma-buf.
--- drivers/gpu/drm/drm_gem_cma_helper.c | 5 +++-- drivers/gpu/drm/drm_prime.c | 2 +- drivers/gpu/drm/msm/msm_drv.h | 2 +- drivers/gpu/drm/msm/msm_gem_prime.c | 4 ++-- drivers/gpu/drm/nouveau/nouveau_gem.h | 2 +- drivers/gpu/drm/nouveau/nouveau_prime.c | 5 +++-- drivers/gpu/drm/qxl/qxl_drv.h | 2 +- drivers/gpu/drm/qxl/qxl_prime.c | 2 +- drivers/gpu/drm/radeon/radeon_drv.c | 2 +- drivers/gpu/drm/radeon/radeon_prime.c | 5 +++-- include/drm/drmP.h | 3 ++- include/drm/drm_gem_cma_helper.h | 3 ++- 12 files changed, 21 insertions(+), 16 deletions(-)
drivers/gpu/drm/drm_gem_cma_helper.c +3 -2
@@ -316,7 +316,8 @@ out:EXPORT_SYMBOL_GPL (drm_gem_cma_prime_get_sg_table);struct drm_gem_object * -drm_gem_cma_prime_import_sg_table (struct drm_device *dev,size_t size, +drm_gem_cma_prime_import_sg_table (struct drm_device *dev, +struct dma_buf_attachment *attach,struct sg_table *sgt) {struct drm_gem_cma_object *cma_obj;
@@ -325,7 +326,7 @@ drm_gem_cma_prime_import_sg_table(struct drm_device *dev, size_t size,return ERR_PTR (-EINVAL );/* Create a CMA GEM buffer. */ - cma_obj =__drm_gem_cma_create (dev, size); + cma_obj =__drm_gem_cma_create (dev, attach->dmabuf->size);if (IS_ERR (cma_obj))return ERR_CAST (cma_obj);
drivers/gpu/drm/drm_prime.c +1 -1
@@ -523,7 +523,7 @@ struct drm_gem_object *drm_gem_prime_import(struct drm_device *dev, goto fail_detach; } - obj = dev->driver->gem_prime_import_sg_table (dev, dma_buf->size, sgt); + obj = dev->driver->gem_prime_import_sg_table (dev, attach, sgt);if (IS_ERR (obj)) { ret =PTR_ERR (obj); goto fail_unmap;
drivers/gpu/drm/msm/msm_drv.h +1 -1
@@ -170,7 +170,7 @@ struct sg_table *msm_gem_prime_get_sg_table(struct drm_gem_object *obj);void *msm_gem_prime_vmap (struct drm_gem_object *obj);void msm_gem_prime_vunmap (struct drm_gem_object *obj,void *vaddr);struct drm_gem_object *msm_gem_prime_import_sg_table (struct drm_device *dev, -size_t size,struct sg_table *sg); +struct dma_buf_attachment *attach,struct sg_table *sg);int msm_gem_prime_pin (struct drm_gem_object *obj);void msm_gem_prime_unpin (struct drm_gem_object *obj);void *msm_gem_vaddr_locked (struct drm_gem_object *obj);
drivers/gpu/drm/msm/msm_gem_prime.c +2 -2
@@ -37,9 +37,9 @@ void msm_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr) }struct drm_gem_object *msm_gem_prime_import_sg_table (struct drm_device *dev, -size_t size,struct sg_table *sg) +struct dma_buf_attachment *attach,struct sg_table *sg) { -return msm_gem_import (dev, size, sg); +return msm_gem_import (dev, attach->dmabuf->size, sg); }int msm_gem_prime_pin (struct drm_gem_object *obj)
drivers/gpu/drm/nouveau/nouveau_gem.h +1 -1
@@ -39,7 +39,7 @@ struct reservation_object *nouveau_gem_prime_res_obj(struct drm_gem_object *);extern void nouveau_gem_prime_unpin (struct drm_gem_object *);extern struct sg_table *nouveau_gem_prime_get_sg_table (struct drm_gem_object *);extern struct drm_gem_object *nouveau_gem_prime_import_sg_table ( -struct drm_device *,size_t size,struct sg_table *); +struct drm_device *,struct dma_buf_attachment *,struct sg_table *);extern void *nouveau_gem_prime_vmap (struct drm_gem_object *);extern void nouveau_gem_prime_vunmap (struct drm_gem_object *,void *);
drivers/gpu/drm/nouveau/nouveau_prime.c +3 -2
@@ -23,6 +23,7 @@*/ #include <drm/drmP.h> +#include <linux/dma-buf.h> #include "nouveau_drm.h" #include "nouveau_gem.h"
@@ -56,7 +57,7 @@ void nouveau_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr) }struct drm_gem_object *nouveau_gem_prime_import_sg_table (struct drm_device *dev, -size_t size, +struct dma_buf_attachment *attach,struct sg_table *sg) {struct nouveau_bo *nvbo;
@@ -65,7 +66,7 @@ struct drm_gem_object *nouveau_gem_prime_import_sg_table(struct drm_device *dev, flags =TTM_PL_FLAG_TT ; - ret =nouveau_bo_new (dev, size,0 , flags,0 ,0 , + ret =nouveau_bo_new (dev, attach->dmabuf->size,0 , flags,0 ,0 , sg, &nvbo);if (ret)return ERR_PTR (ret);
drivers/gpu/drm/qxl/qxl_drv.h +1 -1
@@ -535,7 +535,7 @@ int qxl_gem_prime_pin(struct drm_gem_object *obj);void qxl_gem_prime_unpin (struct drm_gem_object *obj);struct sg_table *qxl_gem_prime_get_sg_table (struct drm_gem_object *obj);struct drm_gem_object *qxl_gem_prime_import_sg_table ( -struct drm_device *dev,size_t size, +struct drm_device *dev,struct dma_buf_attachment *attach,struct sg_table *sgt);void *qxl_gem_prime_vmap (struct drm_gem_object *obj);void qxl_gem_prime_vunmap (struct drm_gem_object *obj,void *vaddr);
drivers/gpu/drm/qxl/qxl_prime.c +1 -1
@@ -46,7 +46,7 @@ struct sg_table *qxl_gem_prime_get_sg_table(struct drm_gem_object *obj) }struct drm_gem_object *qxl_gem_prime_import_sg_table ( -struct drm_device *dev,size_t size, +struct drm_device *dev,struct dma_buf_attachment *attach,struct sg_table *table) {WARN_ONCE (1 ,"not implemented" );
drivers/gpu/drm/radeon/radeon_drv.c +1 -1
@@ -133,7 +133,7 @@ int radeon_mode_dumb_create(struct drm_file *file_priv, struct drm_mode_create_dumb *args);struct sg_table *radeon_gem_prime_get_sg_table (struct drm_gem_object *obj);struct drm_gem_object *radeon_gem_prime_import_sg_table (struct drm_device *dev, -size_t size, +struct dma_buf_attachment *,struct sg_table *sg);int radeon_gem_prime_pin (struct drm_gem_object *obj);void radeon_gem_prime_unpin (struct drm_gem_object *obj);
drivers/gpu/drm/radeon/radeon_prime.c +3 -2
@@ -27,6 +27,7 @@#include "radeon.h" #include <drm/radeon_drm.h> +#include <linux/dma-buf.h> struct sg_table *radeon_gem_prime_get_sg_table (struct drm_gem_object *obj) {
@@ -57,14 +58,14 @@ void radeon_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr) }struct drm_gem_object *radeon_gem_prime_import_sg_table (struct drm_device *dev, -size_t size, +struct dma_buf_attachment *attach,struct sg_table *sg) {struct radeon_device *rdev = dev->dev_private;struct radeon_bo *bo;int ret; - ret =radeon_bo_create (rdev, size,PAGE_SIZE , false, + ret =radeon_bo_create (rdev, attach->dmabuf->size,PAGE_SIZE , false,RADEON_GEM_DOMAIN_GTT ,0 , sg, &bo);if (ret)return ERR_PTR (ret);
include/drm/drmP.h +2 -1
@@ -656,7 +656,8 @@ struct drm_driver { struct drm_gem_object *obj);struct sg_table *(*gem_prime_get_sg_table)(struct drm_gem_object *obj);struct drm_gem_object *(*gem_prime_import_sg_table)( -struct drm_device *dev,size_t size, +struct drm_device *dev, +struct dma_buf_attachment *attach,struct sg_table *sgt);void *(*gem_prime_vmap)(struct drm_gem_object *obj);void (*gem_prime_vunmap)(struct drm_gem_object *obj,void *vaddr);
include/drm/drm_gem_cma_helper.h +2 -1
@@ -44,7 +44,8 @@ void drm_gem_cma_describe(struct drm_gem_cma_object *obj, struct seq_file *m);struct sg_table *drm_gem_cma_prime_get_sg_table (struct drm_gem_object *obj);struct drm_gem_object * -drm_gem_cma_prime_import_sg_table (struct drm_device *dev,size_t size, +drm_gem_cma_prime_import_sg_table (struct drm_device *dev, +struct dma_buf_attachment *attach,struct sg_table *sgt);int drm_gem_cma_prime_mmap (struct drm_gem_object *obj,struct vm_area_struct *vma);
Signature · 1 line
2.0.4
PATCH 4/8 drm/ttm: add reservation_object as argument to ttm_bo_init
In reply to Maarten LankhorstThis allows importing reservation objects from dma-bufs.
--- drivers/gpu/drm/ast/ast_ttm.c | 2 +- drivers/gpu/drm/bochs/bochs_mm.c | 2 +- drivers/gpu/drm/cirrus/cirrus_ttm.c | 2 +- drivers/gpu/drm/mgag200/mgag200_ttm.c | 2 +- drivers/gpu/drm/nouveau/nouveau_bo.c | 2 +- drivers/gpu/drm/qxl/qxl_object.c | 2 +- drivers/gpu/drm/radeon/radeon_object.c | 2 +- drivers/gpu/drm/ttm/ttm_bo.c | 24 ++++++++++++++++++------ drivers/gpu/drm/vmwgfx/vmwgfx_resource.c | 2 +- include/drm/ttm/ttm_bo_api.h | 2 ++ 10 files changed, 28 insertions(+), 14 deletions(-)
drivers/gpu/drm/ast/ast_ttm.c +1 -1
@@ -339,7 +339,7 @@ int ast_bo_create(struct drm_device *dev, int size, int align, ret =ttm_bo_init (&ast->ttm.bdev, &astbo->bo, size, ttm_bo_type_device, &astbo->placement, align >>PAGE_SHIFT , false,NULL , acc_size, -NULL , ast_bo_ttm_destroy); +NULL ,NULL , ast_bo_ttm_destroy);if (ret)return ret;
drivers/gpu/drm/bochs/bochs_mm.c +1 -1
@@ -377,7 +377,7 @@ static int bochs_bo_create(struct drm_device *dev, int size, int align, ret =ttm_bo_init (&bochs->ttm.bdev, &bochsbo->bo, size, ttm_bo_type_device, &bochsbo->placement, align >>PAGE_SHIFT , false,NULL , acc_size, -NULL , bochs_bo_ttm_destroy); +NULL ,NULL , bochs_bo_ttm_destroy);if (ret)return ret;
drivers/gpu/drm/cirrus/cirrus_ttm.c +1 -1
@@ -343,7 +343,7 @@ int cirrus_bo_create(struct drm_device *dev, int size, int align, ret =ttm_bo_init (&cirrus->ttm.bdev, &cirrusbo->bo, size, ttm_bo_type_device, &cirrusbo->placement, align >>PAGE_SHIFT , false,NULL , acc_size, -NULL , cirrus_bo_ttm_destroy); +NULL ,NULL , cirrus_bo_ttm_destroy);if (ret)return ret;
drivers/gpu/drm/mgag200/mgag200_ttm.c +1 -1
@@ -339,7 +339,7 @@ int mgag200_bo_create(struct drm_device *dev, int size, int align, ret =ttm_bo_init (&mdev->ttm.bdev, &mgabo->bo, size, ttm_bo_type_device, &mgabo->placement, align >>PAGE_SHIFT , false,NULL , acc_size, -NULL , mgag200_bo_ttm_destroy); +NULL ,NULL , mgag200_bo_ttm_destroy);if (ret)return ret;
drivers/gpu/drm/nouveau/nouveau_bo.c +1 -1
@@ -230,7 +230,7 @@ nouveau_bo_new(struct drm_device *dev, int size, int align, ret =ttm_bo_init (&drm->ttm.bdev, &nvbo->bo, size, type, &nvbo->placement, align >>PAGE_SHIFT , false,NULL , acc_size, sg, - nouveau_bo_del_ttm); +NULL , nouveau_bo_del_ttm);if (ret) {/* ttm will call nouveau_bo_del_ttm if it fails.. */ return ret;
drivers/gpu/drm/qxl/qxl_object.c +1 -1
@@ -110,7 +110,7 @@ int qxl_bo_create(struct qxl_device *qdev, r =ttm_bo_init (&qdev->mman.bdev, &bo->tbo, size, type, &bo->placement,0 , !kernel,NULL , size, -NULL , &qxl_ttm_bo_destroy); +NULL ,NULL , &qxl_ttm_bo_destroy);if (unlikely (r !=0 )) {if (r != -ERESTARTSYS ) dev_err(qdev->dev,
drivers/gpu/drm/radeon/radeon_object.c +1 -1
@@ -216,7 +216,7 @@ int radeon_bo_create(struct radeon_device *rdev,down_read (&rdev->pm.mclk_lock); r =ttm_bo_init (&rdev->mman.bdev, &bo->tbo, size, type, &bo->placement, page_align, !kernel,NULL , - acc_size, sg, &radeon_ttm_bo_destroy); + acc_size, sg,NULL , &radeon_ttm_bo_destroy);up_read (&rdev->pm.mclk_lock);if (unlikely (r !=0 )) {return r;
drivers/gpu/drm/ttm/ttm_bo.c +18 -6
@@ -1068,6 +1068,7 @@ int ttm_bo_init(struct ttm_bo_device *bdev,struct file *persistent_swap_storage, size_t acc_size, struct sg_table *sg, + struct reservation_object *resv, void (*destroy) (struct ttm_buffer_object *)) { int ret = 0;
@@ -1121,8 +1122,13 @@ int ttm_bo_init(struct ttm_bo_device *bdev, bo->persistent_swap_storage = persistent_swap_storage; bo->acc_size = acc_size; bo->sg = sg; - bo->resv = &bo->ttm_resv; -reservation_object_init (bo->resv); +if (resv) { + bo->resv = resv; +lockdep_assert_held (&bo->resv->lock.base); + }else { + bo->resv = &bo->ttm_resv; +reservation_object_init (&bo->ttm_resv); + }atomic_inc (&bo->glob->bo_count);drm_vma_node_reset (&bo->vma_node);
@@ -1135,13 +1141,19 @@ int ttm_bo_init(struct ttm_bo_device *bdev, ret =drm_vma_offset_add (&bdev->vma_manager, &bo->vma_node, bo->mem.num_pages); - locked =ww_mutex_trylock (&bo->resv->lock); -WARN_ON (!locked); +/* passed reservation objects should already be locked, +* since otherwise lockdep will be angered in radeon. +*/ +if (!resv) { + locked =ww_mutex_trylock (&bo->resv->lock); +WARN_ON (!locked); + }if (likely (!ret)) ret =ttm_bo_validate (bo, placement, interruptible, false); -ttm_bo_unreserve (bo); +if (!resv) +ttm_bo_unreserve (bo);if (unlikely (ret))ttm_bo_unref (&bo);
@@ -1199,7 +1211,7 @@ int ttm_bo_create(struct ttm_bo_device *bdev, acc_size =ttm_bo_acc_size (bdev, size,sizeof (struct ttm_buffer_object )); ret =ttm_bo_init (bdev, bo, size, type, placement, page_alignment, interruptible, persistent_swap_storage, acc_size, -NULL ,NULL ); +NULL ,NULL ,NULL );if (likely (ret ==0 )) *p_bo = bo;
drivers/gpu/drm/vmwgfx/vmwgfx_resource.c +1 -1
@@ -430,7 +430,7 @@ int vmw_dmabuf_init(struct vmw_private *dev_priv, ret =ttm_bo_init (bdev, &vmw_bo->base, size, ttm_bo_type_device, placement,0 , interruptible, -NULL , acc_size,NULL , bo_free); +NULL , acc_size,NULL ,NULL , bo_free);return ret; }
include/drm/ttm/ttm_bo_api.h +2 -0
@@ -460,6 +460,7 @@ size_t ttm_bo_dma_acc_size(struct ttm_bo_device *bdev,* point to the shmem object backing a GEM object if TTM is used to back a * GEM user interface. * @acc_size: Accounted size for this object. +* @resv: Pointer to a reservation_object, or NULL to let ttm allocate one. * @destroy: Destroy function. Use NULL for kfree(). * * This function initializes a pre-allocated struct ttm_buffer_object.
@@ -487,6 +488,7 @@ extern int ttm_bo_init(struct ttm_bo_device *bdev,struct file *persistent_swap_storage, size_t acc_size, struct sg_table *sg, + struct reservation_object *resv, void (*destroy) (struct ttm_buffer_object *));/**
Signature · 1 line
2.0.4
PATCH 5/8 drm/radeon: cope with foreign fences inside display
In reply to Maarten Lankhorst--- drivers/gpu/drm/radeon/radeon.h | 2 +- drivers/gpu/drm/radeon/radeon_display.c | 30 +++++++++++++++++++----------- 2 files changed, 20 insertions(+), 12 deletions(-)
drivers/gpu/drm/radeon/radeon.h +1 -1
@@ -710,7 +710,7 @@ struct radeon_flip_work {uint64_t base;struct drm_pending_vblank_event *event;struct radeon_bo *old_rbo; -struct radeon_fence *fence; +struct fence *fence; };struct r500_irq_stat_regs {
drivers/gpu/drm/radeon/radeon_display.c +19 -11
@@ -402,14 +402,21 @@ static void radeon_flip_work_func(struct work_struct *__work)down_read (&rdev->exclusive_lock);if (work->fence) { - r =radeon_fence_wait (work->fence, false); -if (r == -EDEADLK ) { -up_read (&rdev->exclusive_lock); -do { - r =radeon_gpu_reset (rdev); - }while (r == -EAGAIN ); -down_read (&rdev->exclusive_lock); - } +struct radeon_fence *fence; + + fence =to_radeon_fence (work->fence); +if (fence && fence->rdev == rdev) { + r =radeon_fence_wait (fence, false); +if (r == -EDEADLK ) { +up_read (&rdev->exclusive_lock); +do { + r =radeon_gpu_reset (rdev); + }while (r == -EAGAIN ); +down_read (&rdev->exclusive_lock); + } + }else + r =fence_wait (work->fence, false); +if (r)DRM_ERROR ("failed to wait on page flip fence (%d)!\n" , r);
@@ -418,7 +425,8 @@ static void radeon_flip_work_func(struct work_struct *__work)* confused about which BO the CRTC is scanning out */ -radeon_fence_unref (&work->fence); +fence_put (work->fence); + work->fence =NULL ; }/* We borrow the event spin lock for protecting flip_status */
@@ -494,7 +502,7 @@ static int radeon_crtc_page_flip(struct drm_crtc *crtc,DRM_ERROR ("failed to pin new rbo buffer before flip\n" ); goto cleanup; } - work->fence = (struct radeon_fence *)fence_get (reservation_object_get_excl (new_rbo->tbo.resv)); + work->fence =fence_get (reservation_object_get_excl (new_rbo->tbo.resv));radeon_bo_get_tiling_flags (new_rbo, &tiling_flags,NULL );radeon_bo_unreserve (new_rbo);
@@ -576,7 +584,7 @@ pflip_cleanup: cleanup:drm_gem_object_unreference_unlocked (&work->old_rbo->gem_base); -radeon_fence_unref (&work->fence); +fence_put (work->fence);kfree (work);return r; }
Signature · 1 line
2.0.4
Re: [PATCH 5/8] drm/radeon: cope with foreign fences inside display
Am 17.09.2014 um 14:35 schrieb Maarten Lankhorst:
Quoted 1 line
Signed-off-by Maarten Lankhorst <[email protected]>
Quoted 80 lines
--- drivers/gpu/drm/radeon/radeon.h | 2 +- drivers/gpu/drm/radeon/radeon_display.c | 30 +++++++++++++++++++----------- 2 files changed, 20 insertions(+), 12 deletions(-)drivers/gpu/drm/radeon/radeon.h +1
-1@@ -710,7 +710,7 @@ struct radeon_flip_work {uint64_t base;struct drm_pending_vblank_event *event;struct radeon_bo *old_rbo; -struct radeon_fence *fence; +struct fence *fence; };struct r500_irq_stat_regs {drivers/gpu/drm/radeon/radeon_display.c +19
-11@@ -402,14 +402,21 @@ static void radeon_flip_work_func(struct work_struct *__work)down_read (&rdev->exclusive_lock);if (work->fence) { - r =radeon_fence_wait (work->fence, false); -if (r == -EDEADLK ) { -up_read (&rdev->exclusive_lock); -do { - r =radeon_gpu_reset (rdev); - }while (r == -EAGAIN ); -down_read (&rdev->exclusive_lock); - } +struct radeon_fence *fence; + + fence =to_radeon_fence (work->fence); +if (fence && fence->rdev == rdev) { + r =radeon_fence_wait (fence, false); +if (r == -EDEADLK ) { +up_read (&rdev->exclusive_lock); +do { + r =radeon_gpu_reset (rdev); + }while (r == -EAGAIN ); +down_read (&rdev->exclusive_lock); + } + }else + r =fence_wait (work->fence, false); +if (r)DRM_ERROR ("failed to wait on page flip fence (%d)!\n" , r);@@ -418,7 +425,8 @@ static void radeon_flip_work_func(struct work_struct *__work)* confused about which BO the CRTC is scanning out */ -radeon_fence_unref (&work->fence); +fence_put (work->fence); + work->fence =NULL ; }/* We borrow the event spin lock for protecting flip_status */ @@ -494,7 +502,7 @@ static int radeon_crtc_page_flip(struct drm_crtc *crtc,DRM_ERROR ("failed to pin new rbo buffer before flip\n" ); goto cleanup; } - work->fence = (struct radeon_fence *)fence_get (reservation_object_get_excl (new_rbo->tbo.resv)); + work->fence =fence_get (reservation_object_get_excl (new_rbo->tbo.resv));radeon_bo_get_tiling_flags (new_rbo, &tiling_flags,NULL );radeon_bo_unreserve (new_rbo);@@ -576,7 +584,7 @@ pflip_cleanup: cleanup:drm_gem_object_unreference_unlocked (&work->old_rbo->gem_base); -radeon_fence_unref (&work->fence); +fence_put (work->fence);kfree (work);return r; }
PATCH 6/8 drm/radeon: cope with foreign fences inside the reservation object
In reply to Maarten LankhorstNot the whole world is a radeon! :-)
--- drivers/gpu/drm/radeon/cik.c | 2 +- drivers/gpu/drm/radeon/cik_sdma.c | 2 +- drivers/gpu/drm/radeon/evergreen_dma.c | 2 +- drivers/gpu/drm/radeon/r600.c | 2 +- drivers/gpu/drm/radeon/r600_dma.c | 2 +- drivers/gpu/drm/radeon/radeon.h | 7 ++++--- drivers/gpu/drm/radeon/radeon_cs.c | 27 ++++++++++++++++++++------- drivers/gpu/drm/radeon/radeon_fence.c | 3 +++ drivers/gpu/drm/radeon/radeon_semaphore.c | 24 ++++++++++++++++++------ drivers/gpu/drm/radeon/radeon_vm.c | 4 ++-- drivers/gpu/drm/radeon/rv770_dma.c | 2 +- drivers/gpu/drm/radeon/si_dma.c | 2 +- 12 files changed, 54 insertions(+), 25 deletions(-)
drivers/gpu/drm/radeon/cik.c +1 -1
@@ -3993,7 +3993,7 @@ struct radeon_fence *cik_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/cik_sdma.c +1 -1
@@ -571,7 +571,7 @@ struct radeon_fence *cik_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/evergreen_dma.c +1 -1
@@ -133,7 +133,7 @@ struct radeon_fence *evergreen_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/r600.c +1 -1
@@ -2912,7 +2912,7 @@ struct radeon_fence *r600_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);radeon_ring_write (ring,PACKET3 (PACKET3_SET_CONFIG_REG ,1 ));
drivers/gpu/drm/radeon/r600_dma.c +1 -1
@@ -470,7 +470,7 @@ struct radeon_fence *r600_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/radeon.h +4 -3
@@ -587,9 +587,10 @@ bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring, struct radeon_semaphore *semaphore);void radeon_semaphore_sync_fence (struct radeon_semaphore *semaphore,struct radeon_fence *fence); -void radeon_semaphore_sync_resv (struct radeon_semaphore *semaphore, -struct reservation_object *resv, -bool shared); +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *semaphore, +struct reservation_object *resv, +bool shared,bool intr);int radeon_semaphore_sync_rings (struct radeon_device *rdev,struct radeon_semaphore *semaphore,int waiting_ring);
drivers/gpu/drm/radeon/radeon_cs.c +20 -7
@@ -249,20 +249,21 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priorityreturn 0 ; } -static void radeon_cs_sync_rings (struct radeon_cs_parser *p) +static int radeon_cs_sync_rings (struct radeon_cs_parser *p) { -int i; +int i, r =0 ; -for (i =0 ; i < p->nrelocs; i++) { +for (i =0 ; !r && i < p->nrelocs; i++) {struct reservation_object *resv;if (!p->relocs[i].robj)continue ; resv = p->relocs[i].robj->tbo.resv; -radeon_semaphore_sync_resv (p->ib.semaphore, resv, - p->relocs[i].tv.shared); + r =radeon_semaphore_sync_resv (p->rdev, p->ib.semaphore, resv, + p->relocs[i].tv.shared, true); } +return r; }/* XXX: note that this is called from the legacy UMS CS ioctl as well */
@@ -472,13 +473,19 @@ static int radeon_cs_ib_chunk(struct radeon_device *rdev,return r; } + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); +return r; + } +if (parser->ring ==R600_RING_TYPE_UVD_INDEX )radeon_uvd_note_usage (rdev);else if ((parser->ring ==TN_RING_TYPE_VCE1_INDEX ) || (parser->ring ==TN_RING_TYPE_VCE2_INDEX ))radeon_vce_note_usage (rdev); -radeon_cs_sync_rings (parser); r =radeon_ib_schedule (rdev, &parser->ib,NULL , true);if (r) {DRM_ERROR ("Failed to schedule IB !\n" );
@@ -565,7 +572,13 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev,if (r) { goto out; } -radeon_cs_sync_rings (parser); + + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); + goto out; + }radeon_semaphore_sync_fence (parser->ib.semaphore, vm->fence); if ((rdev->family >= CHIP_TAHITI) &&
drivers/gpu/drm/radeon/radeon_fence.c +3 -0
@@ -541,6 +541,9 @@ int radeon_fence_wait(struct radeon_fence *fence, bool intr)uint64_t seq[RADEON_NUM_RINGS ] = {};long r; +if (unlikely (!to_radeon_fence (&fence->base))) +return fence_wait (&fence->base, intr); + seq[fence->ring] = fence->seq; r =radeon_fence_wait_seq_timeout (fence->rdev, seq, intr,MAX_SCHEDULE_TIMEOUT );if (r <0 ) {
drivers/gpu/drm/radeon/radeon_semaphore.c +18 -6
@@ -124,27 +124,39 @@ void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore,* * Sync to the fence using this semaphore object */ -void radeon_semaphore_sync_resv (struct radeon_semaphore *sema, -struct reservation_object *resv, -bool shared) +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *sema, +struct reservation_object *resv, +bool shared,bool intr) {struct reservation_object_list *flist;struct fence *f; +struct radeon_fence *fence;unsigned i; +int r =0 ;/* always sync to the exclusive fence */ f =reservation_object_get_excl (resv); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence = f ?to_radeon_fence (f) :NULL ; +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else if (f) + r =fence_wait (f, intr); flist =reservation_object_get_list (resv);if (shared || !flist) -return ; +return r;for (i =0 ; i < flist->shared_count; ++i) { f =rcu_dereference_protected (flist->shared[i],reservation_object_held (resv)); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence =to_radeon_fence (f); +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else if (!r) + r =fence_wait (f, intr); } +return r; }/**
drivers/gpu/drm/radeon/radeon_vm.c +2 -2
@@ -698,7 +698,7 @@ int radeon_vm_update_page_directory(struct radeon_device *rdev,if (ib.length_dw !=0 ) {radeon_asic_vm_pad_ib (rdev, &ib); -radeon_semaphore_sync_resv (ib.semaphore, pd->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib.semaphore, pd->tbo.resv, false, false);radeon_semaphore_sync_fence (ib.semaphore, vm->last_id_use);WARN_ON (ib.length_dw > ndw); r =radeon_ib_schedule (rdev, &ib,NULL , false);
@@ -825,7 +825,7 @@ static void radeon_vm_update_ptes(struct radeon_device *rdev,unsigned nptes;uint64_t pte; -radeon_semaphore_sync_resv (ib->semaphore, pt->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib->semaphore, pt->tbo.resv, false, false);if ((addr & ~mask) == (end & ~mask)) nptes = end - addr;
drivers/gpu/drm/radeon/rv770_dma.c +1 -1
@@ -67,7 +67,7 @@ struct radeon_fence *rv770_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/si_dma.c +1 -1
@@ -252,7 +252,7 @@ struct radeon_fence *si_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
Signature · 1 line
2.0.4
Re: [PATCH 6/8] drm/radeon: cope with foreign fences inside the reservation object
Am 17.09.2014 um 14:35 schrieb Maarten Lankhorst:
Quoted 117 lines
Not the whole world is a radeon! :-)
Signed-off-by Maarten Lankhorst <[email protected]>--- drivers/gpu/drm/radeon/cik.c | 2 +- drivers/gpu/drm/radeon/cik_sdma.c | 2 +- drivers/gpu/drm/radeon/evergreen_dma.c | 2 +- drivers/gpu/drm/radeon/r600.c | 2 +- drivers/gpu/drm/radeon/r600_dma.c | 2 +- drivers/gpu/drm/radeon/radeon.h | 7 ++++--- drivers/gpu/drm/radeon/radeon_cs.c | 27 ++++++++++++++++++++------- drivers/gpu/drm/radeon/radeon_fence.c | 3 +++ drivers/gpu/drm/radeon/radeon_semaphore.c | 24 ++++++++++++++++++------ drivers/gpu/drm/radeon/radeon_vm.c | 4 ++-- drivers/gpu/drm/radeon/rv770_dma.c | 2 +- drivers/gpu/drm/radeon/si_dma.c | 2 +- 12 files changed, 54 insertions(+), 25 deletions(-)drivers/gpu/drm/radeon/cik.c +1
-1@@ -3993,7 +3993,7 @@ struct radeon_fence *cik_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/cik_sdma.c +1
-1@@ -571,7 +571,7 @@ struct radeon_fence *cik_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/evergreen_dma.c +1
-1@@ -133,7 +133,7 @@ struct radeon_fence *evergreen_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/r600.c +1
-1@@ -2912,7 +2912,7 @@ struct radeon_fence *r600_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);radeon_ring_write (ring,PACKET3 (PACKET3_SET_CONFIG_REG ,1 ));drivers/gpu/drm/radeon/r600_dma.c +1
-1@@ -470,7 +470,7 @@ struct radeon_fence *r600_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/radeon.h +4
-3@@ -587,9 +587,10 @@ bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring, struct radeon_semaphore *semaphore);void radeon_semaphore_sync_fence (struct radeon_semaphore *semaphore,struct radeon_fence *fence); -void radeon_semaphore_sync_resv (struct radeon_semaphore *semaphore, -struct reservation_object *resv, -bool shared); +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *semaphore, +struct reservation_object *resv, +bool shared,bool intr);int radeon_semaphore_sync_rings (struct radeon_device *rdev,struct radeon_semaphore *semaphore,int waiting_ring);drivers/gpu/drm/radeon/radeon_cs.c +3
-3@@ -249,20 +249,21 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priorityreturn 0 ; } -static void radeon_cs_sync_rings (struct radeon_cs_parser *p) +static int radeon_cs_sync_rings (struct radeon_cs_parser *p) { -int i; +int i, r =0 ; -for (i =0 ; i < p->nrelocs; i++) { +for (i =0 ; !r && i < p->nrelocs; i++) {
drivers/gpu/drm/radeon/radeon_cs.c static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority · line 264
Quoted 10 lines
drivers/gpu/drm/radeon/radeon_cs.c +2
-2struct reservation_object *resv;if (!p->relocs[i].robj)continue ; resv = p->relocs[i].robj->tbo.resv; -radeon_semaphore_sync_resv (p->ib.semaphore, resv, - p->relocs[i].tv.shared); + r =radeon_semaphore_sync_resv (p->rdev, p->ib.semaphore, resv, + p->relocs[i].tv.shared, true);
Break on the first error?
Quoted 75 lines
drivers/gpu/drm/radeon/radeon_cs.c +15
-2} +return r; }/* XXX: note that this is called from the legacy UMS CS ioctl as well */ @@ -472,13 +473,19 @@ static int radeon_cs_ib_chunk(struct radeon_device *rdev,return r; } + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); +return r; + } +if (parser->ring ==R600_RING_TYPE_UVD_INDEX )radeon_uvd_note_usage (rdev);else if ((parser->ring ==TN_RING_TYPE_VCE1_INDEX ) || (parser->ring ==TN_RING_TYPE_VCE2_INDEX ))radeon_vce_note_usage (rdev); -radeon_cs_sync_rings (parser); r =radeon_ib_schedule (rdev, &parser->ib,NULL , true);if (r) {DRM_ERROR ("Failed to schedule IB !\n" );@@ -565,7 +572,13 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev,if (r) { goto out; } -radeon_cs_sync_rings (parser); + + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); + goto out; + }radeon_semaphore_sync_fence (parser->ib.semaphore, vm->fence); if ((rdev->family >= CHIP_TAHITI) &&drivers/gpu/drm/radeon/radeon_fence.c +3
-0@@ -541,6 +541,9 @@ int radeon_fence_wait(struct radeon_fence *fence, bool intr)uint64_t seq[RADEON_NUM_RINGS ] = {};long r; +if (unlikely (!to_radeon_fence (&fence->base))) +return fence_wait (&fence->base, intr); + seq[fence->ring] = fence->seq; r =radeon_fence_wait_seq_timeout (fence->rdev, seq, intr,MAX_SCHEDULE_TIMEOUT );if (r <0 ) {drivers/gpu/drm/radeon/radeon_semaphore.c +5
-3@@ -124,27 +124,39 @@ void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore,* * Sync to the fence using this semaphore object */ -void radeon_semaphore_sync_resv (struct radeon_semaphore *sema, -struct reservation_object *resv, -bool shared) +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *sema, +struct reservation_object *resv, +bool shared,bool intr) {struct reservation_object_list *flist;struct fence *f; +struct radeon_fence *fence;unsigned i;
drivers/gpu/drm/radeon/radeon_semaphore.c void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, · line 144
Quoted 10 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +6
-1+int r =0 ;/* always sync to the exclusive fence */ f =reservation_object_get_excl (resv); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence = f ?to_radeon_fence (f) :NULL ; +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else if (f) + r =fence_wait (f, intr);
Error handling? If anything goes wrong while waiting on the exclusive fence we would overwrite the error with the result of the shared fences.
Quoted 5 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +1
-1flist =reservation_object_get_list (resv);if (shared || !flist) -return ; +return r;
drivers/gpu/drm/radeon/radeon_semaphore.c void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, · line 157
Quoted 10 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +5
-1for (i =0 ; i < flist->shared_count; ++i) { f =rcu_dereference_protected (flist->shared[i],reservation_object_held (resv)); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence =to_radeon_fence (f); +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else if (!r) + r =fence_wait (f, intr);
Again break on the first error.
drivers/gpu/drm/radeon/radeon_semaphore.c void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, · line 157
Quoted 1 line
drivers/gpu/drm/radeon/radeon_semaphore.c +0
-0}
Maybe print the error that something went wrong here instead.
Apart from that approach look good to me,
Christian.
Quoted 52 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +1
-0+return r; }/** drivers/gpu/drm/radeon/radeon_vm.c +2
-2@@ -698,7 +698,7 @@ int radeon_vm_update_page_directory(struct radeon_device *rdev,if (ib.length_dw !=0 ) {radeon_asic_vm_pad_ib (rdev, &ib); -radeon_semaphore_sync_resv (ib.semaphore, pd->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib.semaphore, pd->tbo.resv, false, false);radeon_semaphore_sync_fence (ib.semaphore, vm->last_id_use);WARN_ON (ib.length_dw > ndw); r =radeon_ib_schedule (rdev, &ib,NULL , false);@@ -825,7 +825,7 @@ static void radeon_vm_update_ptes(struct radeon_device *rdev,unsigned nptes;uint64_t pte; -radeon_semaphore_sync_resv (ib->semaphore, pt->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib->semaphore, pt->tbo.resv, false, false);if ((addr & ~mask) == (end & ~mask)) nptes = end - addr;drivers/gpu/drm/radeon/rv770_dma.c +1
-1@@ -67,7 +67,7 @@ struct radeon_fence *rv770_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/si_dma.c +1
-1@@ -252,7 +252,7 @@ struct radeon_fence *si_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
Re: [PATCH 6/8] drm/radeon: cope with foreign fences inside the reservation object
Op 17-09-14 om 15:09 schreef Christian König:
Quoted 108 lines
Am 17.09.2014 um 14:35 schrieb Maarten Lankhorst:
Quoted 107 lines
Not the whole world is a radeon! :-)
Signed-off-by Maarten Lankhorst <[email protected]>--- drivers/gpu/drm/radeon/cik.c | 2 +- drivers/gpu/drm/radeon/cik_sdma.c | 2 +- drivers/gpu/drm/radeon/evergreen_dma.c | 2 +- drivers/gpu/drm/radeon/r600.c | 2 +- drivers/gpu/drm/radeon/r600_dma.c | 2 +- drivers/gpu/drm/radeon/radeon.h | 7 ++++--- drivers/gpu/drm/radeon/radeon_cs.c | 27 ++++++++++++++++++++------- drivers/gpu/drm/radeon/radeon_fence.c | 3 +++ drivers/gpu/drm/radeon/radeon_semaphore.c | 24 ++++++++++++++++++------ drivers/gpu/drm/radeon/radeon_vm.c | 4 ++-- drivers/gpu/drm/radeon/rv770_dma.c | 2 +- drivers/gpu/drm/radeon/si_dma.c | 2 +- 12 files changed, 54 insertions(+), 25 deletions(-)drivers/gpu/drm/radeon/cik.c +1
-0@@ -3993,7 +3993,7 @@ struct radeon_fence *cik_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/cik_sdma.c +1
-0@@ -571,7 +571,7 @@ struct radeon_fence *cik_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/evergreen_dma.c +1
-0@@ -133,7 +133,7 @@ struct radeon_fence *evergreen_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/r600.c +1
-0@@ -2912,7 +2912,7 @@ struct radeon_fence *r600_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);radeon_ring_write (ring,PACKET3 (PACKET3_SET_CONFIG_REG ,1 ));drivers/gpu/drm/radeon/r600_dma.c +1
-0@@ -470,7 +470,7 @@ struct radeon_fence *r600_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/radeon.h +4
-3@@ -587,9 +587,10 @@ bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring, struct radeon_semaphore *semaphore);void radeon_semaphore_sync_fence (struct radeon_semaphore *semaphore,struct radeon_fence *fence); -void radeon_semaphore_sync_resv (struct radeon_semaphore *semaphore, -struct reservation_object *resv, -bool shared); +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *semaphore, +struct reservation_object *resv, +bool shared,bool intr);int radeon_semaphore_sync_rings (struct radeon_device *rdev,struct radeon_semaphore *semaphore,int waiting_ring);drivers/gpu/drm/radeon/radeon_cs.c +3
-1@@ -249,20 +249,21 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priorityreturn 0 ; } -static void radeon_cs_sync_rings(struct radeon_cs_parser *p) +static int radeon_cs_sync_rings(struct radeon_cs_parser *p) { -int i; +int i, r =0 ; - for (i = 0; i < p->nrelocs; i++) { +for (i =0 ; !r && i < p->nrelocs; i++) {struct reservation_object *resv;if (!p->relocs[i].robj)
Quoted 8 lines
drivers/gpu/drm/radeon/radeon_cs.c static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority · line 264
Quoted 6 lines
drivers/gpu/drm/radeon/radeon_cs.c +2
-2continue ; resv = p->relocs[i].robj->tbo.resv; -radeon_semaphore_sync_resv (p->ib.semaphore, resv, - p->relocs[i].tv.shared); + r =radeon_semaphore_sync_resv (p->rdev, p->ib.semaphore, resv, + p->relocs[i].tv.shared, true);Break on the first error?
It does, see the !r in the loop.
Quoted 73 lines
Quoted 73 lines
drivers/gpu/drm/radeon/radeon_cs.c +14
-1} +return r; }/* XXX: note that this is called from the legacy UMS CS ioctl as well */ @@ -472,13 +473,19 @@ static int radeon_cs_ib_chunk(struct radeon_device *rdev,return r; } + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); +return r; + } +if (parser->ring ==R600_RING_TYPE_UVD_INDEX )radeon_uvd_note_usage (rdev);else if ((parser->ring ==TN_RING_TYPE_VCE1_INDEX ) || (parser->ring ==TN_RING_TYPE_VCE2_INDEX ))radeon_vce_note_usage (rdev); -radeon_cs_sync_rings (parser); r =radeon_ib_schedule (rdev, &parser->ib,NULL , true);if (r) {DRM_ERROR ("Failed to schedule IB !\n" );@@ -565,7 +572,13 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev,if (r) { goto out; } -radeon_cs_sync_rings (parser); + + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); + goto out; + }radeon_semaphore_sync_fence (parser->ib.semaphore, vm->fence); if ((rdev->family >= CHIP_TAHITI) &&drivers/gpu/drm/radeon/radeon_fence.c +2
-0@@ -541,6 +541,9 @@ int radeon_fence_wait(struct radeon_fence *fence, bool intr)uint64_t seq[RADEON_NUM_RINGS ] = {};long r; + if (unlikely(!to_radeon_fence(&fence->base))) +return fence_wait (&fence->base, intr); + seq[fence->ring] = fence->seq; r =radeon_fence_wait_seq_timeout (fence->rdev, seq, intr,MAX_SCHEDULE_TIMEOUT );if (r <0 ) {drivers/gpu/drm/radeon/radeon_semaphore.c +6
-3@@ -124,27 +124,39 @@ void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore,* * Sync to the fence using this semaphore object */ -void radeon_semaphore_sync_resv (struct radeon_semaphore *sema, -struct reservation_object *resv, -bool shared) +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *sema, +struct reservation_object *resv, +bool shared,bool intr) {struct reservation_object_list *flist;struct fence *f; +struct radeon_fence *fence;unsigned i; +int r =0 ;/* always sync to the exclusive fence */ f =reservation_object_get_excl (resv);
Quoted 8 lines
drivers/gpu/drm/radeon/radeon_semaphore.c void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, · line 144
Quoted 6 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +5
-1-radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence = f ?to_radeon_fence (f) :NULL ; +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else if (f) + r =fence_wait (f, intr);Error handling? If anything goes wrong while waiting on the exclusive fence we would overwrite the error with the result of the shared fences.
The if (!r) fence_wait in the shared fences takes care of that.
Quoted 14 lines
drivers/gpu/drm/radeon/radeon_semaphore.c void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, · line 157
Quoted 13 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +6
-2flist =reservation_object_get_list (resv);if (shared || !flist) -return ; +return r;for (i =0 ; i < flist->shared_count; ++i) { f =rcu_dereference_protected (flist->shared[i],reservation_object_held (resv)); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence =to_radeon_fence (f); +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else if (!r) + r =fence_wait (f, intr);
Quoted 1 line
Again break on the first error.
Not all the functions check for errors, so if the wait fails I wanted to sync as much as possible.
But investigating them it looks like the places that don't check for errors are the .blit, .copy and radeon_vm calls,
Those bo's cannot be exported as dma-buf, so I think the resv call can't fail in that case, and I can return immediately if the calls fails after all. I should be able to drop the intr parameter too for now, because it means the interruptible wait can be the default here.
~Maarten
PATCH v2 6/8 drm/radeon: cope with foreign fences inside the reservation object
In reply to Christian KönigNot the whole world is a radeon! :-)
---
Changes:
- Removed interruptible parameter, only 1 place has a use for it,
and it's the only place that can hit it.
- Fail faster in radeon_semaphore_sync_resv.
- Make the break on error in radeon_cs.c more explicit.
- Upgrade the unlikely() to a WARN_ON_ONCE() in radeon_fence_wait, with a comment explaining why.
drivers/gpu/drm/radeon/cik.c +1 -1
@@ -3993,7 +3993,7 @@ struct radeon_fence *cik_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/cik_sdma.c +1 -1
@@ -571,7 +571,7 @@ struct radeon_fence *cik_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/evergreen_dma.c +1 -1
@@ -133,7 +133,7 @@ struct radeon_fence *evergreen_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/r600.c +1 -1
@@ -2912,7 +2912,7 @@ struct radeon_fence *r600_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);radeon_ring_write (ring,PACKET3 (PACKET3_SET_CONFIG_REG ,1 ));
drivers/gpu/drm/radeon/r600_dma.c +1 -1
@@ -470,7 +470,7 @@ struct radeon_fence *r600_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/radeon.h +4 -3
@@ -589,9 +589,10 @@ bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring, struct radeon_semaphore *semaphore);void radeon_semaphore_sync_fence (struct radeon_semaphore *semaphore,struct radeon_fence *fence); -void radeon_semaphore_sync_resv (struct radeon_semaphore *semaphore, -struct reservation_object *resv, -bool shared); +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *semaphore, +struct reservation_object *resv, +bool shared);int radeon_semaphore_sync_rings (struct radeon_device *rdev,struct radeon_semaphore *semaphore,int waiting_ring);
drivers/gpu/drm/radeon/radeon_cs.c +22 -6
@@ -249,9 +249,9 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priorityreturn 0 ; } -static void radeon_cs_sync_rings (struct radeon_cs_parser *p) +static int radeon_cs_sync_rings (struct radeon_cs_parser *p) { -int i; +int i, r =0 ;for (i =0 ; i < p->nrelocs; i++) {struct reservation_object *resv;
@@ -260,9 +260,13 @@ static void radeon_cs_sync_rings(struct radeon_cs_parser *p)continue ; resv = p->relocs[i].robj->tbo.resv; -radeon_semaphore_sync_resv (p->ib.semaphore, resv, - p->relocs[i].tv.shared); + r =radeon_semaphore_sync_resv (p->rdev, p->ib.semaphore, resv, + p->relocs[i].tv.shared); + +if (r) +break ; } +return r; }/* XXX: note that this is called from the legacy UMS CS ioctl as well */
@@ -472,13 +476,19 @@ static int radeon_cs_ib_chunk(struct radeon_device *rdev,return r; } + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); +return r; + } +if (parser->ring ==R600_RING_TYPE_UVD_INDEX )radeon_uvd_note_usage (rdev);else if ((parser->ring ==TN_RING_TYPE_VCE1_INDEX ) || (parser->ring ==TN_RING_TYPE_VCE2_INDEX ))radeon_vce_note_usage (rdev); -radeon_cs_sync_rings (parser); r =radeon_ib_schedule (rdev, &parser->ib,NULL , true);if (r) {DRM_ERROR ("Failed to schedule IB !\n" );
@@ -565,7 +575,13 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev,if (r) { goto out; } -radeon_cs_sync_rings (parser); + + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); + goto out; + }radeon_semaphore_sync_fence (parser->ib.semaphore, vm->fence); if ((rdev->family >= CHIP_TAHITI) &&
drivers/gpu/drm/radeon/radeon_fence.c +9 -0
@@ -541,6 +541,15 @@ int radeon_fence_wait(struct radeon_fence *fence, bool intr)uint64_t seq[RADEON_NUM_RINGS ] = {};long r; +/* +* This function should not be called on !radeon fences. +* If this is the case, it would mean this function can +* also be called on radeon fences belonging to another card. +* exclusive_lock is not held in that case. +*/ +if (WARN_ON_ONCE (!to_radeon_fence (&fence->base))) +return fence_wait (&fence->base, intr); + seq[fence->ring] = fence->seq; r =radeon_fence_wait_seq_timeout (fence->rdev, seq, intr,MAX_SCHEDULE_TIMEOUT );if (r <0 ) {
drivers/gpu/drm/radeon/radeon_semaphore.c +22 -7
@@ -124,27 +124,42 @@ void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore,* * Sync to the fence using this semaphore object */ -void radeon_semaphore_sync_resv (struct radeon_semaphore *sema, -struct reservation_object *resv, -bool shared) +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *sema, +struct reservation_object *resv, +bool shared) {struct reservation_object_list *flist;struct fence *f; +struct radeon_fence *fence;unsigned i; +int r =0 ;/* always sync to the exclusive fence */ f =reservation_object_get_excl (resv); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence = f ?to_radeon_fence (f) :NULL ; +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else if (f) + r =fence_wait (f, true); flist =reservation_object_get_list (resv); -if (shared || !flist) -return ; +if (shared || !flist || r) +return r;for (i =0 ; i < flist->shared_count; ++i) { f =rcu_dereference_protected (flist->shared[i],reservation_object_held (resv)); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence =to_radeon_fence (f); +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else + r =fence_wait (f, true); + +if (r) +break ; } +return r; }/**
drivers/gpu/drm/radeon/radeon_vm.c +2 -2
@@ -698,7 +698,7 @@ int radeon_vm_update_page_directory(struct radeon_device *rdev,if (ib.length_dw !=0 ) {radeon_asic_vm_pad_ib (rdev, &ib); -radeon_semaphore_sync_resv (ib.semaphore, pd->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib.semaphore, pd->tbo.resv, false);radeon_semaphore_sync_fence (ib.semaphore, vm->last_id_use);WARN_ON (ib.length_dw > ndw); r =radeon_ib_schedule (rdev, &ib,NULL , false);
@@ -825,7 +825,7 @@ static void radeon_vm_update_ptes(struct radeon_device *rdev,unsigned nptes;uint64_t pte; -radeon_semaphore_sync_resv (ib->semaphore, pt->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib->semaphore, pt->tbo.resv, false);if ((addr & ~mask) == (end & ~mask)) nptes = end - addr;
drivers/gpu/drm/radeon/rv770_dma.c +1 -1
@@ -67,7 +67,7 @@ struct radeon_fence *rv770_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
drivers/gpu/drm/radeon/si_dma.c +1 -1
@@ -252,7 +252,7 @@ struct radeon_fence *si_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
Re: [PATCH v2 6/8] drm/radeon: cope with foreign fences inside the reservation object
Op 25-09-14 om 12:39 schreef Maarten Lankhorst:
Quoted 10 lines
Not the whole world is a radeon! :-)
Signed-off-by Maarten Lankhorst <[email protected]>---
Changes:
- Removed interruptible parameter, only 1 place has a use for it,
and it's the only place that can hit it.
- Fail faster inradeon_semaphore_sync_resv.
- Make the break on error inradeon_cs.cmore explicit.
- Upgrade theunlikely()to aWARN_ON_ONCE()inradeon_fence_wait, with a comment explaining why.
Ping, can you review?
Quoted 274 lines
drivers/gpu/drm/radeon/cik.c +1
-1@@ -3993,7 +3993,7 @@ struct radeon_fence *cik_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/cik_sdma.c +1
-1@@ -571,7 +571,7 @@ struct radeon_fence *cik_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/evergreen_dma.c +1
-1@@ -133,7 +133,7 @@ struct radeon_fence *evergreen_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/r600.c +1
-1@@ -2912,7 +2912,7 @@ struct radeon_fence *r600_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);radeon_ring_write (ring,PACKET3 (PACKET3_SET_CONFIG_REG ,1 ));drivers/gpu/drm/radeon/r600_dma.c +1
-1@@ -470,7 +470,7 @@ struct radeon_fence *r600_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/radeon.h +4
-3@@ -589,9 +589,10 @@ bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring, struct radeon_semaphore *semaphore);void radeon_semaphore_sync_fence (struct radeon_semaphore *semaphore,struct radeon_fence *fence); -void radeon_semaphore_sync_resv (struct radeon_semaphore *semaphore, -struct reservation_object *resv, -bool shared); +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *semaphore, +struct reservation_object *resv, +bool shared);int radeon_semaphore_sync_rings (struct radeon_device *rdev,struct radeon_semaphore *semaphore,int waiting_ring);drivers/gpu/drm/radeon/radeon_cs.c +22
-6@@ -249,9 +249,9 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priorityreturn 0 ; } -static void radeon_cs_sync_rings (struct radeon_cs_parser *p) +static int radeon_cs_sync_rings (struct radeon_cs_parser *p) { -int i; +int i, r =0 ;for (i =0 ; i < p->nrelocs; i++) {struct reservation_object *resv;@@ -260,9 +260,13 @@ static void radeon_cs_sync_rings(struct radeon_cs_parser *p)continue ; resv = p->relocs[i].robj->tbo.resv; -radeon_semaphore_sync_resv (p->ib.semaphore, resv, - p->relocs[i].tv.shared); + r =radeon_semaphore_sync_resv (p->rdev, p->ib.semaphore, resv, + p->relocs[i].tv.shared); + +if (r) +break ; } +return r; }/* XXX: note that this is called from the legacy UMS CS ioctl as well */ @@ -472,13 +476,19 @@ static int radeon_cs_ib_chunk(struct radeon_device *rdev,return r; } + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); +return r; + } +if (parser->ring ==R600_RING_TYPE_UVD_INDEX )radeon_uvd_note_usage (rdev);else if ((parser->ring ==TN_RING_TYPE_VCE1_INDEX ) || (parser->ring ==TN_RING_TYPE_VCE2_INDEX ))radeon_vce_note_usage (rdev); -radeon_cs_sync_rings (parser); r =radeon_ib_schedule (rdev, &parser->ib,NULL , true);if (r) {DRM_ERROR ("Failed to schedule IB !\n" );@@ -565,7 +575,13 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev,if (r) { goto out; } -radeon_cs_sync_rings (parser); + + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); + goto out; + }radeon_semaphore_sync_fence (parser->ib.semaphore, vm->fence); if ((rdev->family >= CHIP_TAHITI) &&drivers/gpu/drm/radeon/radeon_fence.c +9
-0@@ -541,6 +541,15 @@ int radeon_fence_wait(struct radeon_fence *fence, bool intr)uint64_t seq[RADEON_NUM_RINGS ] = {};long r; +/* +* This function should not be called on !radeon fences. +* If this is the case, it would mean this function can +* also be called on radeon fences belonging to another card. +* exclusive_lock is not held in that case. +*/ +if (WARN_ON_ONCE (!to_radeon_fence (&fence->base))) +return fence_wait (&fence->base, intr); + seq[fence->ring] = fence->seq; r =radeon_fence_wait_seq_timeout (fence->rdev, seq, intr,MAX_SCHEDULE_TIMEOUT );if (r <0 ) {drivers/gpu/drm/radeon/radeon_semaphore.c +22
-7@@ -124,27 +124,42 @@ void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore,* * Sync to the fence using this semaphore object */ -void radeon_semaphore_sync_resv (struct radeon_semaphore *sema, -struct reservation_object *resv, -bool shared) +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *sema, +struct reservation_object *resv, +bool shared) {struct reservation_object_list *flist;struct fence *f; +struct radeon_fence *fence;unsigned i; +int r =0 ;/* always sync to the exclusive fence */ f =reservation_object_get_excl (resv); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence = f ?to_radeon_fence (f) :NULL ; +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else if (f) + r =fence_wait (f, true); flist =reservation_object_get_list (resv); -if (shared || !flist) -return ; +if (shared || !flist || r) +return r;for (i =0 ; i < flist->shared_count; ++i) { f =rcu_dereference_protected (flist->shared[i],reservation_object_held (resv)); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence =to_radeon_fence (f); +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else + r =fence_wait (f, true); + +if (r) +break ; } +return r; }/** drivers/gpu/drm/radeon/radeon_vm.c +2
-2@@ -698,7 +698,7 @@ int radeon_vm_update_page_directory(struct radeon_device *rdev,if (ib.length_dw !=0 ) {radeon_asic_vm_pad_ib (rdev, &ib); -radeon_semaphore_sync_resv (ib.semaphore, pd->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib.semaphore, pd->tbo.resv, false);radeon_semaphore_sync_fence (ib.semaphore, vm->last_id_use);WARN_ON (ib.length_dw > ndw); r =radeon_ib_schedule (rdev, &ib,NULL , false);@@ -825,7 +825,7 @@ static void radeon_vm_update_ptes(struct radeon_device *rdev,unsigned nptes;uint64_t pte; -radeon_semaphore_sync_resv (ib->semaphore, pt->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib->semaphore, pt->tbo.resv, false);if ((addr & ~mask) == (end & ~mask)) nptes = end - addr;drivers/gpu/drm/radeon/rv770_dma.c +1
-1@@ -67,7 +67,7 @@ struct radeon_fence *rv770_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/si_dma.c +1
-1@@ -252,7 +252,7 @@ struct radeon_fence *si_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
Re: [PATCH v2 6/8] drm/radeon: cope with foreign fences inside the reservation object
Am 01.10.2014 um 10:57 schrieb Maarten Lankhorst:
Quoted 11 lines
Op 25-09-14 om 12:39 schreef Maarten Lankhorst:
Quoted 10 lines
Not the whole world is a radeon! :-)
Signed-off-by Maarten Lankhorst <[email protected]>---
Changes:
- Removed interruptible parameter, only 1 place has a use for it,
and it's the only place that can hit it.
- Fail faster inradeon_semaphore_sync_resv.
- Make the break on error inradeon_cs.cmore explicit.
- Upgrade theunlikely()to aWARN_ON_ONCE()inradeon_fence_wait, with a comment explaining why.
Quoted 1 line
Ping, can you review?
LGTM, patch is Reviewed-by: Christian König <[email protected]>
But we might want to pull it in through Alex drm-next-3.18 (or -3.19) branch to avoid merge conflicts.
Regards,
Christian.
Quoted 274 lines
Quoted 274 lines
drivers/gpu/drm/radeon/cik.c +1
-1@@ -3993,7 +3993,7 @@ struct radeon_fence *cik_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/cik_sdma.c +1
-1@@ -571,7 +571,7 @@ struct radeon_fence *cik_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/evergreen_dma.c +1
-1@@ -133,7 +133,7 @@ struct radeon_fence *evergreen_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/r600.c +1
-1@@ -2912,7 +2912,7 @@ struct radeon_fence *r600_copy_cpdma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);radeon_ring_write (ring,PACKET3 (PACKET3_SET_CONFIG_REG ,1 ));drivers/gpu/drm/radeon/r600_dma.c +1
-1@@ -470,7 +470,7 @@ struct radeon_fence *r600_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/radeon.h +4
-3@@ -589,9 +589,10 @@ bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring, struct radeon_semaphore *semaphore);void radeon_semaphore_sync_fence (struct radeon_semaphore *semaphore,struct radeon_fence *fence); -void radeon_semaphore_sync_resv (struct radeon_semaphore *semaphore, -struct reservation_object *resv, -bool shared); +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *semaphore, +struct reservation_object *resv, +bool shared);int radeon_semaphore_sync_rings (struct radeon_device *rdev,struct radeon_semaphore *semaphore,int waiting_ring);drivers/gpu/drm/radeon/radeon_cs.c +22
-6@@ -249,9 +249,9 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priorityreturn 0 ; } -static void radeon_cs_sync_rings (struct radeon_cs_parser *p) +static int radeon_cs_sync_rings (struct radeon_cs_parser *p) { -int i; +int i, r =0 ;for (i =0 ; i < p->nrelocs; i++) {struct reservation_object *resv;@@ -260,9 +260,13 @@ static void radeon_cs_sync_rings(struct radeon_cs_parser *p)continue ; resv = p->relocs[i].robj->tbo.resv; -radeon_semaphore_sync_resv (p->ib.semaphore, resv, - p->relocs[i].tv.shared); + r =radeon_semaphore_sync_resv (p->rdev, p->ib.semaphore, resv, + p->relocs[i].tv.shared); + +if (r) +break ; } +return r; }/* XXX: note that this is called from the legacy UMS CS ioctl as well */ @@ -472,13 +476,19 @@ static int radeon_cs_ib_chunk(struct radeon_device *rdev,return r; } + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); +return r; + } +if (parser->ring ==R600_RING_TYPE_UVD_INDEX )radeon_uvd_note_usage (rdev);else if ((parser->ring ==TN_RING_TYPE_VCE1_INDEX ) || (parser->ring ==TN_RING_TYPE_VCE2_INDEX ))radeon_vce_note_usage (rdev); -radeon_cs_sync_rings (parser); r =radeon_ib_schedule (rdev, &parser->ib,NULL , true);if (r) {DRM_ERROR ("Failed to schedule IB !\n" );@@ -565,7 +575,13 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev,if (r) { goto out; } -radeon_cs_sync_rings (parser); + + r =radeon_cs_sync_rings (parser); +if (r) { +if (r != -ERESTARTSYS ) +DRM_ERROR ("Failed to sync rings: %i\n" , r); + goto out; + }radeon_semaphore_sync_fence (parser->ib.semaphore, vm->fence); if ((rdev->family >= CHIP_TAHITI) &&drivers/gpu/drm/radeon/radeon_fence.c +9
-0@@ -541,6 +541,15 @@ int radeon_fence_wait(struct radeon_fence *fence, bool intr)uint64_t seq[RADEON_NUM_RINGS ] = {};long r; +/* +* This function should not be called on !radeon fences. +* If this is the case, it would mean this function can +* also be called on radeon fences belonging to another card. +* exclusive_lock is not held in that case. +*/ +if (WARN_ON_ONCE (!to_radeon_fence (&fence->base))) +return fence_wait (&fence->base, intr); + seq[fence->ring] = fence->seq; r =radeon_fence_wait_seq_timeout (fence->rdev, seq, intr,MAX_SCHEDULE_TIMEOUT );if (r <0 ) {drivers/gpu/drm/radeon/radeon_semaphore.c +22
-7@@ -124,27 +124,42 @@ void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore,* * Sync to the fence using this semaphore object */ -void radeon_semaphore_sync_resv (struct radeon_semaphore *sema, -struct reservation_object *resv, -bool shared) +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *sema, +struct reservation_object *resv, +bool shared) {struct reservation_object_list *flist;struct fence *f; +struct radeon_fence *fence;unsigned i; +int r =0 ;/* always sync to the exclusive fence */ f =reservation_object_get_excl (resv); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence = f ?to_radeon_fence (f) :NULL ; +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else if (f) + r =fence_wait (f, true); flist =reservation_object_get_list (resv); -if (shared || !flist) -return ; +if (shared || !flist || r) +return r;for (i =0 ; i < flist->shared_count; ++i) { f =rcu_dereference_protected (flist->shared[i],reservation_object_held (resv)); -radeon_semaphore_sync_fence (sema, (struct radeon_fence *)f); + fence =to_radeon_fence (f); +if (fence && fence->rdev == rdev) +radeon_semaphore_sync_fence (sema, fence); +else + r =fence_wait (f, true); + +if (r) +break ; } +return r; }/** drivers/gpu/drm/radeon/radeon_vm.c +2
-2@@ -698,7 +698,7 @@ int radeon_vm_update_page_directory(struct radeon_device *rdev,if (ib.length_dw !=0 ) {radeon_asic_vm_pad_ib (rdev, &ib); -radeon_semaphore_sync_resv (ib.semaphore, pd->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib.semaphore, pd->tbo.resv, false);radeon_semaphore_sync_fence (ib.semaphore, vm->last_id_use);WARN_ON (ib.length_dw > ndw); r =radeon_ib_schedule (rdev, &ib,NULL , false);@@ -825,7 +825,7 @@ static void radeon_vm_update_ptes(struct radeon_device *rdev,unsigned nptes;uint64_t pte; -radeon_semaphore_sync_resv (ib->semaphore, pt->tbo.resv, false); +radeon_semaphore_sync_resv (rdev, ib->semaphore, pt->tbo.resv, false);if ((addr & ~mask) == (end & ~mask)) nptes = end - addr;drivers/gpu/drm/radeon/rv770_dma.c +1
-1@@ -67,7 +67,7 @@ struct radeon_fence *rv770_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {drivers/gpu/drm/radeon/si_dma.c +1
-1@@ -252,7 +252,7 @@ struct radeon_fence *si_copy_dma(struct radeon_device *rdev,return ERR_PTR (r); } -radeon_semaphore_sync_resv (sem, resv, false); +radeon_semaphore_sync_resv (rdev, sem, resv, false);radeon_semaphore_sync_rings (rdev, sem, ring->idx);for (i =0 ; i < num_loops; i++) {
Re: [PATCH v2 6/8] drm/radeon: cope with foreign fences inside the reservation object
Hey,
Op 01-10-14 om 11:04 schreef Christian König:
Quoted 11 lines
Am 01.10.2014 um 10:57 schrieb Maarten Lankhorst:
Quoted 10 lines
Op 25-09-14 om 12:39 schreef Maarten Lankhorst:
Quoted 9 lines
Not the whole world is a radeon! :-)
Signed-off-by Maarten Lankhorst <[email protected]>---
Changes:
- Removed interruptible parameter, only 1 place has a use for it,
and it's the only place that can hit it.
- Fail faster inradeon_semaphore_sync_resv.
- Make the break on error inradeon_cs.cmore explicit.
- Upgrade theunlikely()to aWARN_ON_ONCE()inradeon_fence_wait, with a comment explaining why.
Quoted 5 lines
Quoted 1 line
Ping, can you review?
LGTM, patch is Reviewed-by: Christian König <[email protected]>
But we might want to pull it in through Alex drm-next-3.18 (or -3.19) branch to avoid merge conflicts.
I can push nouveau through my own tree and once the patches are merged get the radeon changes through agd5f's tree?
~Maarten
Re: [PATCH v2 6/8] drm/radeon: cope with foreign fences inside the reservation object
Quoted 14 lines · Maarten Lankhorst
Hey,
Op 01-10-14 om 11:04 schreef Christian König:
Quoted 11 lines
Am 01.10.2014 um 10:57 schrieb Maarten Lankhorst:
Quoted 10 lines
Op 25-09-14 om 12:39 schreef Maarten Lankhorst:
Quoted 9 lines
Not the whole world is a radeon! :-)
Signed-off-by Maarten Lankhorst <[email protected]>---
Changes:
- Removed interruptible parameter, only 1 place has a use for it,
and it's the only place that can hit it.
- Fail faster inradeon_semaphore_sync_resv.
- Make the break on error inradeon_cs.cmore explicit.
- Upgrade theunlikely()to aWARN_ON_ONCE()inradeon_fence_wait, with a comment explaining why.
Quoted 6 lines
Quoted 5 lines
Quoted 1 line
Ping, can you review?
LGTM, patch is Reviewed-by: Christian König <[email protected]>
But we might want to pull it in through Alex drm-next-3.18 (or -3.19) branch to avoid merge conflicts.
I can push nouveau through my own tree and once the patches are merged get the radeon changes through agd5f's tree?
Either way works for me. Are we trying to get this in for 3.18 or 3.19?
Alex
_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [PATCH 6/8] drm/radeon: cope with foreign fences inside the reservation object
In reply to Maarten LankhorstQuoted 5 lines · Maarten Lankhorst
drivers/gpu/drm/radeon/radeon_semaphore.c +0
-0
drivers/gpu/drm/radeon/radeon_semaphore.c void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, · line 130
Quoted 10 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +4
-3@@ -124,27 +124,39 @@ void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore,* * Sync to the fence using this semaphore object */ -void radeon_semaphore_sync_resv (struct radeon_semaphore *sema, -struct reservation_object *resv, -bool shared) +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *sema, +struct reservation_object *resv, +bool shared,bool intr)
The callers of this function would be more readable if it took flags instead of the shared and intr bools.
Signature · 2 lines
Earthling Michel Dänzer | http://www.amd.com Libre software enthusiast | Mesa and X developer
Re: [PATCH 6/8] drm/radeon: cope with foreign fences inside the reservation object
Hey,
Op 18-09-14 om 05:26 schreef Michel Dänzer:
Quoted 10 lines · Maarten Lankhorst
Quoted 9 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +0
-1@@ -124,27 +124,39 @@ void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore,* * Sync to the fence using this semaphore object */ -void radeon_semaphore_sync_resv(struct radeon_semaphore *sema,
Quoted 8 lines
drivers/gpu/drm/radeon/radeon_semaphore.c void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, · line 130
Quoted 6 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +4
-2-struct reservation_object *resv, - bool shared) +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *sema, +struct reservation_object *resv, +bool shared,bool intr)The callers of this function would be more readable if it took flags instead of the shared and intr bools.
This does not match the rest of the TTM design. Things like ttm_bo_reserve take separate bools, not flags.
~Maarten
Re: [PATCH 6/8] drm/radeon: cope with foreign fences inside the reservation object
Quoted 19 lines · Maarten Lankhorst
Op 18-09-14 om 05:26 schreef Michel Dänzer:
Quoted 10 lines · Maarten Lankhorst
Quoted 9 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +0
-1@@ -124,27 +124,39 @@ void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore,* * Sync to the fence using this semaphore object */ -void radeon_semaphore_sync_resv(struct radeon_semaphore *sema,Quoted 8 lines
drivers/gpu/drm/radeon/radeon_semaphore.c void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, · line 130
Quoted 6 lines
drivers/gpu/drm/radeon/radeon_semaphore.c +4
-2-struct reservation_object *resv, - bool shared) +int radeon_semaphore_sync_resv (struct radeon_device *rdev, +struct radeon_semaphore *sema, +struct reservation_object *resv, +bool shared,bool intr)The callers of this function would be more readable if it took flags instead of the shared and intr bools.
Quoted 2 lines
This does not match the rest of the TTM design. Things like
ttm_bo_reservetake separate bools, not flags.
So? :)
Signature · 2 lines
Earthling Michel Dänzer | http://www.amd.com Libre software enthusiast | Mesa and X developer
PATCH 7/8 drm/radeon: export reservation_object from dmabuf to ttm
In reply to Maarten LankhorstAdds an extra argument to ttm_bo_create, which is only used in radeon_prime.c.
--- drivers/gpu/drm/radeon/cik.c | 4 ++-- drivers/gpu/drm/radeon/evergreen.c | 6 +++--- drivers/gpu/drm/radeon/r600.c | 4 ++-- drivers/gpu/drm/radeon/radeon_benchmark.c | 4 ++-- drivers/gpu/drm/radeon/radeon_device.c | 2 +- drivers/gpu/drm/radeon/radeon_gart.c | 2 +- drivers/gpu/drm/radeon/radeon_gem.c | 2 +- drivers/gpu/drm/radeon/radeon_object.c | 8 +++++--- drivers/gpu/drm/radeon/radeon_object.h | 1 + drivers/gpu/drm/radeon/radeon_prime.c | 5 ++++- drivers/gpu/drm/radeon/radeon_ring.c | 2 +- drivers/gpu/drm/radeon/radeon_sa.c | 2 +- drivers/gpu/drm/radeon/radeon_test.c | 5 +++-- drivers/gpu/drm/radeon/radeon_ttm.c | 2 +- drivers/gpu/drm/radeon/radeon_uvd.c | 3 ++- drivers/gpu/drm/radeon/radeon_vce.c | 3 ++- drivers/gpu/drm/radeon/radeon_vm.c | 5 +++-- 17 files changed, 35 insertions(+), 25 deletions(-)
drivers/gpu/drm/radeon/cik.c +2 -2
@@ -4690,7 +4690,7 @@ static int cik_mec_init(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->mec.num_mec *rdev->mec.num_pipe *MEC_HPD_SIZE *2 ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &rdev->mec.hpd_eop_obj);if (r) {dev_warn (rdev->dev,"(%d) create HDP EOP bo failed\n" , r);
@@ -4861,7 +4861,7 @@ static int cik_cp_compute_resume(struct radeon_device *rdev)sizeof (struct bonaire_mqd ),PAGE_SIZE , true,RADEON_GEM_DOMAIN_GTT ,0 ,NULL , - &rdev->ring[idx].mqd_obj); + NULL, &rdev->ring[idx].mqd_obj);if (r) {dev_warn (rdev->dev,"(%d) create MQD bo failed\n" , r);return r;
drivers/gpu/drm/radeon/evergreen.c +3 -3
@@ -4023,7 +4023,7 @@ int sumo_rlc_init(struct radeon_device *rdev)if (rdev->rlc.save_restore_obj ==NULL ) { r =radeon_bo_create (rdev, dws *4 ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.save_restore_obj); +NULL , &rdev->rlc.save_restore_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC sr bo failed\n" , r);return r;
@@ -4102,7 +4102,7 @@ int sumo_rlc_init(struct radeon_device *rdev)if (rdev->rlc.clear_state_obj ==NULL ) { r =radeon_bo_create (rdev, dws *4 ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.clear_state_obj); +NULL , &rdev->rlc.clear_state_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC c bo failed\n" , r);sumo_rlc_fini (rdev);
@@ -4179,7 +4179,7 @@ int sumo_rlc_init(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->rlc.cp_table_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.cp_table_obj); +NULL , &rdev->rlc.cp_table_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC cp table bo failed\n" , r);sumo_rlc_fini (rdev);
drivers/gpu/drm/radeon/r600.c +2 -2
@@ -1430,7 +1430,7 @@ int r600_vram_scratch_init(struct radeon_device *rdev)if (rdev->vram_scratch.robj ==NULL ) { r =radeon_bo_create (rdev,RADEON_GPU_PAGE_SIZE ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &rdev->vram_scratch.robj); +0 ,NULL ,NULL , &rdev->vram_scratch.robj);if (r) {return r; }
@@ -3368,7 +3368,7 @@ int r600_ih_ring_alloc(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->ih.ring_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_GTT ,0 , -NULL , &rdev->ih.ring_obj); +NULL ,NULL , &rdev->ih.ring_obj);if (r) {DRM_ERROR ("radeon: failed to create ih ring buffer (%d).\n" , r);return r;
drivers/gpu/drm/radeon/radeon_benchmark.c +2 -2
@@ -93,7 +93,7 @@ static void radeon_benchmark_move(struct radeon_device *rdev, unsigned size,int time; n =RADEON_BENCHMARK_ITERATIONS ; - r =radeon_bo_create (rdev, size,PAGE_SIZE , true, sdomain,0 ,NULL , &sobj); + r =radeon_bo_create (rdev, size,PAGE_SIZE , true, sdomain,0 ,NULL ,NULL , &sobj);if (r) { goto out_cleanup; }
@@ -105,7 +105,7 @@ static void radeon_benchmark_move(struct radeon_device *rdev, unsigned size,if (r) { goto out_cleanup; } - r =radeon_bo_create (rdev, size,PAGE_SIZE , true, ddomain,0 ,NULL , &dobj); + r =radeon_bo_create (rdev, size,PAGE_SIZE , true, ddomain,0 ,NULL ,NULL , &dobj);if (r) { goto out_cleanup; }
drivers/gpu/drm/radeon/radeon_device.c +1 -1
@@ -430,7 +430,7 @@ int radeon_wb_init(struct radeon_device *rdev)if (rdev->wb.wb_obj ==NULL ) { r =radeon_bo_create (rdev,RADEON_GPU_PAGE_SIZE ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &rdev->wb.wb_obj);if (r) {dev_warn (rdev->dev,"(%d) create WB bo failed\n" , r);
drivers/gpu/drm/radeon/radeon_gart.c +1 -1
@@ -128,7 +128,7 @@ int radeon_gart_table_vram_alloc(struct radeon_device *rdev)if (rdev->gart.robj ==NULL ) { r =radeon_bo_create (rdev, rdev->gart.table_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &rdev->gart.robj); +0 ,NULL ,NULL , &rdev->gart.robj);if (r) {return r; }
drivers/gpu/drm/radeon/radeon_gem.c +1 -1
@@ -67,7 +67,7 @@ int radeon_gem_object_create(struct radeon_device *rdev, unsigned long size, retry: r =radeon_bo_create (rdev, size, alignment, kernel, initial_domain, - flags,NULL , &robj); + flags,NULL ,NULL , &robj);if (r) {if (r != -ERESTARTSYS ) {if (initial_domain ==RADEON_GEM_DOMAIN_VRAM ) {
drivers/gpu/drm/radeon/radeon_object.c +5 -3
@@ -167,8 +167,10 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) }int radeon_bo_create (struct radeon_device *rdev, -unsigned long size,int byte_align,bool kernel,u32 domain, -u32 flags,struct sg_table *sg,struct radeon_bo **bo_ptr) +unsigned long size,int byte_align,bool kernel, +u32 domain,u32 flags,struct sg_table *sg, +struct reservation_object *robj, +struct radeon_bo **bo_ptr) {struct radeon_bo *bo;enum ttm_bo_type type;
@@ -216,7 +218,7 @@ int radeon_bo_create(struct radeon_device *rdev,down_read (&rdev->pm.mclk_lock); r =ttm_bo_init (&rdev->mman.bdev, &bo->tbo, size, type, &bo->placement, page_align, !kernel,NULL , - acc_size, sg,NULL , &radeon_ttm_bo_destroy); + acc_size, sg, robj, &radeon_ttm_bo_destroy);up_read (&rdev->pm.mclk_lock);if (unlikely (r !=0 )) {return r;
drivers/gpu/drm/radeon/radeon_object.h +1 -0
@@ -126,6 +126,7 @@ extern int radeon_bo_create(struct radeon_device *rdev, unsigned long size, int byte_align, bool kernel, u32 domain, u32 flags, struct sg_table *sg, + struct reservation_object *robj, struct radeon_bo **bo_ptr);extern int radeon_bo_kmap (struct radeon_bo *bo,void **ptr);extern void radeon_bo_kunmap (struct radeon_bo *bo);
drivers/gpu/drm/radeon/radeon_prime.c +4 -1
@@ -61,12 +61,15 @@ struct drm_gem_object *radeon_gem_prime_import_sg_table(struct drm_device *dev,struct dma_buf_attachment *attach, struct sg_table *sg) { + struct reservation_object *robj = attach->dmabuf->resv;struct radeon_device *rdev = dev->dev_private;struct radeon_bo *bo;int ret; +ww_mutex_lock (&robj->lock,NULL ); ret =radeon_bo_create (rdev, attach->dmabuf->size,PAGE_SIZE , false, -RADEON_GEM_DOMAIN_GTT ,0 , sg, &bo); +RADEON_GEM_DOMAIN_GTT ,0 , sg, robj, &bo); +ww_mutex_unlock (&robj->lock);if (ret)return ERR_PTR (ret);
drivers/gpu/drm/radeon/radeon_ring.c +1 -1
@@ -383,7 +383,7 @@ int radeon_ring_init(struct radeon_device *rdev, struct radeon_ring *ring, unsig/* Allocate ring buffer */ if (ring->ring_obj ==NULL ) { r =radeon_bo_create (rdev, ring->ring_size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &ring->ring_obj);if (r) {dev_err (rdev->dev,"(%d) ring create failed\n" , r);
drivers/gpu/drm/radeon/radeon_sa.c +1 -1
@@ -65,7 +65,7 @@ int radeon_sa_bo_manager_init(struct radeon_device *rdev, } r =radeon_bo_create (rdev, size, align, true, - domain, flags,NULL , &sa_manager->bo); + domain, flags,NULL ,NULL , &sa_manager->bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate bo for manager\n" , r);return r;
drivers/gpu/drm/radeon/radeon_test.c +3 -2
@@ -67,7 +67,7 @@ static void radeon_do_test_moves(struct radeon_device *rdev, int flag) } r =radeon_bo_create (rdev, size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &vram_obj); +0 ,NULL ,NULL , &vram_obj);if (r) {DRM_ERROR ("Failed to create VRAM object\n" ); goto out_cleanup;
@@ -87,7 +87,8 @@ static void radeon_do_test_moves(struct radeon_device *rdev, int flag)struct radeon_fence *fence =NULL ; r =radeon_bo_create (rdev, size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , gtt_obj + i); +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , + gtt_obj + i);if (r) {DRM_ERROR ("Failed to create GTT object %d\n" , i); goto out_lclean;
drivers/gpu/drm/radeon/radeon_ttm.c +1 -1
@@ -858,7 +858,7 @@ int radeon_ttm_init(struct radeon_device *rdev)radeon_ttm_set_active_vram_size (rdev, rdev->mc.visible_vram_size); r =radeon_bo_create (rdev,256 *1024 ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 , +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL ,NULL , &rdev->stollen_vga_memory);if (r) {return r;
drivers/gpu/drm/radeon/radeon_uvd.c +2 -1
@@ -141,7 +141,8 @@ int radeon_uvd_init(struct radeon_device *rdev)RADEON_UVD_STACK_SIZE +RADEON_UVD_HEAP_SIZE +RADEON_GPU_PAGE_SIZE ; r =radeon_bo_create (rdev, bo_size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &rdev->uvd.vcpu_bo); +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , +NULL , &rdev->uvd.vcpu_bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate UVD bo\n" , r);return r;
drivers/gpu/drm/radeon/radeon_vce.c +2 -1
@@ -126,7 +126,8 @@ int radeon_vce_init(struct radeon_device *rdev) size =RADEON_GPU_PAGE_ALIGN (rdev->vce_fw->size) +RADEON_VCE_STACK_SIZE +RADEON_VCE_HEAP_SIZE ; r =radeon_bo_create (rdev, size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &rdev->vce.vcpu_bo); +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL ,NULL , + &rdev->vce.vcpu_bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate VCE bo\n" , r);return r;
drivers/gpu/drm/radeon/radeon_vm.c +3 -2
@@ -548,7 +548,8 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, r =radeon_bo_create (rdev,RADEON_VM_PTE_COUNT *8 ,RADEON_GPU_PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &pt); +RADEON_GEM_DOMAIN_VRAM ,0 , +NULL ,NULL , &pt);if (r)return r;
@@ -1127,7 +1128,7 @@ int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) r =radeon_bo_create (rdev, pd_size, align, true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &vm->page_directory); +NULL , &vm->page_directory);if (r)return r;
Signature · 1 line
2.0.4
Re: [PATCH 7/8] drm/radeon: export reservation_object from dmabuf to ttm
Am 17.09.2014 um 14:35 schrieb Maarten Lankhorst:
Quoted 164 lines
Adds an extra argument to
ttm_bo_create, which is only used inradeon_prime.c.Signed-off-by Maarten Lankhorst <[email protected]>--- drivers/gpu/drm/radeon/cik.c | 4 ++-- drivers/gpu/drm/radeon/evergreen.c | 6 +++--- drivers/gpu/drm/radeon/r600.c | 4 ++-- drivers/gpu/drm/radeon/radeon_benchmark.c | 4 ++-- drivers/gpu/drm/radeon/radeon_device.c | 2 +- drivers/gpu/drm/radeon/radeon_gart.c | 2 +- drivers/gpu/drm/radeon/radeon_gem.c | 2 +- drivers/gpu/drm/radeon/radeon_object.c | 8 +++++--- drivers/gpu/drm/radeon/radeon_object.h | 1 + drivers/gpu/drm/radeon/radeon_prime.c | 5 ++++- drivers/gpu/drm/radeon/radeon_ring.c | 2 +- drivers/gpu/drm/radeon/radeon_sa.c | 2 +- drivers/gpu/drm/radeon/radeon_test.c | 5 +++-- drivers/gpu/drm/radeon/radeon_ttm.c | 2 +- drivers/gpu/drm/radeon/radeon_uvd.c | 3 ++- drivers/gpu/drm/radeon/radeon_vce.c | 3 ++- drivers/gpu/drm/radeon/radeon_vm.c | 5 +++-- 17 files changed, 35 insertions(+), 25 deletions(-)drivers/gpu/drm/radeon/cik.c +2
-2@@ -4690,7 +4690,7 @@ static int cik_mec_init(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->mec.num_mec *rdev->mec.num_pipe *MEC_HPD_SIZE *2 ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &rdev->mec.hpd_eop_obj);if (r) {dev_warn (rdev->dev,"(%d) create HDP EOP bo failed\n" , r);@@ -4861,7 +4861,7 @@ static int cik_cp_compute_resume(struct radeon_device *rdev)sizeof (struct bonaire_mqd ),PAGE_SIZE , true,RADEON_GEM_DOMAIN_GTT ,0 ,NULL , - &rdev->ring[idx].mqd_obj); + NULL, &rdev->ring[idx].mqd_obj);if (r) {dev_warn (rdev->dev,"(%d) create MQD bo failed\n" , r);return r;drivers/gpu/drm/radeon/evergreen.c +3
-3@@ -4023,7 +4023,7 @@ int sumo_rlc_init(struct radeon_device *rdev)if (rdev->rlc.save_restore_obj ==NULL ) { r =radeon_bo_create (rdev, dws *4 ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.save_restore_obj); +NULL , &rdev->rlc.save_restore_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC sr bo failed\n" , r);return r;@@ -4102,7 +4102,7 @@ int sumo_rlc_init(struct radeon_device *rdev)if (rdev->rlc.clear_state_obj ==NULL ) { r =radeon_bo_create (rdev, dws *4 ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.clear_state_obj); +NULL , &rdev->rlc.clear_state_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC c bo failed\n" , r);sumo_rlc_fini (rdev);@@ -4179,7 +4179,7 @@ int sumo_rlc_init(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->rlc.cp_table_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.cp_table_obj); +NULL , &rdev->rlc.cp_table_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC cp table bo failed\n" , r);sumo_rlc_fini (rdev);drivers/gpu/drm/radeon/r600.c +2
-2@@ -1430,7 +1430,7 @@ int r600_vram_scratch_init(struct radeon_device *rdev)if (rdev->vram_scratch.robj ==NULL ) { r =radeon_bo_create (rdev,RADEON_GPU_PAGE_SIZE ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &rdev->vram_scratch.robj); +0 ,NULL ,NULL , &rdev->vram_scratch.robj);if (r) {return r; }@@ -3368,7 +3368,7 @@ int r600_ih_ring_alloc(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->ih.ring_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_GTT ,0 , -NULL , &rdev->ih.ring_obj); +NULL ,NULL , &rdev->ih.ring_obj);if (r) {DRM_ERROR ("radeon: failed to create ih ring buffer (%d).\n" , r);return r;drivers/gpu/drm/radeon/radeon_benchmark.c +2
-2@@ -93,7 +93,7 @@ static void radeon_benchmark_move(struct radeon_device *rdev, unsigned size,int time; n =RADEON_BENCHMARK_ITERATIONS ; - r =radeon_bo_create (rdev, size,PAGE_SIZE , true, sdomain,0 ,NULL , &sobj); + r =radeon_bo_create (rdev, size,PAGE_SIZE , true, sdomain,0 ,NULL ,NULL , &sobj);if (r) { goto out_cleanup; }@@ -105,7 +105,7 @@ static void radeon_benchmark_move(struct radeon_device *rdev, unsigned size,if (r) { goto out_cleanup; } - r =radeon_bo_create (rdev, size,PAGE_SIZE , true, ddomain,0 ,NULL , &dobj); + r =radeon_bo_create (rdev, size,PAGE_SIZE , true, ddomain,0 ,NULL ,NULL , &dobj);if (r) { goto out_cleanup; }drivers/gpu/drm/radeon/radeon_device.c +1
-1@@ -430,7 +430,7 @@ int radeon_wb_init(struct radeon_device *rdev)if (rdev->wb.wb_obj ==NULL ) { r =radeon_bo_create (rdev,RADEON_GPU_PAGE_SIZE ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &rdev->wb.wb_obj);if (r) {dev_warn (rdev->dev,"(%d) create WB bo failed\n" , r);drivers/gpu/drm/radeon/radeon_gart.c +1
-1@@ -128,7 +128,7 @@ int radeon_gart_table_vram_alloc(struct radeon_device *rdev)if (rdev->gart.robj ==NULL ) { r =radeon_bo_create (rdev, rdev->gart.table_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &rdev->gart.robj); +0 ,NULL ,NULL , &rdev->gart.robj);if (r) {return r; }drivers/gpu/drm/radeon/radeon_gem.c +1
-1@@ -67,7 +67,7 @@ int radeon_gem_object_create(struct radeon_device *rdev, unsigned long size, retry: r =radeon_bo_create (rdev, size, alignment, kernel, initial_domain, - flags,NULL , &robj); + flags,NULL ,NULL , &robj);if (r) {if (r != -ERESTARTSYS ) {if (initial_domain ==RADEON_GEM_DOMAIN_VRAM ) {drivers/gpu/drm/radeon/radeon_object.c +0
-0
drivers/gpu/drm/radeon/radeon_object.c void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) · line 173
Quoted 9 lines
drivers/gpu/drm/radeon/radeon_object.c +4
-2@@ -167,8 +167,10 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) }int radeon_bo_create (struct radeon_device *rdev, -unsigned long size,int byte_align,bool kernel,u32 domain, -u32 flags,struct sg_table *sg,struct radeon_bo **bo_ptr) +unsigned long size,int byte_align,bool kernel, +u32 domain,u32 flags,struct sg_table *sg, +struct reservation_object *robj, +struct radeon_bo **bo_ptr)
We should sooner or later split the function into multiple ones, we try to handle to many things with it.
Anyway, for now the patch looks good. Just don't name the new parameter "robj", cause that sounds to much like short for "radeon object". "resv" for example sounds fine to me.
Christian.
Quoted 158 lines
drivers/gpu/drm/radeon/radeon_object.c +1
-1{struct radeon_bo *bo;enum ttm_bo_type type;@@ -216,7 +218,7 @@ int radeon_bo_create(struct radeon_device *rdev,down_read (&rdev->pm.mclk_lock); r =ttm_bo_init (&rdev->mman.bdev, &bo->tbo, size, type, &bo->placement, page_align, !kernel,NULL , - acc_size, sg,NULL , &radeon_ttm_bo_destroy); + acc_size, sg, robj, &radeon_ttm_bo_destroy);up_read (&rdev->pm.mclk_lock);if (unlikely (r !=0 )) {return r;drivers/gpu/drm/radeon/radeon_object.h +1
-0@@ -126,6 +126,7 @@ extern int radeon_bo_create(struct radeon_device *rdev, unsigned long size, int byte_align, bool kernel, u32 domain, u32 flags, struct sg_table *sg, + struct reservation_object *robj, struct radeon_bo **bo_ptr);extern int radeon_bo_kmap (struct radeon_bo *bo,void **ptr);extern void radeon_bo_kunmap (struct radeon_bo *bo);drivers/gpu/drm/radeon/radeon_prime.c +4
-1@@ -61,12 +61,15 @@ struct drm_gem_object *radeon_gem_prime_import_sg_table(struct drm_device *dev,struct dma_buf_attachment *attach, struct sg_table *sg) { + struct reservation_object *robj = attach->dmabuf->resv;struct radeon_device *rdev = dev->dev_private;struct radeon_bo *bo;int ret; +ww_mutex_lock (&robj->lock,NULL ); ret =radeon_bo_create (rdev, attach->dmabuf->size,PAGE_SIZE , false, -RADEON_GEM_DOMAIN_GTT ,0 , sg, &bo); +RADEON_GEM_DOMAIN_GTT ,0 , sg, robj, &bo); +ww_mutex_unlock (&robj->lock);if (ret)return ERR_PTR (ret);drivers/gpu/drm/radeon/radeon_ring.c +1
-1@@ -383,7 +383,7 @@ int radeon_ring_init(struct radeon_device *rdev, struct radeon_ring *ring, unsig/* Allocate ring buffer */ if (ring->ring_obj ==NULL ) { r =radeon_bo_create (rdev, ring->ring_size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &ring->ring_obj);if (r) {dev_err (rdev->dev,"(%d) ring create failed\n" , r);drivers/gpu/drm/radeon/radeon_sa.c +1
-1@@ -65,7 +65,7 @@ int radeon_sa_bo_manager_init(struct radeon_device *rdev, } r =radeon_bo_create (rdev, size, align, true, - domain, flags,NULL , &sa_manager->bo); + domain, flags,NULL ,NULL , &sa_manager->bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate bo for manager\n" , r);return r;drivers/gpu/drm/radeon/radeon_test.c +3
-2@@ -67,7 +67,7 @@ static void radeon_do_test_moves(struct radeon_device *rdev, int flag) } r =radeon_bo_create (rdev, size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &vram_obj); +0 ,NULL ,NULL , &vram_obj);if (r) {DRM_ERROR ("Failed to create VRAM object\n" ); goto out_cleanup;@@ -87,7 +87,8 @@ static void radeon_do_test_moves(struct radeon_device *rdev, int flag)struct radeon_fence *fence =NULL ; r =radeon_bo_create (rdev, size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , gtt_obj + i); +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , + gtt_obj + i);if (r) {DRM_ERROR ("Failed to create GTT object %d\n" , i); goto out_lclean;drivers/gpu/drm/radeon/radeon_ttm.c +1
-1@@ -858,7 +858,7 @@ int radeon_ttm_init(struct radeon_device *rdev)radeon_ttm_set_active_vram_size (rdev, rdev->mc.visible_vram_size); r =radeon_bo_create (rdev,256 *1024 ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 , +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL ,NULL , &rdev->stollen_vga_memory);if (r) {return r;drivers/gpu/drm/radeon/radeon_uvd.c +2
-1@@ -141,7 +141,8 @@ int radeon_uvd_init(struct radeon_device *rdev)RADEON_UVD_STACK_SIZE +RADEON_UVD_HEAP_SIZE +RADEON_GPU_PAGE_SIZE ; r =radeon_bo_create (rdev, bo_size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &rdev->uvd.vcpu_bo); +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , +NULL , &rdev->uvd.vcpu_bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate UVD bo\n" , r);return r;drivers/gpu/drm/radeon/radeon_vce.c +2
-1@@ -126,7 +126,8 @@ int radeon_vce_init(struct radeon_device *rdev) size =RADEON_GPU_PAGE_ALIGN (rdev->vce_fw->size) +RADEON_VCE_STACK_SIZE +RADEON_VCE_HEAP_SIZE ; r =radeon_bo_create (rdev, size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &rdev->vce.vcpu_bo); +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL ,NULL , + &rdev->vce.vcpu_bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate VCE bo\n" , r);return r;drivers/gpu/drm/radeon/radeon_vm.c +3
-2@@ -548,7 +548,8 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, r =radeon_bo_create (rdev,RADEON_VM_PTE_COUNT *8 ,RADEON_GPU_PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &pt); +RADEON_GEM_DOMAIN_VRAM ,0 , +NULL ,NULL , &pt);if (r)return r;@@ -1127,7 +1128,7 @@ int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) r =radeon_bo_create (rdev, pd_size, align, true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &vm->page_directory); +NULL , &vm->page_directory);if (r)return r;
Re: [PATCH 7/8] drm/radeon: export reservation_object from dmabuf to ttm
In reply to Maarten LankhorstQuoted 1 line · Maarten Lankhorst
Adds an extra argument to
ttm_bo_create, which is only used inradeon_prime.c.
Typo: 'ttm_bo_create' -> 'radeon_bo_create'
Signature · 2 lines
Earthling Michel Dänzer | http://www.amd.com Libre software enthusiast | Mesa and X developer
PATCH v2 7/8 drm/radeon: export reservation_object from dmabuf to ttm
In reply to Maarten Lankhorstdrm/radeon: export reservation_object from dmabuf to ttm
Adds an extra argument to radeon_bo_create, which is only used in radeon_prime.c.
---
v2:
robj -> resv
fix typo in commit description
drivers/gpu/drm/radeon/cik.c +2 -2
@@ -4690,7 +4690,7 @@ static int cik_mec_init(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->mec.num_mec *rdev->mec.num_pipe *MEC_HPD_SIZE *2 ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &rdev->mec.hpd_eop_obj);if (r) {dev_warn (rdev->dev,"(%d) create HDP EOP bo failed\n" , r);
@@ -4861,7 +4861,7 @@ static int cik_cp_compute_resume(struct radeon_device *rdev)sizeof (struct bonaire_mqd ),PAGE_SIZE , true,RADEON_GEM_DOMAIN_GTT ,0 ,NULL , - &rdev->ring[idx].mqd_obj); + NULL, &rdev->ring[idx].mqd_obj);if (r) {dev_warn (rdev->dev,"(%d) create MQD bo failed\n" , r);return r;
drivers/gpu/drm/radeon/evergreen.c +3 -3
@@ -4023,7 +4023,7 @@ int sumo_rlc_init(struct radeon_device *rdev)if (rdev->rlc.save_restore_obj ==NULL ) { r =radeon_bo_create (rdev, dws *4 ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.save_restore_obj); +NULL , &rdev->rlc.save_restore_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC sr bo failed\n" , r);return r;
@@ -4102,7 +4102,7 @@ int sumo_rlc_init(struct radeon_device *rdev)if (rdev->rlc.clear_state_obj ==NULL ) { r =radeon_bo_create (rdev, dws *4 ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.clear_state_obj); +NULL , &rdev->rlc.clear_state_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC c bo failed\n" , r);sumo_rlc_fini (rdev);
@@ -4179,7 +4179,7 @@ int sumo_rlc_init(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->rlc.cp_table_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.cp_table_obj); +NULL , &rdev->rlc.cp_table_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC cp table bo failed\n" , r);sumo_rlc_fini (rdev);
drivers/gpu/drm/radeon/r600.c +2 -2
@@ -1430,7 +1430,7 @@ int r600_vram_scratch_init(struct radeon_device *rdev)if (rdev->vram_scratch.robj ==NULL ) { r =radeon_bo_create (rdev,RADEON_GPU_PAGE_SIZE ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &rdev->vram_scratch.robj); +0 ,NULL ,NULL , &rdev->vram_scratch.robj);if (r) {return r; }
@@ -3368,7 +3368,7 @@ int r600_ih_ring_alloc(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->ih.ring_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_GTT ,0 , -NULL , &rdev->ih.ring_obj); +NULL ,NULL , &rdev->ih.ring_obj);if (r) {DRM_ERROR ("radeon: failed to create ih ring buffer (%d).\n" , r);return r;
drivers/gpu/drm/radeon/radeon_benchmark.c +2 -2
@@ -93,7 +93,7 @@ static void radeon_benchmark_move(struct radeon_device *rdev, unsigned size,int time; n =RADEON_BENCHMARK_ITERATIONS ; - r =radeon_bo_create (rdev, size,PAGE_SIZE , true, sdomain,0 ,NULL , &sobj); + r =radeon_bo_create (rdev, size,PAGE_SIZE , true, sdomain,0 ,NULL ,NULL , &sobj);if (r) { goto out_cleanup; }
@@ -105,7 +105,7 @@ static void radeon_benchmark_move(struct radeon_device *rdev, unsigned size,if (r) { goto out_cleanup; } - r =radeon_bo_create (rdev, size,PAGE_SIZE , true, ddomain,0 ,NULL , &dobj); + r =radeon_bo_create (rdev, size,PAGE_SIZE , true, ddomain,0 ,NULL ,NULL , &dobj);if (r) { goto out_cleanup; }
drivers/gpu/drm/radeon/radeon_device.c +1 -1
@@ -430,7 +430,7 @@ int radeon_wb_init(struct radeon_device *rdev)if (rdev->wb.wb_obj ==NULL ) { r =radeon_bo_create (rdev,RADEON_GPU_PAGE_SIZE ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &rdev->wb.wb_obj);if (r) {dev_warn (rdev->dev,"(%d) create WB bo failed\n" , r);
drivers/gpu/drm/radeon/radeon_gart.c +1 -1
@@ -128,7 +128,7 @@ int radeon_gart_table_vram_alloc(struct radeon_device *rdev)if (rdev->gart.robj ==NULL ) { r =radeon_bo_create (rdev, rdev->gart.table_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &rdev->gart.robj); +0 ,NULL ,NULL , &rdev->gart.robj);if (r) {return r; }
drivers/gpu/drm/radeon/radeon_gem.c +1 -1
@@ -67,7 +67,7 @@ int radeon_gem_object_create(struct radeon_device *rdev, unsigned long size, retry: r =radeon_bo_create (rdev, size, alignment, kernel, initial_domain, - flags,NULL , &robj); + flags,NULL ,NULL , &robj);if (r) {if (r != -ERESTARTSYS ) {if (initial_domain ==RADEON_GEM_DOMAIN_VRAM ) {
drivers/gpu/drm/radeon/radeon_object.c +5 -3
@@ -167,8 +167,10 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) }int radeon_bo_create (struct radeon_device *rdev, -unsigned long size,int byte_align,bool kernel,u32 domain, -u32 flags,struct sg_table *sg,struct radeon_bo **bo_ptr) +unsigned long size,int byte_align,bool kernel, +u32 domain,u32 flags,struct sg_table *sg, +struct reservation_object *resv, +struct radeon_bo **bo_ptr) {struct radeon_bo *bo;enum ttm_bo_type type;
@@ -216,7 +218,7 @@ int radeon_bo_create(struct radeon_device *rdev,down_read (&rdev->pm.mclk_lock); r =ttm_bo_init (&rdev->mman.bdev, &bo->tbo, size, type, &bo->placement, page_align, !kernel,NULL , - acc_size, sg,NULL , &radeon_ttm_bo_destroy); + acc_size, sg, resv, &radeon_ttm_bo_destroy);up_read (&rdev->pm.mclk_lock);if (unlikely (r !=0 )) {return r;
drivers/gpu/drm/radeon/radeon_object.h +1 -0
@@ -126,6 +126,7 @@ extern int radeon_bo_create(struct radeon_device *rdev, unsigned long size, int byte_align, bool kernel, u32 domain, u32 flags, struct sg_table *sg, + struct reservation_object *resv, struct radeon_bo **bo_ptr);extern int radeon_bo_kmap (struct radeon_bo *bo,void **ptr);extern void radeon_bo_kunmap (struct radeon_bo *bo);
drivers/gpu/drm/radeon/radeon_prime.c +4 -1
@@ -61,12 +61,15 @@ struct drm_gem_object *radeon_gem_prime_import_sg_table(struct drm_device *dev,struct dma_buf_attachment *attach, struct sg_table *sg) { + struct reservation_object *resv = attach->dmabuf->resv;struct radeon_device *rdev = dev->dev_private;struct radeon_bo *bo;int ret; +ww_mutex_lock (&resv->lock,NULL ); ret =radeon_bo_create (rdev, attach->dmabuf->size,PAGE_SIZE , false, -RADEON_GEM_DOMAIN_GTT ,0 , sg, &bo); +RADEON_GEM_DOMAIN_GTT ,0 , sg, resv, &bo); +ww_mutex_unlock (&resv->lock);if (ret)return ERR_PTR (ret);
drivers/gpu/drm/radeon/radeon_ring.c +1 -1
@@ -383,7 +383,7 @@ int radeon_ring_init(struct radeon_device *rdev, struct radeon_ring *ring, unsig/* Allocate ring buffer */ if (ring->ring_obj ==NULL ) { r =radeon_bo_create (rdev, ring->ring_size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &ring->ring_obj);if (r) {dev_err (rdev->dev,"(%d) ring create failed\n" , r);
drivers/gpu/drm/radeon/radeon_sa.c +1 -1
@@ -65,7 +65,7 @@ int radeon_sa_bo_manager_init(struct radeon_device *rdev, } r =radeon_bo_create (rdev, size, align, true, - domain, flags,NULL , &sa_manager->bo); + domain, flags,NULL ,NULL , &sa_manager->bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate bo for manager\n" , r);return r;
drivers/gpu/drm/radeon/radeon_test.c +3 -2
@@ -67,7 +67,7 @@ static void radeon_do_test_moves(struct radeon_device *rdev, int flag) } r =radeon_bo_create (rdev, size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &vram_obj); +0 ,NULL ,NULL , &vram_obj);if (r) {DRM_ERROR ("Failed to create VRAM object\n" ); goto out_cleanup;
@@ -87,7 +87,8 @@ static void radeon_do_test_moves(struct radeon_device *rdev, int flag)struct radeon_fence *fence =NULL ; r =radeon_bo_create (rdev, size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , gtt_obj + i); +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , + gtt_obj + i);if (r) {DRM_ERROR ("Failed to create GTT object %d\n" , i); goto out_lclean;
drivers/gpu/drm/radeon/radeon_ttm.c +1 -1
@@ -858,7 +858,7 @@ int radeon_ttm_init(struct radeon_device *rdev)radeon_ttm_set_active_vram_size (rdev, rdev->mc.visible_vram_size); r =radeon_bo_create (rdev,256 *1024 ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 , +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL ,NULL , &rdev->stollen_vga_memory);if (r) {return r;
drivers/gpu/drm/radeon/radeon_uvd.c +2 -1
@@ -141,7 +141,8 @@ int radeon_uvd_init(struct radeon_device *rdev)RADEON_UVD_STACK_SIZE +RADEON_UVD_HEAP_SIZE +RADEON_GPU_PAGE_SIZE ; r =radeon_bo_create (rdev, bo_size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &rdev->uvd.vcpu_bo); +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , +NULL , &rdev->uvd.vcpu_bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate UVD bo\n" , r);return r;
drivers/gpu/drm/radeon/radeon_vce.c +2 -1
@@ -126,7 +126,8 @@ int radeon_vce_init(struct radeon_device *rdev) size =RADEON_GPU_PAGE_ALIGN (rdev->vce_fw->size) +RADEON_VCE_STACK_SIZE +RADEON_VCE_HEAP_SIZE ; r =radeon_bo_create (rdev, size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &rdev->vce.vcpu_bo); +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL ,NULL , + &rdev->vce.vcpu_bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate VCE bo\n" , r);return r;
drivers/gpu/drm/radeon/radeon_vm.c +3 -2
@@ -548,7 +548,8 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, r =radeon_bo_create (rdev,RADEON_VM_PTE_COUNT *8 ,RADEON_GPU_PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &pt); +RADEON_GEM_DOMAIN_VRAM ,0 , +NULL ,NULL , &pt);if (r)return r;
@@ -1127,7 +1128,7 @@ int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) r =radeon_bo_create (rdev, pd_size, align, true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &vm->page_directory); +NULL , &vm->page_directory);if (r)return r;
Re: [PATCH v2 7/8] drm/radeon: export reservation_object from dmabuf to ttm
Am 18.09.2014 um 14:11 schrieb Maarten Lankhorst:
Quoted 5 lines
drm/radeon: export
reservation_objectfrom dmabuf to ttmAdds an extra argument to
radeon_bo_create, which is only used inradeon_prime.c.Signed-off-by Maarten Lankhorst <[email protected]>
Quoted 306 lines
---
v2:
robj -> resv
fix typo in commit descriptiondrivers/gpu/drm/radeon/cik.c +2
-2@@ -4690,7 +4690,7 @@ static int cik_mec_init(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->mec.num_mec *rdev->mec.num_pipe *MEC_HPD_SIZE *2 ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &rdev->mec.hpd_eop_obj);if (r) {dev_warn (rdev->dev,"(%d) create HDP EOP bo failed\n" , r);@@ -4861,7 +4861,7 @@ static int cik_cp_compute_resume(struct radeon_device *rdev)sizeof (struct bonaire_mqd ),PAGE_SIZE , true,RADEON_GEM_DOMAIN_GTT ,0 ,NULL , - &rdev->ring[idx].mqd_obj); + NULL, &rdev->ring[idx].mqd_obj);if (r) {dev_warn (rdev->dev,"(%d) create MQD bo failed\n" , r);return r;drivers/gpu/drm/radeon/evergreen.c +3
-3@@ -4023,7 +4023,7 @@ int sumo_rlc_init(struct radeon_device *rdev)if (rdev->rlc.save_restore_obj ==NULL ) { r =radeon_bo_create (rdev, dws *4 ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.save_restore_obj); +NULL , &rdev->rlc.save_restore_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC sr bo failed\n" , r);return r;@@ -4102,7 +4102,7 @@ int sumo_rlc_init(struct radeon_device *rdev)if (rdev->rlc.clear_state_obj ==NULL ) { r =radeon_bo_create (rdev, dws *4 ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.clear_state_obj); +NULL , &rdev->rlc.clear_state_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC c bo failed\n" , r);sumo_rlc_fini (rdev);@@ -4179,7 +4179,7 @@ int sumo_rlc_init(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->rlc.cp_table_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &rdev->rlc.cp_table_obj); +NULL , &rdev->rlc.cp_table_obj);if (r) {dev_warn (rdev->dev,"(%d) create RLC cp table bo failed\n" , r);sumo_rlc_fini (rdev);drivers/gpu/drm/radeon/r600.c +2
-2@@ -1430,7 +1430,7 @@ int r600_vram_scratch_init(struct radeon_device *rdev)if (rdev->vram_scratch.robj ==NULL ) { r =radeon_bo_create (rdev,RADEON_GPU_PAGE_SIZE ,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &rdev->vram_scratch.robj); +0 ,NULL ,NULL , &rdev->vram_scratch.robj);if (r) {return r; }@@ -3368,7 +3368,7 @@ int r600_ih_ring_alloc(struct radeon_device *rdev) r =radeon_bo_create (rdev, rdev->ih.ring_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_GTT ,0 , -NULL , &rdev->ih.ring_obj); +NULL ,NULL , &rdev->ih.ring_obj);if (r) {DRM_ERROR ("radeon: failed to create ih ring buffer (%d).\n" , r);return r;drivers/gpu/drm/radeon/radeon_benchmark.c +2
-2@@ -93,7 +93,7 @@ static void radeon_benchmark_move(struct radeon_device *rdev, unsigned size,int time; n =RADEON_BENCHMARK_ITERATIONS ; - r =radeon_bo_create (rdev, size,PAGE_SIZE , true, sdomain,0 ,NULL , &sobj); + r =radeon_bo_create (rdev, size,PAGE_SIZE , true, sdomain,0 ,NULL ,NULL , &sobj);if (r) { goto out_cleanup; }@@ -105,7 +105,7 @@ static void radeon_benchmark_move(struct radeon_device *rdev, unsigned size,if (r) { goto out_cleanup; } - r =radeon_bo_create (rdev, size,PAGE_SIZE , true, ddomain,0 ,NULL , &dobj); + r =radeon_bo_create (rdev, size,PAGE_SIZE , true, ddomain,0 ,NULL ,NULL , &dobj);if (r) { goto out_cleanup; }drivers/gpu/drm/radeon/radeon_device.c +1
-1@@ -430,7 +430,7 @@ int radeon_wb_init(struct radeon_device *rdev)if (rdev->wb.wb_obj ==NULL ) { r =radeon_bo_create (rdev,RADEON_GPU_PAGE_SIZE ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &rdev->wb.wb_obj);if (r) {dev_warn (rdev->dev,"(%d) create WB bo failed\n" , r);drivers/gpu/drm/radeon/radeon_gart.c +1
-1@@ -128,7 +128,7 @@ int radeon_gart_table_vram_alloc(struct radeon_device *rdev)if (rdev->gart.robj ==NULL ) { r =radeon_bo_create (rdev, rdev->gart.table_size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &rdev->gart.robj); +0 ,NULL ,NULL , &rdev->gart.robj);if (r) {return r; }drivers/gpu/drm/radeon/radeon_gem.c +1
-1@@ -67,7 +67,7 @@ int radeon_gem_object_create(struct radeon_device *rdev, unsigned long size, retry: r =radeon_bo_create (rdev, size, alignment, kernel, initial_domain, - flags,NULL , &robj); + flags,NULL ,NULL , &robj);if (r) {if (r != -ERESTARTSYS ) {if (initial_domain ==RADEON_GEM_DOMAIN_VRAM ) {drivers/gpu/drm/radeon/radeon_object.c +5
-3@@ -167,8 +167,10 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) }int radeon_bo_create (struct radeon_device *rdev, -unsigned long size,int byte_align,bool kernel,u32 domain, -u32 flags,struct sg_table *sg,struct radeon_bo **bo_ptr) +unsigned long size,int byte_align,bool kernel, +u32 domain,u32 flags,struct sg_table *sg, +struct reservation_object *resv, +struct radeon_bo **bo_ptr) {struct radeon_bo *bo;enum ttm_bo_type type;@@ -216,7 +218,7 @@ int radeon_bo_create(struct radeon_device *rdev,down_read (&rdev->pm.mclk_lock); r =ttm_bo_init (&rdev->mman.bdev, &bo->tbo, size, type, &bo->placement, page_align, !kernel,NULL , - acc_size, sg,NULL , &radeon_ttm_bo_destroy); + acc_size, sg, resv, &radeon_ttm_bo_destroy);up_read (&rdev->pm.mclk_lock);if (unlikely (r !=0 )) {return r;drivers/gpu/drm/radeon/radeon_object.h +1
-0@@ -126,6 +126,7 @@ extern int radeon_bo_create(struct radeon_device *rdev, unsigned long size, int byte_align, bool kernel, u32 domain, u32 flags, struct sg_table *sg, + struct reservation_object *resv, struct radeon_bo **bo_ptr);extern int radeon_bo_kmap (struct radeon_bo *bo,void **ptr);extern void radeon_bo_kunmap (struct radeon_bo *bo);drivers/gpu/drm/radeon/radeon_prime.c +4
-1@@ -61,12 +61,15 @@ struct drm_gem_object *radeon_gem_prime_import_sg_table(struct drm_device *dev,struct dma_buf_attachment *attach, struct sg_table *sg) { + struct reservation_object *resv = attach->dmabuf->resv;struct radeon_device *rdev = dev->dev_private;struct radeon_bo *bo;int ret; +ww_mutex_lock (&resv->lock,NULL ); ret =radeon_bo_create (rdev, attach->dmabuf->size,PAGE_SIZE , false, -RADEON_GEM_DOMAIN_GTT ,0 , sg, &bo); +RADEON_GEM_DOMAIN_GTT ,0 , sg, resv, &bo); +ww_mutex_unlock (&resv->lock);if (ret)return ERR_PTR (ret);drivers/gpu/drm/radeon/radeon_ring.c +1
-1@@ -383,7 +383,7 @@ int radeon_ring_init(struct radeon_device *rdev, struct radeon_ring *ring, unsig/* Allocate ring buffer */ if (ring->ring_obj ==NULL ) { r =radeon_bo_create (rdev, ring->ring_size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 , +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , &ring->ring_obj);if (r) {dev_err (rdev->dev,"(%d) ring create failed\n" , r);drivers/gpu/drm/radeon/radeon_sa.c +1
-1@@ -65,7 +65,7 @@ int radeon_sa_bo_manager_init(struct radeon_device *rdev, } r =radeon_bo_create (rdev, size, align, true, - domain, flags,NULL , &sa_manager->bo); + domain, flags,NULL ,NULL , &sa_manager->bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate bo for manager\n" , r);return r;drivers/gpu/drm/radeon/radeon_test.c +3
-2@@ -67,7 +67,7 @@ static void radeon_do_test_moves(struct radeon_device *rdev, int flag) } r =radeon_bo_create (rdev, size,PAGE_SIZE , true,RADEON_GEM_DOMAIN_VRAM , -0 ,NULL , &vram_obj); +0 ,NULL ,NULL , &vram_obj);if (r) {DRM_ERROR ("Failed to create VRAM object\n" ); goto out_cleanup;@@ -87,7 +87,8 @@ static void radeon_do_test_moves(struct radeon_device *rdev, int flag)struct radeon_fence *fence =NULL ; r =radeon_bo_create (rdev, size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_GTT ,0 ,NULL , gtt_obj + i); +RADEON_GEM_DOMAIN_GTT ,0 ,NULL ,NULL , + gtt_obj + i);if (r) {DRM_ERROR ("Failed to create GTT object %d\n" , i); goto out_lclean;drivers/gpu/drm/radeon/radeon_ttm.c +1
-1@@ -858,7 +858,7 @@ int radeon_ttm_init(struct radeon_device *rdev)radeon_ttm_set_active_vram_size (rdev, rdev->mc.visible_vram_size); r =radeon_bo_create (rdev,256 *1024 ,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 , +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL ,NULL , &rdev->stollen_vga_memory);if (r) {return r;drivers/gpu/drm/radeon/radeon_uvd.c +2
-1@@ -141,7 +141,8 @@ int radeon_uvd_init(struct radeon_device *rdev)RADEON_UVD_STACK_SIZE +RADEON_UVD_HEAP_SIZE +RADEON_GPU_PAGE_SIZE ; r =radeon_bo_create (rdev, bo_size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &rdev->uvd.vcpu_bo); +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , +NULL , &rdev->uvd.vcpu_bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate UVD bo\n" , r);return r;drivers/gpu/drm/radeon/radeon_vce.c +2
-1@@ -126,7 +126,8 @@ int radeon_vce_init(struct radeon_device *rdev) size =RADEON_GPU_PAGE_ALIGN (rdev->vce_fw->size) +RADEON_VCE_STACK_SIZE +RADEON_VCE_HEAP_SIZE ; r =radeon_bo_create (rdev, size,PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &rdev->vce.vcpu_bo); +RADEON_GEM_DOMAIN_VRAM ,0 ,NULL ,NULL , + &rdev->vce.vcpu_bo);if (r) {dev_err (rdev->dev,"(%d) failed to allocate VCE bo\n" , r);return r;drivers/gpu/drm/radeon/radeon_vm.c +2
-1@@ -548,7 +548,8 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, r =radeon_bo_create (rdev,RADEON_VM_PTE_COUNT *8 ,RADEON_GPU_PAGE_SIZE , true, -RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , &pt); +RADEON_GEM_DOMAIN_VRAM ,0 , +NULL ,NULL , &pt);if (r)return r;
drivers/gpu/drm/radeon/radeon_vm.c int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) · line 1134
Quoted 8 lines
drivers/gpu/drm/radeon/radeon_vm.c +1
-1@@ -1127,7 +1128,7 @@ int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) r =radeon_bo_create (rdev, pd_size, align, true,RADEON_GEM_DOMAIN_VRAM ,0 ,NULL , - &vm->page_directory); +NULL , &vm->page_directory);if (r)return r;
_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel
PATCH 8/8 drm/nouveau: export reservation_object from dmabuf to ttm
In reply to Maarten LankhorstAdds an extra argument to nouveau_bo_new, which is only used in nouveau_prime.c.
--- drivers/gpu/drm/nouveau/dispnv04/crtc.c | 2 +- drivers/gpu/drm/nouveau/nouveau_bo.c | 4 ++-- drivers/gpu/drm/nouveau/nouveau_bo.h | 1 + drivers/gpu/drm/nouveau/nouveau_chan.c | 2 +- drivers/gpu/drm/nouveau/nouveau_fence.c | 6 +++++- drivers/gpu/drm/nouveau/nouveau_gem.c | 2 +- drivers/gpu/drm/nouveau/nouveau_prime.c | 5 ++++- drivers/gpu/drm/nouveau/nv17_fence.c | 2 +- drivers/gpu/drm/nouveau/nv50_display.c | 6 +++--- drivers/gpu/drm/nouveau/nv50_fence.c | 2 +- drivers/gpu/drm/nouveau/nv84_fence.c | 4 ++-- 11 files changed, 22 insertions(+), 14 deletions(-)
drivers/gpu/drm/nouveau/dispnv04/crtc.c +1 -1
@@ -1127,7 +1127,7 @@ nv04_crtc_create(struct drm_device *dev, int crtc_num)drm_mode_crtc_set_gamma_size (&nv_crtc->base,256 ); ret =nouveau_bo_new (dev,64 *64 *4 ,0x100 ,TTM_PL_FLAG_VRAM , -0 ,0x0000 ,NULL , &nv_crtc->cursor.nvbo); +0 ,0x0000 ,NULL ,NULL , &nv_crtc->cursor.nvbo);if (!ret) { ret =nouveau_bo_pin (nv_crtc->cursor.nvbo,TTM_PL_FLAG_VRAM );if (!ret) {
drivers/gpu/drm/nouveau/nouveau_bo.c +2 -2
@@ -181,7 +181,7 @@ nouveau_bo_fixup_align(struct nouveau_bo *nvbo, u32 flags,int nouveau_bo_new (struct drm_device *dev,int size,int align,uint32_t flags,uint32_t tile_mode,uint32_t tile_flags, -struct sg_table *sg, +struct sg_table *sg,struct reservation_object *robj,struct nouveau_bo **pnvbo) {struct nouveau_drm *drm =nouveau_drm (dev);
@@ -230,7 +230,7 @@ nouveau_bo_new(struct drm_device *dev, int size, int align, ret =ttm_bo_init (&drm->ttm.bdev, &nvbo->bo, size, type, &nvbo->placement, align >>PAGE_SHIFT , false,NULL , acc_size, sg, -NULL , nouveau_bo_del_ttm); + robj, nouveau_bo_del_ttm);if (ret) {/* ttm will call nouveau_bo_del_ttm if it fails.. */ return ret;
drivers/gpu/drm/nouveau/nouveau_bo.h +1 -0
@@ -68,6 +68,7 @@ extern struct ttm_bo_driver nouveau_bo_driver;void nouveau_bo_move_init (struct nouveau_drm *);int nouveau_bo_new (struct drm_device *,int size,int align,u32 flags,u32 tile_mode,u32 tile_flags,struct sg_table *sg, +struct reservation_object *robj,struct nouveau_bo **);int nouveau_bo_pin (struct nouveau_bo *,u32 flags);int nouveau_bo_unpin (struct nouveau_bo *);
drivers/gpu/drm/nouveau/nouveau_chan.c +1 -1
@@ -106,7 +106,7 @@ nouveau_channel_prep(struct nouveau_drm *drm, struct nvif_device *device,if (nouveau_vram_pushbuf) target =TTM_PL_FLAG_VRAM ; - ret =nouveau_bo_new (drm->dev, size,0 , target,0 ,0 ,NULL , + ret =nouveau_bo_new (drm->dev, size,0 , target,0 ,0 ,NULL ,NULL , &chan->push.buffer);if (ret ==0 ) { ret =nouveau_bo_pin (chan->push.buffer, target);
drivers/gpu/drm/nouveau/nouveau_fence.c +5 -1
@@ -195,8 +195,12 @@ nouveau_fence_work(struct fence *fence, work =kmalloc (sizeof (*work),GFP_KERNEL );if (!work) { +/* +* this might not be a nouveau fence any more, +* so force a lazy wait here +*/ WARN_ON (nouveau_fence_wait ((struct nouveau_fence *)fence, - false, false)); + true, false)); goto err; }
drivers/gpu/drm/nouveau/nouveau_gem.c +1 -1
@@ -165,7 +165,7 @@ nouveau_gem_new(struct drm_device *dev, int size, int align, uint32_t domain, flags |=TTM_PL_FLAG_SYSTEM ; ret =nouveau_bo_new (dev, size, align, flags, tile_mode, - tile_flags,NULL , pnvbo); + tile_flags,NULL ,NULL , pnvbo);if (ret)return ret; nvbo = *pnvbo;
drivers/gpu/drm/nouveau/nouveau_prime.c +4 -1
@@ -61,13 +61,16 @@ struct drm_gem_object *nouveau_gem_prime_import_sg_table(struct drm_device *dev, struct sg_table *sg) {struct nouveau_bo *nvbo; +struct reservation_object *robj = attach->dmabuf->resv;u32 flags =0 ;int ret; flags =TTM_PL_FLAG_TT ; +ww_mutex_lock (&robj->lock,NULL ); ret =nouveau_bo_new (dev, attach->dmabuf->size,0 , flags,0 ,0 , - sg, &nvbo); + sg, robj, &nvbo); +ww_mutex_unlock (&robj->lock);if (ret)return ERR_PTR (ret);
drivers/gpu/drm/nouveau/nv17_fence.c +1 -1
@@ -129,7 +129,7 @@ nv17_fence_create(struct nouveau_drm *drm)spin_lock_init (&priv->lock); ret =nouveau_bo_new (drm->dev,4096 ,0x1000 ,TTM_PL_FLAG_VRAM , -0 ,0x0000 ,NULL , &priv->bo); +0 ,0x0000 ,NULL ,NULL , &priv->bo);if (!ret) { ret =nouveau_bo_pin (priv->bo,TTM_PL_FLAG_VRAM );if (!ret) {
drivers/gpu/drm/nouveau/nv50_display.c +3 -3
@@ -1383,7 +1383,7 @@ nv50_crtc_create(struct drm_device *dev, int index)drm_mode_crtc_set_gamma_size (crtc,256 ); ret =nouveau_bo_new (dev,8192 ,0x100 ,TTM_PL_FLAG_VRAM , -0 ,0x0000 ,NULL , &head->base.lut.nvbo); +0 ,0x0000 ,NULL ,NULL , &head->base.lut.nvbo);if (!ret) { ret =nouveau_bo_pin (head->base.lut.nvbo,TTM_PL_FLAG_VRAM );if (!ret) {
@@ -1406,7 +1406,7 @@ nv50_crtc_create(struct drm_device *dev, int index) goto out; ret =nouveau_bo_new (dev,64 *64 *4 ,0x100 ,TTM_PL_FLAG_VRAM , -0 ,0x0000 ,NULL , &head->base.cursor.nvbo); +0 ,0x0000 ,NULL ,NULL , &head->base.cursor.nvbo);if (!ret) { ret =nouveau_bo_pin (head->base.cursor.nvbo,TTM_PL_FLAG_VRAM );if (!ret) {
@@ -2468,7 +2468,7 @@ nv50_display_create(struct drm_device *dev)/* small shared memory area we use for notifiers and semaphores */ ret =nouveau_bo_new (dev,4096 ,0x1000 ,TTM_PL_FLAG_VRAM , -0 ,0x0000 ,NULL , &disp->sync); +0 ,0x0000 ,NULL ,NULL , &disp->sync);if (!ret) { ret =nouveau_bo_pin (disp->sync,TTM_PL_FLAG_VRAM );if (!ret) {
drivers/gpu/drm/nouveau/nv50_fence.c +1 -1
@@ -100,7 +100,7 @@ nv50_fence_create(struct nouveau_drm *drm)spin_lock_init (&priv->lock); ret =nouveau_bo_new (drm->dev,4096 ,0x1000 ,TTM_PL_FLAG_VRAM , -0 ,0x0000 ,NULL , &priv->bo); +0 ,0x0000 ,NULL ,NULL , &priv->bo);if (!ret) { ret =nouveau_bo_pin (priv->bo,TTM_PL_FLAG_VRAM );if (!ret) {
drivers/gpu/drm/nouveau/nv84_fence.c +2 -2
@@ -233,7 +233,7 @@ nv84_fence_create(struct nouveau_drm *drm) priv->base.uevent = true; ret =nouveau_bo_new (drm->dev,16 * priv->base.contexts,0 , -TTM_PL_FLAG_VRAM ,0 ,0 ,NULL , &priv->bo); +TTM_PL_FLAG_VRAM ,0 ,0 ,NULL ,NULL , &priv->bo);if (ret ==0 ) { ret =nouveau_bo_pin (priv->bo,TTM_PL_FLAG_VRAM );if (ret ==0 ) {
@@ -247,7 +247,7 @@ nv84_fence_create(struct nouveau_drm *drm)if (ret ==0 ) ret =nouveau_bo_new (drm->dev,16 * priv->base.contexts,0 , -TTM_PL_FLAG_TT ,0 ,0 ,NULL , +TTM_PL_FLAG_TT ,0 ,0 ,NULL ,NULL , &priv->bo_gart);if (ret ==0 ) { ret =nouveau_bo_pin (priv->bo_gart,TTM_PL_FLAG_TT );
Signature · 1 line
2.0.4