From e922cfb1c410d8cb200a08f1b2a527d92d95480a Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Thu, 11 Jun 2015 11:26:03 -0400 Subject: drm/amdgpu: fix hpd range check in dce_v8_0_hpd_irq() Spotted by Dan Carpenter. This is a slight variant of his fix. Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c index 72c27ac..aaca8d6 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c @@ -3379,7 +3379,7 @@ static int dce_v8_0_hpd_irq(struct amdgpu_device *adev, uint32_t disp_int, mask, int_control, tmp; unsigned hpd; - if (entry->src_data > 6) { + if (entry->src_data >= adev->mode_info.num_hpd) { DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data); return 0; } -- cgit v0.10.2 From 69ee2410238e6ea6fd5d0a91b04510bba8abc7b5 Mon Sep 17 00:00:00 2001 From: Maninder Singh Date: Fri, 19 Jun 2015 09:35:23 +0530 Subject: drm/radeon: use kzalloc for allocating one thing MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Use kzalloc for allocating one thing rather than kcalloc(1... The semantic patch that makes this change is as follows: // @@ @@ - kcalloc(1, + kzalloc( ...) // Signed-off-by: Maninder Singh Reviewed-by: Vaneet Narang Reviewed-by: Christian König Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index edafd3c..06ac59f 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -719,7 +719,7 @@ static int radeon_ttm_tt_populate(struct ttm_tt *ttm) return 0; if (gtt && gtt->userptr) { - ttm->sg = kcalloc(1, sizeof(struct sg_table), GFP_KERNEL); + ttm->sg = kzalloc(sizeof(struct sg_table), GFP_KERNEL); if (!ttm->sg) return -ENOMEM; -- cgit v0.10.2 From cd06bf687b89f7532d19b0e4484299372e1bf70c Mon Sep 17 00:00:00 2001 From: Ben Goz Date: Wed, 24 Jun 2015 22:39:21 +0300 Subject: drm/amdgpu: Initialize compute sdma and memory from kgd v2: add missing MTYPE_NONCACHED enum Signed-off-by: Ben Goz Acked-by: Oded Gabbay Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/cikd.h b/drivers/gpu/drm/amd/amdgpu/cikd.h index 220865a..d19085a 100644 --- a/drivers/gpu/drm/amd/amdgpu/cikd.h +++ b/drivers/gpu/drm/amd/amdgpu/cikd.h @@ -552,4 +552,10 @@ #define VCE_CMD_IB_AUTO 0x00000005 #define VCE_CMD_SEMAPHORE 0x00000006 +/* valid for both DEFAULT_MTYPE and APE1_MTYPE */ +enum { + MTYPE_CACHED = 0, + MTYPE_NONCACHED = 3 +}; + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index cb790744..2c188fb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -2010,6 +2010,46 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev, } /** + * gmc_v7_0_init_compute_vmid - gart enable + * + * @rdev: amdgpu_device pointer + * + * Initialize compute vmid sh_mem registers + * + */ +#define DEFAULT_SH_MEM_BASES (0x6000) +#define FIRST_COMPUTE_VMID (8) +#define LAST_COMPUTE_VMID (16) +static void gmc_v7_0_init_compute_vmid(struct amdgpu_device *adev) +{ + int i; + uint32_t sh_mem_config; + uint32_t sh_mem_bases; + + /* + * Configure apertures: + * LDS: 0x60000000'00000000 - 0x60000001'00000000 (4GB) + * Scratch: 0x60000001'00000000 - 0x60000002'00000000 (4GB) + * GPUVM: 0x60010000'00000000 - 0x60020000'00000000 (1TB) + */ + sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16); + sh_mem_config = SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT; + sh_mem_config |= MTYPE_NONCACHED << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT; + mutex_lock(&adev->srbm_mutex); + for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) { + cik_srbm_select(adev, 0, 0, 0, i); + /* CP and shaders */ + WREG32(mmSH_MEM_CONFIG, sh_mem_config); + WREG32(mmSH_MEM_APE1_BASE, 1); + WREG32(mmSH_MEM_APE1_LIMIT, 0); + WREG32(mmSH_MEM_BASES, sh_mem_bases); + } + cik_srbm_select(adev, 0, 0, 0, 0); + mutex_unlock(&adev->srbm_mutex); +} + +/** * gfx_v7_0_gpu_init - setup the 3D engine * * @adev: amdgpu_device pointer @@ -2230,6 +2270,8 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev) cik_srbm_select(adev, 0, 0, 0, 0); mutex_unlock(&adev->srbm_mutex); + gmc_v7_0_init_compute_vmid(adev); + WREG32(mmSX_DEBUG_1, 0x20); WREG32(mmTA_CNTL_AUX, 0x00010000); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 14242bd..e4aeb74 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -1894,6 +1894,51 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev, mutex_unlock(&adev->grbm_idx_mutex); } +/** + * gmc_v8_0_init_compute_vmid - gart enable + * + * @rdev: amdgpu_device pointer + * + * Initialize compute vmid sh_mem registers + * + */ +#define DEFAULT_SH_MEM_BASES (0x6000) +#define FIRST_COMPUTE_VMID (8) +#define LAST_COMPUTE_VMID (16) +static void gmc_v8_0_init_compute_vmid(struct amdgpu_device *adev) +{ + int i; + uint32_t sh_mem_config; + uint32_t sh_mem_bases; + + /* + * Configure apertures: + * LDS: 0x60000000'00000000 - 0x60000001'00000000 (4GB) + * Scratch: 0x60000001'00000000 - 0x60000002'00000000 (4GB) + * GPUVM: 0x60010000'00000000 - 0x60020000'00000000 (1TB) + */ + sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16); + + sh_mem_config = SH_MEM_ADDRESS_MODE_HSA64 << + SH_MEM_CONFIG__ADDRESS_MODE__SHIFT | + SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT | + MTYPE_CC << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT | + SH_MEM_CONFIG__PRIVATE_ATC_MASK; + + mutex_lock(&adev->srbm_mutex); + for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) { + vi_srbm_select(adev, 0, 0, 0, i); + /* CP and shaders */ + WREG32(mmSH_MEM_CONFIG, sh_mem_config); + WREG32(mmSH_MEM_APE1_BASE, 1); + WREG32(mmSH_MEM_APE1_LIMIT, 0); + WREG32(mmSH_MEM_BASES, sh_mem_bases); + } + vi_srbm_select(adev, 0, 0, 0, 0); + mutex_unlock(&adev->srbm_mutex); +} + static void gfx_v8_0_gpu_init(struct amdgpu_device *adev) { u32 gb_addr_config; @@ -2113,6 +2158,8 @@ static void gfx_v8_0_gpu_init(struct amdgpu_device *adev) vi_srbm_select(adev, 0, 0, 0, 0); mutex_unlock(&adev->srbm_mutex); + gmc_v8_0_init_compute_vmid(adev); + mutex_lock(&adev->grbm_idx_mutex); /* * making sure that the following register writes will be broadcasted diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index e3c1fde..7bb37b9 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -439,6 +439,31 @@ static void sdma_v3_0_rlc_stop(struct amdgpu_device *adev) } /** + * sdma_v3_0_ctx_switch_enable - stop the async dma engines context switch + * + * @adev: amdgpu_device pointer + * @enable: enable/disable the DMA MEs context switch. + * + * Halt or unhalt the async dma engines context switch (VI). + */ +static void sdma_v3_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable) +{ + u32 f32_cntl; + int i; + + for (i = 0; i < SDMA_MAX_INSTANCE; i++) { + f32_cntl = RREG32(mmSDMA0_CNTL + sdma_offsets[i]); + if (enable) + f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL, + AUTO_CTXSW_ENABLE, 1); + else + f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL, + AUTO_CTXSW_ENABLE, 0); + WREG32(mmSDMA0_CNTL + sdma_offsets[i], f32_cntl); + } +} + +/** * sdma_v3_0_enable - stop the async dma engines * * @adev: amdgpu_device pointer @@ -648,6 +673,8 @@ static int sdma_v3_0_start(struct amdgpu_device *adev) /* unhalt the MEs */ sdma_v3_0_enable(adev, true); + /* enable sdma ring preemption */ + sdma_v3_0_ctx_switch_enable(adev, true); /* start the gfx rings and rlc compute queues */ r = sdma_v3_0_gfx_resume(adev); @@ -1079,6 +1106,7 @@ static int sdma_v3_0_hw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + sdma_v3_0_ctx_switch_enable(adev, false); sdma_v3_0_enable(adev, false); return 0; -- cgit v0.10.2 From 78ad5cdd21f0d614983fc397338944e797ec70b9 Mon Sep 17 00:00:00 2001 From: Ben Goz Date: Fri, 12 Jun 2015 11:08:58 +0300 Subject: drm/amdgpu: Configure doorbell to maximum slots Signed-off-by: Ben Goz Acked-by: Oded Gabbay Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index e4aeb74..7683d7f 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -3128,7 +3128,7 @@ static int gfx_v8_0_cp_compute_resume(struct amdgpu_device *adev) WREG32(mmCP_MEC_DOORBELL_RANGE_LOWER, AMDGPU_DOORBELL_KIQ << 2); WREG32(mmCP_MEC_DOORBELL_RANGE_UPPER, - AMDGPU_DOORBELL_MEC_RING7 << 2); + 0x7FFFF << 2); } tmp = RREG32(mmCP_HQD_PQ_DOORBELL_CONTROL); tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, -- cgit v0.10.2 From 161569deaa03cf3c00ed63352006193f250b0648 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?J=C3=A9r=C3=B4me=20Glisse?= Date: Fri, 19 Jun 2015 10:32:15 -0400 Subject: drm/radeon: compute ring fix hibernation (CI GPU family) v2. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit In order for hibernation to reliably work we need to cleanup more thoroughly the compute ring. Hibernation is different from suspend resume as when we resume from hibernation the hardware is first fully initialize by regular kernel then freeze callback happens (which correspond to a suspend inside the radeon kernel driver) and turn off each of the block. It turns out we were not cleanly shutting down the compute ring. This patch fix that. Hibernation and suspend to ram were tested (several times) on : Bonaire Hawaii Mullins Kaveri Kabini Changed since v1: - Factor the ring stop logic into a function taking ring as arg. Cc: stable@vger.kernel.org Signed-off-by: Jérôme Glisse Reviewed-by: Christian König Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/radeon/cik.c b/drivers/gpu/drm/radeon/cik.c index b0688b0..4ecf5ca 100644 --- a/drivers/gpu/drm/radeon/cik.c +++ b/drivers/gpu/drm/radeon/cik.c @@ -4604,6 +4604,31 @@ void cik_compute_set_wptr(struct radeon_device *rdev, WDOORBELL32(ring->doorbell_index, ring->wptr); } +static void cik_compute_stop(struct radeon_device *rdev, + struct radeon_ring *ring) +{ + u32 j, tmp; + + cik_srbm_select(rdev, ring->me, ring->pipe, ring->queue, 0); + /* Disable wptr polling. */ + tmp = RREG32(CP_PQ_WPTR_POLL_CNTL); + tmp &= ~WPTR_POLL_EN; + WREG32(CP_PQ_WPTR_POLL_CNTL, tmp); + /* Disable HQD. */ + if (RREG32(CP_HQD_ACTIVE) & 1) { + WREG32(CP_HQD_DEQUEUE_REQUEST, 1); + for (j = 0; j < rdev->usec_timeout; j++) { + if (!(RREG32(CP_HQD_ACTIVE) & 1)) + break; + udelay(1); + } + WREG32(CP_HQD_DEQUEUE_REQUEST, 0); + WREG32(CP_HQD_PQ_RPTR, 0); + WREG32(CP_HQD_PQ_WPTR, 0); + } + cik_srbm_select(rdev, 0, 0, 0, 0); +} + /** * cik_cp_compute_enable - enable/disable the compute CP MEs * @@ -4617,6 +4642,15 @@ static void cik_cp_compute_enable(struct radeon_device *rdev, bool enable) if (enable) WREG32(CP_MEC_CNTL, 0); else { + /* + * To make hibernation reliable we need to clear compute ring + * configuration before halting the compute ring. + */ + mutex_lock(&rdev->srbm_mutex); + cik_compute_stop(rdev,&rdev->ring[CAYMAN_RING_TYPE_CP1_INDEX]); + cik_compute_stop(rdev,&rdev->ring[CAYMAN_RING_TYPE_CP2_INDEX]); + mutex_unlock(&rdev->srbm_mutex); + WREG32(CP_MEC_CNTL, (MEC_ME1_HALT | MEC_ME2_HALT)); rdev->ring[CAYMAN_RING_TYPE_CP1_INDEX].ready = false; rdev->ring[CAYMAN_RING_TYPE_CP2_INDEX].ready = false; -- cgit v0.10.2 From 2ba8d1bb8f6b589037f7db1f01144fc80750e8f7 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?J=C3=A9r=C3=B4me=20Glisse?= Date: Fri, 19 Jun 2015 10:32:16 -0400 Subject: drm/radeon: SDMA fix hibernation (CI GPU family). MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit In order for hibernation to reliably work we need to properly turn off the SDMA block, sadly after numerous attemps i haven't not found proper sequence for clean and full shutdown. So simply reset both SDMA block, this makes hibernation works reliably on sea island GPU family (CI) Hibernation and suspend to ram were tested (several times) on : Bonaire Hawaii Mullins Kaveri Kabini Cc: stable@vger.kernel.org Signed-off-by: Jérôme Glisse Reviewed-by: Christian König Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/radeon/cik_sdma.c b/drivers/gpu/drm/radeon/cik_sdma.c index f86eb54..d16f2ee 100644 --- a/drivers/gpu/drm/radeon/cik_sdma.c +++ b/drivers/gpu/drm/radeon/cik_sdma.c @@ -268,6 +268,17 @@ static void cik_sdma_gfx_stop(struct radeon_device *rdev) } rdev->ring[R600_RING_TYPE_DMA_INDEX].ready = false; rdev->ring[CAYMAN_RING_TYPE_DMA1_INDEX].ready = false; + + /* FIXME use something else than big hammer but after few days can not + * seem to find good combination so reset SDMA blocks as it seems we + * do not shut them down properly. This fix hibernation and does not + * affect suspend to ram. + */ + WREG32(SRBM_SOFT_RESET, SOFT_RESET_SDMA | SOFT_RESET_SDMA1); + (void)RREG32(SRBM_SOFT_RESET); + udelay(50); + WREG32(SRBM_SOFT_RESET, 0); + (void)RREG32(SRBM_SOFT_RESET); } /** -- cgit v0.10.2 From 344c19f9558d69faec75869b33d558258429bcd9 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Tue, 2 Jun 2015 15:47:16 +0200 Subject: drm/amdgpu: simplify fence debugfs output a bit MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Christian König Reviewed-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index 5c9918d..8f0fc22 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -1023,7 +1023,7 @@ static int amdgpu_debugfs_fence_info(struct seq_file *m, void *data) amdgpu_fence_process(ring); - seq_printf(m, "--- ring %d ---\n", i); + seq_printf(m, "--- ring %d (%s) ---\n", i, ring->name); seq_printf(m, "Last signaled fence 0x%016llx\n", (unsigned long long)atomic64_read(&ring->fence_drv.last_seq)); seq_printf(m, "Last emitted 0x%016llx\n", @@ -1031,7 +1031,8 @@ static int amdgpu_debugfs_fence_info(struct seq_file *m, void *data) for (j = 0; j < AMDGPU_MAX_RINGS; ++j) { struct amdgpu_ring *other = adev->rings[j]; - if (i != j && other && other->fence_drv.initialized) + if (i != j && other && other->fence_drv.initialized && + ring->fence_drv.sync_seq[j]) seq_printf(m, "Last sync to ring %d 0x%016llx\n", j, ring->fence_drv.sync_seq[j]); } -- cgit v0.10.2 From 93e3e4385b69d8885e0dd64d1c5b8c351a151e91 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Tue, 9 Jun 2015 16:58:33 +0200 Subject: drm/amdgpu: add BO map/unmap trace point MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Christian König Acked-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h index b56dd64..9fe4f3c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h @@ -61,6 +61,54 @@ TRACE_EVENT(amdgpu_vm_grab_id, TP_printk("vmid=%u, ring=%u", __entry->vmid, __entry->ring) ); +TRACE_EVENT(amdgpu_vm_bo_map, + TP_PROTO(struct amdgpu_bo_va *bo_va, + struct amdgpu_bo_va_mapping *mapping), + TP_ARGS(bo_va, mapping), + TP_STRUCT__entry( + __field(struct amdgpu_bo *, bo) + __field(long, start) + __field(long, last) + __field(u64, offset) + __field(u32, flags) + ), + + TP_fast_assign( + __entry->bo = bo_va->bo; + __entry->start = mapping->it.start; + __entry->last = mapping->it.last; + __entry->offset = mapping->offset; + __entry->flags = mapping->flags; + ), + TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%08x", + __entry->bo, __entry->start, __entry->last, + __entry->offset, __entry->flags) +); + +TRACE_EVENT(amdgpu_vm_bo_unmap, + TP_PROTO(struct amdgpu_bo_va *bo_va, + struct amdgpu_bo_va_mapping *mapping), + TP_ARGS(bo_va, mapping), + TP_STRUCT__entry( + __field(struct amdgpu_bo *, bo) + __field(long, start) + __field(long, last) + __field(u64, offset) + __field(u32, flags) + ), + + TP_fast_assign( + __entry->bo = bo_va->bo; + __entry->start = mapping->it.start; + __entry->last = mapping->it.last; + __entry->offset = mapping->offset; + __entry->flags = mapping->flags; + ), + TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%08x", + __entry->bo, __entry->start, __entry->last, + __entry->offset, __entry->flags) +); + TRACE_EVENT(amdgpu_vm_bo_update, TP_PROTO(struct amdgpu_bo_va_mapping *mapping), TP_ARGS(mapping), diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 407882b..9a4e3b6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -1001,6 +1001,7 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev, list_add(&mapping->list, &bo_va->mappings); interval_tree_insert(&mapping->it, &vm->va); + trace_amdgpu_vm_bo_map(bo_va, mapping); bo_va->addr = 0; @@ -1058,6 +1059,7 @@ error_free: mutex_lock(&vm->mutex); list_del(&mapping->list); interval_tree_remove(&mapping->it, &vm->va); + trace_amdgpu_vm_bo_unmap(bo_va, mapping); kfree(mapping); error_unlock: @@ -1099,6 +1101,7 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev, mutex_lock(&vm->mutex); list_del(&mapping->list); interval_tree_remove(&mapping->it, &vm->va); + trace_amdgpu_vm_bo_unmap(bo_va, mapping); if (bo_va->addr) { /* clear the old address */ @@ -1139,6 +1142,7 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev, list_for_each_entry_safe(mapping, next, &bo_va->mappings, list) { list_del(&mapping->list); interval_tree_remove(&mapping->it, &vm->va); + trace_amdgpu_vm_bo_unmap(bo_va, mapping); if (bo_va->addr) list_add(&mapping->list, &vm->freed); else -- cgit v0.10.2 From ec74407ac047c150080758f2a4cec47393a6c0f8 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Wed, 10 Jun 2015 14:45:21 +0200 Subject: drm/amdgpu: add amdgpu_bo_list_set trace point MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Christian König Acked-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c index 36d34e0..f82a2dd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c @@ -30,6 +30,7 @@ #include #include "amdgpu.h" +#include "amdgpu_trace.h" static int amdgpu_bo_list_create(struct amdgpu_fpriv *fpriv, struct amdgpu_bo_list **result, @@ -124,6 +125,8 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev, gws_obj = entry->robj; if (entry->prefered_domains == AMDGPU_GEM_DOMAIN_OA) oa_obj = entry->robj; + + trace_amdgpu_bo_list_set(list, entry->robj); } for (i = 0; i < list->num_entries; ++i) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h index 9fe4f3c..ed20f44 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h @@ -169,6 +169,21 @@ TRACE_EVENT(amdgpu_vm_flush, __entry->pd_addr, __entry->ring, __entry->id) ); +TRACE_EVENT(amdgpu_bo_list_set, + TP_PROTO(struct amdgpu_bo_list *list, struct amdgpu_bo *bo), + TP_ARGS(list, bo), + TP_STRUCT__entry( + __field(struct amdgpu_bo_list *, list) + __field(struct amdgpu_bo *, bo) + ), + + TP_fast_assign( + __entry->list = list; + __entry->bo = bo; + ), + TP_printk("list=%p, bo=%p", __entry->list, __entry->bo) +); + DECLARE_EVENT_CLASS(amdgpu_fence_request, TP_PROTO(struct drm_device *dev, int ring, u32 seqno), -- cgit v0.10.2 From e30590e6e36992a6db28f03409286c164868939f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Wed, 10 Jun 2015 19:21:14 +0200 Subject: drm/amdgpu: print the bo_list in the CS trace point as well MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Christian König Acked-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h index ed20f44..961d726 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h @@ -30,19 +30,21 @@ TRACE_EVENT(amdgpu_cs, TP_PROTO(struct amdgpu_cs_parser *p, int i), TP_ARGS(p, i), TP_STRUCT__entry( + __field(struct amdgpu_bo_list *, bo_list) __field(u32, ring) __field(u32, dw) __field(u32, fences) ), TP_fast_assign( + __entry->bo_list = p->bo_list; __entry->ring = p->ibs[i].ring->idx; __entry->dw = p->ibs[i].length_dw; __entry->fences = amdgpu_fence_count_emitted( p->ibs[i].ring); ), - TP_printk("ring=%u, dw=%u, fences=%u", - __entry->ring, __entry->dw, + TP_printk("bo_list=%p, ring=%u, dw=%u, fences=%u", + __entry->bo_list, __entry->ring, __entry->dw, __entry->fences) ); -- cgit v0.10.2 From 332300b97e700e000739c8db782406b6fd8f092d Mon Sep 17 00:00:00 2001 From: "monk.liu" Date: Mon, 8 Jun 2015 14:48:15 +0800 Subject: drm/amdgpu: fix wrong type MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: monk.liu Reviewed-by: Christian König diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index 8f0fc22..f2d885c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -522,7 +522,8 @@ long amdgpu_fence_wait_seq_timeout(struct amdgpu_device *adev, u64 *target_seq, { uint64_t last_seq[AMDGPU_MAX_RINGS]; bool signaled; - int i, r; + int i; + long r; if (timeout == 0) { return amdgpu_fence_any_seq_signaled(adev, target_seq); -- cgit v0.10.2 From 68fdd3df79ee4bff4d63dab920b01d9ed5731115 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Tue, 16 Jun 2015 14:50:02 +0200 Subject: drm/amdgpu: silence invalid error message MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Christian König Reviewed-by: Leo Liu diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index 0ec2222..c3ea363 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -496,7 +496,7 @@ error_unreserve: error_free: drm_free_large(vm_bos); - if (r) + if (r && r != -ERESTARTSYS) DRM_ERROR("Couldn't update BO_VA (%d)\n", r); } -- cgit v0.10.2 From f1689ec1b0b1256d0e69653cd4aaeee44aafdf5c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Thu, 11 Jun 2015 20:56:18 +0200 Subject: drm/amdgpu: check VCE relocation buffer range MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit port of radeon commit 2fc5703abda201f138faf63bdca743d04dbf4b1a. Signed-off-by: Christian König Reviewed-by: Leo Liu diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 22866d1..963c4ba 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1622,6 +1622,7 @@ struct amdgpu_vce { unsigned fb_version; atomic_t handles[AMDGPU_MAX_VCE_HANDLES]; struct drm_file *filp[AMDGPU_MAX_VCE_HANDLES]; + uint32_t img_size[AMDGPU_MAX_VCE_HANDLES]; struct delayed_work idle_work; const struct firmware *fw; /* VCE firmware */ struct amdgpu_ring ring[AMDGPU_MAX_VCE_RINGS]; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index 1127a50..cb1bff7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -464,10 +464,12 @@ int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle, * @p: parser context * @lo: address of lower dword * @hi: address of higher dword + * @size: minimum size * * Patch relocation inside command stream with real buffer address */ -int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, int lo, int hi) +static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, + int lo, int hi, unsigned size) { struct amdgpu_bo_va_mapping *mapping; struct amdgpu_ib *ib = &p->ibs[ib_idx]; @@ -484,6 +486,13 @@ int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, int lo, int return -EINVAL; } + if ((addr + (uint64_t)size) > + ((uint64_t)mapping->it.last + 1) * AMDGPU_GPU_PAGE_SIZE) { + DRM_ERROR("BO to small for addr 0x%010Lx %d %d\n", + addr, lo, hi); + return -EINVAL; + } + addr -= ((uint64_t)mapping->it.start) * AMDGPU_GPU_PAGE_SIZE; addr += amdgpu_bo_gpu_offset(bo); @@ -494,6 +503,39 @@ int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, int lo, int } /** + * amdgpu_vce_validate_handle - validate stream handle + * + * @p: parser context + * @handle: handle to validate + * + * Validates the handle and return the found session index or -EINVAL + * we we don't have another free session index. + */ +static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p, + uint32_t handle) +{ + unsigned i; + + /* validate the handle */ + for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) { + if (atomic_read(&p->adev->vce.handles[i]) == handle) + return i; + } + + /* handle not found try to alloc a new one */ + for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) { + if (!atomic_cmpxchg(&p->adev->vce.handles[i], 0, handle)) { + p->adev->vce.filp[i] = p->filp; + p->adev->vce.img_size[i] = 0; + return i; + } + } + + DRM_ERROR("No more free VCE handles!\n"); + return -EINVAL; +} + +/** * amdgpu_vce_cs_parse - parse and validate the command stream * * @p: parser context @@ -501,10 +543,12 @@ int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, int lo, int */ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) { - uint32_t handle = 0; - bool destroy = false; - int i, r, idx = 0; struct amdgpu_ib *ib = &p->ibs[ib_idx]; + int session_idx = -1; + bool destroyed = false; + uint32_t tmp, handle = 0; + uint32_t *size = &tmp; + int i, r, idx = 0; amdgpu_vce_note_usage(p->adev); @@ -517,13 +561,29 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) return -EINVAL; } + if (destroyed) { + DRM_ERROR("No other command allowed after destroy!\n"); + return -EINVAL; + } + switch (cmd) { case 0x00000001: // session handle = amdgpu_get_ib_value(p, ib_idx, idx + 2); + session_idx = amdgpu_vce_validate_handle(p, handle); + if (session_idx < 0) + return session_idx; + size = &p->adev->vce.img_size[session_idx]; break; case 0x00000002: // task info + break; + case 0x01000001: // create + *size = amdgpu_get_ib_value(p, ib_idx, idx + 8) * + amdgpu_get_ib_value(p, ib_idx, idx + 10) * + 8 * 3 / 2; + break; + case 0x04000001: // config extension case 0x04000002: // pic control case 0x04000005: // rate control @@ -534,23 +594,39 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) break; case 0x03000001: // encode - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 10, idx + 9); + r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 10, idx + 9, + *size); if (r) return r; - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 12, idx + 11); + r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 12, idx + 11, + *size / 3); if (r) return r; break; case 0x02000001: // destroy - destroy = true; + destroyed = true; break; case 0x05000001: // context buffer + r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, + *size * 2); + if (r) + return r; + break; + case 0x05000004: // video bitstream buffer + tmp = amdgpu_get_ib_value(p, ib_idx, idx + 4); + r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, + tmp); + if (r) + return r; + break; + case 0x05000005: // feedback buffer - r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2); + r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, + 4096); if (r) return r; break; @@ -560,34 +636,21 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) return -EINVAL; } + if (session_idx == -1) { + DRM_ERROR("no session command at start of IB\n"); + return -EINVAL; + } + idx += len / 4; } - if (destroy) { + if (destroyed) { /* IB contains a destroy msg, free the handle */ for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) atomic_cmpxchg(&p->adev->vce.handles[i], handle, 0); - - return 0; - } - - /* create or encode, validate the handle */ - for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) { - if (atomic_read(&p->adev->vce.handles[i]) == handle) - return 0; - } - - /* handle not found try to alloc a new one */ - for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) { - if (!atomic_cmpxchg(&p->adev->vce.handles[i], 0, handle)) { - p->adev->vce.filp[i] = p->filp; - return 0; - } } - DRM_ERROR("No more free VCE handles!\n"); - - return -EINVAL; + return 0; } /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h index b6a9d09..7ccdb59 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h @@ -33,7 +33,6 @@ int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle, int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle, struct amdgpu_fence **fence); void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp); -int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, int lo, int hi); int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx); bool amdgpu_vce_ring_emit_semaphore(struct amdgpu_ring *ring, struct amdgpu_semaphore *semaphore, -- cgit v0.10.2 From 2f4b936869e6432a3361217da7e071d4c0c662f6 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Thu, 11 Jun 2015 21:33:55 +0200 Subject: drm/amdgpu: make VCE handle check more strict MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Port of radeon commit 29c63fe22a17c64e54016040cd882481bd45ee5a. Signed-off-by: Christian König Reviewed-by: Leo Liu diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index cb1bff7..70181c1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -507,19 +507,27 @@ static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, * * @p: parser context * @handle: handle to validate + * @allocated: allocated a new handle? * * Validates the handle and return the found session index or -EINVAL * we we don't have another free session index. */ static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p, - uint32_t handle) + uint32_t handle, bool *allocated) { unsigned i; + *allocated = false; + /* validate the handle */ for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) { - if (atomic_read(&p->adev->vce.handles[i]) == handle) + if (atomic_read(&p->adev->vce.handles[i]) == handle) { + if (p->adev->vce.filp[i] != p->filp) { + DRM_ERROR("VCE handle collision detected!\n"); + return -EINVAL; + } return i; + } } /* handle not found try to alloc a new one */ @@ -527,6 +535,7 @@ static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p, if (!atomic_cmpxchg(&p->adev->vce.handles[i], 0, handle)) { p->adev->vce.filp[i] = p->filp; p->adev->vce.img_size[i] = 0; + *allocated = true; return i; } } @@ -546,9 +555,11 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) struct amdgpu_ib *ib = &p->ibs[ib_idx]; int session_idx = -1; bool destroyed = false; + bool created = false; + bool allocated = false; uint32_t tmp, handle = 0; uint32_t *size = &tmp; - int i, r, idx = 0; + int i, r = 0, idx = 0; amdgpu_vce_note_usage(p->adev); @@ -558,18 +569,21 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) if ((len < 8) || (len & 3)) { DRM_ERROR("invalid VCE command length (%d)!\n", len); - return -EINVAL; + r = -EINVAL; + goto out; } if (destroyed) { DRM_ERROR("No other command allowed after destroy!\n"); - return -EINVAL; + r = -EINVAL; + goto out; } switch (cmd) { case 0x00000001: // session handle = amdgpu_get_ib_value(p, ib_idx, idx + 2); - session_idx = amdgpu_vce_validate_handle(p, handle); + session_idx = amdgpu_vce_validate_handle(p, handle, + &allocated); if (session_idx < 0) return session_idx; size = &p->adev->vce.img_size[session_idx]; @@ -579,6 +593,13 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) break; case 0x01000001: // create + created = true; + if (!allocated) { + DRM_ERROR("Handle already in use!\n"); + r = -EINVAL; + goto out; + } + *size = amdgpu_get_ib_value(p, ib_idx, idx + 8) * amdgpu_get_ib_value(p, ib_idx, idx + 10) * 8 * 3 / 2; @@ -597,12 +618,12 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 10, idx + 9, *size); if (r) - return r; + goto out; r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 12, idx + 11, *size / 3); if (r) - return r; + goto out; break; case 0x02000001: // destroy @@ -613,7 +634,7 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, *size * 2); if (r) - return r; + goto out; break; case 0x05000004: // video bitstream buffer @@ -621,36 +642,47 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, tmp); if (r) - return r; + goto out; break; case 0x05000005: // feedback buffer r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, 4096); if (r) - return r; + goto out; break; default: DRM_ERROR("invalid VCE command (0x%x)!\n", cmd); - return -EINVAL; + r = -EINVAL; + goto out; } if (session_idx == -1) { DRM_ERROR("no session command at start of IB\n"); - return -EINVAL; + r = -EINVAL; + goto out; } idx += len / 4; } - if (destroyed) { - /* IB contains a destroy msg, free the handle */ + if (allocated && !created) { + DRM_ERROR("New session without create command!\n"); + r = -ENOENT; + } + +out: + if ((!r && destroyed) || (r && allocated)) { + /* + * IB contains a destroy msg or we have allocated an + * handle and got an error, anyway free the handle + */ for (i = 0; i < AMDGPU_MAX_VCE_HANDLES; ++i) atomic_cmpxchg(&p->adev->vce.handles[i], handle, 0); } - return 0; + return r; } /** -- cgit v0.10.2 From dc78330a8e07285a26f165242c3a86744ef3ad20 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Fri, 12 Jun 2015 14:16:20 +0200 Subject: drm/amdgpu: check VCE feedback and bitstream index MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Christian König Reviewed-by: Leo Liu diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index 70181c1..d3ca730 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -469,20 +469,24 @@ int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle, * Patch relocation inside command stream with real buffer address */ static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, - int lo, int hi, unsigned size) + int lo, int hi, unsigned size, uint32_t index) { struct amdgpu_bo_va_mapping *mapping; struct amdgpu_ib *ib = &p->ibs[ib_idx]; struct amdgpu_bo *bo; uint64_t addr; + if (index == 0xffffffff) + index = 0; + addr = ((uint64_t)amdgpu_get_ib_value(p, ib_idx, lo)) | ((uint64_t)amdgpu_get_ib_value(p, ib_idx, hi)) << 32; + addr += ((uint64_t)size) * ((uint64_t)index); mapping = amdgpu_cs_find_mapping(p, addr, &bo); if (mapping == NULL) { - DRM_ERROR("Can't find BO for addr 0x%010Lx %d %d\n", - addr, lo, hi); + DRM_ERROR("Can't find BO for addr 0x%010Lx %d %d %d %d\n", + addr, lo, hi, size, index); return -EINVAL; } @@ -495,6 +499,7 @@ static int amdgpu_vce_cs_reloc(struct amdgpu_cs_parser *p, uint32_t ib_idx, addr -= ((uint64_t)mapping->it.start) * AMDGPU_GPU_PAGE_SIZE; addr += amdgpu_bo_gpu_offset(bo); + addr -= ((uint64_t)size) * ((uint64_t)index); ib->ptr[lo] = addr & 0xFFFFFFFF; ib->ptr[hi] = addr >> 32; @@ -553,6 +558,7 @@ static int amdgpu_vce_validate_handle(struct amdgpu_cs_parser *p, int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) { struct amdgpu_ib *ib = &p->ibs[ib_idx]; + unsigned fb_idx = 0, bs_idx = 0; int session_idx = -1; bool destroyed = false; bool created = false; @@ -590,6 +596,8 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) break; case 0x00000002: // task info + fb_idx = amdgpu_get_ib_value(p, ib_idx, idx + 6); + bs_idx = amdgpu_get_ib_value(p, ib_idx, idx + 7); break; case 0x01000001: // create @@ -616,12 +624,12 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) case 0x03000001: // encode r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 10, idx + 9, - *size); + *size, 0); if (r) goto out; r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 12, idx + 11, - *size / 3); + *size / 3, 0); if (r) goto out; break; @@ -632,7 +640,7 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) case 0x05000001: // context buffer r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, - *size * 2); + *size * 2, 0); if (r) goto out; break; @@ -640,14 +648,14 @@ int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx) case 0x05000004: // video bitstream buffer tmp = amdgpu_get_ib_value(p, ib_idx, idx + 4); r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, - tmp); + tmp, bs_idx); if (r) goto out; break; case 0x05000005: // feedback buffer r = amdgpu_vce_cs_reloc(p, ib_idx, idx + 3, idx + 2, - 4096); + 4096, fb_idx); if (r) goto out; break; -- cgit v0.10.2 From 845253e78a7871ae87cdc5c135a914c58d6d4f64 Mon Sep 17 00:00:00 2001 From: Sonny Jiang Date: Tue, 23 Jun 2015 11:59:55 -0400 Subject: drm/amdgpu: reset wptr at cp compute resume (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This patch is to resolve compute hang at resume time. v2: (agd5f) squash in second fix Signed-off-by: Sonny Jiang Reviewed-by: Christian König diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 7683d7f..7b683fb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -3144,6 +3144,12 @@ static int gfx_v8_0_cp_compute_resume(struct amdgpu_device *adev) WREG32(mmCP_HQD_PQ_DOORBELL_CONTROL, mqd->cp_hqd_pq_doorbell_control); + /* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */ + ring->wptr = 0; + mqd->cp_hqd_pq_wptr = ring->wptr; + WREG32(mmCP_HQD_PQ_WPTR, mqd->cp_hqd_pq_wptr); + mqd->cp_hqd_pq_rptr = RREG32(mmCP_HQD_PQ_RPTR); + /* set the vmid for the queue */ mqd->cp_hqd_vmid = 0; WREG32(mmCP_HQD_VMID, mqd->cp_hqd_vmid); -- cgit v0.10.2 From b8682ac253bdab6c15678f926bdc028b2e5dc0d2 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Mon, 22 Jun 2015 14:54:32 +0200 Subject: drm/amdgpu: fix crash on invalid CS IOCTL MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Christian König Reviewed-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index f09b2cb..7cd567b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -445,8 +445,9 @@ static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, bo for (i = 0; i < parser->nchunks; i++) drm_free_large(parser->chunks[i].kdata); kfree(parser->chunks); - for (i = 0; i < parser->num_ibs; i++) - amdgpu_ib_free(parser->adev, &parser->ibs[i]); + if (parser->ibs) + for (i = 0; i < parser->num_ibs; i++) + amdgpu_ib_free(parser->adev, &parser->ibs[i]); kfree(parser->ibs); if (parser->uf.bo) drm_gem_object_unreference_unlocked(&parser->uf.bo->gem_base); -- cgit v0.10.2 From 9a5e8fb1c8372ea4b58a76bbb82cb97683bb204b Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Tue, 23 Jun 2015 17:07:03 +0200 Subject: drm/amdgpu: add chunk id validity check MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Christian König Reviewed-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 7cd567b..86b78c7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -181,8 +181,6 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data) } p->chunks[i].chunk_id = user_chunk.chunk_id; p->chunks[i].length_dw = user_chunk.length_dw; - if (p->chunks[i].chunk_id == AMDGPU_CHUNK_ID_IB) - p->num_ibs++; size = p->chunks[i].length_dw; cdata = (void __user *)(unsigned long)user_chunk.chunk_data; @@ -199,7 +197,12 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data) goto out; } - if (p->chunks[i].chunk_id == AMDGPU_CHUNK_ID_FENCE) { + switch (p->chunks[i].chunk_id) { + case AMDGPU_CHUNK_ID_IB: + p->num_ibs++; + break; + + case AMDGPU_CHUNK_ID_FENCE: size = sizeof(struct drm_amdgpu_cs_chunk_fence); if (p->chunks[i].length_dw * sizeof(uint32_t) >= size) { uint32_t handle; @@ -221,6 +224,11 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data) r = -EINVAL; goto out; } + break; + + default: + r = -EINVAL; + goto out; } } -- cgit v0.10.2 From b13e22aeba06aea2acdae657e988c93c22080858 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Fri, 26 Jun 2015 12:31:29 +0200 Subject: drm/radeon: fix adding all VAs to the freed list on remove v2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We only should do so when the BO_VA was actually mapped. Otherwise we get a nice error message on the next CS. v2: It actually doesn't matter if it was invalidated or not, if it was mapped we need to clear the area where it was mapped. Signed-off-by: Christian König Tested-by: Michel Dänzer (v1) Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/radeon/radeon_vm.c b/drivers/gpu/drm/radeon/radeon_vm.c index 9739ded..618ab5e 100644 --- a/drivers/gpu/drm/radeon/radeon_vm.c +++ b/drivers/gpu/drm/radeon/radeon_vm.c @@ -1122,12 +1122,12 @@ void radeon_vm_bo_rmv(struct radeon_device *rdev, interval_tree_remove(&bo_va->it, &vm->va); spin_lock(&vm->status_lock); - if (list_empty(&bo_va->vm_status)) { + list_del(&bo_va->vm_status); + if (bo_va->it.start || bo_va->it.last) { bo_va->bo = radeon_bo_ref(bo_va->bo); list_add(&bo_va->vm_status, &vm->freed); } else { radeon_fence_unref(&bo_va->last_pt_update); - list_del(&bo_va->vm_status); kfree(bo_va); } spin_unlock(&vm->status_lock); -- cgit v0.10.2 From 5f0b34cc72aa627fd28230be3de469321e0af03e Mon Sep 17 00:00:00 2001 From: Maninder Singh Date: Fri, 26 Jun 2015 13:28:50 +0530 Subject: drm/amdgpu: use kzalloc for allocating one thing Use kzalloc rather than kcalloc(1.. for allocating one thing. Signed-off-by: Maninder Singh Reviewed-by: Vaneet Narang Reviewed-by: Christian Konig Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index d3706a4..dd3415d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -674,7 +674,7 @@ static int amdgpu_ttm_tt_populate(struct ttm_tt *ttm) return 0; if (gtt && gtt->userptr) { - ttm->sg = kcalloc(1, sizeof(struct sg_table), GFP_KERNEL); + ttm->sg = kzalloc(sizeof(struct sg_table), GFP_KERNEL); if (!ttm->sg) return -ENOMEM; -- cgit v0.10.2 From a08c1d516fb3eb302ff14a10eedda844e75c625f Mon Sep 17 00:00:00 2001 From: Maninder Singh Date: Fri, 26 Jun 2015 13:26:57 +0530 Subject: drm/amdgpu: remove unnecessary check before kfree kfree(NULL) is safe and this check is probably not required Signed-off-by: Maninder Singh Reviewed-by: Vaneet Narang Reviewed-by: Christian Konig Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index fec487d..a85cd08 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -1575,8 +1575,7 @@ void amdgpu_device_fini(struct amdgpu_device *adev) amdgpu_fence_driver_fini(adev); amdgpu_fbdev_fini(adev); r = amdgpu_fini(adev); - if (adev->ip_block_enabled) - kfree(adev->ip_block_enabled); + kfree(adev->ip_block_enabled); adev->ip_block_enabled = NULL; adev->accel_working = false; /* free i2c buses */ -- cgit v0.10.2 From d8d090b711bc56e095e59f9b536ed3c73c47b36d Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Fri, 26 Jun 2015 13:02:57 -0400 Subject: drm/amdgpu: allocate ip_block_enabled memory in common code MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Remove duplication across asic families and make it symmetric with the freeing of the code in amdgpu_device.c Reviewed-by: Christian König Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index a85cd08..e190975 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -1191,7 +1191,9 @@ static int amdgpu_early_init(struct amdgpu_device *adev) return -EINVAL; } - + adev->ip_block_enabled = kcalloc(adev->num_ip_blocks, sizeof(bool), GFP_KERNEL); + if (adev->ip_block_enabled == NULL) + return -ENOMEM; if (adev->ip_blocks == NULL) { DRM_ERROR("No IP blocks found!\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c index 5dab578..341c566 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik.c +++ b/drivers/gpu/drm/amd/amdgpu/cik.c @@ -2256,10 +2256,6 @@ int cik_set_ip_blocks(struct amdgpu_device *adev) return -EINVAL; } - adev->ip_block_enabled = kcalloc(adev->num_ip_blocks, sizeof(bool), GFP_KERNEL); - if (adev->ip_block_enabled == NULL) - return -ENOMEM; - return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index 90fc93c..fa5a444 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -1189,10 +1189,6 @@ int vi_set_ip_blocks(struct amdgpu_device *adev) return -EINVAL; } - adev->ip_block_enabled = kcalloc(adev->num_ip_blocks, sizeof(bool), GFP_KERNEL); - if (adev->ip_block_enabled == NULL) - return -ENOMEM; - return 0; } -- cgit v0.10.2 From 966c62fb40610e3cbda7bcd9a0193d8cbb3b77e3 Mon Sep 17 00:00:00 2001 From: Sonny Jiang Date: Fri, 26 Jun 2015 12:26:45 -0400 Subject: drm/amdgpu: correct define SMU_EnabledFeatureScoreboard_SclkDpmOn Signed-off-by: Sonny Jiang Reviewed-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.h b/drivers/gpu/drm/amd/amdgpu/cz_dpm.h index 782a741..99e1afc 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.h +++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.h @@ -46,7 +46,7 @@ /* Do not change the following, it is also defined in SMU8.h */ #define SMU_EnabledFeatureScoreboard_AcpDpmOn 0x00000001 -#define SMU_EnabledFeatureScoreboard_SclkDpmOn 0x00100000 +#define SMU_EnabledFeatureScoreboard_SclkDpmOn 0x00200000 #define SMU_EnabledFeatureScoreboard_UvdDpmOn 0x00800000 #define SMU_EnabledFeatureScoreboard_VceDpmOn 0x01000000 -- cgit v0.10.2 From 9dcabece15d16da48bb7c7071306bf6158b42ba3 Mon Sep 17 00:00:00 2001 From: Sonny Jiang Date: Fri, 26 Jun 2015 12:31:46 -0400 Subject: drm/amdgpu: disable enable_nb_ps_policy temporarily Fixes a hang on resume. Signed-off-by: Sonny Jiang Reviewed-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c index e4936a4..f75a31d 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c +++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c @@ -425,7 +425,7 @@ static int cz_dpm_init(struct amdgpu_device *adev) pi->mgcg_cgtt_local1 = 0x0; pi->clock_slow_down_step = 25000; pi->skip_clock_slow_down = 1; - pi->enable_nb_ps_policy = 1; + pi->enable_nb_ps_policy = 0; pi->caps_power_containment = true; pi->caps_cac = true; pi->didt_enabled = false; -- cgit v0.10.2 From ac9134906b3f5c2b45dc80dab0fee792bd516d52 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Mon, 29 Jun 2015 11:09:11 -0400 Subject: Revert "drm/radeon: dont switch vt on suspend" This reverts commit b9729b17a414f99c61f4db9ac9f9ed987fa0cbfe. This seems to break the cursor on resume for lots of systems. Cc: stable@vger.kernel.org diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c index aeb6767..634793e 100644 --- a/drivers/gpu/drm/radeon/radeon_fb.c +++ b/drivers/gpu/drm/radeon/radeon_fb.c @@ -257,7 +257,6 @@ static int radeonfb_create(struct drm_fb_helper *helper, } info->par = rfbdev; - info->skip_vt_switch = true; ret = radeon_framebuffer_init(rdev->ddev, &rfbdev->rfb, &mode_cmd, gobj); if (ret) { -- cgit v0.10.2 From 7cebc728174424d67df91dfb14f8b6dc13bed993 Mon Sep 17 00:00:00 2001 From: Alexander Kuleshov Date: Sat, 27 Jun 2015 13:16:05 +0600 Subject: gpu/drm/amdgpu: Fix build when CONFIG_DEBUG_FS is not set MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit If the CONFIG_DEBUG_FS is not selected, compilation of the drivers/gpu/drm/amd/amdgpu/amdgpu_device.c provides two warnings that amdgpu_debugfs_regs_init and amdgpu_debugfs_regs_cleanup are used but never defined. And as result: ERROR: "amdgpu_debugfs_regs_cleanup" [drivers/gpu/drm/amd/amdgpu/amdgpu.ko] undefined! ERROR: "amdgpu_debugfs_regs_init" [drivers/gpu/drm/amd/amdgpu/amdgpu.ko] undefined! ^ Reviewed-by: Christian König Signed-off-by: Alexander Kuleshov Signed-off-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index e190975..ba46be3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -2001,4 +2001,10 @@ int amdgpu_debugfs_init(struct drm_minor *minor) void amdgpu_debugfs_cleanup(struct drm_minor *minor) { } +#else +static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev) +{ + return 0; +} +static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev) { } #endif -- cgit v0.10.2 From 03507c4f2f63d8d98c2455cf4d192589fac553c7 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Fri, 19 Jun 2015 17:00:19 +0200 Subject: drm/amdgpu: recreate fence from user seq MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit And use common fence infrastructure for the wait. Signed-off-by: Christian König Reviewed-by: Alex Deucher Reviewed-by: Chunming Zhou diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 963c4ba..0165783 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -425,6 +425,8 @@ int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring, unsigned irq_type); int amdgpu_fence_emit(struct amdgpu_ring *ring, void *owner, struct amdgpu_fence **fence); +int amdgpu_fence_recreate(struct amdgpu_ring *ring, void *owner, + uint64_t seq, struct amdgpu_fence **fence); void amdgpu_fence_process(struct amdgpu_ring *ring); int amdgpu_fence_wait_next(struct amdgpu_ring *ring); int amdgpu_fence_wait_empty(struct amdgpu_ring *ring); @@ -435,9 +437,6 @@ int amdgpu_fence_wait(struct amdgpu_fence *fence, bool interruptible); int amdgpu_fence_wait_any(struct amdgpu_device *adev, struct amdgpu_fence **fences, bool intr); -long amdgpu_fence_wait_seq_timeout(struct amdgpu_device *adev, - u64 *target_seq, bool intr, - long timeout); struct amdgpu_fence *amdgpu_fence_ref(struct amdgpu_fence *fence); void amdgpu_fence_unref(struct amdgpu_fence **fence); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 86b78c7..84ba1d1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -739,9 +739,9 @@ int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data, { union drm_amdgpu_wait_cs *wait = data; struct amdgpu_device *adev = dev->dev_private; - uint64_t seq[AMDGPU_MAX_RINGS] = {0}; - struct amdgpu_ring *ring = NULL; unsigned long timeout = amdgpu_gem_timeout(wait->in.timeout); + struct amdgpu_fence *fence = NULL; + struct amdgpu_ring *ring = NULL; struct amdgpu_ctx *ctx; long r; @@ -754,9 +754,12 @@ int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data, if (r) return r; - seq[ring->idx] = wait->in.handle; + r = amdgpu_fence_recreate(ring, filp, wait->in.handle, &fence); + if (r) + return r; - r = amdgpu_fence_wait_seq_timeout(adev, seq, true, timeout); + r = fence_wait_timeout(&fence->base, true, timeout); + amdgpu_fence_unref(&fence); amdgpu_ctx_put(ctx); if (r < 0) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index f2d885c..a7189a1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -136,6 +136,38 @@ int amdgpu_fence_emit(struct amdgpu_ring *ring, void *owner, } /** + * amdgpu_fence_recreate - recreate a fence from an user fence + * + * @ring: ring the fence is associated with + * @owner: creator of the fence + * @seq: user fence sequence number + * @fence: resulting amdgpu fence object + * + * Recreates a fence command from the user fence sequence number (all asics). + * Returns 0 on success, -ENOMEM on failure. + */ +int amdgpu_fence_recreate(struct amdgpu_ring *ring, void *owner, + uint64_t seq, struct amdgpu_fence **fence) +{ + struct amdgpu_device *adev = ring->adev; + + if (seq > ring->fence_drv.sync_seq[ring->idx]) + return -EINVAL; + + *fence = kmalloc(sizeof(struct amdgpu_fence), GFP_KERNEL); + if ((*fence) == NULL) + return -ENOMEM; + + (*fence)->seq = seq; + (*fence)->ring = ring; + (*fence)->owner = owner; + fence_init(&(*fence)->base, &amdgpu_fence_ops, + &adev->fence_queue.lock, adev->fence_context + ring->idx, + (*fence)->seq); + return 0; +} + +/** * amdgpu_fence_check_signaled - callback from fence_queue * * this function is called with fence_queue lock held, which is also used @@ -517,8 +549,9 @@ static bool amdgpu_fence_any_seq_signaled(struct amdgpu_device *adev, u64 *seq) * the wait timeout, or an error for all other cases. * -EDEADLK is returned when a GPU lockup has been detected. */ -long amdgpu_fence_wait_seq_timeout(struct amdgpu_device *adev, u64 *target_seq, - bool intr, long timeout) +static long amdgpu_fence_wait_seq_timeout(struct amdgpu_device *adev, + u64 *target_seq, bool intr, + long timeout) { uint64_t last_seq[AMDGPU_MAX_RINGS]; bool signaled; -- cgit v0.10.2 From 2b48d323b26c37555df3447e11ab9e962eccdc26 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Fri, 19 Jun 2015 17:31:29 +0200 Subject: drm/amdgpu: add optional dependencies to the CS IOCTL v2 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit v2: remove unrelated whitespace change, fix C comment Signed-off-by: Christian König Reviewed-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 84ba1d1..d63135b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -226,6 +226,9 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data) } break; + case AMDGPU_CHUNK_ID_DEPENDENCIES: + break; + default: r = -EINVAL; goto out; @@ -663,6 +666,55 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev, return 0; } +static int amdgpu_cs_dependencies(struct amdgpu_device *adev, + struct amdgpu_cs_parser *p) +{ + struct amdgpu_ib *ib; + int i, j, r; + + if (!p->num_ibs) + return 0; + + /* Add dependencies to first IB */ + ib = &p->ibs[0]; + for (i = 0; i < p->nchunks; ++i) { + struct drm_amdgpu_cs_chunk_dep *deps; + struct amdgpu_cs_chunk *chunk; + unsigned num_deps; + + chunk = &p->chunks[i]; + + if (chunk->chunk_id != AMDGPU_CHUNK_ID_DEPENDENCIES) + continue; + + deps = (struct drm_amdgpu_cs_chunk_dep *)chunk->kdata; + num_deps = chunk->length_dw * 4 / + sizeof(struct drm_amdgpu_cs_chunk_dep); + + for (j = 0; j < num_deps; ++j) { + struct amdgpu_fence *fence; + struct amdgpu_ring *ring; + + r = amdgpu_cs_get_ring(adev, deps[j].ip_type, + deps[j].ip_instance, + deps[j].ring, &ring); + if (r) + return r; + + r = amdgpu_fence_recreate(ring, p->filp, + deps[j].handle, + &fence); + if (r) + return r; + + amdgpu_sync_fence(&ib->sync, fence); + amdgpu_fence_unref(&fence); + } + } + + return 0; +} + int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) { struct amdgpu_device *adev = dev->dev_private; @@ -697,11 +749,16 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) else DRM_ERROR("Failed to process the buffer list %d!\n", r); } - } else { + } + + if (!r) { reserved_buffers = true; r = amdgpu_cs_ib_fill(adev, &parser); } + if (!r) + r = amdgpu_cs_dependencies(adev, &parser); + if (r) { amdgpu_cs_parser_fini(&parser, r, reserved_buffers); up_read(&adev->exclusive_lock); diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h index d3f4832..8edf104 100644 --- a/include/uapi/drm/amdgpu_drm.h +++ b/include/uapi/drm/amdgpu_drm.h @@ -348,6 +348,7 @@ struct drm_amdgpu_gem_va { #define AMDGPU_CHUNK_ID_IB 0x01 #define AMDGPU_CHUNK_ID_FENCE 0x02 +#define AMDGPU_CHUNK_ID_DEPENDENCIES 0x03 struct drm_amdgpu_cs_chunk { uint32_t chunk_id; @@ -399,6 +400,14 @@ struct drm_amdgpu_cs_chunk_ib { uint32_t ring; }; +struct drm_amdgpu_cs_chunk_dep { + uint32_t ip_type; + uint32_t ip_instance; + uint32_t ring; + uint32_t ctx_id; + uint64_t handle; +}; + struct drm_amdgpu_cs_chunk_fence { uint32_t handle; uint32_t offset; -- cgit v0.10.2 From fc220f6580be80c36dfba9964b2fd71ceb1d3d97 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Christian=20K=C3=B6nig?= Date: Mon, 29 Jun 2015 17:12:20 +0200 Subject: drm/amdgpu: add flag to delay VM updates MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Christian König Reviewed-by: Alex Deucher diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index c3ea363..975edb1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -525,8 +525,8 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data, return -EINVAL; } - invalid_flags = ~(AMDGPU_VM_PAGE_READABLE | AMDGPU_VM_PAGE_WRITEABLE | - AMDGPU_VM_PAGE_EXECUTABLE); + invalid_flags = ~(AMDGPU_VM_DELAY_UPDATE | AMDGPU_VM_PAGE_READABLE | + AMDGPU_VM_PAGE_WRITEABLE | AMDGPU_VM_PAGE_EXECUTABLE); if ((args->flags & invalid_flags)) { dev_err(&dev->pdev->dev, "invalid flags 0x%08X vs 0x%08X\n", args->flags, invalid_flags); @@ -579,7 +579,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data, break; } - if (!r) + if (!r && !(args->flags & AMDGPU_VM_DELAY_UPDATE)) amdgpu_gem_va_update_vm(adev, bo_va); drm_gem_object_unreference_unlocked(gobj); diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h index 8edf104..b6fce90 100644 --- a/include/uapi/drm/amdgpu_drm.h +++ b/include/uapi/drm/amdgpu_drm.h @@ -313,6 +313,9 @@ struct drm_amdgpu_gem_op { #define AMDGPU_VA_OP_MAP 1 #define AMDGPU_VA_OP_UNMAP 2 +/* Delay the page table update till the next CS */ +#define AMDGPU_VM_DELAY_UPDATE (1 << 0) + /* Mapping flags */ /* readable mapping */ #define AMDGPU_VM_PAGE_READABLE (1 << 1) -- cgit v0.10.2 From 479e9a95120aaae0bf0d3e0b5b26b36ac4a347b6 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Tue, 30 Jun 2015 09:30:01 -0400 Subject: drm/radeon: only check the sink type on DP connectors Avoids a crash on pre-DP asics that support HDMI. Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org diff --git a/drivers/gpu/drm/radeon/radeon_audio.c b/drivers/gpu/drm/radeon/radeon_audio.c index c892152..fa719c5 100644 --- a/drivers/gpu/drm/radeon/radeon_audio.c +++ b/drivers/gpu/drm/radeon/radeon_audio.c @@ -469,22 +469,22 @@ void radeon_audio_detect(struct drm_connector *connector, dig = radeon_encoder->enc_priv; if (status == connector_status_connected) { - struct radeon_connector *radeon_connector; - int sink_type; - if (!drm_detect_monitor_audio(radeon_connector_edid(connector))) { radeon_encoder->audio = NULL; return; } - radeon_connector = to_radeon_connector(connector); - sink_type = radeon_dp_getsinktype(radeon_connector); + if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); - if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort && - sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) - radeon_encoder->audio = rdev->audio.dp_funcs; - else + if (radeon_dp_getsinktype(radeon_connector) == + CONNECTOR_OBJECT_ID_DISPLAYPORT) + radeon_encoder->audio = rdev->audio.dp_funcs; + else + radeon_encoder->audio = rdev->audio.hdmi_funcs; + } else { radeon_encoder->audio = rdev->audio.hdmi_funcs; + } dig->afmt->pin = radeon_audio_get_pin(connector->encoder); radeon_audio_enable(rdev, dig->afmt->pin, 0xf); -- cgit v0.10.2