aboutsummaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c13
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h6
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/mes_v11_0.c49
-rw-r--r--drivers/gpu/drm/amd/amdgpu/mes_v12_0.c9
-rw-r--r--drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c18
11 files changed, 81 insertions, 32 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
index ac108fca64fe..4bd61c169ca8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
@@ -278,7 +278,7 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev,
msg = RREG32(mmMP0_SMN_C2PMSG_33);
if (msg & 0x80000000)
break;
- usleep_range(1000, 1100);
+ msleep(1);
}
}
@@ -1500,6 +1500,7 @@ union gc_info {
struct gc_info_v1_0 v1;
struct gc_info_v1_1 v1_1;
struct gc_info_v1_2 v1_2;
+ struct gc_info_v1_3 v1_3;
struct gc_info_v2_0 v2;
struct gc_info_v2_1 v2_1;
};
@@ -1558,6 +1559,16 @@ static int amdgpu_discovery_get_gfx_info(struct amdgpu_device *adev)
adev->gfx.config.gc_gl1c_size_per_instance = le32_to_cpu(gc_info->v1_2.gc_gl1c_size_per_instance);
adev->gfx.config.gc_gl2c_per_gpu = le32_to_cpu(gc_info->v1_2.gc_gl2c_per_gpu);
}
+ if (le16_to_cpu(gc_info->v1.header.version_minor) >= 3) {
+ adev->gfx.config.gc_tcp_size_per_cu = le32_to_cpu(gc_info->v1_3.gc_tcp_size_per_cu);
+ adev->gfx.config.gc_tcp_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_tcp_cache_line_size);
+ adev->gfx.config.gc_instruction_cache_size_per_sqc = le32_to_cpu(gc_info->v1_3.gc_instruction_cache_size_per_sqc);
+ adev->gfx.config.gc_instruction_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_instruction_cache_line_size);
+ adev->gfx.config.gc_scalar_data_cache_size_per_sqc = le32_to_cpu(gc_info->v1_3.gc_scalar_data_cache_size_per_sqc);
+ adev->gfx.config.gc_scalar_data_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_scalar_data_cache_line_size);
+ adev->gfx.config.gc_tcc_size = le32_to_cpu(gc_info->v1_3.gc_tcc_size);
+ adev->gfx.config.gc_tcc_cache_line_size = le32_to_cpu(gc_info->v1_3.gc_tcc_cache_line_size);
+ }
break;
case 2:
adev->gfx.config.max_shader_engines = le32_to_cpu(gc_info->v2.gc_num_se);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index aad2027e5c7c..0e617dff8765 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -348,6 +348,9 @@ int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
return -EINVAL;
}
+ /* always clear VRAM */
+ flags |= AMDGPU_GEM_CREATE_VRAM_CLEARED;
+
/* create a gem object to contain this object in */
if (args->in.domains & (AMDGPU_GEM_DOMAIN_GDS |
AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA)) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index c770cb201e64..1849510a308a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -657,7 +657,7 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id)
uint64_t queue_mask = 0;
int r, i, j;
- if (adev->enable_mes)
+ if (adev->mes.enable_legacy_queue_map)
return amdgpu_gfx_mes_enable_kcq(adev, xcc_id);
if (!kiq->pmf || !kiq->pmf->kiq_map_queues || !kiq->pmf->kiq_set_resources)
@@ -719,7 +719,7 @@ int amdgpu_gfx_enable_kgq(struct amdgpu_device *adev, int xcc_id)
amdgpu_device_flush_hdp(adev, NULL);
- if (adev->enable_mes) {
+ if (adev->mes.enable_legacy_queue_map) {
for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
j = i + xcc_id * adev->gfx.num_gfx_rings;
r = amdgpu_mes_map_legacy_queue(adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
index ddda94e49db4..56cc58edbb4e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
@@ -240,6 +240,12 @@ struct amdgpu_gfx_config {
uint32_t gc_tcp_size_per_cu;
uint32_t gc_num_cu_per_sqc;
uint32_t gc_tcc_size;
+ uint32_t gc_tcp_cache_line_size;
+ uint32_t gc_instruction_cache_size_per_sqc;
+ uint32_t gc_instruction_cache_line_size;
+ uint32_t gc_scalar_data_cache_size_per_sqc;
+ uint32_t gc_scalar_data_cache_line_size;
+ uint32_t gc_tcc_cache_line_size;
};
struct amdgpu_cu_info {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
index 0bc837dab578..bcce1add4ef6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h
@@ -75,6 +75,7 @@ struct amdgpu_mes {
uint32_t sched_version;
uint32_t kiq_version;
+ bool enable_legacy_queue_map;
uint32_t total_max_queue;
uint32_t max_doorbell_slices;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
index 0c856005df6b..38face981c3e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
@@ -166,6 +166,9 @@ static ssize_t ta_if_load_debugfs_write(struct file *fp, const char *buf, size_t
if (ret)
return -EFAULT;
+ if (ta_bin_len > PSP_1_MEG)
+ return -EINVAL;
+
copy_pos += sizeof(uint32_t);
ta_bin = kzalloc(ta_bin_len, GFP_KERNEL);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
index 2957702fca0c..e444e621ddaa 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -4116,6 +4116,7 @@ static void gfx_v10_0_check_gfxoff_flag(struct amdgpu_device *adev)
static int gfx_v10_0_init_microcode(struct amdgpu_device *adev)
{
+ char fw_name[53];
char ucode_prefix[30];
const char *wks = "";
int err;
@@ -4149,8 +4150,8 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev)
amdgpu_gfx_cp_init_microcode(adev, AMDGPU_UCODE_ID_CP_CE);
if (!amdgpu_sriov_vf(adev)) {
- err = amdgpu_ucode_request(adev, &adev->gfx.rlc_fw,
- "amdgpu/%s_rlc.bin", ucode_prefix);
+ snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", ucode_prefix);
+ err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
if (err)
goto out;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
index 2c611b8577a7..e45d23e82878 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
@@ -3005,7 +3005,7 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m,
(order_base_2(prop->queue_size / 4) - 1));
tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE,
(order_base_2(AMDGPU_GPU_PAGE_SIZE / 4) - 1));
- tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 0);
+ tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, UNORD_DISPATCH, 1);
tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, TUNNEL_DISPATCH, 0);
tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, PRIV_STATE, 1);
tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, KMD_QUEUE, 1);
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
index 2ea8223eb969..8aded0a67037 100644
--- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c
@@ -693,6 +693,28 @@ static void mes_v11_0_free_ucode_buffers(struct amdgpu_device *adev,
(void **)&adev->mes.ucode_fw_ptr[pipe]);
}
+static void mes_v11_0_get_fw_version(struct amdgpu_device *adev)
+{
+ int pipe;
+
+ /* get MES scheduler/KIQ versions */
+ mutex_lock(&adev->srbm_mutex);
+
+ for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) {
+ soc21_grbm_select(adev, 3, pipe, 0, 0);
+
+ if (pipe == AMDGPU_MES_SCHED_PIPE)
+ adev->mes.sched_version =
+ RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
+ else if (pipe == AMDGPU_MES_KIQ_PIPE && adev->enable_mes_kiq)
+ adev->mes.kiq_version =
+ RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
+ }
+
+ soc21_grbm_select(adev, 0, 0, 0, 0);
+ mutex_unlock(&adev->srbm_mutex);
+}
+
static void mes_v11_0_enable(struct amdgpu_device *adev, bool enable)
{
uint64_t ucode_addr;
@@ -1062,18 +1084,6 @@ static int mes_v11_0_queue_init(struct amdgpu_device *adev,
mes_v11_0_queue_init_register(ring);
}
- /* get MES scheduler/KIQ versions */
- mutex_lock(&adev->srbm_mutex);
- soc21_grbm_select(adev, 3, pipe, 0, 0);
-
- if (pipe == AMDGPU_MES_SCHED_PIPE)
- adev->mes.sched_version = RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
- else if (pipe == AMDGPU_MES_KIQ_PIPE && adev->enable_mes_kiq)
- adev->mes.kiq_version = RREG32_SOC15(GC, 0, regCP_MES_GP3_LO);
-
- soc21_grbm_select(adev, 0, 0, 0, 0);
- mutex_unlock(&adev->srbm_mutex);
-
return 0;
}
@@ -1320,15 +1330,24 @@ static int mes_v11_0_kiq_hw_init(struct amdgpu_device *adev)
mes_v11_0_enable(adev, true);
+ mes_v11_0_get_fw_version(adev);
+
mes_v11_0_kiq_setting(&adev->gfx.kiq[0].ring);
r = mes_v11_0_queue_init(adev, AMDGPU_MES_KIQ_PIPE);
if (r)
goto failure;
- r = mes_v11_0_hw_init(adev);
- if (r)
- goto failure;
+ if ((adev->mes.sched_version & AMDGPU_MES_VERSION_MASK) >= 0x47)
+ adev->mes.enable_legacy_queue_map = true;
+ else
+ adev->mes.enable_legacy_queue_map = false;
+
+ if (adev->mes.enable_legacy_queue_map) {
+ r = mes_v11_0_hw_init(adev);
+ if (r)
+ goto failure;
+ }
return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c
index e39a58d262c9..a79a8adc3aa5 100644
--- a/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c
@@ -1266,6 +1266,7 @@ static int mes_v12_0_sw_init(void *handle)
adev->mes.funcs = &mes_v12_0_funcs;
adev->mes.kiq_hw_init = &mes_v12_0_kiq_hw_init;
adev->mes.kiq_hw_fini = &mes_v12_0_kiq_hw_fini;
+ adev->mes.enable_legacy_queue_map = true;
adev->mes.event_log_size = AMDGPU_MES_LOG_BUFFER_SIZE;
@@ -1422,9 +1423,11 @@ static int mes_v12_0_kiq_hw_init(struct amdgpu_device *adev)
mes_v12_0_set_hw_resources_1(&adev->mes, AMDGPU_MES_KIQ_PIPE);
}
- r = mes_v12_0_hw_init(adev);
- if (r)
- goto failure;
+ if (adev->mes.enable_legacy_queue_map) {
+ r = mes_v12_0_hw_init(adev);
+ if (r)
+ goto failure;
+ }
return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
index af1e90159ce3..2e72d445415f 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
@@ -176,14 +176,16 @@ static void sdma_v5_2_ring_set_wptr(struct amdgpu_ring *ring)
DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n",
ring->doorbell_index, ring->wptr << 2);
WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
- /* SDMA seems to miss doorbells sometimes when powergating kicks in.
- * Updating the wptr directly will wake it. This is only safe because
- * we disallow gfxoff in begin_use() and then allow it again in end_use().
- */
- WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR),
- lower_32_bits(ring->wptr << 2));
- WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),
- upper_32_bits(ring->wptr << 2));
+ if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(5, 2, 1)) {
+ /* SDMA seems to miss doorbells sometimes when powergating kicks in.
+ * Updating the wptr directly will wake it. This is only safe because
+ * we disallow gfxoff in begin_use() and then allow it again in end_use().
+ */
+ WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR),
+ lower_32_bits(ring->wptr << 2));
+ WREG32(sdma_v5_2_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),
+ upper_32_bits(ring->wptr << 2));
+ }
} else {
DRM_DEBUG("Not using doorbell -- "
"mmSDMA%i_GFX_RB_WPTR == 0x%08x "
Powered by cgit, Gruvboxed by Blaster4385.