diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 244 |
1 files changed, 124 insertions, 120 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c index 653ce5e..f325fd8 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c @@ -335,12 +335,8 @@ out: */ static uint32_t sdma_v3_0_ring_get_rptr(struct amdgpu_ring *ring) { - u32 rptr; - /* XXX check if swapping is necessary on BE */ - rptr = ring->adev->wb.wb[ring->rptr_offs] >> 2; - - return rptr; + return ring->adev->wb.wb[ring->rptr_offs] >> 2; } /** @@ -499,31 +495,6 @@ static void sdma_v3_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 se amdgpu_ring_write(ring, SDMA_PKT_TRAP_INT_CONTEXT_INT_CONTEXT(0)); } -unsigned init_cond_exec(struct amdgpu_ring *ring) -{ - unsigned ret; - amdgpu_ring_write(ring, SDMA_PKT_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, 1); - ret = ring->wptr;/* this is the offset we need patch later */ - amdgpu_ring_write(ring, 0x55aa55aa);/* insert dummy here and patch it later */ - return ret; -} - -void patch_cond_exec(struct amdgpu_ring *ring, unsigned offset) -{ - unsigned cur; - BUG_ON(ring->ring[offset] != 0x55aa55aa); - - cur = ring->wptr - 1; - if (likely(cur > offset)) - ring->ring[offset] = cur - offset; - else - ring->ring[offset] = (ring->ring_size>>2) - offset + cur; -} - - /** * sdma_v3_0_gfx_stop - stop the gfx async dma engines * @@ -976,24 +947,16 @@ static void sdma_v3_0_vm_copy_pte(struct amdgpu_ib *ib, uint64_t pe, uint64_t src, unsigned count) { - while (count) { - unsigned bytes = count * 8; - if (bytes > 0x1FFFF8) - bytes = 0x1FFFF8; - - ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | - SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); - ib->ptr[ib->length_dw++] = bytes; - ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ - ib->ptr[ib->length_dw++] = lower_32_bits(src); - ib->ptr[ib->length_dw++] = upper_32_bits(src); - ib->ptr[ib->length_dw++] = lower_32_bits(pe); - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - - pe += bytes; - src += bytes; - count -= bytes / 8; - } + unsigned bytes = count * 8; + + ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_COPY) | + SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); + ib->ptr[ib->length_dw++] = bytes; + ib->ptr[ib->length_dw++] = 0; /* src/dst endian swap */ + ib->ptr[ib->length_dw++] = lower_32_bits(src); + ib->ptr[ib->length_dw++] = upper_32_bits(src); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); + ib->ptr[ib->length_dw++] = upper_32_bits(pe); } /** @@ -1001,39 +964,27 @@ static void sdma_v3_0_vm_copy_pte(struct amdgpu_ib *ib, * * @ib: indirect buffer to fill with commands * @pe: addr of the page entry - * @addr: dst addr to write into pe + * @value: dst addr to write into pe * @count: number of page entries to update * @incr: increase next addr by incr bytes - * @flags: access flags * * Update PTEs by writing them manually using sDMA (CIK). */ -static void sdma_v3_0_vm_write_pte(struct amdgpu_ib *ib, - const dma_addr_t *pages_addr, uint64_t pe, - uint64_t addr, unsigned count, - uint32_t incr, uint32_t flags) -{ - uint64_t value; - unsigned ndw; - - while (count) { - ndw = count * 2; - if (ndw > 0xFFFFE) - ndw = 0xFFFFE; - - /* for non-physically contiguous pages (system) */ - ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | - SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); - ib->ptr[ib->length_dw++] = pe; - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - ib->ptr[ib->length_dw++] = ndw; - for (; ndw > 0; ndw -= 2, --count, pe += 8) { - value = amdgpu_vm_map_gart(pages_addr, addr); - addr += incr; - value |= flags; - ib->ptr[ib->length_dw++] = value; - ib->ptr[ib->length_dw++] = upper_32_bits(value); - } +static void sdma_v3_0_vm_write_pte(struct amdgpu_ib *ib, uint64_t pe, + uint64_t value, unsigned count, + uint32_t incr) +{ + unsigned ndw = count * 2; + + ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_WRITE) | + SDMA_PKT_HEADER_SUB_OP(SDMA_SUBOP_COPY_LINEAR); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); + ib->ptr[ib->length_dw++] = upper_32_bits(pe); + ib->ptr[ib->length_dw++] = ndw; + for (; ndw > 0; ndw -= 2, --count, pe += 8) { + ib->ptr[ib->length_dw++] = lower_32_bits(value); + ib->ptr[ib->length_dw++] = upper_32_bits(value); + value += incr; } } @@ -1049,40 +1000,21 @@ static void sdma_v3_0_vm_write_pte(struct amdgpu_ib *ib, * * Update the page tables using sDMA (CIK). */ -static void sdma_v3_0_vm_set_pte_pde(struct amdgpu_ib *ib, - uint64_t pe, +static void sdma_v3_0_vm_set_pte_pde(struct amdgpu_ib *ib, uint64_t pe, uint64_t addr, unsigned count, uint32_t incr, uint32_t flags) { - uint64_t value; - unsigned ndw; - - while (count) { - ndw = count; - if (ndw > 0x7FFFF) - ndw = 0x7FFFF; - - if (flags & AMDGPU_PTE_VALID) - value = addr; - else - value = 0; - - /* for physically contiguous pages (vram) */ - ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE); - ib->ptr[ib->length_dw++] = pe; /* dst addr */ - ib->ptr[ib->length_dw++] = upper_32_bits(pe); - ib->ptr[ib->length_dw++] = flags; /* mask */ - ib->ptr[ib->length_dw++] = 0; - ib->ptr[ib->length_dw++] = value; /* value */ - ib->ptr[ib->length_dw++] = upper_32_bits(value); - ib->ptr[ib->length_dw++] = incr; /* increment size */ - ib->ptr[ib->length_dw++] = 0; - ib->ptr[ib->length_dw++] = ndw; /* number of entries */ - - pe += ndw * 8; - addr += ndw * incr; - count -= ndw; - } + /* for physically contiguous pages (vram) */ + ib->ptr[ib->length_dw++] = SDMA_PKT_HEADER_OP(SDMA_OP_GEN_PTEPDE); + ib->ptr[ib->length_dw++] = lower_32_bits(pe); /* dst addr */ + ib->ptr[ib->length_dw++] = upper_32_bits(pe); + ib->ptr[ib->length_dw++] = flags; /* mask */ + ib->ptr[ib->length_dw++] = 0; + ib->ptr[ib->length_dw++] = lower_32_bits(addr); /* value */ + ib->ptr[ib->length_dw++] = upper_32_bits(addr); + ib->ptr[ib->length_dw++] = incr; /* increment size */ + ib->ptr[ib->length_dw++] = 0; + ib->ptr[ib->length_dw++] = count; /* number of entries */ } /** @@ -1172,6 +1104,22 @@ static void sdma_v3_0_ring_emit_vm_flush(struct amdgpu_ring *ring, SDMA_PKT_POLL_REGMEM_DW5_INTERVAL(10)); /* retry count, poll interval */ } +static unsigned sdma_v3_0_ring_get_emit_ib_size(struct amdgpu_ring *ring) +{ + return + 7 + 6; /* sdma_v3_0_ring_emit_ib */ +} + +static unsigned sdma_v3_0_ring_get_dma_frame_size(struct amdgpu_ring *ring) +{ + return + 6 + /* sdma_v3_0_ring_emit_hdp_flush */ + 3 + /* sdma_v3_0_ring_emit_hdp_invalidate */ + 6 + /* sdma_v3_0_ring_emit_pipeline_sync */ + 12 + /* sdma_v3_0_ring_emit_vm_flush */ + 10 + 10 + 10; /* sdma_v3_0_ring_emit_fence x3 for user fence, vm fence */ +} + static int sdma_v3_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -1320,28 +1268,79 @@ static int sdma_v3_0_wait_for_idle(void *handle) return -ETIMEDOUT; } -static int sdma_v3_0_soft_reset(void *handle) +static int sdma_v3_0_check_soft_reset(void *handle) { - u32 srbm_soft_reset = 0; struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; u32 tmp = RREG32(mmSRBM_STATUS2); - if (tmp & SRBM_STATUS2__SDMA_BUSY_MASK) { - /* sdma0 */ - tmp = RREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET); - tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0); - WREG32(mmSDMA0_F32_CNTL + SDMA0_REGISTER_OFFSET, tmp); + if ((tmp & SRBM_STATUS2__SDMA_BUSY_MASK) || + (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK)) { srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA_MASK; - } - if (tmp & SRBM_STATUS2__SDMA1_BUSY_MASK) { - /* sdma1 */ - tmp = RREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET); - tmp = REG_SET_FIELD(tmp, SDMA0_F32_CNTL, HALT, 0); - WREG32(mmSDMA0_F32_CNTL + SDMA1_REGISTER_OFFSET, tmp); srbm_soft_reset |= SRBM_SOFT_RESET__SOFT_RESET_SDMA1_MASK; } if (srbm_soft_reset) { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang = true; + adev->sdma.srbm_soft_reset = srbm_soft_reset; + } else { + adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang = false; + adev->sdma.srbm_soft_reset = 0; + } + + return 0; +} + +static int sdma_v3_0_pre_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang) + return 0; + + srbm_soft_reset = adev->sdma.srbm_soft_reset; + + if (REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA) || + REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA1)) { + sdma_v3_0_ctx_switch_enable(adev, false); + sdma_v3_0_enable(adev, false); + } + + return 0; +} + +static int sdma_v3_0_post_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang) + return 0; + + srbm_soft_reset = adev->sdma.srbm_soft_reset; + + if (REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA) || + REG_GET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_SDMA1)) { + sdma_v3_0_gfx_resume(adev); + sdma_v3_0_rlc_resume(adev); + } + + return 0; +} + +static int sdma_v3_0_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; + u32 tmp; + + if (!adev->ip_block_status[AMD_IP_BLOCK_TYPE_SDMA].hang) + return 0; + + srbm_soft_reset = adev->sdma.srbm_soft_reset; + + if (srbm_soft_reset) { tmp = RREG32(mmSRBM_SOFT_RESET); tmp |= srbm_soft_reset; dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp); @@ -1559,6 +1558,9 @@ const struct amd_ip_funcs sdma_v3_0_ip_funcs = { .resume = sdma_v3_0_resume, .is_idle = sdma_v3_0_is_idle, .wait_for_idle = sdma_v3_0_wait_for_idle, + .check_soft_reset = sdma_v3_0_check_soft_reset, + .pre_soft_reset = sdma_v3_0_pre_soft_reset, + .post_soft_reset = sdma_v3_0_post_soft_reset, .soft_reset = sdma_v3_0_soft_reset, .set_clockgating_state = sdma_v3_0_set_clockgating_state, .set_powergating_state = sdma_v3_0_set_powergating_state, @@ -1579,6 +1581,8 @@ static const struct amdgpu_ring_funcs sdma_v3_0_ring_funcs = { .test_ib = sdma_v3_0_ring_test_ib, .insert_nop = sdma_v3_0_ring_insert_nop, .pad_ib = sdma_v3_0_ring_pad_ib, + .get_emit_ib_size = sdma_v3_0_ring_get_emit_ib_size, + .get_dma_frame_size = sdma_v3_0_ring_get_dma_frame_size, }; static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev) |
