diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c | 824 |
1 files changed, 515 insertions, 309 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c index 45be0af2570b..217040044987 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c @@ -43,18 +43,80 @@ #include "sdma_common.h" #include "sdma_v6_0.h" #include "v11_structs.h" +#include "mes_userqueue.h" +#include "amdgpu_userq_fence.h" MODULE_FIRMWARE("amdgpu/sdma_6_0_0.bin"); MODULE_FIRMWARE("amdgpu/sdma_6_0_1.bin"); MODULE_FIRMWARE("amdgpu/sdma_6_0_2.bin"); MODULE_FIRMWARE("amdgpu/sdma_6_0_3.bin"); MODULE_FIRMWARE("amdgpu/sdma_6_1_0.bin"); +MODULE_FIRMWARE("amdgpu/sdma_6_1_1.bin"); +MODULE_FIRMWARE("amdgpu/sdma_6_1_2.bin"); +MODULE_FIRMWARE("amdgpu/sdma_6_1_3.bin"); #define SDMA1_REG_OFFSET 0x600 #define SDMA0_HYP_DEC_REG_START 0x5880 #define SDMA0_HYP_DEC_REG_END 0x589a #define SDMA1_HYP_DEC_REG_OFFSET 0x20 +static const struct amdgpu_hwip_reg_entry sdma_reg_list_6_0[] = { + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS1_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS2_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS3_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS4_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS5_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_STATUS6_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UCODE_CHECKSUM), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_RB_RPTR_FETCH_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_RB_RPTR_FETCH), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_RD_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_WR_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_RD_XNACK0), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_RD_XNACK1), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_WR_XNACK0), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_UTCL1_WR_XNACK1), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_CNTL), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_RPTR), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_RPTR_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_WPTR), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_RB_WPTR_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_OFFSET), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_BASE_LO), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_BASE_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_CNTL), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_RPTR), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_IB_SUB_REMAIN), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE0_DUMMY_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE_STATUS0), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_CNTL), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_RPTR), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_RPTR_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_WPTR), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_RB_WPTR_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_OFFSET), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_BASE_LO), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_BASE_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_RPTR), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_IB_SUB_REMAIN), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE1_DUMMY_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_CNTL), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_RPTR), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_RPTR_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_WPTR), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_RB_WPTR_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_OFFSET), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_BASE_LO), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_BASE_HI), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_RPTR), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_IB_SUB_REMAIN), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_QUEUE2_DUMMY_REG), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_INT_STATUS), + SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS2), + SOC15_REG_ENTRY_STR(GC, 0, regSDMA0_CHICKEN_BITS), +}; + static void sdma_v6_0_set_ring_funcs(struct amdgpu_device *adev); static void sdma_v6_0_set_buffer_funcs(struct amdgpu_device *adev); static void sdma_v6_0_set_vm_pte_funcs(struct amdgpu_device *adev); @@ -79,35 +141,23 @@ static u32 sdma_v6_0_get_reg_offset(struct amdgpu_device *adev, u32 instance, u3 return base + internal_offset; } -static unsigned sdma_v6_0_ring_init_cond_exec(struct amdgpu_ring *ring) +static unsigned sdma_v6_0_ring_init_cond_exec(struct amdgpu_ring *ring, + uint64_t addr) { unsigned ret; amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COND_EXE)); - amdgpu_ring_write(ring, lower_32_bits(ring->cond_exe_gpu_addr)); - amdgpu_ring_write(ring, upper_32_bits(ring->cond_exe_gpu_addr)); + amdgpu_ring_write(ring, lower_32_bits(addr)); + amdgpu_ring_write(ring, upper_32_bits(addr)); amdgpu_ring_write(ring, 1); - ret = ring->wptr & ring->buf_mask;/* this is the offset we need patch later */ - amdgpu_ring_write(ring, 0x55aa55aa);/* insert dummy here and patch it later */ + /* this is the offset we need patch later */ + ret = ring->wptr & ring->buf_mask; + /* insert dummy here and patch it later */ + amdgpu_ring_write(ring, 0); return ret; } -static void sdma_v6_0_ring_patch_cond_exec(struct amdgpu_ring *ring, - unsigned offset) -{ - unsigned cur; - - BUG_ON(offset > ring->buf_mask); - BUG_ON(ring->ring[offset] != 0x55aa55aa); - - cur = (ring->wptr - 1) & ring->buf_mask; - if (cur > offset) - ring->ring[offset] = cur - offset; - else - ring->ring[offset] = (ring->buf_mask + 1) - offset + cur; -} - /** * sdma_v6_0_ring_get_rptr - get the current read pointer * @@ -156,68 +206,35 @@ static uint64_t sdma_v6_0_ring_get_wptr(struct amdgpu_ring *ring) static void sdma_v6_0_ring_set_wptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - uint32_t *wptr_saved; - uint32_t *is_queue_unmap; - uint64_t aggregated_db_index; - uint32_t mqd_size = adev->mqds[AMDGPU_HW_IP_DMA].mqd_size; - - DRM_DEBUG("Setting write pointer\n"); - - if (ring->is_mes_queue) { - wptr_saved = (uint32_t *)(ring->mqd_ptr + mqd_size); - is_queue_unmap = (uint32_t *)(ring->mqd_ptr + mqd_size + - sizeof(uint32_t)); - aggregated_db_index = - amdgpu_mes_get_aggregated_doorbell_index(adev, - ring->hw_prio); + if (ring->use_doorbell) { + DRM_DEBUG("Using doorbell -- " + "wptr_offs == 0x%08x " + "lower_32_bits(ring->wptr) << 2 == 0x%08x " + "upper_32_bits(ring->wptr) << 2 == 0x%08x\n", + ring->wptr_offs, + lower_32_bits(ring->wptr << 2), + upper_32_bits(ring->wptr << 2)); + /* XXX check if swapping is necessary on BE */ atomic64_set((atomic64_t *)ring->wptr_cpu_addr, ring->wptr << 2); - *wptr_saved = ring->wptr << 2; - if (*is_queue_unmap) { - WDOORBELL64(aggregated_db_index, ring->wptr << 2); - DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n", - ring->doorbell_index, ring->wptr << 2); - WDOORBELL64(ring->doorbell_index, ring->wptr << 2); - } else { - DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n", - ring->doorbell_index, ring->wptr << 2); - WDOORBELL64(ring->doorbell_index, ring->wptr << 2); - - if (*is_queue_unmap) - WDOORBELL64(aggregated_db_index, - ring->wptr << 2); - } + DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n", + ring->doorbell_index, ring->wptr << 2); + WDOORBELL64(ring->doorbell_index, ring->wptr << 2); } else { - if (ring->use_doorbell) { - DRM_DEBUG("Using doorbell -- " - "wptr_offs == 0x%08x " - "lower_32_bits(ring->wptr) << 2 == 0x%08x " - "upper_32_bits(ring->wptr) << 2 == 0x%08x\n", - ring->wptr_offs, - lower_32_bits(ring->wptr << 2), - upper_32_bits(ring->wptr << 2)); - /* XXX check if swapping is necessary on BE */ - atomic64_set((atomic64_t *)ring->wptr_cpu_addr, - ring->wptr << 2); - DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n", - ring->doorbell_index, ring->wptr << 2); - WDOORBELL64(ring->doorbell_index, ring->wptr << 2); - } else { - DRM_DEBUG("Not using doorbell -- " - "regSDMA%i_GFX_RB_WPTR == 0x%08x " - "regSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n", - ring->me, - lower_32_bits(ring->wptr << 2), - ring->me, - upper_32_bits(ring->wptr << 2)); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, - ring->me, regSDMA0_QUEUE0_RB_WPTR), - lower_32_bits(ring->wptr << 2)); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, - ring->me, regSDMA0_QUEUE0_RB_WPTR_HI), - upper_32_bits(ring->wptr << 2)); - } + DRM_DEBUG("Not using doorbell -- " + "regSDMA%i_GFX_RB_WPTR == 0x%08x " + "regSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n", + ring->me, + lower_32_bits(ring->wptr << 2), + ring->me, + upper_32_bits(ring->wptr << 2)); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, + ring->me, regSDMA0_QUEUE0_RB_WPTR), + lower_32_bits(ring->wptr << 2)); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, + ring->me, regSDMA0_QUEUE0_RB_WPTR_HI), + upper_32_bits(ring->wptr << 2)); } } @@ -234,7 +251,7 @@ static void sdma_v6_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) amdgpu_ring_write(ring, ring->funcs->nop); } -/** +/* * sdma_v6_0_ring_emit_ib - Schedule an IB on the DMA engine * * @ring: amdgpu ring pointer @@ -361,11 +378,9 @@ static void sdma_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 se } if (flags & AMDGPU_FENCE_FLAG_INT) { - uint32_t ctx = ring->is_mes_queue ? - (ring->hw_queue_id | AMDGPU_FENCE_MES_QUEUE_FLAG) : 0; /* generate an interrupt */ amdgpu_ring_write(ring, SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_TRAP)); - amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(ctx)); + amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0)); } } @@ -381,8 +396,6 @@ static void sdma_v6_0_gfx_stop(struct amdgpu_device *adev) u32 rb_cntl, ib_cntl; int i; - amdgpu_sdma_unset_buffer_funcs_helper(adev); - for (i = 0; i < adev->sdma.num_instances; i++) { rb_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL)); rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_ENABLE, 0); @@ -457,14 +470,16 @@ static void sdma_v6_0_enable(struct amdgpu_device *adev, bool enable) } /** - * sdma_v6_0_gfx_resume - setup and start the async dma engines + * sdma_v6_0_gfx_resume_instance - start/restart a certain sdma engine * * @adev: amdgpu_device pointer + * @i: instance + * @restore: used to restore wptr when restart * - * Set up the gfx DMA ring buffers and enable them. - * Returns 0 for success, error for failure. + * Set up the gfx DMA ring buffers and enable them. On restart, we will restore wptr and rptr. + * Return 0 for success. */ -static int sdma_v6_0_gfx_resume(struct amdgpu_device *adev) +static int sdma_v6_0_gfx_resume_instance(struct amdgpu_device *adev, int i, bool restore) { struct amdgpu_ring *ring; u32 rb_cntl, ib_cntl; @@ -473,130 +488,154 @@ static int sdma_v6_0_gfx_resume(struct amdgpu_device *adev) u32 doorbell_offset; u32 temp; u64 wptr_gpu_addr; - int i, r; - for (i = 0; i < adev->sdma.num_instances; i++) { - ring = &adev->sdma.instance[i].ring; + ring = &adev->sdma.instance[i].ring; + if (!amdgpu_sriov_vf(adev)) + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0); - if (!amdgpu_sriov_vf(adev)) - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0); - - /* Set ring buffer size in dwords */ - rb_bufsz = order_base_2(ring->ring_size / 4); - rb_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL)); - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_SIZE, rb_bufsz); + /* Set ring buffer size in dwords */ + rb_bufsz = order_base_2(ring->ring_size / 4); + rb_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL)); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_SIZE, rb_bufsz); #ifdef __BIG_ENDIAN - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_SWAP_ENABLE, 1); - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, - RPTR_WRITEBACK_SWAP_ENABLE, 1); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_SWAP_ENABLE, 1); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, + RPTR_WRITEBACK_SWAP_ENABLE, 1); #endif - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_PRIV, 1); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl); - - /* Initialize the ring buffer's read and write pointers */ + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_PRIV, 1); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl); + + /* Initialize the ring buffer's read and write pointers */ + if (restore) { + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR), lower_32_bits(ring->wptr << 2)); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_HI), upper_32_bits(ring->wptr << 2)); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), lower_32_bits(ring->wptr << 2)); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), upper_32_bits(ring->wptr << 2)); + } else { WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR), 0); WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_HI), 0); WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), 0); WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), 0); + } + /* setup the wptr shadow polling */ + wptr_gpu_addr = ring->wptr_gpu_addr; + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_POLL_ADDR_LO), + lower_32_bits(wptr_gpu_addr)); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_POLL_ADDR_HI), + upper_32_bits(wptr_gpu_addr)); + + /* set the wb address whether it's enabled or not */ + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_ADDR_HI), + upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_ADDR_LO), + lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC); + + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, WPTR_POLL_ENABLE, 0); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, F32_WPTR_POLL_ENABLE, 1); + + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_BASE), ring->gpu_addr >> 8); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_BASE_HI), ring->gpu_addr >> 40); + + if (!restore) + ring->wptr = 0; - /* setup the wptr shadow polling */ - wptr_gpu_addr = ring->wptr_gpu_addr; - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_POLL_ADDR_LO), - lower_32_bits(wptr_gpu_addr)); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_POLL_ADDR_HI), - upper_32_bits(wptr_gpu_addr)); - - /* set the wb address whether it's enabled or not */ - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_ADDR_HI), - upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_RPTR_ADDR_LO), - lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC); - - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1); - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, WPTR_POLL_ENABLE, 0); - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, F32_WPTR_POLL_ENABLE, 1); + /* before programing wptr to a less value, need set minor_ptr_update first */ + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_MINOR_PTR_UPDATE), 1); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_BASE), ring->gpu_addr >> 8); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_BASE_HI), ring->gpu_addr >> 40); + if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */ + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), lower_32_bits(ring->wptr) << 2); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2); + } - ring->wptr = 0; + doorbell = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL)); + doorbell_offset = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL_OFFSET)); - /* before programing wptr to a less value, need set minor_ptr_update first */ - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_MINOR_PTR_UPDATE), 1); + if (ring->use_doorbell) { + doorbell = REG_SET_FIELD(doorbell, SDMA0_QUEUE0_DOORBELL, ENABLE, 1); + doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_QUEUE0_DOORBELL_OFFSET, + OFFSET, ring->doorbell_index); + } else { + doorbell = REG_SET_FIELD(doorbell, SDMA0_QUEUE0_DOORBELL, ENABLE, 0); + } + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL), doorbell); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL_OFFSET), doorbell_offset); - if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */ - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR), lower_32_bits(ring->wptr) << 2); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2); - } + if (i == 0) + adev->nbio.funcs->sdma_doorbell_range(adev, i, ring->use_doorbell, + ring->doorbell_index, + adev->doorbell_index.sdma_doorbell_range * adev->sdma.num_instances); - doorbell = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL)); - doorbell_offset = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL_OFFSET)); + if (amdgpu_sriov_vf(adev)) + sdma_v6_0_ring_set_wptr(ring); + + /* set minor_ptr_update to 0 after wptr programed */ + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_MINOR_PTR_UPDATE), 0); + + /* Set up sdma hang watchdog */ + temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_WATCHDOG_CNTL)); + /* 100ms per unit */ + temp = REG_SET_FIELD(temp, SDMA0_WATCHDOG_CNTL, QUEUE_HANG_COUNT, + max(adev->usec_timeout/100000, 1)); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_WATCHDOG_CNTL), temp); + + /* Set up RESP_MODE to non-copy addresses */ + temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_CNTL)); + temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3); + temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, REDO_DELAY, 9); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_CNTL), temp); + + /* program default cache read and write policy */ + temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_PAGE)); + /* clean read policy and write policy bits */ + temp &= 0xFF0FFF; + temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) | + (CACHE_WRITE_POLICY_L2__DEFAULT << 14) | + SDMA0_UTCL1_PAGE__LLC_NOALLOC_MASK); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_PAGE), temp); - if (ring->use_doorbell) { - doorbell = REG_SET_FIELD(doorbell, SDMA0_QUEUE0_DOORBELL, ENABLE, 1); - doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_QUEUE0_DOORBELL_OFFSET, - OFFSET, ring->doorbell_index); - } else { - doorbell = REG_SET_FIELD(doorbell, SDMA0_QUEUE0_DOORBELL, ENABLE, 0); - } - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL), doorbell); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_DOORBELL_OFFSET), doorbell_offset); - - if (i == 0) - adev->nbio.funcs->sdma_doorbell_range(adev, i, ring->use_doorbell, - ring->doorbell_index, - adev->doorbell_index.sdma_doorbell_range * adev->sdma.num_instances); - - if (amdgpu_sriov_vf(adev)) - sdma_v6_0_ring_set_wptr(ring); - - /* set minor_ptr_update to 0 after wptr programed */ - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_MINOR_PTR_UPDATE), 0); - - /* Set up RESP_MODE to non-copy addresses */ - temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_CNTL)); - temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3); - temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, REDO_DELAY, 9); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_CNTL), temp); - - /* program default cache read and write policy */ - temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_PAGE)); - /* clean read policy and write policy bits */ - temp &= 0xFF0FFF; - temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) | - (CACHE_WRITE_POLICY_L2__DEFAULT << 14) | - SDMA0_UTCL1_PAGE__LLC_NOALLOC_MASK); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_UTCL1_PAGE), temp); - - if (!amdgpu_sriov_vf(adev)) { - /* unhalt engine */ - temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL)); - temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0); - temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, TH1_RESET, 0); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL), temp); - } + if (!amdgpu_sriov_vf(adev)) { + /* unhalt engine */ + temp = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL)); + temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0); + temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, TH1_RESET, 0); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL), temp); + } - /* enable DMA RB */ - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_ENABLE, 1); - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl); + /* enable DMA RB */ + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_QUEUE0_RB_CNTL, RB_ENABLE, 1); + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_RB_CNTL), rb_cntl); - ib_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL)); - ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_ENABLE, 1); + ib_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL)); + ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_ENABLE, 1); #ifdef __BIG_ENDIAN - ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_SWAP_ENABLE, 1); + ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_QUEUE0_IB_CNTL, IB_SWAP_ENABLE, 1); #endif - /* enable DMA IBs */ - WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL), ib_cntl); + /* enable DMA IBs */ + WREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_QUEUE0_IB_CNTL), ib_cntl); - if (amdgpu_sriov_vf(adev)) - sdma_v6_0_enable(adev, true); + if (amdgpu_sriov_vf(adev)) + sdma_v6_0_enable(adev, true); + + return amdgpu_ring_test_helper(ring); +} - r = amdgpu_ring_test_helper(ring); +/** + * sdma_v6_0_gfx_resume - setup and start the async dma engines + * + * @adev: amdgpu_device pointer + * + * Set up the gfx DMA ring buffers and enable them. + * Returns 0 for success, error for failure. + */ +static int sdma_v6_0_gfx_resume(struct amdgpu_device *adev) +{ + int i, r; + + for (i = 0; i < adev->sdma.num_instances; i++) { + r = sdma_v6_0_gfx_resume_instance(adev, i, false); if (r) return r; - - if (adev->mman.buffer_funcs_ring == ring) - amdgpu_ttm_set_buffer_funcs_status(adev, true); } return 0; @@ -717,9 +756,9 @@ static int sdma_v6_0_load_microcode(struct amdgpu_device *adev) return 0; } -static int sdma_v6_0_soft_reset(void *handle) +static int sdma_v6_0_soft_reset(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; u32 tmp; int i; @@ -753,9 +792,9 @@ static int sdma_v6_0_soft_reset(void *handle) return sdma_v6_0_start(adev); } -static bool sdma_v6_0_check_soft_reset(void *handle) +static bool sdma_v6_0_check_soft_reset(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; struct amdgpu_ring *ring; int i, r; long tmo = msecs_to_jiffies(1000); @@ -852,6 +891,12 @@ static int sdma_v6_0_mqd_init(struct amdgpu_device *adev, void *mqd, m->sdmax_rlcx_rb_aql_cntl = regSDMA0_QUEUE0_RB_AQL_CNTL_DEFAULT; m->sdmax_rlcx_dummy_reg = regSDMA0_QUEUE0_DUMMY_REG_DEFAULT; + m->sdmax_rlcx_csa_addr_lo = lower_32_bits(prop->csa_addr); + m->sdmax_rlcx_csa_addr_hi = upper_32_bits(prop->csa_addr); + + m->sdmax_rlcx_f32_dbg0 = lower_32_bits(prop->fence_address); + m->sdmax_rlcx_f32_dbg1 = upper_32_bits(prop->fence_address); + return 0; } @@ -878,28 +923,18 @@ static int sdma_v6_0_ring_test_ring(struct amdgpu_ring *ring) int r; u32 tmp; u64 gpu_addr; - volatile uint32_t *cpu_ptr = NULL; tmp = 0xCAFEDEAD; - if (ring->is_mes_queue) { - uint32_t offset = 0; - offset = amdgpu_mes_ctx_get_offs(ring, - AMDGPU_MES_CTX_PADDING_OFFS); - gpu_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset); - cpu_ptr = amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset); - *cpu_ptr = tmp; - } else { - r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r); - return r; - } - - gpu_addr = adev->wb.gpu_addr + (index * 4); - adev->wb.wb[index] = cpu_to_le32(tmp); + r = amdgpu_device_wb_get(adev, &index); + if (r) { + dev_err(adev->dev, "(%d) failed to allocate wb slot\n", r); + return r; } + gpu_addr = adev->wb.gpu_addr + (index * 4); + adev->wb.wb[index] = cpu_to_le32(tmp); + r = amdgpu_ring_alloc(ring, 5); if (r) { DRM_ERROR("amdgpu: dma failed to lock ring %d (%d).\n", ring->idx, r); @@ -916,10 +951,7 @@ static int sdma_v6_0_ring_test_ring(struct amdgpu_ring *ring) amdgpu_ring_commit(ring); for (i = 0; i < adev->usec_timeout; i++) { - if (ring->is_mes_queue) - tmp = le32_to_cpu(*cpu_ptr); - else - tmp = le32_to_cpu(adev->wb.wb[index]); + tmp = le32_to_cpu(adev->wb.wb[index]); if (tmp == 0xDEADBEEF) break; if (amdgpu_emu_mode == 1) @@ -931,13 +963,12 @@ static int sdma_v6_0_ring_test_ring(struct amdgpu_ring *ring) if (i >= adev->usec_timeout) r = -ETIMEDOUT; - if (!ring->is_mes_queue) - amdgpu_device_wb_free(adev, index); + amdgpu_device_wb_free(adev, index); return r; } -/** +/* * sdma_v6_0_ring_test_ib - test an IB on the DMA engine * * @ring: amdgpu_ring structure holding ring information @@ -955,37 +986,23 @@ static int sdma_v6_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) long r; u32 tmp = 0; u64 gpu_addr; - volatile uint32_t *cpu_ptr = NULL; tmp = 0xCAFEDEAD; memset(&ib, 0, sizeof(ib)); - if (ring->is_mes_queue) { - uint32_t offset = 0; - offset = amdgpu_mes_ctx_get_offs(ring, AMDGPU_MES_CTX_IB_OFFS); - ib.gpu_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset); - ib.ptr = (void *)amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset); - - offset = amdgpu_mes_ctx_get_offs(ring, - AMDGPU_MES_CTX_PADDING_OFFS); - gpu_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset); - cpu_ptr = amdgpu_mes_ctx_get_offs_cpu_addr(ring, offset); - *cpu_ptr = tmp; - } else { - r = amdgpu_device_wb_get(adev, &index); - if (r) { - dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r); - return r; - } + r = amdgpu_device_wb_get(adev, &index); + if (r) { + dev_err(adev->dev, "(%ld) failed to allocate wb slot\n", r); + return r; + } - gpu_addr = adev->wb.gpu_addr + (index * 4); - adev->wb.wb[index] = cpu_to_le32(tmp); + gpu_addr = adev->wb.gpu_addr + (index * 4); + adev->wb.wb[index] = cpu_to_le32(tmp); - r = amdgpu_ib_get(adev, NULL, 256, AMDGPU_IB_POOL_DIRECT, &ib); - if (r) { - DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); - goto err0; - } + r = amdgpu_ib_get(adev, NULL, 256, AMDGPU_IB_POOL_DIRECT, &ib); + if (r) { + DRM_ERROR("amdgpu: failed to get ib (%ld).\n", r); + goto err0; } ib.ptr[0] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_WRITE) | @@ -1013,10 +1030,7 @@ static int sdma_v6_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) goto err1; } - if (ring->is_mes_queue) - tmp = le32_to_cpu(*cpu_ptr); - else - tmp = le32_to_cpu(adev->wb.wb[index]); + tmp = le32_to_cpu(adev->wb.wb[index]); if (tmp == 0xDEADBEEF) r = 0; @@ -1024,11 +1038,10 @@ static int sdma_v6_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = -EINVAL; err1: - amdgpu_ib_free(adev, &ib, NULL); + amdgpu_ib_free(&ib, NULL); dma_fence_put(f); err0: - if (!ring->is_mes_queue) - amdgpu_device_wb_free(adev, index); + amdgpu_device_wb_free(adev, index); return r; } @@ -1119,7 +1132,7 @@ static void sdma_v6_0_vm_set_pte_pde(struct amdgpu_ib *ib, ib->ptr[ib->length_dw++] = count - 1; /* number of entries */ } -/** +/* * sdma_v6_0_ring_pad_ib - pad the IB * @ib: indirect buffer to fill with padding * @ring: amdgpu ring pointer @@ -1168,7 +1181,7 @@ static void sdma_v6_0_ring_emit_pipeline_sync(struct amdgpu_ring *ring) SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(4)); /* retry count, poll interval */ } -/** +/* * sdma_v6_0_ring_emit_vm_flush - vm flush using sDMA * * @ring: amdgpu_ring pointer @@ -1246,19 +1259,40 @@ static struct amdgpu_sdma_ras sdma_v6_0_3_ras = { static void sdma_v6_0_set_ras_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(6, 0, 3): adev->sdma.ras = &sdma_v6_0_3_ras; break; default: break; } - } -static int sdma_v6_0_early_init(void *handle) +static int sdma_v6_0_early_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; + int r; + + switch (amdgpu_user_queue) { + case -1: + case 0: + default: + adev->sdma.no_user_submission = false; + adev->sdma.disable_uq = true; + break; + case 1: + adev->sdma.no_user_submission = false; + adev->sdma.disable_uq = false; + break; + case 2: + adev->sdma.no_user_submission = true; + adev->sdma.disable_uq = false; + break; + } + + r = amdgpu_sdma_init_microcode(adev, 0, true); + if (r) + return r; sdma_v6_0_set_ring_funcs(adev); sdma_v6_0_set_buffer_funcs(adev); @@ -1270,11 +1304,13 @@ static int sdma_v6_0_early_init(void *handle) return 0; } -static int sdma_v6_0_sw_init(void *handle) +static int sdma_v6_0_sw_init(struct amdgpu_ip_block *ip_block) { struct amdgpu_ring *ring; int r, i; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; + uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_6_0); + uint32_t *ptr; /* SDMA trap event */ r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX, @@ -1283,17 +1319,19 @@ static int sdma_v6_0_sw_init(void *handle) if (r) return r; - r = amdgpu_sdma_init_microcode(adev, 0, true); - if (r) { - DRM_ERROR("Failed to load sdma firmware!\n"); + /* SDMA user fence event */ + r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX, + GFX_11_0_0__SRCID__SDMA_FENCE, + &adev->sdma.fence_irq); + if (r) return r; - } for (i = 0; i < adev->sdma.num_instances; i++) { ring = &adev->sdma.instance[i].ring; ring->ring_obj = NULL; ring->use_doorbell = true; ring->me = i; + ring->no_user_submission = adev->sdma.no_user_submission; DRM_DEBUG("SDMA %d use_doorbell being set to: [%s]\n", i, ring->use_doorbell?"true":"false"); @@ -1311,67 +1349,155 @@ static int sdma_v6_0_sw_init(void *handle) return r; } + adev->sdma.supported_reset = + amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring); + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { + case IP_VERSION(6, 0, 0): + case IP_VERSION(6, 0, 2): + case IP_VERSION(6, 0, 3): + if ((adev->sdma.instance[0].fw_version >= 21) && + !amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) + adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; + break; + default: + break; + } + if (amdgpu_sdma_ras_sw_init(adev)) { dev_err(adev->dev, "Failed to initialize sdma ras block!\n"); return -EINVAL; } + /* Allocate memory for SDMA IP Dump buffer */ + ptr = kcalloc(adev->sdma.num_instances * reg_count, sizeof(uint32_t), GFP_KERNEL); + if (ptr) + adev->sdma.ip_dump = ptr; + else + DRM_ERROR("Failed to allocated memory for SDMA IP Dump\n"); + + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { + case IP_VERSION(6, 0, 0): + if ((adev->sdma.instance[0].fw_version >= 27) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 0, 1): + if ((adev->sdma.instance[0].fw_version >= 18) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 0, 2): + if ((adev->sdma.instance[0].fw_version >= 23) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 0, 3): + if (adev->sdma.instance[0].fw_version >= 29 && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 1, 0): + if ((adev->sdma.instance[0].fw_version >= 14) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 1, 1): + if ((adev->sdma.instance[0].fw_version >= 17) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 1, 2): + if ((adev->sdma.instance[0].fw_version >= 15) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + case IP_VERSION(6, 1, 3): + if ((adev->sdma.instance[0].fw_version >= 10) && !adev->sdma.disable_uq) + adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; + break; + default: + break; + } + + r = amdgpu_sdma_sysfs_reset_mask_init(adev); + if (r) + return r; + return r; } -static int sdma_v6_0_sw_fini(void *handle) +static int sdma_v6_0_sw_fini(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i; for (i = 0; i < adev->sdma.num_instances; i++) amdgpu_ring_fini(&adev->sdma.instance[i].ring); + amdgpu_sdma_sysfs_reset_mask_fini(adev); amdgpu_sdma_destroy_inst_ctx(adev, true); + kfree(adev->sdma.ip_dump); + return 0; } -static int sdma_v6_0_hw_init(void *handle) +static int sdma_v6_0_set_userq_trap_interrupts(struct amdgpu_device *adev, + bool enable) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + unsigned int irq_type; + int i, r; - return sdma_v6_0_start(adev); + if (adev->userq_funcs[AMDGPU_HW_IP_DMA]) { + for (i = 0; i < adev->sdma.num_instances; i++) { + irq_type = AMDGPU_SDMA_IRQ_INSTANCE0 + i; + if (enable) + r = amdgpu_irq_get(adev, &adev->sdma.trap_irq, + irq_type); + else + r = amdgpu_irq_put(adev, &adev->sdma.trap_irq, + irq_type); + if (r) + return r; + } + } + + return 0; } -static int sdma_v6_0_hw_fini(void *handle) +static int sdma_v6_0_hw_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; + int r; - if (amdgpu_sriov_vf(adev)) { - /* disable the scheduler for SDMA */ - amdgpu_sdma_unset_buffer_funcs_helper(adev); + r = sdma_v6_0_start(adev); + if (r) + return r; + + return sdma_v6_0_set_userq_trap_interrupts(adev, true); +} + +static int sdma_v6_0_hw_fini(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + + if (amdgpu_sriov_vf(adev)) return 0; - } sdma_v6_0_ctxempty_int_enable(adev, false); sdma_v6_0_enable(adev, false); + sdma_v6_0_set_userq_trap_interrupts(adev, false); return 0; } -static int sdma_v6_0_suspend(void *handle) +static int sdma_v6_0_suspend(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; - - return sdma_v6_0_hw_fini(adev); + return sdma_v6_0_hw_fini(ip_block); } -static int sdma_v6_0_resume(void *handle) +static int sdma_v6_0_resume(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; - - return sdma_v6_0_hw_init(adev); + return sdma_v6_0_hw_init(ip_block); } -static bool sdma_v6_0_is_idle(void *handle) +static bool sdma_v6_0_is_idle(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; u32 i; for (i = 0; i < adev->sdma.num_instances; i++) { @@ -1384,11 +1510,11 @@ static bool sdma_v6_0_is_idle(void *handle) return true; } -static int sdma_v6_0_wait_for_idle(void *handle) +static int sdma_v6_0_wait_for_idle(struct amdgpu_ip_block *ip_block) { unsigned i; u32 sdma0, sdma1; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; for (i = 0; i < adev->usec_timeout; i++) { sdma0 = RREG32(sdma_v6_0_get_reg_offset(adev, 0, regSDMA0_STATUS_REG)); @@ -1446,6 +1572,31 @@ static int sdma_v6_0_ring_preempt_ib(struct amdgpu_ring *ring) return r; } +static int sdma_v6_0_reset_queue(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) +{ + struct amdgpu_device *adev = ring->adev; + int r; + + if (ring->me >= adev->sdma.num_instances) { + dev_err(adev->dev, "sdma instance not found\n"); + return -EINVAL; + } + + amdgpu_ring_reset_helper_begin(ring, timedout_fence); + + r = amdgpu_mes_reset_legacy_queue(adev, ring, vmid, true); + if (r) + return r; + + r = sdma_v6_0_gfx_resume_instance(adev, ring->me, true); + if (r) + return r; + + return amdgpu_ring_reset_helper_end(ring, timedout_fence); +} + static int sdma_v6_0_set_trap_irq_state(struct amdgpu_device *adev, struct amdgpu_irq_src *source, unsigned type, @@ -1470,25 +1621,9 @@ static int sdma_v6_0_process_trap_irq(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) { int instances, queue; - uint32_t mes_queue_id = entry->src_data[0]; DRM_DEBUG("IH: SDMA trap\n"); - if (adev->enable_mes && (mes_queue_id & AMDGPU_FENCE_MES_QUEUE_FLAG)) { - struct amdgpu_mes_queue *queue; - - mes_queue_id &= AMDGPU_FENCE_MES_QUEUE_ID_MASK; - - spin_lock(&adev->mes.queue_id_lock); - queue = idr_find(&adev->mes.queue_id_idr, mes_queue_id); - if (queue) { - DRM_DEBUG("process smda queue id = %d\n", mes_queue_id); - amdgpu_fence_process(queue->ring); - } - spin_unlock(&adev->mes.queue_id_lock); - return 0; - } - queue = entry->ring_id & 0xf; instances = (entry->ring_id & 0xf0) >> 4; if (instances > 1) { @@ -1510,6 +1645,29 @@ static int sdma_v6_0_process_trap_irq(struct amdgpu_device *adev, return 0; } +static int sdma_v6_0_process_fence_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + u32 doorbell_offset = entry->src_data[0]; + + if (adev->enable_mes && doorbell_offset) { + struct amdgpu_userq_fence_driver *fence_drv = NULL; + struct xarray *xa = &adev->userq_xa; + unsigned long flags; + + doorbell_offset >>= SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT; + + xa_lock_irqsave(xa, flags); + fence_drv = xa_load(xa, doorbell_offset); + if (fence_drv) + amdgpu_userq_fence_driver_process(fence_drv); + xa_unlock_irqrestore(xa, flags); + } + + return 0; +} + static int sdma_v6_0_process_illegal_inst_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) @@ -1517,26 +1675,67 @@ static int sdma_v6_0_process_illegal_inst_irq(struct amdgpu_device *adev, return 0; } -static int sdma_v6_0_set_clockgating_state(void *handle, +static int sdma_v6_0_set_clockgating_state(struct amdgpu_ip_block *ip_block, enum amd_clockgating_state state) { return 0; } -static int sdma_v6_0_set_powergating_state(void *handle, +static int sdma_v6_0_set_powergating_state(struct amdgpu_ip_block *ip_block, enum amd_powergating_state state) { return 0; } -static void sdma_v6_0_get_clockgating_state(void *handle, u64 *flags) +static void sdma_v6_0_get_clockgating_state(struct amdgpu_ip_block *ip_block, u64 *flags) +{ +} + +static void sdma_v6_0_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p) +{ + struct amdgpu_device *adev = ip_block->adev; + int i, j; + uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_6_0); + uint32_t instance_offset; + + if (!adev->sdma.ip_dump) + return; + + drm_printf(p, "num_instances:%d\n", adev->sdma.num_instances); + for (i = 0; i < adev->sdma.num_instances; i++) { + instance_offset = i * reg_count; + drm_printf(p, "\nInstance:%d\n", i); + + for (j = 0; j < reg_count; j++) + drm_printf(p, "%-50s \t 0x%08x\n", sdma_reg_list_6_0[j].reg_name, + adev->sdma.ip_dump[instance_offset + j]); + } +} + +static void sdma_v6_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { + struct amdgpu_device *adev = ip_block->adev; + int i, j; + uint32_t instance_offset; + uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_6_0); + + if (!adev->sdma.ip_dump) + return; + + amdgpu_gfx_off_ctrl(adev, false); + for (i = 0; i < adev->sdma.num_instances; i++) { + instance_offset = i * reg_count; + for (j = 0; j < reg_count; j++) + adev->sdma.ip_dump[instance_offset + j] = + RREG32(sdma_v6_0_get_reg_offset(adev, i, + sdma_reg_list_6_0[j].reg_offset)); + } + amdgpu_gfx_off_ctrl(adev, true); } const struct amd_ip_funcs sdma_v6_0_ip_funcs = { .name = "sdma_v6_0", .early_init = sdma_v6_0_early_init, - .late_init = NULL, .sw_init = sdma_v6_0_sw_init, .sw_fini = sdma_v6_0_sw_fini, .hw_init = sdma_v6_0_hw_init, @@ -1550,6 +1749,8 @@ const struct amd_ip_funcs sdma_v6_0_ip_funcs = { .set_clockgating_state = sdma_v6_0_set_clockgating_state, .set_powergating_state = sdma_v6_0_set_powergating_state, .get_clockgating_state = sdma_v6_0_get_clockgating_state, + .dump_ip_state = sdma_v6_0_dump_ip_state, + .print_ip_state = sdma_v6_0_print_ip_state, }; static const struct amdgpu_ring_funcs sdma_v6_0_ring_funcs = { @@ -1584,8 +1785,8 @@ static const struct amdgpu_ring_funcs sdma_v6_0_ring_funcs = { .emit_reg_wait = sdma_v6_0_ring_emit_reg_wait, .emit_reg_write_reg_wait = sdma_v6_0_ring_emit_reg_write_reg_wait, .init_cond_exec = sdma_v6_0_ring_init_cond_exec, - .patch_cond_exec = sdma_v6_0_ring_patch_cond_exec, .preempt_ib = sdma_v6_0_ring_preempt_ib, + .reset = sdma_v6_0_reset_queue, }; static void sdma_v6_0_set_ring_funcs(struct amdgpu_device *adev) @@ -1603,6 +1804,10 @@ static const struct amdgpu_irq_src_funcs sdma_v6_0_trap_irq_funcs = { .process = sdma_v6_0_process_trap_irq, }; +static const struct amdgpu_irq_src_funcs sdma_v6_0_fence_irq_funcs = { + .process = sdma_v6_0_process_fence_irq, +}; + static const struct amdgpu_irq_src_funcs sdma_v6_0_illegal_inst_irq_funcs = { .process = sdma_v6_0_process_illegal_inst_irq, }; @@ -1612,6 +1817,7 @@ static void sdma_v6_0_set_irq_funcs(struct amdgpu_device *adev) adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE0 + adev->sdma.num_instances; adev->sdma.trap_irq.funcs = &sdma_v6_0_trap_irq_funcs; + adev->sdma.fence_irq.funcs = &sdma_v6_0_fence_irq_funcs; adev->sdma.illegal_inst_irq.funcs = &sdma_v6_0_illegal_inst_irq_funcs; } @@ -1622,7 +1828,7 @@ static void sdma_v6_0_set_irq_funcs(struct amdgpu_device *adev) * @src_offset: src GPU address * @dst_offset: dst GPU address * @byte_count: number of bytes to xfer - * @tmz: if a secure copy should be used + * @copy_flags: copy flags for the buffers * * Copy GPU buffers using the DMA engine. * Used by the amdgpu ttm implementation to move pages if @@ -1632,11 +1838,11 @@ static void sdma_v6_0_emit_copy_buffer(struct amdgpu_ib *ib, uint64_t src_offset, uint64_t dst_offset, uint32_t byte_count, - bool tmz) + uint32_t copy_flags) { ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_COPY) | SDMA_PKT_COPY_LINEAR_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR) | - SDMA_PKT_COPY_LINEAR_HEADER_TMZ(tmz ? 1 : 0); + SDMA_PKT_COPY_LINEAR_HEADER_TMZ((copy_flags & AMDGPU_COPY_FLAGS_TMZ) ? 1 : 0); ib->ptr[ib->length_dw++] = byte_count - 1; ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ ib->ptr[ib->length_dw++] = lower_32_bits(src_offset); @@ -1660,7 +1866,7 @@ static void sdma_v6_0_emit_fill_buffer(struct amdgpu_ib *ib, uint64_t dst_offset, uint32_t byte_count) { - ib->ptr[ib->length_dw++] = SDMA_PKT_COPY_LINEAR_HEADER_OP(SDMA_OP_CONST_FILL); + ib->ptr[ib->length_dw++] = SDMA_PKT_CONSTANT_FILL_HEADER_OP(SDMA_OP_CONST_FILL); ib->ptr[ib->length_dw++] = lower_32_bits(dst_offset); ib->ptr[ib->length_dw++] = upper_32_bits(dst_offset); ib->ptr[ib->length_dw++] = src_data; |