From bfaf1883605fd0c0dbabacd67ed49708470d5ea4 Mon Sep 17 00:00:00 2001 From: Srinivasan Shanmugam Date: Tue, 8 Oct 2024 19:01:48 +0530 Subject: [PATCH 01/16] drm/amd/amdgpu: Fix double unlock in amdgpu_mes_add_ring MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit This patch addresses a double unlock issue in the amdgpu_mes_add_ring function. The mutex was being unlocked twice under certain error conditions, which could lead to undefined behavior. The fix ensures that the mutex is unlocked only once before jumping to the clean_up_memory label. The unlock operation is moved to just before the goto statement within the conditional block that checks the return value of amdgpu_ring_init. This prevents the second unlock attempt after the clean_up_memory label, which is no longer necessary as the mutex is already unlocked by this point in the code flow. This change resolves the potential double unlock and maintains the correct mutex handling throughout the function. Fixes below: Commit d0c423b64765 ("drm/amdgpu/mes: use ring for kernel queue submission"), leads to the following Smatch static checker warning: drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c:1240 amdgpu_mes_add_ring() warn: double unlock '&adev->mes.mutex_hidden' (orig line 1213) drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c 1143 int amdgpu_mes_add_ring(struct amdgpu_device *adev, int gang_id, 1144 int queue_type, int idx, 1145 struct amdgpu_mes_ctx_data *ctx_data, 1146 struct amdgpu_ring **out) 1147 { 1148 struct amdgpu_ring *ring; 1149 struct amdgpu_mes_gang *gang; 1150 struct amdgpu_mes_queue_properties qprops = {0}; 1151 int r, queue_id, pasid; 1152 1153 /* 1154 * Avoid taking any other locks under MES lock to avoid circular 1155 * lock dependencies. 1156 */ 1157 amdgpu_mes_lock(&adev->mes); 1158 gang = idr_find(&adev->mes.gang_id_idr, gang_id); 1159 if (!gang) { 1160 DRM_ERROR("gang id %d doesn't exist\n", gang_id); 1161 amdgpu_mes_unlock(&adev->mes); 1162 return -EINVAL; 1163 } 1164 pasid = gang->process->pasid; 1165 1166 ring = kzalloc(sizeof(struct amdgpu_ring), GFP_KERNEL); 1167 if (!ring) { 1168 amdgpu_mes_unlock(&adev->mes); 1169 return -ENOMEM; 1170 } 1171 1172 ring->ring_obj = NULL; 1173 ring->use_doorbell = true; 1174 ring->is_mes_queue = true; 1175 ring->mes_ctx = ctx_data; 1176 ring->idx = idx; 1177 ring->no_scheduler = true; 1178 1179 if (queue_type == AMDGPU_RING_TYPE_COMPUTE) { 1180 int offset = offsetof(struct amdgpu_mes_ctx_meta_data, 1181 compute[ring->idx].mec_hpd); 1182 ring->eop_gpu_addr = 1183 amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset); 1184 } 1185 1186 switch (queue_type) { 1187 case AMDGPU_RING_TYPE_GFX: 1188 ring->funcs = adev->gfx.gfx_ring[0].funcs; 1189 ring->me = adev->gfx.gfx_ring[0].me; 1190 ring->pipe = adev->gfx.gfx_ring[0].pipe; 1191 break; 1192 case AMDGPU_RING_TYPE_COMPUTE: 1193 ring->funcs = adev->gfx.compute_ring[0].funcs; 1194 ring->me = adev->gfx.compute_ring[0].me; 1195 ring->pipe = adev->gfx.compute_ring[0].pipe; 1196 break; 1197 case AMDGPU_RING_TYPE_SDMA: 1198 ring->funcs = adev->sdma.instance[0].ring.funcs; 1199 break; 1200 default: 1201 BUG(); 1202 } 1203 1204 r = amdgpu_ring_init(adev, ring, 1024, NULL, 0, 1205 AMDGPU_RING_PRIO_DEFAULT, NULL); 1206 if (r) 1207 goto clean_up_memory; 1208 1209 amdgpu_mes_ring_to_queue_props(adev, ring, &qprops); 1210 1211 dma_fence_wait(gang->process->vm->last_update, false); 1212 dma_fence_wait(ctx_data->meta_data_va->last_pt_update, false); 1213 amdgpu_mes_unlock(&adev->mes); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1214 1215 r = amdgpu_mes_add_hw_queue(adev, gang_id, &qprops, &queue_id); 1216 if (r) 1217 goto clean_up_ring; ^^^^^^^^^^^^^^^^^^ 1218 1219 ring->hw_queue_id = queue_id; 1220 ring->doorbell_index = qprops.doorbell_off; 1221 1222 if (queue_type == AMDGPU_RING_TYPE_GFX) 1223 sprintf(ring->name, "gfx_%d.%d.%d", pasid, gang_id, queue_id); 1224 else if (queue_type == AMDGPU_RING_TYPE_COMPUTE) 1225 sprintf(ring->name, "compute_%d.%d.%d", pasid, gang_id, 1226 queue_id); 1227 else if (queue_type == AMDGPU_RING_TYPE_SDMA) 1228 sprintf(ring->name, "sdma_%d.%d.%d", pasid, gang_id, 1229 queue_id); 1230 else 1231 BUG(); 1232 1233 *out = ring; 1234 return 0; 1235 1236 clean_up_ring: 1237 amdgpu_ring_fini(ring); 1238 clean_up_memory: 1239 kfree(ring); --> 1240 amdgpu_mes_unlock(&adev->mes); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1241 return r; 1242 } Fixes: d0c423b64765 ("drm/amdgpu/mes: use ring for kernel queue submission") Cc: Christian König Cc: Alex Deucher Cc: Hawking Zhang Suggested-by: Jack Xiao Reported by: Dan Carpenter Signed-off-by: Srinivasan Shanmugam Reviewed-by: Jack Xiao Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c index aab9632a06ce..6909af56fcad 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c @@ -1203,8 +1203,10 @@ int amdgpu_mes_add_ring(struct amdgpu_device *adev, int gang_id, r = amdgpu_ring_init(adev, ring, 1024, NULL, 0, AMDGPU_RING_PRIO_DEFAULT, NULL); - if (r) + if (r) { + amdgpu_mes_unlock(&adev->mes); goto clean_up_memory; + } amdgpu_mes_ring_to_queue_props(adev, ring, &qprops); @@ -1237,7 +1239,6 @@ clean_up_ring: amdgpu_ring_fini(ring); clean_up_memory: kfree(ring); - amdgpu_mes_unlock(&adev->mes); return r; } -- 2.51.0 From 98c0b0efcc11f2a5ddf3ce33af1e48eedf808b04 Mon Sep 17 00:00:00 2001 From: Philip Yang Date: Fri, 4 Oct 2024 16:28:07 -0400 Subject: [PATCH 02/16] drm/amdkfd: Accounting pdd vram_usage for svm Process device data pdd->vram_usage is read by rocm-smi via sysfs, this is currently missing the svm_bo usage accounting, so "rocm-smi --showpids" per process VRAM usage report is incorrect. Add pdd->vram_usage accounting when svm_bo allocation and release, change to atomic64_t type because it is updated outside process mutex now. Signed-off-by: Philip Yang Reviewed-by: Felix Kuehling Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdkfd/kfd_chardev.c | 6 +++--- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 2 +- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 4 ++-- drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 26 ++++++++++++++++++++++++ 4 files changed, 32 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index a1f191a5984b..065d87841459 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -1148,7 +1148,7 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep, if (flags & KFD_IOC_ALLOC_MEM_FLAGS_AQL_QUEUE_MEM) size >>= 1; - WRITE_ONCE(pdd->vram_usage, pdd->vram_usage + PAGE_ALIGN(size)); + atomic64_add(PAGE_ALIGN(size), &pdd->vram_usage); } mutex_unlock(&p->mutex); @@ -1219,7 +1219,7 @@ static int kfd_ioctl_free_memory_of_gpu(struct file *filep, kfd_process_device_remove_obj_handle( pdd, GET_IDR_HANDLE(args->handle)); - WRITE_ONCE(pdd->vram_usage, pdd->vram_usage - size); + atomic64_sub(size, &pdd->vram_usage); err_unlock: err_pdd: @@ -2347,7 +2347,7 @@ static int criu_restore_memory_of_gpu(struct kfd_process_device *pdd, } else if (bo_bucket->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) { bo_bucket->restored_offset = offset; /* Update the VRAM usage count */ - WRITE_ONCE(pdd->vram_usage, pdd->vram_usage + bo_bucket->size); + atomic64_add(bo_bucket->size, &pdd->vram_usage); } return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index 6a5bf88cc232..9e5ca0b93b2a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -775,7 +775,7 @@ struct kfd_process_device { enum kfd_pdd_bound bound; /* VRAM usage */ - uint64_t vram_usage; + atomic64_t vram_usage; struct attribute attr_vram; char vram_filename[MAX_SYSFS_FILENAME_LEN]; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index d665ecdcd12f..d4aa843aacfd 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -332,7 +332,7 @@ static ssize_t kfd_procfs_show(struct kobject *kobj, struct attribute *attr, } else if (strncmp(attr->name, "vram_", 5) == 0) { struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device, attr_vram); - return snprintf(buffer, PAGE_SIZE, "%llu\n", READ_ONCE(pdd->vram_usage)); + return snprintf(buffer, PAGE_SIZE, "%llu\n", atomic64_read(&pdd->vram_usage)); } else if (strncmp(attr->name, "sdma_", 5) == 0) { struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device, attr_sdma); @@ -1625,7 +1625,7 @@ struct kfd_process_device *kfd_create_process_device_data(struct kfd_node *dev, pdd->bound = PDD_UNBOUND; pdd->already_dequeued = false; pdd->runtime_inuse = false; - pdd->vram_usage = 0; + atomic64_set(&pdd->vram_usage, 0); pdd->sdma_past_activity_counter = 0; pdd->user_gpu_id = dev->id; atomic64_set(&pdd->evict_duration_counter, 0); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 857ec6f23bba..3e2911895c74 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -405,6 +405,27 @@ static void svm_range_bo_release(struct kref *kref) spin_lock(&svm_bo->list_lock); } spin_unlock(&svm_bo->list_lock); + + if (mmget_not_zero(svm_bo->eviction_fence->mm)) { + struct kfd_process_device *pdd; + struct kfd_process *p; + struct mm_struct *mm; + + mm = svm_bo->eviction_fence->mm; + /* + * The forked child process takes svm_bo device pages ref, svm_bo could be + * released after parent process is gone. + */ + p = kfd_lookup_process_by_mm(mm); + if (p) { + pdd = kfd_get_process_device_data(svm_bo->node, p); + if (pdd) + atomic64_sub(amdgpu_bo_size(svm_bo->bo), &pdd->vram_usage); + kfd_unref_process(p); + } + mmput(mm); + } + if (!dma_fence_is_signaled(&svm_bo->eviction_fence->base)) /* We're not in the eviction worker. Signal the fence. */ dma_fence_signal(&svm_bo->eviction_fence->base); @@ -532,6 +553,7 @@ int svm_range_vram_node_new(struct kfd_node *node, struct svm_range *prange, bool clear) { + struct kfd_process_device *pdd; struct amdgpu_bo_param bp; struct svm_range_bo *svm_bo; struct amdgpu_bo_user *ubo; @@ -623,6 +645,10 @@ svm_range_vram_node_new(struct kfd_node *node, struct svm_range *prange, list_add(&prange->svm_bo_list, &svm_bo->range_list); spin_unlock(&svm_bo->list_lock); + pdd = svm_range_get_pdd_by_node(prange, node); + if (pdd) + atomic64_add(amdgpu_bo_size(bo), &pdd->vram_usage); + return 0; reserve_bo_failed: -- 2.51.0 From d25d26b8a87a97ce3bd7f5db840f9ca72ba5d422 Mon Sep 17 00:00:00 2001 From: Lijo Lazar Date: Mon, 7 Oct 2024 13:49:45 +0530 Subject: [PATCH 03/16] drm/amdgpu: Wait for reset on init completion When reset on initialization is requested, wait for the reset to finish. In cases where module is loaded after boot, this makes sure all initialization work is done after a successful return of modprobe. Signed-off-by: Lijo Lazar Reviewed-by: Ramesh Errabolu Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 63503719303c..fcdbcff57632 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -1548,6 +1548,7 @@ static void amdgpu_xgmi_schedule_reset_on_init(struct amdgpu_hive_info *hive) int amdgpu_xgmi_reset_on_init(struct amdgpu_device *adev) { struct amdgpu_hive_info *hive; + bool reset_scheduled; int num_devs; hive = amdgpu_get_xgmi_hive(adev); @@ -1556,12 +1557,18 @@ int amdgpu_xgmi_reset_on_init(struct amdgpu_device *adev) mutex_lock(&hive->hive_lock); num_devs = atomic_read(&hive->number_devices); - if (num_devs == adev->gmc.xgmi.num_physical_nodes) + reset_scheduled = false; + if (num_devs == adev->gmc.xgmi.num_physical_nodes) { amdgpu_xgmi_schedule_reset_on_init(hive); + reset_scheduled = true; + } mutex_unlock(&hive->hive_lock); amdgpu_put_xgmi_hive(hive); + if (reset_scheduled) + flush_work(&hive->reset_on_init_work); + return 0; } -- 2.51.0 From 204f55736eb36b530cbcd6959477b44ed87a2213 Mon Sep 17 00:00:00 2001 From: Alex Hung Date: Fri, 20 Sep 2024 16:02:05 -0600 Subject: [PATCH 04/16] drm/amd/display: Check returns from drm_dp_dpcd_write [WHAT & HOW] drm_dp_dpcd_write() returns negative error on failure and thus returned values need to be checked. This fixes 3 UNUSED_VALUE issues reported by Coverity. Reviewed-by: Rodrigo Siqueira Signed-off-by: Alex Hung Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../amd/display/amdgpu_dm/amdgpu_dm_helpers.c | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index 6923184a6536..8eaf292bc4eb 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -643,6 +643,8 @@ static bool execute_synaptics_rc_command(struct drm_dp_aux *aux, // write rc data memmove(rc_data, data, length); ret = drm_dp_dpcd_write(aux, SYNAPTICS_RC_DATA, rc_data, sizeof(rc_data)); + if (ret < 0) + goto err; } // write rc offset @@ -651,20 +653,21 @@ static bool execute_synaptics_rc_command(struct drm_dp_aux *aux, rc_offset[2] = (unsigned char) (offset >> 16) & 0xFF; rc_offset[3] = (unsigned char) (offset >> 24) & 0xFF; ret = drm_dp_dpcd_write(aux, SYNAPTICS_RC_OFFSET, rc_offset, sizeof(rc_offset)); + if (ret < 0) + goto err; // write rc length rc_length[0] = (unsigned char) length & 0xFF; rc_length[1] = (unsigned char) (length >> 8) & 0xFF; ret = drm_dp_dpcd_write(aux, SYNAPTICS_RC_LENGTH, rc_length, sizeof(rc_length)); + if (ret < 0) + goto err; // write rc cmd rc_cmd = cmd | 0x80; ret = drm_dp_dpcd_write(aux, SYNAPTICS_RC_COMMAND, &rc_cmd, sizeof(rc_cmd)); - - if (ret < 0) { - DRM_ERROR("%s: write cmd ..., err = %d\n", __func__, ret); - return false; - } + if (ret < 0) + goto err; // poll until active is 0 for (i = 0; i < 10; i++) { @@ -687,6 +690,10 @@ static bool execute_synaptics_rc_command(struct drm_dp_aux *aux, drm_dbg_dp(aux->drm_dev, "success = %d\n", success); return success; + +err: + DRM_ERROR("%s: write cmd ..., err = %d\n", __func__, ret); + return false; } static void apply_synaptics_fifo_reset_wa(struct drm_dp_aux *aux) -- 2.51.0 From 9237c9504b0e4f738365f023539c6f3dc5a8bb49 Mon Sep 17 00:00:00 2001 From: Alex Hung Date: Mon, 30 Sep 2024 11:45:12 -0600 Subject: [PATCH 05/16] drm/amd/display: Check status from dpcd_get_tunneling_device_data [WHAT & HOW] dpcd_get_tunneling_device_data calls core_link_read_dpcd which can fail. The status from core_link_read_dpcd should be checked and error messages is printed in case of failures. This fixes 1 UNUSED_VALUE issue reported by Coverity. Reviewed-by: Rodrigo Siqueira Signed-off-by: Alex Hung Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../drm/amd/display/dc/link/protocols/link_dp_capability.c | 2 ++ .../gpu/drm/amd/display/dc/link/protocols/link_dp_dpia.c | 7 +++++++ 2 files changed, 9 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c index 286841456157..2ec968f239f7 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c @@ -1636,6 +1636,8 @@ static bool retrieve_link_cap(struct dc_link *link) /* Read DP tunneling information. */ status = dpcd_get_tunneling_device_data(link); + if (status != DC_OK) + dm_error("%s: Read tunndeling device data failed.\n", __func__); dpcd_set_source_specific_data(link); /* Sink may need to configure internals based on vendor, so allow some diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia.c index 6af42ba9885c..0d123e647652 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia.c @@ -59,12 +59,18 @@ enum dc_status dpcd_get_tunneling_device_data(struct dc_link *link) dpcd_dp_tun_data, sizeof(dpcd_dp_tun_data)); + if (status != DC_OK) + goto err; + status = core_link_read_dpcd( link, DP_USB4_ROUTER_TOPOLOGY_ID, dpcd_topology_data, sizeof(dpcd_topology_data)); + if (status != DC_OK) + goto err; + link->dpcd_caps.usb4_dp_tun_info.dp_tun_cap.raw = dpcd_dp_tun_data[DP_TUNNELING_CAPABILITIES_SUPPORT - DP_TUNNELING_CAPABILITIES_SUPPORT]; link->dpcd_caps.usb4_dp_tun_info.dpia_info.raw = @@ -75,6 +81,7 @@ enum dc_status dpcd_get_tunneling_device_data(struct dc_link *link) for (i = 0; i < DPCD_USB4_TOPOLOGY_ID_LEN; i++) link->dpcd_caps.usb4_dp_tun_info.usb4_topology_id[i] = dpcd_topology_data[i]; +err: return status; } -- 2.51.0 From e421808cbe6d0ed124d255b684eebb21437df14a Mon Sep 17 00:00:00 2001 From: Alex Hung Date: Mon, 30 Sep 2024 11:29:02 -0600 Subject: [PATCH 06/16] drm/amd/display: Remove useless assignments [WHAT & HOW] "split_pipe" are assigned to test_pipe and then immediately are updated to other values. The same also applies to "status" as well. Similarly, "id", "dwb" and "unused_dpps" are assigned but the functions immediately return, and thus they have no effects. As a results, the assignments removed. This fixes 5 UNUSED_VALUE issues reported by Coverity. Reviewed-by: Rodrigo Siqueira Signed-off-by: Alex Hung Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c | 2 -- drivers/gpu/drm/amd/display/dc/gpio/gpio_service.c | 1 - drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c | 1 - drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c | 1 - drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 1 - .../dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c | 1 - 6 files changed, 7 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c index 866b0abcff1b..9190c1328d5b 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c @@ -209,8 +209,6 @@ static bool optimize_configuration(struct dml2_context *dml2, struct dml2_wrappe p->cur_display_config->output.OutputEncoder[0], p->cur_mode_support_info->DSCEnabled[0]) - 1; if (odms_needed <= unused_dpps) { - unused_dpps -= odms_needed; - if (odms_needed == 1) { p->new_policy->ODMUse[0] = dml_odm_use_policy_combine_2to1; optimization_done = true; diff --git a/drivers/gpu/drm/amd/display/dc/gpio/gpio_service.c b/drivers/gpu/drm/amd/display/dc/gpio/gpio_service.c index f344478e9bd4..b099989d9364 100644 --- a/drivers/gpu/drm/amd/display/dc/gpio/gpio_service.c +++ b/drivers/gpu/drm/amd/display/dc/gpio/gpio_service.c @@ -443,7 +443,6 @@ struct gpio *dal_gpio_create_irq( case GPIO_ID_GPIO_PAD: break; default: - id = GPIO_ID_HPD; ASSERT_CRITICAL(false); return NULL; } diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c index a6a1db5ba8ba..681bb92c6069 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c @@ -3453,7 +3453,6 @@ static bool dcn10_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx) r2 = test_pipe->plane_res.scl_data.recout; r2_r = r2.x + r2.width; r2_b = r2.y + r2.height; - split_pipe = test_pipe; /** * There is another half plane on same layer because of diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c index bf9b5daea2af..9e1777f727c4 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c @@ -486,7 +486,6 @@ bool dcn30_mmhubbub_warmup( } /*following is the original: warmup each DWB's mcif buffer*/ for (i = 0; i < num_dwb; i++) { - dwb = dc->res_pool->dwbc[wb_info[i].dwb_pipe_inst]; mcif_wb = dc->res_pool->mcif_wb[wb_info[i].dwb_pipe_inst]; /*warmup is for VM mode only*/ if (wb_info[i].mcif_buf_params.p_vmid == 0) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 49a37f5ee28e..65176b59dcb2 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -1068,7 +1068,6 @@ static bool dcn401_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx) r2 = test_pipe->plane_res.scl_data.recout; r2_r = r2.x + r2.width; r2_b = r2.y + r2.height; - split_pipe = test_pipe; /** * There is another half plane on same layer because of diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c index b5cf75975fff..ccf8096dde29 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_fixed_vs_pe_retimer.c @@ -412,7 +412,6 @@ enum link_training_result dp_perform_fixed_vs_pe_training_sequence( /* 5. check CR done*/ if (dp_is_cr_done(lane_count, dpcd_lane_status)) { - status = LINK_TRAINING_SUCCESS; break; } -- 2.51.0 From 60612f75992d96955fb7154468c58d5d168cf1ab Mon Sep 17 00:00:00 2001 From: Roman Li Date: Mon, 30 Sep 2024 18:07:16 -0400 Subject: [PATCH 07/16] drm/amd/display: Increase idle worker HPD detection time [Why] Idle worker thread waits HPD_DETECTION_TIME for HPD processing complete. Some displays require longer time for that. [How] Increase HPD_DETECTION_TIME to 100ms. Reviewed-by: Sun peng Li Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c index 32fa643cd899..ffa4d3965b4b 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c @@ -36,7 +36,7 @@ #include "amdgpu_dm_debugfs.h" #define HPD_DETECTION_PERIOD_uS 5000000 -#define HPD_DETECTION_TIME_uS 1000 +#define HPD_DETECTION_TIME_uS 100000 void amdgpu_dm_crtc_handle_vblank(struct amdgpu_crtc *acrtc) { -- 2.51.0 From f79f4dd6001acd1e6ea6aae8e219060a566409b2 Mon Sep 17 00:00:00 2001 From: Roman Li Date: Mon, 30 Sep 2024 18:18:35 -0400 Subject: [PATCH 08/16] drm/amd/display: Add check for headless for idle optimization [Why] Currently idle worker thread that checks for HPD while system is in IPS2 only supports headless and static screen use-cases. In other display-off scenarios hotplug may not work. [How] For display-off only allow idle optimization when no display is connected. Reviewed-by: Sun peng Li Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 894442d2233c..8fdef03b04a1 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -9589,7 +9589,7 @@ static void amdgpu_dm_commit_streams(struct drm_atomic_state *state, WARN_ON(!dc_commit_streams(dm->dc, ¶ms)); /* Allow idle optimization when vblank count is 0 for display off */ - if (dm->active_vblank_irq_count == 0) + if ((dm->active_vblank_irq_count == 0) && amdgpu_dm_is_headless(dm->adev)) dc_allow_idle_optimizations(dm->dc, true); mutex_unlock(&dm->dc_lock); -- 2.51.0 From df60dcf5b58a642e531609d0d673bb63a11ed06e Mon Sep 17 00:00:00 2001 From: Ilya Bakoulin Date: Mon, 16 Sep 2024 14:38:19 -0400 Subject: [PATCH 09/16] drm/amd/display: Add 3DLUT FL HW bug workaround [Why] There is a known HW bug that causes the internal 3DLUT fetch signal to be lost at VREADY, regardless of whether the OTG lock is being held or not. A workaround is necessary to make sure that this internal signal stays up after OTG unlock. [How] Set the 3DLUT_ENABLE bit immediately before and after the unlock. Also use VUPDATE_KEEPOUT to prevent lock transition in the region between VSTARTUP and VREADY, which could cause issues with this WA sequence. Also including misc. 3DLUT DMA-related sequence fixes to address a few regressions causing corruption. Reviewed-by: Dillon Varone Signed-off-by: Ilya Bakoulin Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 30 +++++++--- .../amd/display/dc/hwss/dcn20/dcn20_hwseq.c | 8 ++- .../amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 60 ++++++++++++++++++- .../amd/display/dc/hwss/dcn401/dcn401_hwseq.h | 2 + .../amd/display/dc/hwss/dcn401/dcn401_init.c | 1 + .../display/dc/hwss/hw_sequencer_private.h | 1 + .../amd/display/dc/inc/hw/timing_generator.h | 2 + .../amd/display/dc/optc/dcn401/dcn401_optc.c | 31 ++++++++++ 8 files changed, 122 insertions(+), 13 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index d1e397d5f84e..743f3292d98e 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -2650,7 +2650,8 @@ static enum surface_update_type det_surface_update(const struct dc *dc, elevate_update_type(&overall_type, type); } - if (update_flags->bits.lut_3d) { + if (update_flags->bits.lut_3d && + u->surface->mcm_luts.lut3d_data.lut3d_src != DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM) { type = UPDATE_TYPE_FULL; elevate_update_type(&overall_type, type); } @@ -2926,10 +2927,20 @@ static void copy_surface_update_to_plane( sizeof(struct dc_transfer_func_distributed_points)); } - if (srf_update->func_shaper) + if (srf_update->cm2_params) { + surface->mcm_shaper_3dlut_setting = srf_update->cm2_params->component_settings.shaper_3dlut_setting; + surface->mcm_lut1d_enable = srf_update->cm2_params->component_settings.lut1d_enable; + surface->mcm_luts = srf_update->cm2_params->cm2_luts; + } + + if (srf_update->func_shaper) { memcpy(&surface->in_shaper_func, srf_update->func_shaper, sizeof(surface->in_shaper_func)); + if (surface->mcm_shaper_3dlut_setting >= DC_CM2_SHAPER_3DLUT_SETTING_ENABLE_SHAPER) + surface->mcm_luts.shaper = &surface->in_shaper_func; + } + if (srf_update->lut3d_func) memcpy(&surface->lut3d_func, srf_update->lut3d_func, sizeof(surface->lut3d_func)); @@ -2942,10 +2953,17 @@ static void copy_surface_update_to_plane( surface->sdr_white_level_nits = srf_update->sdr_white_level_nits; - if (srf_update->blend_tf) + if (srf_update->blend_tf) { memcpy(&surface->blend_tf, srf_update->blend_tf, sizeof(surface->blend_tf)); + if (surface->mcm_lut1d_enable) + surface->mcm_luts.lut1d_func = &surface->blend_tf; + } + + if (srf_update->cm2_params || srf_update->blend_tf) + surface->lut_bank_a = !surface->lut_bank_a; + if (srf_update->input_csc_color_matrix) surface->input_csc_color_matrix = *srf_update->input_csc_color_matrix; @@ -2957,11 +2975,7 @@ static void copy_surface_update_to_plane( if (srf_update->gamut_remap_matrix) surface->gamut_remap_matrix = *srf_update->gamut_remap_matrix; - if (srf_update->cm2_params) { - surface->mcm_shaper_3dlut_setting = srf_update->cm2_params->component_settings.shaper_3dlut_setting; - surface->mcm_lut1d_enable = srf_update->cm2_params->component_settings.lut1d_enable; - surface->mcm_luts = srf_update->cm2_params->cm2_luts; - } + if (srf_update->cursor_csc_color_matrix) surface->cursor_csc_color_matrix = *srf_update->cursor_csc_color_matrix; diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c index 1a32e53c1b22..9da5b50bea8a 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c @@ -1458,8 +1458,12 @@ void dcn20_pipe_control_lock( } else { if (lock) pipe->stream_res.tg->funcs->lock(pipe->stream_res.tg); - else - pipe->stream_res.tg->funcs->unlock(pipe->stream_res.tg); + else { + if (dc->hwseq->funcs.perform_3dlut_wa_unlock) + dc->hwseq->funcs.perform_3dlut_wa_unlock(pipe); + else + pipe->stream_res.tg->funcs->unlock(pipe->stream_res.tg); + } } } diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 65176b59dcb2..413dcbf4decf 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -506,7 +506,7 @@ void dcn401_populate_mcm_luts(struct dc *dc, dcn401_get_mcm_lut_xable_from_pipe_ctx(dc, pipe_ctx, &shaper_xable, &lut3d_xable, &lut1d_xable); /* 1D LUT */ - if (mcm_luts.lut1d_func && lut3d_xable != MCM_LUT_DISABLE) { + if (mcm_luts.lut1d_func) { memset(&m_lut_params, 0, sizeof(m_lut_params)); if (mcm_luts.lut1d_func->type == TF_TYPE_HWPWL) m_lut_params.pwl = &mcm_luts.lut1d_func->pwl; @@ -521,7 +521,7 @@ void dcn401_populate_mcm_luts(struct dc *dc, mpc->funcs->populate_lut(mpc, MCM_LUT_1DLUT, m_lut_params, lut_bank_a, mpcc_id); } if (mpc->funcs->program_lut_mode) - mpc->funcs->program_lut_mode(mpc, MCM_LUT_1DLUT, lut1d_xable, lut_bank_a, mpcc_id); + mpc->funcs->program_lut_mode(mpc, MCM_LUT_1DLUT, lut1d_xable && m_lut_params.pwl, lut_bank_a, mpcc_id); } /* Shaper */ @@ -669,11 +669,17 @@ bool dcn401_set_mcm_luts(struct pipe_ctx *pipe_ctx, { struct dpp *dpp_base = pipe_ctx->plane_res.dpp; int mpcc_id = pipe_ctx->plane_res.hubp->inst; - struct mpc *mpc = pipe_ctx->stream_res.opp->ctx->dc->res_pool->mpc; + struct dc *dc = pipe_ctx->stream_res.opp->ctx->dc; + struct mpc *mpc = dc->res_pool->mpc; bool result; const struct pwl_params *lut_params = NULL; bool rval; + if (plane_state->mcm_luts.lut3d_data.lut3d_src == DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM) { + dcn401_populate_mcm_luts(dc, pipe_ctx, plane_state->mcm_luts, plane_state->lut_bank_a); + return true; + } + mpc->funcs->set_movable_cm_location(mpc, MPCC_MOVABLE_CM_LOCATION_BEFORE, mpcc_id); pipe_ctx->plane_state->mcm_location = MPCC_MOVABLE_CM_LOCATION_BEFORE; // 1D LUT @@ -1814,6 +1820,54 @@ void dcn401_interdependent_update_lock(struct dc *dc, } } +void dcn401_perform_3dlut_wa_unlock(struct pipe_ctx *pipe_ctx) +{ + /* If 3DLUT FL is enabled and 3DLUT is in use, follow the workaround sequence for pipe unlock to make sure that + * HUBP will properly fetch 3DLUT contents after unlock. + * + * This is meant to work around a known HW issue where VREADY will cancel the pending 3DLUT_ENABLE signal regardless + * of whether OTG lock is currently being held or not. + */ + struct pipe_ctx *wa_pipes[MAX_PIPES] = { NULL }; + struct pipe_ctx *odm_pipe, *mpc_pipe; + int i, wa_pipe_ct = 0; + + for (odm_pipe = pipe_ctx; odm_pipe != NULL; odm_pipe = odm_pipe->next_odm_pipe) { + for (mpc_pipe = odm_pipe; mpc_pipe != NULL; mpc_pipe = mpc_pipe->bottom_pipe) { + if (mpc_pipe->plane_state && mpc_pipe->plane_state->mcm_luts.lut3d_data.lut3d_src + == DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM + && mpc_pipe->plane_state->mcm_shaper_3dlut_setting + == DC_CM2_SHAPER_3DLUT_SETTING_ENABLE_SHAPER_3DLUT) { + wa_pipes[wa_pipe_ct++] = mpc_pipe; + } + } + } + + if (wa_pipe_ct > 0) { + if (pipe_ctx->stream_res.tg->funcs->set_vupdate_keepout) + pipe_ctx->stream_res.tg->funcs->set_vupdate_keepout(pipe_ctx->stream_res.tg, true); + + for (i = 0; i < wa_pipe_ct; ++i) { + if (wa_pipes[i]->plane_res.hubp->funcs->hubp_enable_3dlut_fl) + wa_pipes[i]->plane_res.hubp->funcs->hubp_enable_3dlut_fl(wa_pipes[i]->plane_res.hubp, true); + } + + pipe_ctx->stream_res.tg->funcs->unlock(pipe_ctx->stream_res.tg); + if (pipe_ctx->stream_res.tg->funcs->wait_update_lock_status) + pipe_ctx->stream_res.tg->funcs->wait_update_lock_status(pipe_ctx->stream_res.tg, false); + + for (i = 0; i < wa_pipe_ct; ++i) { + if (wa_pipes[i]->plane_res.hubp->funcs->hubp_enable_3dlut_fl) + wa_pipes[i]->plane_res.hubp->funcs->hubp_enable_3dlut_fl(wa_pipes[i]->plane_res.hubp, true); + } + + if (pipe_ctx->stream_res.tg->funcs->set_vupdate_keepout) + pipe_ctx->stream_res.tg->funcs->set_vupdate_keepout(pipe_ctx->stream_res.tg, false); + } else { + pipe_ctx->stream_res.tg->funcs->unlock(pipe_ctx->stream_res.tg); + } +} + void dcn401_program_outstanding_updates(struct dc *dc, struct dc_state *context) { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h index 66d679080c44..28a513dfc005 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h @@ -95,4 +95,6 @@ void dcn401_reset_back_end_for_pipe( void dcn401_reset_hw_ctx_wrap( struct dc *dc, struct dc_state *context); +void dcn401_perform_3dlut_wa_unlock(struct pipe_ctx *pipe_ctx); + #endif /* __DC_HWSS_DCN401_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c index a1392e776709..c73305e57d39 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c @@ -138,6 +138,7 @@ static const struct hwseq_private_funcs dcn401_private_funcs = { .apply_single_controller_ctx_to_hw = dce110_apply_single_controller_ctx_to_hw, .reset_back_end_for_pipe = dcn401_reset_back_end_for_pipe, .populate_mcm_luts = NULL, + .perform_3dlut_wa_unlock = dcn401_perform_3dlut_wa_unlock, }; void dcn401_hw_sequencer_init_functions(struct dc *dc) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h index 0ac675456979..22a5d4a03c98 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h @@ -182,6 +182,7 @@ struct hwseq_private_funcs { struct pipe_ctx *pipe_ctx, struct dc_cm2_func_luts mcm_luts, bool lut_bank_a); + void (*perform_3dlut_wa_unlock)(struct pipe_ctx *pipe_ctx); }; struct dce_hwseq { diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h index 4e08e80eafe8..b74e18cc1e66 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h @@ -345,6 +345,8 @@ struct timing_generator_funcs { bool (*get_optc_double_buffer_pending)(struct timing_generator *tg); bool (*get_otg_double_buffer_pending)(struct timing_generator *tg); bool (*get_pipe_update_pending)(struct timing_generator *tg); + void (*set_vupdate_keepout)(struct timing_generator *tg, bool enable); + bool (*wait_update_lock_status)(struct timing_generator *tg, bool locked); }; #endif diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c index db670fc17264..783ca9acc762 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c @@ -430,6 +430,35 @@ static void optc401_program_global_sync( REG_UPDATE(OTG_PSTATE_REGISTER, OTG_PSTATE_KEEPOUT_START, pstate_keepout); } +static void optc401_set_vupdate_keepout(struct timing_generator *tg, bool enable) +{ + struct optc *optc1 = DCN10TG_FROM_TG(tg); + + REG_SET_3(OTG_VUPDATE_KEEPOUT, 0, + MASTER_UPDATE_LOCK_VUPDATE_KEEPOUT_START_OFFSET, 0, + MASTER_UPDATE_LOCK_VUPDATE_KEEPOUT_END_OFFSET, optc1->vready_offset + 10, + OTG_MASTER_UPDATE_LOCK_VUPDATE_KEEPOUT_EN, enable); + + return; +} + +static bool optc401_wait_update_lock_status(struct timing_generator *tg, bool locked) +{ + struct optc *optc1 = DCN10TG_FROM_TG(tg); + uint32_t lock_status = 0; + + REG_WAIT(OTG_MASTER_UPDATE_LOCK, + UPDATE_LOCK_STATUS, locked, + 1, 150000); + + REG_GET(OTG_MASTER_UPDATE_LOCK, UPDATE_LOCK_STATUS, &lock_status); + + if (lock_status != locked) + return false; + + return true; +} + static struct timing_generator_funcs dcn401_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, @@ -496,6 +525,8 @@ static struct timing_generator_funcs dcn401_tg_funcs = { .get_optc_double_buffer_pending = optc3_get_optc_double_buffer_pending, .get_otg_double_buffer_pending = optc3_get_otg_update_pending, .get_pipe_update_pending = optc3_get_pipe_update_pending, + .set_vupdate_keepout = optc401_set_vupdate_keepout, + .wait_update_lock_status = optc401_wait_update_lock_status, }; void dcn401_timing_generator_init(struct optc *optc1) -- 2.51.0 From 74a55071dd9c1f75abc942bda964cab3684c780c Mon Sep 17 00:00:00 2001 From: "Leo (Hanghong) Ma" Date: Thu, 19 Sep 2024 15:19:29 -0400 Subject: [PATCH 10/16] drm/amd/display: Fix Coverity change for visual confirm [Why && How] Previous change for Coverity has caused regression on visual confirm so fix it by reverting the part that affects visual confirm. Reviewed-by: Chris Park Signed-off-by: Leo (Hanghong) Ma Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c index 9291f078ffde..8486c144d81c 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c @@ -519,7 +519,8 @@ void dc_dmub_srv_get_visual_confirm_color_cmd(struct dc *dc, struct pipe_ctx *pi union dmub_rb_cmd cmd = { 0 }; unsigned int panel_inst = 0; - if (!dc_get_edp_link_panel_inst(dc, pipe_ctx->stream->link, &panel_inst)) + if (!dc_get_edp_link_panel_inst(dc, pipe_ctx->stream->link, &panel_inst) && + dc->debug.visual_confirm == VISUAL_CONFIRM_DISABLE) return; memset(&cmd, 0, sizeof(cmd)); -- 2.51.0 From 0fe33e115fec305c35c66b78ad26e3755ab54b9c Mon Sep 17 00:00:00 2001 From: Leo Chen Date: Thu, 3 Oct 2024 12:20:23 -0400 Subject: [PATCH 11/16] drm/amd/display: Full exit out of IPS2 when all allow signals have been cleared [Why] A race condition occurs between cursor movement and vertical interrupt control thread from OS, with both threads trying to exit IPS2. Vertical interrupt control thread clears the prev driver allow signal while not fully finishing the IPS2 exit process. [How] We want to detect all the allow signals have been cleared before we perform the full exit. Reviewed-by: Nicholas Kazlauskas Signed-off-by: Leo Chen Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c | 6 ++++-- drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 3 ++- 2 files changed, 6 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c index 8486c144d81c..f0417ee6fcf8 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c @@ -1295,6 +1295,8 @@ static void dc_dmub_srv_notify_idle(const struct dc *dc, bool allow_idle) memset(&new_signals, 0, sizeof(new_signals)); + new_signals.bits.allow_idle = 1; /* always set */ + if (dc->config.disable_ips == DMUB_IPS_ENABLE || dc->config.disable_ips == DMUB_IPS_DISABLE_DYNAMIC) { new_signals.bits.allow_pg = 1; @@ -1390,7 +1392,7 @@ static void dc_dmub_srv_exit_low_power_state(const struct dc *dc) */ dc_dmub_srv->needs_idle_wake = false; - if (prev_driver_signals.bits.allow_ips2 && + if ((prev_driver_signals.bits.allow_ips2 || prev_driver_signals.all == 0) && (!dc->debug.optimize_ips_handshake || ips_fw->signals.bits.ips2_commit || !ips_fw->signals.bits.in_idle)) { DC_LOG_IPS( @@ -1451,7 +1453,7 @@ static void dc_dmub_srv_exit_low_power_state(const struct dc *dc) } dc_dmub_srv_notify_idle(dc, false); - if (prev_driver_signals.bits.allow_ips1) { + if (prev_driver_signals.bits.allow_ips1 || prev_driver_signals.all == 0) { DC_LOG_IPS( "wait for IPS1 commit clear (ips1_commit=%u ips2_commit=%u)", ips_fw->signals.bits.ips1_commit, diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h index f9e82c4f8deb..6edd3d34c7b5 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h @@ -752,7 +752,8 @@ union dmub_shared_state_ips_driver_signals { uint32_t allow_ips1 : 1; /**< 1 is IPS1 is allowed */ uint32_t allow_ips2 : 1; /**< 1 is IPS1 is allowed */ uint32_t allow_z10 : 1; /**< 1 if Z10 is allowed */ - uint32_t reserved_bits : 28; /**< Reversed bits */ + uint32_t allow_idle : 1; /**< 1 if driver is allowing idle */ + uint32_t reserved_bits : 27; /**< Reversed bits */ } bits; uint32_t all; }; -- 2.51.0 From ad6b76cd9145cd798655c18781736a2534f378c6 Mon Sep 17 00:00:00 2001 From: Alex Hung Date: Wed, 2 Oct 2024 17:55:11 -0600 Subject: [PATCH 12/16] drm/amd/display: print messages when programming shaper/3dlut fails [WHAT & HOW] Print error messages when programming shaper lut or 3dlut fails. This fixes 5 UNUSED_VALUE issues reported by Coverity. Reviewed-by: Rodrigo Siqueira Signed-off-by: Alex Hung Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c | 7 +++++++ .../gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c | 13 +++++++------ .../display/dc/link/protocols/link_dp_training.c | 11 +++++++++++ 3 files changed, 25 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c index 9e1777f727c4..e89ebfda4873 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c @@ -245,6 +245,7 @@ static bool dcn30_set_mpc_shaper_3dlut(struct pipe_ctx *pipe_ctx, { struct dpp *dpp_base = pipe_ctx->plane_res.dpp; int mpcc_id = pipe_ctx->plane_res.hubp->inst; + struct dc *dc = pipe_ctx->stream->ctx->dc; struct mpc *mpc = pipe_ctx->stream_res.opp->ctx->dc->res_pool->mpc; bool result = false; int acquired_rmu = 0; @@ -283,8 +284,14 @@ static bool dcn30_set_mpc_shaper_3dlut(struct pipe_ctx *pipe_ctx, result = mpc->funcs->program_3dlut(mpc, &stream->lut3d_func->lut_3d, stream->lut3d_func->state.bits.rmu_mux_num); + if (!result) + DC_LOG_ERROR("%s: program_3dlut failed\n", __func__); + result = mpc->funcs->program_shaper(mpc, shaper_lut, stream->lut3d_func->state.bits.rmu_mux_num); + if (!result) + DC_LOG_ERROR("%s: program_shaper failed\n", __func__); + } else { // loop through the available mux and release the requested mpcc_id mpc->funcs->release_rmu(mpc, mpcc_id); diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c index 119c6d08200f..d7f8b2dcaa6b 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c @@ -439,6 +439,7 @@ bool dcn32_set_mpc_shaper_3dlut( { struct dpp *dpp_base = pipe_ctx->plane_res.dpp; int mpcc_id = pipe_ctx->plane_res.hubp->inst; + struct dc *dc = pipe_ctx->stream->ctx->dc; struct mpc *mpc = pipe_ctx->stream_res.opp->ctx->dc->res_pool->mpc; bool result = false; @@ -458,13 +459,13 @@ bool dcn32_set_mpc_shaper_3dlut( if (stream->lut3d_func && stream->lut3d_func->state.bits.initialized == 1) { - result = mpc->funcs->program_3dlut(mpc, - &stream->lut3d_func->lut_3d, - mpcc_id); + result = mpc->funcs->program_3dlut(mpc, &stream->lut3d_func->lut_3d, mpcc_id); + if (!result) + DC_LOG_ERROR("%s: program_3dlut failed\n", __func__); - result = mpc->funcs->program_shaper(mpc, - shaper_lut, - mpcc_id); + result = mpc->funcs->program_shaper(mpc, shaper_lut, mpcc_id); + if (!result) + DC_LOG_ERROR("%s: program_shaper failed\n", __func__); } return result; diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c index 27b881f947e8..a42c56a57253 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c @@ -1107,9 +1107,13 @@ enum dc_status dpcd_set_link_settings( status = core_link_write_dpcd(link, DP_DOWNSPREAD_CTRL, &downspread.raw, sizeof(downspread)); + if (status != DC_OK) + DC_LOG_ERROR("%s:%d: core_link_write_dpcd (DP_DOWNSPREAD_CTRL) failed\n", __func__, __LINE__); status = core_link_write_dpcd(link, DP_LANE_COUNT_SET, &lane_count_set.raw, 1); + if (status != DC_OK) + DC_LOG_ERROR("%s:%d: core_link_write_dpcd (DP_LANE_COUNT_SET) failed\n", __func__, __LINE__); if (link->dpcd_caps.dpcd_rev.raw >= DPCD_REV_13 && lt_settings->link_settings.use_link_rate_set == true) { @@ -1125,12 +1129,19 @@ enum dc_status dpcd_set_link_settings( supported_link_rates, sizeof(supported_link_rates)); } status = core_link_write_dpcd(link, DP_LINK_BW_SET, &rate, 1); + if (status != DC_OK) + DC_LOG_ERROR("%s:%d: core_link_write_dpcd (DP_LINK_BW_SET) failed\n", __func__, __LINE__); + status = core_link_write_dpcd(link, DP_LINK_RATE_SET, <_settings->link_settings.link_rate_set, 1); + if (status != DC_OK) + DC_LOG_ERROR("%s:%d: core_link_write_dpcd (DP_LINK_RATE_SET) failed\n", __func__, __LINE__); } else { rate = get_dpcd_link_rate(<_settings->link_settings); status = core_link_write_dpcd(link, DP_LINK_BW_SET, &rate, 1); + if (status != DC_OK) + DC_LOG_ERROR("%s:%d: core_link_write_dpcd (DP_LINK_BW_SET) failed\n", __func__, __LINE__); } if (rate) { -- 2.51.0 From 668d6808e9f23305e0eb108dbf75fc13ab344ae8 Mon Sep 17 00:00:00 2001 From: Robin Chen Date: Thu, 12 Sep 2024 20:59:36 +0800 Subject: [PATCH 13/16] drm/amd/display: Read Sink emission rate capability [WHY] To get sink emission rate information for future supported refresh rate calculation. Reviewed-by: ChunTao Tso Signed-off-by: Robin Chen Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc_dp_types.h | 3 ++- .../dc/link/protocols/link_dp_capability.c | 8 ++++++++ .../gpu/drm/amd/display/include/dpcd_defs.h | 19 ++++++++++--------- 3 files changed, 20 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h index 50fa8547d718..8dd6eb044829 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h @@ -1192,6 +1192,7 @@ struct dpcd_caps { struct edp_psr_info psr_info; struct replay_info pr_info; + uint16_t edp_oled_emission_rate; }; union dpcd_sink_ext_caps { @@ -1205,7 +1206,7 @@ union dpcd_sink_ext_caps { uint8_t oled : 1; uint8_t reserved_2 : 1; uint8_t miniled : 1; - uint8_t reserved : 1; + uint8_t emission_output : 1; } bits; uint8_t raw; }; diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c index 2ec968f239f7..952e787493c4 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c @@ -2067,6 +2067,14 @@ void detect_edp_sink_caps(struct dc_link *link) core_link_read_dpcd(link, DP_SINK_PR_MAX_NUMBER_OF_DEVIATION_LINE, &link->dpcd_caps.pr_info.max_deviation_line, sizeof(link->dpcd_caps.pr_info.max_deviation_line)); + + /* + * OLED Emission Rate info + */ + if (link->dpcd_sink_ext_caps.bits.emission_output) + core_link_read_dpcd(link, DP_SINK_EMISSION_RATE, + (uint8_t *)&link->dpcd_caps.edp_oled_emission_rate, + sizeof(link->dpcd_caps.edp_oled_emission_rate)); } bool dp_get_max_link_enc_cap(const struct dc_link *link, struct dc_link_settings *max_link_enc_cap) diff --git a/drivers/gpu/drm/amd/display/include/dpcd_defs.h b/drivers/gpu/drm/amd/display/include/dpcd_defs.h index aee5170f5fb2..de8f3cfed6c8 100644 --- a/drivers/gpu/drm/amd/display/include/dpcd_defs.h +++ b/drivers/gpu/drm/amd/display/include/dpcd_defs.h @@ -164,18 +164,19 @@ enum dpcd_psr_sink_states { PSR_SINK_STATE_SINK_INTERNAL_ERROR = 7, }; -#define DP_SOURCE_SEQUENCE 0x30c -#define DP_SOURCE_TABLE_REVISION 0x310 -#define DP_SOURCE_PAYLOAD_SIZE 0x311 -#define DP_SOURCE_SINK_CAP 0x317 -#define DP_SOURCE_BACKLIGHT_LEVEL 0x320 -#define DP_SOURCE_BACKLIGHT_CURRENT_PEAK 0x326 -#define DP_SOURCE_BACKLIGHT_CONTROL 0x32E -#define DP_SOURCE_BACKLIGHT_ENABLE 0x32F -#define DP_SOURCE_MINIMUM_HBLANK_SUPPORTED 0x340 +#define DP_SOURCE_SEQUENCE 0x30C +#define DP_SOURCE_TABLE_REVISION 0x310 +#define DP_SOURCE_PAYLOAD_SIZE 0x311 +#define DP_SOURCE_SINK_CAP 0x317 +#define DP_SOURCE_BACKLIGHT_LEVEL 0x320 +#define DP_SOURCE_BACKLIGHT_CURRENT_PEAK 0x326 +#define DP_SOURCE_BACKLIGHT_CONTROL 0x32E +#define DP_SOURCE_BACKLIGHT_ENABLE 0x32F +#define DP_SOURCE_MINIMUM_HBLANK_SUPPORTED 0x340 #define DP_SINK_PR_REPLAY_STATUS 0x378 #define DP_SINK_PR_PIXEL_DEVIATION_PER_LINE 0x379 #define DP_SINK_PR_MAX_NUMBER_OF_DEVIATION_LINE 0x37A +#define DP_SINK_EMISSION_RATE 0x37E /* Remove once drm_dp_helper.h is updated upstream */ #ifndef DP_TOTAL_LTTPR_CNT -- 2.51.0 From c5a332f98da56faf54d6a647d903d7272de83320 Mon Sep 17 00:00:00 2001 From: Dillon Varone Date: Wed, 25 Sep 2024 20:18:07 -0400 Subject: [PATCH 14/16] drm/amd/display: resolve correct MALL size for dcn401 [WHY] Code for dcn401 to calculate available MALL size for display was shared with dcn32 and did not provide the correct result for all ASICs. [HOW] Add dcn401 specific function to properly calculate the available MALL for display. Reviewed-by: Chris Park Signed-off-by: Dillon Varone Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../dc/resource/dcn32/dcn32_resource.c | 4 +++ .../dc/resource/dcn401/dcn401_resource.c | 27 ++++++++++++++++--- 2 files changed, 28 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c index a1890df29f5c..aaaa888d112d 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c @@ -1990,6 +1990,10 @@ unsigned int dcn32_calculate_mall_ways_from_bytes(const struct dc *dc, unsigned return 0; } + if (dc->caps.max_cab_allocation_bytes == 0) { + return 0xffffffff; + } + /* add 2 lines for worst case alignment */ cache_lines_used = total_size_in_mall_bytes / dc->caps.cache_line_size + 2; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c index 59184abab1a7..7da81966b4df 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c @@ -1704,6 +1704,29 @@ static int dcn401_get_power_profile(const struct dc_state *context) return dpm_level; } +static unsigned int dcn401_calc_num_avail_chans_for_mall(struct dc *dc, unsigned int num_chans) +{ + unsigned int num_available_chans = 1; + + /* channels for MALL must be a power of 2 */ + while (num_chans > 1) { + num_available_chans = (num_available_chans << 1); + num_chans = (num_chans >> 1); + } + + /* cannot be odd */ + num_available_chans &= ~1; + + /* clamp to max available channels for MALL per ASIC */ + if (ASICREV_IS_GC_12_0_0_A0(dc->ctx->asic_id.hw_internal_rev)) { + num_available_chans = num_available_chans > 16 ? 16 : num_available_chans; + } else if (ASICREV_IS_GC_12_0_1_A0(dc->ctx->asic_id.hw_internal_rev)) { + num_available_chans = num_available_chans > 8 ? 8 : num_available_chans; + } + + return num_available_chans; +} + static struct resource_funcs dcn401_res_pool_funcs = { .destroy = dcn401_destroy_resource_pool, .link_enc_create = dcn401_link_encoder_create, @@ -1812,14 +1835,12 @@ static bool dcn401_resource_construct( dc->caps.min_horizontal_blanking_period = 80; dc->caps.dmdata_alloc_size = 2048; dc->caps.mall_size_per_mem_channel = 4; - /* total size = mall per channel * num channels * 1024 * 1024 */ - dc->caps.mall_size_total = dc->caps.mall_size_per_mem_channel * dc->ctx->dc_bios->vram_info.num_chans * 1048576; dc->caps.cursor_cache_size = dc->caps.max_cursor_size * dc->caps.max_cursor_size * 8; dc->caps.cache_line_size = 64; dc->caps.cache_num_ways = 16; /* Calculate the available MALL space */ - dc->caps.max_cab_allocation_bytes = dcn32_calc_num_avail_chans_for_mall( + dc->caps.max_cab_allocation_bytes = dcn401_calc_num_avail_chans_for_mall( dc, dc->ctx->dc_bios->vram_info.num_chans) * dc->caps.mall_size_per_mem_channel * 1024 * 1024; dc->caps.mall_size_total = dc->caps.max_cab_allocation_bytes; -- 2.51.0 From c6335dad37c1332124f8d6ed5dcb046614fc9902 Mon Sep 17 00:00:00 2001 From: Samson Tam Date: Wed, 2 Oct 2024 21:51:18 -0400 Subject: [PATCH 15/16] drm/amd/display: add sharpening policy to plane state [Why] Pass in sharpening policy through plane state from control side [How] Add sharpener support through dc_caps. Add sharpen policy to plane state and move to spl_input. Pass sharpen policy from plane state to SPL. Reviewed-by: Aric Cyr Signed-off-by: Samson Tam Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc.h | 6 ++++++ drivers/gpu/drm/amd/display/dc/dc_spl_translate.c | 2 +- .../drm/amd/display/dc/resource/dcn401/dcn401_resource.c | 1 + drivers/gpu/drm/amd/display/dc/spl/dc_spl.c | 6 +++--- drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h | 2 +- 5 files changed, 12 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 621444a1f190..ad8e2a91c306 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -227,6 +227,10 @@ struct dc_dmub_caps { uint8_t fams_ver; }; +struct dc_scl_caps { + bool sharpener_support; +}; + struct dc_caps { uint32_t max_streams; uint32_t max_links; @@ -292,6 +296,7 @@ struct dc_caps { bool sequential_ono; /* Conservative limit for DCC cases which require ODM4:1 to support*/ uint32_t dcc_plane_width_limit; + struct dc_scl_caps scl_caps; }; struct dc_bug_wa { @@ -1355,6 +1360,7 @@ struct dc_plane_state { enum mpcc_movable_cm_location mcm_location; struct dc_csc_transform cursor_csc_color_matrix; bool adaptive_sharpness_en; + int adaptive_sharpness_policy; int sharpness_level; enum linear_light_scaling linear_light_scaling; unsigned int sdr_white_level_nits; diff --git a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c index 20fd0afca8cb..eeffe44be9e3 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c +++ b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c @@ -188,7 +188,7 @@ void translate_SPL_in_params_from_pipe_ctx(struct pipe_ctx *pipe_ctx, struct spl spl_in->h_active = pipe_ctx->plane_res.scl_data.h_active; spl_in->v_active = pipe_ctx->plane_res.scl_data.v_active; - spl_in->debug.sharpen_policy = (enum sharpen_policy)pipe_ctx->stream->ctx->dc->debug.sharpen_policy; + spl_in->sharpen_policy = (enum sharpen_policy)plane_state->adaptive_sharpness_policy; spl_in->debug.scale_to_sharpness_policy = (enum scale_to_sharpness_policy)pipe_ctx->stream->ctx->dc->debug.scale_to_sharpness_policy; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c index 7da81966b4df..46feff9348d9 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c @@ -2171,6 +2171,7 @@ static bool dcn401_resource_construct( /* SPL */ spl_init_easf_filter_coeffs(); spl_init_blur_scale_coeffs(); + dc->caps.scl_caps.sharpener_support = true; return true; diff --git a/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c b/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c index c38a5c8646e8..f043c7e32e16 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c +++ b/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c @@ -848,13 +848,13 @@ static bool spl_get_isharp_en(struct spl_in *spl_in, * surfaces based on policy setting */ if (!spl_is_yuv420(spl_in->basic_in.format) && - (spl_in->debug.sharpen_policy == SHARPEN_YUV)) + (spl_in->sharpen_policy == SHARPEN_YUV)) return enable_isharp; else if ((spl_is_yuv420(spl_in->basic_in.format) && !fullscreen) && - (spl_in->debug.sharpen_policy == SHARPEN_RGB_FULLSCREEN_YUV)) + (spl_in->sharpen_policy == SHARPEN_RGB_FULLSCREEN_YUV)) return enable_isharp; else if (!spl_in->is_fullscreen && - spl_in->debug.sharpen_policy == SHARPEN_FULLSCREEN_ALL) + spl_in->sharpen_policy == SHARPEN_FULLSCREEN_ALL) return enable_isharp; /* diff --git a/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h b/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h index 48e4217555f8..fcb5d389592b 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h +++ b/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h @@ -508,7 +508,6 @@ struct spl_funcs { struct spl_debug { int visual_confirm_base_offset; int visual_confirm_dpp_offset; - enum sharpen_policy sharpen_policy; enum scale_to_sharpness_policy scale_to_sharpness_policy; }; @@ -530,6 +529,7 @@ struct spl_in { int h_active; int v_active; int sdr_white_level_nits; + enum sharpen_policy sharpen_policy; }; // end of SPL inputs -- 2.51.0 From c0933f1de2a5eab2fefd07372b415583675cfe9f Mon Sep 17 00:00:00 2001 From: Joshua Aberback Date: Thu, 3 Oct 2024 17:28:11 -0400 Subject: [PATCH 16/16] drm/amd/display: Clear pipe pointers on pipe reset [Why] We want to clean up unnecessary asserts, one of which is an assert in resource_is_pipe_type that fires if a pipe has no stream and still has pointers to other pipes ("dangling state"). This gets hit because pipes are not properly cleaned up in reset_back_end_for_pipe. When resetting a pipe, the existing MPCC / ODM combine pointers are no longer valid, especially when we put ODM in bypass. [How] - reset pipe pointers in reset_back_end_for_pipe - remove useless code to avoid confusion (a long time ago it had a reason to be there, not anymore) Reviewed-by: Alvin Lee Signed-off-by: Joshua Aberback Signed-off-by: Roman Li Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c | 12 ++++-------- .../drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 12 ++++-------- 2 files changed, 8 insertions(+), 16 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c index 9da5b50bea8a..05424a9af58b 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c @@ -2766,7 +2766,6 @@ void dcn20_reset_back_end_for_pipe( struct pipe_ctx *pipe_ctx, struct dc_state *context) { - int i; struct dc_link *link = pipe_ctx->stream->link; const struct link_hwss *link_hwss = get_link_hwss(link, &pipe_ctx->link_res); @@ -2833,19 +2832,16 @@ void dcn20_reset_back_end_for_pipe( } } - for (i = 0; i < dc->res_pool->pipe_count; i++) - if (&dc->current_state->res_ctx.pipe_ctx[i] == pipe_ctx) - break; - - if (i == dc->res_pool->pipe_count) - return; - /* * In case of a dangling plane, setting this to NULL unconditionally * causes failures during reset hw ctx where, if stream is NULL, * it is expected that the pipe_ctx pointers to pipes and plane are NULL. */ pipe_ctx->stream = NULL; + pipe_ctx->top_pipe = NULL; + pipe_ctx->bottom_pipe = NULL; + pipe_ctx->next_odm_pipe = NULL; + pipe_ctx->prev_odm_pipe = NULL; DC_LOG_DEBUG("Reset back end for pipe %d, tg:%d\n", pipe_ctx->pipe_idx, pipe_ctx->stream_res.tg->inst); } diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 413dcbf4decf..3c70f40bf047 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -1883,7 +1883,6 @@ void dcn401_reset_back_end_for_pipe( struct pipe_ctx *pipe_ctx, struct dc_state *context) { - int i; struct dc_link *link = pipe_ctx->stream->link; const struct link_hwss *link_hwss = get_link_hwss(link, &pipe_ctx->link_res); @@ -1954,19 +1953,16 @@ void dcn401_reset_back_end_for_pipe( dc->res_pool->dccg->funcs->set_dtbclk_p_src(dc->res_pool->dccg, REFCLK, pipe_ctx->stream_res.tg->inst); } - for (i = 0; i < dc->res_pool->pipe_count; i++) - if (&dc->current_state->res_ctx.pipe_ctx[i] == pipe_ctx) - break; - - if (i == dc->res_pool->pipe_count) - return; - /* * In case of a dangling plane, setting this to NULL unconditionally * causes failures during reset hw ctx where, if stream is NULL, * it is expected that the pipe_ctx pointers to pipes and plane are NULL. */ pipe_ctx->stream = NULL; + pipe_ctx->top_pipe = NULL; + pipe_ctx->bottom_pipe = NULL; + pipe_ctx->next_odm_pipe = NULL; + pipe_ctx->prev_odm_pipe = NULL; DC_LOG_DEBUG("Reset back end for pipe %d, tg:%d\n", pipe_ctx->pipe_idx, pipe_ctx->stream_res.tg->inst); } -- 2.51.0