diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c | 243 |
1 files changed, 82 insertions, 161 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index 8c174c11eaee..2b7aba22ecc1 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -55,6 +55,9 @@ #define GFX7_NUM_GFX_RINGS 1 #define GFX7_MEC_HPD_SIZE 2048 +#define BONAIRE_GB_ADDR_CONFIG_GOLDEN 0x12010001 +#define HAWAII_GB_ADDR_CONFIG_GOLDEN 0x12011003 + static void gfx_v7_0_set_ring_funcs(struct amdgpu_device *adev); static void gfx_v7_0_set_irq_funcs(struct amdgpu_device *adev); static void gfx_v7_0_set_gds_init(struct amdgpu_device *adev); @@ -90,8 +93,7 @@ MODULE_FIRMWARE("amdgpu/mullins_ce.bin"); MODULE_FIRMWARE("amdgpu/mullins_rlc.bin"); MODULE_FIRMWARE("amdgpu/mullins_mec.bin"); -static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] = -{ +static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] = { {mmGDS_VMID0_BASE, mmGDS_VMID0_SIZE, mmGDS_GWS_VMID0, mmGDS_OA_VMID0}, {mmGDS_VMID1_BASE, mmGDS_VMID1_SIZE, mmGDS_GWS_VMID1, mmGDS_OA_VMID1}, {mmGDS_VMID2_BASE, mmGDS_VMID2_SIZE, mmGDS_GWS_VMID2, mmGDS_OA_VMID2}, @@ -110,8 +112,7 @@ static const struct amdgpu_gds_reg_offset amdgpu_gds_reg_offset[] = {mmGDS_VMID15_BASE, mmGDS_VMID15_SIZE, mmGDS_GWS_VMID15, mmGDS_OA_VMID15} }; -static const u32 spectre_rlc_save_restore_register_list[] = -{ +static const u32 spectre_rlc_save_restore_register_list[] = { (0x0e00 << 16) | (0xc12c >> 2), 0x00000000, (0x0e00 << 16) | (0xc140 >> 2), @@ -557,8 +558,7 @@ static const u32 spectre_rlc_save_restore_register_list[] = (0x0e00 << 16) | (0x9600 >> 2), }; -static const u32 kalindi_rlc_save_restore_register_list[] = -{ +static const u32 kalindi_rlc_save_restore_register_list[] = { (0x0e00 << 16) | (0xc12c >> 2), 0x00000000, (0x0e00 << 16) | (0xc140 >> 2), @@ -883,7 +883,7 @@ static const u32 kalindi_rlc_save_restore_register_list[] = }; static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev); -static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, volatile u32 *buffer); +static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer); static void gfx_v7_0_init_pg(struct amdgpu_device *adev); static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev); @@ -912,7 +912,6 @@ static void gfx_v7_0_free_microcode(struct amdgpu_device *adev) static int gfx_v7_0_init_microcode(struct amdgpu_device *adev) { const char *chip_name; - char fw_name[30]; int err; DRM_DEBUG("\n"); @@ -933,43 +932,48 @@ static int gfx_v7_0_init_microcode(struct amdgpu_device *adev) case CHIP_MULLINS: chip_name = "mullins"; break; - default: BUG(); + default: + BUG(); } - snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_pfp.bin", chip_name); - err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw, fw_name); + err = amdgpu_ucode_request(adev, &adev->gfx.pfp_fw, + AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_pfp.bin", chip_name); if (err) goto out; - snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_me.bin", chip_name); - err = amdgpu_ucode_request(adev, &adev->gfx.me_fw, fw_name); + err = amdgpu_ucode_request(adev, &adev->gfx.me_fw, + AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_me.bin", chip_name); if (err) goto out; - snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ce.bin", chip_name); - err = amdgpu_ucode_request(adev, &adev->gfx.ce_fw, fw_name); + err = amdgpu_ucode_request(adev, &adev->gfx.ce_fw, + AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_ce.bin", chip_name); if (err) goto out; - snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec.bin", chip_name); - err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw, fw_name); + err = amdgpu_ucode_request(adev, &adev->gfx.mec_fw, + AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_mec.bin", chip_name); if (err) goto out; if (adev->asic_type == CHIP_KAVERI) { - snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mec2.bin", chip_name); - err = amdgpu_ucode_request(adev, &adev->gfx.mec2_fw, fw_name); + err = amdgpu_ucode_request(adev, &adev->gfx.mec2_fw, + AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_mec2.bin", chip_name); if (err) goto out; } - snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name); - err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw, fw_name); - if (err) - goto out; + err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw, + AMDGPU_UCODE_REQUIRED, + "amdgpu/%s_rlc.bin", chip_name); out: if (err) { - pr_err("gfx7: Failed to load firmware \"%s\"\n", fw_name); + pr_err("gfx7: Failed to load firmware %s gfx firmware\n", chip_name); gfx_v7_0_free_microcode(adev); } return err; @@ -2119,6 +2123,8 @@ static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr, { bool write64bit = flags & AMDGPU_FENCE_FLAG_64BIT; bool int_sel = flags & AMDGPU_FENCE_FLAG_INT; + bool exec = flags & AMDGPU_FENCE_FLAG_EXEC; + /* Workaround for cache flush problems. First send a dummy EOP * event down the pipe with seq one below. */ @@ -2138,7 +2144,8 @@ static void gfx_v7_0_ring_emit_fence_gfx(struct amdgpu_ring *ring, u64 addr, amdgpu_ring_write(ring, (EOP_TCL1_ACTION_EN | EOP_TC_ACTION_EN | EVENT_TYPE(CACHE_FLUSH_AND_INV_TS_EVENT) | - EVENT_INDEX(5))); + EVENT_INDEX(5) | + (exec ? EOP_EXEC : 0))); amdgpu_ring_write(ring, addr & 0xfffffffc); amdgpu_ring_write(ring, (upper_32_bits(addr) & 0xffff) | DATA_SEL(write64bit ? 2 : 1) | INT_SEL(int_sel ? 2 : 0)); @@ -2326,7 +2333,7 @@ static int gfx_v7_0_ring_test_ib(struct amdgpu_ring *ring, long timeout) r = -EINVAL; error: - amdgpu_ib_free(adev, &ib, NULL); + amdgpu_ib_free(&ib, NULL); dma_fence_put(f); return r; } @@ -2561,7 +2568,7 @@ static int gfx_v7_0_cp_gfx_resume(struct amdgpu_device *adev) ring->wptr = 0; WREG32(mmCP_RB0_WPTR, lower_32_bits(ring->wptr)); - /* set the wb address wether it's enabled or not */ + /* set the wb address whether it's enabled or not */ rptr_addr = ring->rptr_gpu_addr; WREG32(mmCP_RB0_RPTR_ADDR, lower_32_bits(rptr_addr)); WREG32(mmCP_RB0_RPTR_ADDR_HI, upper_32_bits(rptr_addr) & 0xFF); @@ -2759,45 +2766,6 @@ static int gfx_v7_0_mec_init(struct amdgpu_device *adev) return 0; } -struct hqd_registers -{ - u32 cp_mqd_base_addr; - u32 cp_mqd_base_addr_hi; - u32 cp_hqd_active; - u32 cp_hqd_vmid; - u32 cp_hqd_persistent_state; - u32 cp_hqd_pipe_priority; - u32 cp_hqd_queue_priority; - u32 cp_hqd_quantum; - u32 cp_hqd_pq_base; - u32 cp_hqd_pq_base_hi; - u32 cp_hqd_pq_rptr; - u32 cp_hqd_pq_rptr_report_addr; - u32 cp_hqd_pq_rptr_report_addr_hi; - u32 cp_hqd_pq_wptr_poll_addr; - u32 cp_hqd_pq_wptr_poll_addr_hi; - u32 cp_hqd_pq_doorbell_control; - u32 cp_hqd_pq_wptr; - u32 cp_hqd_pq_control; - u32 cp_hqd_ib_base_addr; - u32 cp_hqd_ib_base_addr_hi; - u32 cp_hqd_ib_rptr; - u32 cp_hqd_ib_control; - u32 cp_hqd_iq_timer; - u32 cp_hqd_iq_rptr; - u32 cp_hqd_dequeue_request; - u32 cp_hqd_dma_offload; - u32 cp_hqd_sema_cmd; - u32 cp_hqd_msg_type; - u32 cp_hqd_atomic0_preop_lo; - u32 cp_hqd_atomic0_preop_hi; - u32 cp_hqd_atomic1_preop_lo; - u32 cp_hqd_atomic1_preop_hi; - u32 cp_hqd_hq_scheduler0; - u32 cp_hqd_hq_scheduler1; - u32 cp_mqd_control; -}; - static void gfx_v7_0_compute_pipe_init(struct amdgpu_device *adev, int mec, int pipe) { @@ -2917,7 +2885,7 @@ static void gfx_v7_0_mqd_init(struct amdgpu_device *adev, mqd->cp_hqd_pq_wptr_poll_addr_lo = wb_gpu_addr & 0xfffffffc; mqd->cp_hqd_pq_wptr_poll_addr_hi = upper_32_bits(wb_gpu_addr) & 0xffff; - /* set the wb address wether it's enabled or not */ + /* set the wb address whether it's enabled or not */ wb_gpu_addr = ring->rptr_gpu_addr; mqd->cp_hqd_pq_rptr_report_addr_lo = wb_gpu_addr & 0xfffffffc; mqd->cp_hqd_pq_rptr_report_addr_hi = @@ -3277,7 +3245,7 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev) /* init spm vmid with 0xf */ if (adev->gfx.rlc.funcs->update_spm_vmid) - adev->gfx.rlc.funcs->update_spm_vmid(adev, 0xf); + adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf); return 0; } @@ -3503,7 +3471,7 @@ static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev) return 0; } -static void gfx_v7_0_update_spm_vmid(struct amdgpu_device *adev, unsigned vmid) +static void gfx_v7_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned vmid) { u32 data; @@ -3914,70 +3882,24 @@ static u32 gfx_v7_0_get_csb_size(struct amdgpu_device *adev) return count; } -static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, - volatile u32 *buffer) +static void gfx_v7_0_get_csb_buffer(struct amdgpu_device *adev, u32 *buffer) { - u32 count = 0, i; - const struct cs_section_def *sect = NULL; - const struct cs_extent_def *ext = NULL; + u32 count = 0; if (adev->gfx.rlc.cs_data == NULL) return; if (buffer == NULL) return; - buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0)); - buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_BEGIN_CLEAR_STATE); - - buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CONTEXT_CONTROL, 1)); - buffer[count++] = cpu_to_le32(0x80000000); - buffer[count++] = cpu_to_le32(0x80000000); - - for (sect = adev->gfx.rlc.cs_data; sect->section != NULL; ++sect) { - for (ext = sect->section; ext->extent != NULL; ++ext) { - if (sect->id == SECT_CONTEXT) { - buffer[count++] = - cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, ext->reg_count)); - buffer[count++] = cpu_to_le32(ext->reg_index - PACKET3_SET_CONTEXT_REG_START); - for (i = 0; i < ext->reg_count; i++) - buffer[count++] = cpu_to_le32(ext->extent[i]); - } else { - return; - } - } - } + count = amdgpu_gfx_csb_preamble_start(buffer); + count = amdgpu_gfx_csb_data_parser(adev, buffer, count); buffer[count++] = cpu_to_le32(PACKET3(PACKET3_SET_CONTEXT_REG, 2)); buffer[count++] = cpu_to_le32(mmPA_SC_RASTER_CONFIG - PACKET3_SET_CONTEXT_REG_START); - switch (adev->asic_type) { - case CHIP_BONAIRE: - buffer[count++] = cpu_to_le32(0x16000012); - buffer[count++] = cpu_to_le32(0x00000000); - break; - case CHIP_KAVERI: - buffer[count++] = cpu_to_le32(0x00000000); /* XXX */ - buffer[count++] = cpu_to_le32(0x00000000); - break; - case CHIP_KABINI: - case CHIP_MULLINS: - buffer[count++] = cpu_to_le32(0x00000000); /* XXX */ - buffer[count++] = cpu_to_le32(0x00000000); - break; - case CHIP_HAWAII: - buffer[count++] = cpu_to_le32(0x3a00161a); - buffer[count++] = cpu_to_le32(0x0000002e); - break; - default: - buffer[count++] = cpu_to_le32(0x00000000); - buffer[count++] = cpu_to_le32(0x00000000); - break; - } + buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config); + buffer[count++] = cpu_to_le32(adev->gfx.config.rb_config[0][0].raster_config_1); - buffer[count++] = cpu_to_le32(PACKET3(PACKET3_PREAMBLE_CNTL, 0)); - buffer[count++] = cpu_to_le32(PACKET3_PREAMBLE_END_CLEAR_STATE); - - buffer[count++] = cpu_to_le32(PACKET3(PACKET3_CLEAR_STATE, 0)); - buffer[count++] = cpu_to_le32(0); + amdgpu_gfx_csb_preamble_end(buffer, count); } static void gfx_v7_0_init_pg(struct amdgpu_device *adev) @@ -4175,9 +4097,9 @@ static const struct amdgpu_rlc_funcs gfx_v7_0_rlc_funcs = { .update_spm_vmid = gfx_v7_0_update_spm_vmid }; -static int gfx_v7_0_early_init(void *handle) +static int gfx_v7_0_early_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; adev->gfx.xcc_mask = 1; adev->gfx.num_gfx_rings = GFX7_NUM_GFX_RINGS; @@ -4192,9 +4114,9 @@ static int gfx_v7_0_early_init(void *handle) return 0; } -static int gfx_v7_0_late_init(void *handle) +static int gfx_v7_0_late_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int r; r = amdgpu_irq_get(adev, &adev->gfx.priv_reg_irq, 0); @@ -4384,10 +4306,10 @@ static int gfx_v7_0_compute_ring_init(struct amdgpu_device *adev, int ring_id, return 0; } -static int gfx_v7_0_sw_init(void *handle) +static int gfx_v7_0_sw_init(struct amdgpu_ip_block *ip_block) { struct amdgpu_ring *ring; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j, k, r, ring_id; switch (adev->asic_type) { @@ -4477,12 +4399,17 @@ static int gfx_v7_0_sw_init(void *handle) gfx_v7_0_gpu_early_init(adev); + adev->gfx.gfx_supported_reset = + amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]); + adev->gfx.compute_supported_reset = + amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]); + return r; } -static int gfx_v7_0_sw_fini(void *handle) +static int gfx_v7_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->gfx.num_gfx_rings; i++) @@ -4506,10 +4433,10 @@ static int gfx_v7_0_sw_fini(void *handle) return 0; } -static int gfx_v7_0_hw_init(void *handle) +static int gfx_v7_0_hw_init(struct amdgpu_ip_block *ip_block) { int r; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; gfx_v7_0_constants_init(adev); @@ -4527,9 +4454,9 @@ static int gfx_v7_0_hw_init(void *handle) return r; } -static int gfx_v7_0_hw_fini(void *handle) +static int gfx_v7_0_hw_fini(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0); amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0); @@ -4540,23 +4467,19 @@ static int gfx_v7_0_hw_fini(void *handle) return 0; } -static int gfx_v7_0_suspend(void *handle) +static int gfx_v7_0_suspend(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; - - return gfx_v7_0_hw_fini(adev); + return gfx_v7_0_hw_fini(ip_block); } -static int gfx_v7_0_resume(void *handle) +static int gfx_v7_0_resume(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; - - return gfx_v7_0_hw_init(adev); + return gfx_v7_0_hw_init(ip_block); } -static bool gfx_v7_0_is_idle(void *handle) +static bool gfx_v7_0_is_idle(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; if (RREG32(mmGRBM_STATUS) & GRBM_STATUS__GUI_ACTIVE_MASK) return false; @@ -4564,11 +4487,11 @@ static bool gfx_v7_0_is_idle(void *handle) return true; } -static int gfx_v7_0_wait_for_idle(void *handle) +static int gfx_v7_0_wait_for_idle(struct amdgpu_ip_block *ip_block) { unsigned i; u32 tmp; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; for (i = 0; i < adev->usec_timeout; i++) { /* read MC_STATUS */ @@ -4581,11 +4504,11 @@ static int gfx_v7_0_wait_for_idle(void *handle) return -ETIMEDOUT; } -static int gfx_v7_0_soft_reset(void *handle) +static int gfx_v7_0_soft_reset(struct amdgpu_ip_block *ip_block) { u32 grbm_soft_reset = 0, srbm_soft_reset = 0; u32 tmp; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; /* GRBM_STATUS */ tmp = RREG32(mmGRBM_STATUS); @@ -4891,11 +4814,11 @@ static int gfx_v7_0_priv_inst_irq(struct amdgpu_device *adev, return 0; } -static int gfx_v7_0_set_clockgating_state(void *handle, +static int gfx_v7_0_set_clockgating_state(struct amdgpu_ip_block *ip_block, enum amd_clockgating_state state) { bool gate = false; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; if (state == AMD_CG_STATE_GATE) gate = true; @@ -4914,11 +4837,11 @@ static int gfx_v7_0_set_clockgating_state(void *handle, return 0; } -static int gfx_v7_0_set_powergating_state(void *handle, +static int gfx_v7_0_set_powergating_state(struct amdgpu_ip_block *ip_block, enum amd_powergating_state state) { bool gate = false; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; if (state == AMD_PG_STATE_GATE) gate = true; @@ -5044,6 +4967,7 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = { .insert_nop = amdgpu_ring_insert_nop, .pad_ib = amdgpu_ring_generic_pad_ib, .emit_wreg = gfx_v7_0_ring_emit_wreg, + .soft_recovery = gfx_v7_0_ring_soft_recovery, .emit_mem_sync = gfx_v7_0_emit_mem_sync_compute, }; @@ -5122,13 +5046,13 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev) gfx_v7_0_set_user_cu_inactive_bitmap( adev, disable_masks[i * 2 + j]); bitmap = gfx_v7_0_get_cu_active_bitmap(adev); - cu_info->bitmap[i][j] = bitmap; + cu_info->bitmap[0][i][j] = bitmap; - for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) { + for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) { if (bitmap & mask) { if (counter < ao_cu_num) ao_bitmap |= mask; - counter ++; + counter++; } mask <<= 1; } @@ -5150,8 +5074,7 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev) cu_info->lds_size = 64; } -const struct amdgpu_ip_block_version gfx_v7_1_ip_block = -{ +const struct amdgpu_ip_block_version gfx_v7_1_ip_block = { .type = AMD_IP_BLOCK_TYPE_GFX, .major = 7, .minor = 1, @@ -5159,8 +5082,7 @@ const struct amdgpu_ip_block_version gfx_v7_1_ip_block = .funcs = &gfx_v7_0_ip_funcs, }; -const struct amdgpu_ip_block_version gfx_v7_2_ip_block = -{ +const struct amdgpu_ip_block_version gfx_v7_2_ip_block = { .type = AMD_IP_BLOCK_TYPE_GFX, .major = 7, .minor = 2, @@ -5168,8 +5090,7 @@ const struct amdgpu_ip_block_version gfx_v7_2_ip_block = .funcs = &gfx_v7_0_ip_funcs, }; -const struct amdgpu_ip_block_version gfx_v7_3_ip_block = -{ +const struct amdgpu_ip_block_version gfx_v7_3_ip_block = { .type = AMD_IP_BLOCK_TYPE_GFX, .major = 7, .minor = 3, |
