diff options
Diffstat (limited to 'drivers/gpu/drm/nouveau')
-rw-r--r-- | drivers/gpu/drm/nouveau/dispnv04/crtc.c | 9 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_abi16.c | 21 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_bo.c | 8 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_chan.c | 21 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_dma.h | 3 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_dmem.c | 9 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_drv.h | 19 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_exec.c | 20 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_exec.h | 10 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_fence.c | 32 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_fence.h | 5 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_gem.c | 5 | ||||
-rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_sched.c | 12 |
13 files changed, 105 insertions, 69 deletions
diff --git a/drivers/gpu/drm/nouveau/dispnv04/crtc.c b/drivers/gpu/drm/nouveau/dispnv04/crtc.c index a34924523133..a34917b048f9 100644 --- a/drivers/gpu/drm/nouveau/dispnv04/crtc.c +++ b/drivers/gpu/drm/nouveau/dispnv04/crtc.c @@ -1122,18 +1122,11 @@ nv04_page_flip_emit(struct nouveau_channel *chan, PUSH_NVSQ(push, NV_SW, NV_SW_PAGE_FLIP, 0x00000000); PUSH_KICK(push); - ret = nouveau_fence_new(pfence); + ret = nouveau_fence_new(pfence, chan); if (ret) goto fail; - ret = nouveau_fence_emit(*pfence, chan); - if (ret) - goto fail_fence_unref; - return 0; - -fail_fence_unref: - nouveau_fence_unref(pfence); fail: spin_lock_irqsave(&dev->event_lock, flags); list_del(&s->head); diff --git a/drivers/gpu/drm/nouveau/nouveau_abi16.c b/drivers/gpu/drm/nouveau/nouveau_abi16.c index 30afbec9e3b1..2edd7bb13fae 100644 --- a/drivers/gpu/drm/nouveau/nouveau_abi16.c +++ b/drivers/gpu/drm/nouveau/nouveau_abi16.c @@ -31,6 +31,7 @@ #include "nouveau_drv.h" #include "nouveau_dma.h" +#include "nouveau_exec.h" #include "nouveau_gem.h" #include "nouveau_chan.h" #include "nouveau_abi16.h" @@ -183,6 +184,20 @@ nouveau_abi16_fini(struct nouveau_abi16 *abi16) cli->abi16 = NULL; } +static inline int +getparam_dma_ib_max(struct nvif_device *device) +{ + const struct nvif_mclass dmas[] = { + { NV03_CHANNEL_DMA, 0 }, + { NV10_CHANNEL_DMA, 0 }, + { NV17_CHANNEL_DMA, 0 }, + { NV40_CHANNEL_DMA, 0 }, + {} + }; + + return nvif_mclass(&device->object, dmas) < 0 ? NV50_DMA_IB_MAX : 0; +} + int nouveau_abi16_ioctl_getparam(ABI16_IOCTL_ARGS) { @@ -247,6 +262,12 @@ nouveau_abi16_ioctl_getparam(ABI16_IOCTL_ARGS) case NOUVEAU_GETPARAM_GRAPH_UNITS: getparam->value = nvkm_gr_units(gr); break; + case NOUVEAU_GETPARAM_EXEC_PUSH_MAX: { + int ib_max = getparam_dma_ib_max(device); + + getparam->value = nouveau_exec_push_max_from_ib_max(ib_max); + break; + } default: NV_PRINTK(dbg, cli, "unknown parameter %lld\n", getparam->param); return -EINVAL; diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 19cab37ac69c..0f3bd187ede6 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -875,16 +875,10 @@ nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, if (ret) goto out_unlock; - ret = nouveau_fence_new(&fence); + ret = nouveau_fence_new(&fence, chan); if (ret) goto out_unlock; - ret = nouveau_fence_emit(fence, chan); - if (ret) { - nouveau_fence_unref(&fence); - goto out_unlock; - } - /* TODO: figure out a better solution here * * wait on the fence here explicitly as going through diff --git a/drivers/gpu/drm/nouveau/nouveau_chan.c b/drivers/gpu/drm/nouveau/nouveau_chan.c index 1fd5ccf41128..7c97b2886807 100644 --- a/drivers/gpu/drm/nouveau/nouveau_chan.c +++ b/drivers/gpu/drm/nouveau/nouveau_chan.c @@ -70,11 +70,9 @@ nouveau_channel_idle(struct nouveau_channel *chan) struct nouveau_fence *fence = NULL; int ret; - ret = nouveau_fence_new(&fence); + ret = nouveau_fence_new(&fence, chan); if (!ret) { - ret = nouveau_fence_emit(fence, chan); - if (!ret) - ret = nouveau_fence_wait(fence, false, false); + ret = nouveau_fence_wait(fence, false, false); nouveau_fence_unref(&fence); } @@ -259,10 +257,7 @@ static int nouveau_channel_ctor(struct nouveau_drm *drm, struct nvif_device *device, bool priv, u64 runm, struct nouveau_channel **pchan) { - static const struct { - s32 oclass; - int version; - } hosts[] = { + const struct nvif_mclass hosts[] = { { AMPERE_CHANNEL_GPFIFO_B, 0 }, { AMPERE_CHANNEL_GPFIFO_A, 0 }, { TURING_CHANNEL_GPFIFO_A, 0 }, @@ -445,9 +440,11 @@ nouveau_channel_init(struct nouveau_channel *chan, u32 vram, u32 gart) } /* initialise dma tracking parameters */ - switch (chan->user.oclass & 0x00ff) { - case 0x006b: - case 0x006e: + switch (chan->user.oclass) { + case NV03_CHANNEL_DMA: + case NV10_CHANNEL_DMA: + case NV17_CHANNEL_DMA: + case NV40_CHANNEL_DMA: chan->user_put = 0x40; chan->user_get = 0x44; chan->dma.max = (0x10000 / 4) - 2; @@ -457,7 +454,7 @@ nouveau_channel_init(struct nouveau_channel *chan, u32 vram, u32 gart) chan->user_get = 0x44; chan->user_get_hi = 0x60; chan->dma.ib_base = 0x10000 / 4; - chan->dma.ib_max = (0x02000 / 8) - 1; + chan->dma.ib_max = NV50_DMA_IB_MAX; chan->dma.ib_put = 0; chan->dma.ib_free = chan->dma.ib_max - chan->dma.ib_put; chan->dma.max = chan->dma.ib_base; diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.h b/drivers/gpu/drm/nouveau/nouveau_dma.h index 1744d95b233e..c52cda82353e 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dma.h +++ b/drivers/gpu/drm/nouveau/nouveau_dma.h @@ -49,6 +49,9 @@ void nv50_dma_push(struct nouveau_channel *, u64 addr, u32 length, /* Maximum push buffer size. */ #define NV50_DMA_PUSH_MAX_LENGTH 0x7fffff +/* Maximum IBs per ring. */ +#define NV50_DMA_IB_MAX ((0x02000 / 8) - 1) + /* Object handles - for stuff that's doesn't use handle == oclass. */ enum { NvDmaFB = 0x80000002, diff --git a/drivers/gpu/drm/nouveau/nouveau_dmem.c b/drivers/gpu/drm/nouveau/nouveau_dmem.c index 61e84562094a..12feecf71e75 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dmem.c +++ b/drivers/gpu/drm/nouveau/nouveau_dmem.c @@ -209,8 +209,7 @@ static vm_fault_t nouveau_dmem_migrate_to_ram(struct vm_fault *vmf) goto done; } - if (!nouveau_fence_new(&fence)) - nouveau_fence_emit(fence, dmem->migrate.chan); + nouveau_fence_new(&fence, dmem->migrate.chan); migrate_vma_pages(&args); nouveau_dmem_fence_done(&fence); dma_unmap_page(drm->dev->dev, dma_addr, PAGE_SIZE, DMA_BIDIRECTIONAL); @@ -403,8 +402,7 @@ nouveau_dmem_evict_chunk(struct nouveau_dmem_chunk *chunk) } } - if (!nouveau_fence_new(&fence)) - nouveau_fence_emit(fence, chunk->drm->dmem->migrate.chan); + nouveau_fence_new(&fence, chunk->drm->dmem->migrate.chan); migrate_device_pages(src_pfns, dst_pfns, npages); nouveau_dmem_fence_done(&fence); migrate_device_finalize(src_pfns, dst_pfns, npages); @@ -677,8 +675,7 @@ static void nouveau_dmem_migrate_chunk(struct nouveau_drm *drm, addr += PAGE_SIZE; } - if (!nouveau_fence_new(&fence)) - nouveau_fence_emit(fence, drm->dmem->migrate.chan); + nouveau_fence_new(&fence, drm->dmem->migrate.chan); migrate_vma_pages(args); nouveau_dmem_fence_done(&fence); nouveau_pfns_map(svmm, args->vma->vm_mm, args->start, pfns, i); diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h index 1fe17ff95f5e..e73a233c6572 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drv.h +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h @@ -189,21 +189,12 @@ u_free(void *addr) static inline void * u_memcpya(uint64_t user, unsigned int nmemb, unsigned int size) { - void *mem; - void __user *userptr = (void __force __user *)(uintptr_t)user; + void __user *userptr = u64_to_user_ptr(user); + size_t bytes; - size *= nmemb; - - mem = kvmalloc(size, GFP_KERNEL); - if (!mem) - return ERR_PTR(-ENOMEM); - - if (copy_from_user(mem, userptr, size)) { - u_free(mem); - return ERR_PTR(-EFAULT); - } - - return mem; + if (unlikely(check_mul_overflow(nmemb, size, &bytes))) + return ERR_PTR(-EOVERFLOW); + return vmemdup_user(userptr, bytes); } #include <nvif/object.h> diff --git a/drivers/gpu/drm/nouveau/nouveau_exec.c b/drivers/gpu/drm/nouveau/nouveau_exec.c index a90c4cd8cbb2..c1837ba95fb5 100644 --- a/drivers/gpu/drm/nouveau/nouveau_exec.c +++ b/drivers/gpu/drm/nouveau/nouveau_exec.c @@ -96,7 +96,8 @@ nouveau_exec_job_submit(struct nouveau_job *job) unsigned long index; int ret; - ret = nouveau_fence_new(&exec_job->fence); + /* Create a new fence, but do not emit yet. */ + ret = nouveau_fence_create(&exec_job->fence, exec_job->chan); if (ret) return ret; @@ -170,13 +171,17 @@ nouveau_exec_job_run(struct nouveau_job *job) nv50_dma_push(chan, p->va, p->va_len, no_prefetch); } - ret = nouveau_fence_emit(fence, chan); + ret = nouveau_fence_emit(fence); if (ret) { + nouveau_fence_unref(&exec_job->fence); NV_PRINTK(err, job->cli, "error fencing pushbuf: %d\n", ret); WIND_RING(chan); return ERR_PTR(ret); } + /* The fence was emitted successfully, set the job's fence pointer to + * NULL in order to avoid freeing it up when the job is cleaned up. + */ exec_job->fence = NULL; return &fence->base; @@ -189,7 +194,7 @@ nouveau_exec_job_free(struct nouveau_job *job) nouveau_job_free(job); - nouveau_fence_unref(&exec_job->fence); + kfree(exec_job->fence); kfree(exec_job->push.s); kfree(exec_job); } @@ -208,7 +213,7 @@ nouveau_exec_job_timeout(struct nouveau_job *job) nouveau_sched_entity_fini(job->entity); - return DRM_GPU_SCHED_STAT_ENODEV; + return DRM_GPU_SCHED_STAT_NOMINAL; } static struct nouveau_job_ops nouveau_exec_job_ops = { @@ -374,7 +379,7 @@ nouveau_exec_ioctl_exec(struct drm_device *dev, struct nouveau_channel *chan = NULL; struct nouveau_exec_job_args args = {}; struct drm_nouveau_exec *req = data; - int ret = 0; + int push_max, ret = 0; if (unlikely(!abi16)) return -ENOMEM; @@ -399,9 +404,10 @@ nouveau_exec_ioctl_exec(struct drm_device *dev, if (!chan->dma.ib_max) return nouveau_abi16_put(abi16, -ENOSYS); - if (unlikely(req->push_count > NOUVEAU_GEM_MAX_PUSH)) { + push_max = nouveau_exec_push_max_from_ib_max(chan->dma.ib_max); + if (unlikely(req->push_count > push_max)) { NV_PRINTK(err, cli, "pushbuf push count exceeds limit: %d max %d\n", - req->push_count, NOUVEAU_GEM_MAX_PUSH); + req->push_count, push_max); return nouveau_abi16_put(abi16, -EINVAL); } diff --git a/drivers/gpu/drm/nouveau/nouveau_exec.h b/drivers/gpu/drm/nouveau/nouveau_exec.h index 778cacd90f65..5488d337bcc0 100644 --- a/drivers/gpu/drm/nouveau/nouveau_exec.h +++ b/drivers/gpu/drm/nouveau/nouveau_exec.h @@ -51,4 +51,14 @@ int nouveau_exec_job_init(struct nouveau_exec_job **job, int nouveau_exec_ioctl_exec(struct drm_device *dev, void *data, struct drm_file *file_priv); +static inline unsigned int +nouveau_exec_push_max_from_ib_max(int ib_max) +{ + /* Limit the number of IBs per job to half the size of the ring in order + * to avoid the ring running dry between submissions and preserve one + * more slot for the job's HW fence. + */ + return ib_max > 1 ? ib_max / 2 - 1 : 0; +} + #endif diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.c b/drivers/gpu/drm/nouveau/nouveau_fence.c index 77c739a55b19..ca762ea55413 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fence.c +++ b/drivers/gpu/drm/nouveau/nouveau_fence.c @@ -205,16 +205,13 @@ nouveau_fence_context_new(struct nouveau_channel *chan, struct nouveau_fence_cha } int -nouveau_fence_emit(struct nouveau_fence *fence, struct nouveau_channel *chan) +nouveau_fence_emit(struct nouveau_fence *fence) { + struct nouveau_channel *chan = unrcu_pointer(fence->channel); struct nouveau_fence_chan *fctx = chan->fence; struct nouveau_fence_priv *priv = (void*)chan->drm->fence; int ret; - if (unlikely(!chan->fence)) - return -ENODEV; - - fence->channel = chan; fence->timeout = jiffies + (15 * HZ); if (priv->uevent) @@ -406,18 +403,41 @@ nouveau_fence_unref(struct nouveau_fence **pfence) } int -nouveau_fence_new(struct nouveau_fence **pfence) +nouveau_fence_create(struct nouveau_fence **pfence, + struct nouveau_channel *chan) { struct nouveau_fence *fence; + if (unlikely(!chan->fence)) + return -ENODEV; + fence = kzalloc(sizeof(*fence), GFP_KERNEL); if (!fence) return -ENOMEM; + fence->channel = chan; + *pfence = fence; return 0; } +int +nouveau_fence_new(struct nouveau_fence **pfence, + struct nouveau_channel *chan) +{ + int ret = 0; + + ret = nouveau_fence_create(pfence, chan); + if (ret) + return ret; + + ret = nouveau_fence_emit(*pfence); + if (ret) + nouveau_fence_unref(pfence); + + return ret; +} + static const char *nouveau_fence_get_get_driver_name(struct dma_fence *fence) { return "nouveau"; diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.h b/drivers/gpu/drm/nouveau/nouveau_fence.h index 2c72d96ef17d..64d33ae7f356 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fence.h +++ b/drivers/gpu/drm/nouveau/nouveau_fence.h @@ -17,10 +17,11 @@ struct nouveau_fence { unsigned long timeout; }; -int nouveau_fence_new(struct nouveau_fence **); +int nouveau_fence_create(struct nouveau_fence **, struct nouveau_channel *); +int nouveau_fence_new(struct nouveau_fence **, struct nouveau_channel *); void nouveau_fence_unref(struct nouveau_fence **); -int nouveau_fence_emit(struct nouveau_fence *, struct nouveau_channel *); +int nouveau_fence_emit(struct nouveau_fence *); bool nouveau_fence_done(struct nouveau_fence *); int nouveau_fence_wait(struct nouveau_fence *, bool lazy, bool intr); int nouveau_fence_sync(struct nouveau_bo *, struct nouveau_channel *, bool exclusive, bool intr); diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c index c0b10d8d3d03..a0d303e5ce3d 100644 --- a/drivers/gpu/drm/nouveau/nouveau_gem.c +++ b/drivers/gpu/drm/nouveau/nouveau_gem.c @@ -914,11 +914,8 @@ revalidate: } } - ret = nouveau_fence_new(&fence); - if (!ret) - ret = nouveau_fence_emit(fence, chan); + ret = nouveau_fence_new(&fence, chan); if (ret) { - nouveau_fence_unref(&fence); NV_PRINTK(err, cli, "error fencing pushbuf: %d\n", ret); WIND_RING(chan); goto out; diff --git a/drivers/gpu/drm/nouveau/nouveau_sched.c b/drivers/gpu/drm/nouveau/nouveau_sched.c index 88217185e0f3..3b7ea5221226 100644 --- a/drivers/gpu/drm/nouveau/nouveau_sched.c +++ b/drivers/gpu/drm/nouveau/nouveau_sched.c @@ -375,14 +375,20 @@ nouveau_sched_run_job(struct drm_sched_job *sched_job) static enum drm_gpu_sched_stat nouveau_sched_timedout_job(struct drm_sched_job *sched_job) { + struct drm_gpu_scheduler *sched = sched_job->sched; struct nouveau_job *job = to_nouveau_job(sched_job); + enum drm_gpu_sched_stat stat = DRM_GPU_SCHED_STAT_NOMINAL; - NV_PRINTK(warn, job->cli, "Job timed out.\n"); + drm_sched_stop(sched, sched_job); if (job->ops->timeout) - return job->ops->timeout(job); + stat = job->ops->timeout(job); + else + NV_PRINTK(warn, job->cli, "Generic job timeout.\n"); + + drm_sched_start(sched, true); - return DRM_GPU_SCHED_STAT_ENODEV; + return stat; } static void |