From 084e2640e51626f413f85663e3ba7e32d4272477 Mon Sep 17 00:00:00 2001 From: Luben Tuikov Date: Thu, 11 Mar 2021 19:11:01 -0500 Subject: drm/amdgpu: Fix check for RAS support Use positive logic to check for RAS support. Rename the function to actually indicate what it is testing for. Essentially, make the function a predicate with the correct name. Cc: Stanley Yang Cc: Alexander Deucher Signed-off-by: Luben Tuikov Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 15 ++++++--------- 1 file changed, 6 insertions(+), 9 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 0e16683876aa..17652972fd49 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1933,15 +1933,12 @@ int amdgpu_ras_request_reset_on_boot(struct amdgpu_device *adev, return 0; } -static int amdgpu_ras_check_asic_type(struct amdgpu_device *adev) +static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev) { - if (adev->asic_type != CHIP_VEGA10 && - adev->asic_type != CHIP_VEGA20 && - adev->asic_type != CHIP_ARCTURUS && - adev->asic_type != CHIP_SIENNA_CICHLID) - return 1; - else - return 0; + return adev->asic_type == CHIP_VEGA10 || + adev->asic_type == CHIP_VEGA20 || + adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_SIENNA_CICHLID; } /* @@ -1960,7 +1957,7 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev, *supported = 0; if (amdgpu_sriov_vf(adev) || !adev->is_atom_fw || - amdgpu_ras_check_asic_type(adev)) + !amdgpu_ras_asic_supported(adev)) return; if (amdgpu_atomfirmware_mem_ecc_supported(adev)) { -- cgit v1.2.3 From 36000c7a51080840902d79e1558851076ecb7a96 Mon Sep 17 00:00:00 2001 From: Tian Tao Date: Wed, 24 Mar 2021 17:17:40 +0800 Subject: drm/amdgpu: Convert sysfs sprintf/snprintf family to sysfs_emit Fix the following coccicheck warning: drivers/gpu//drm/amd/amdgpu/amdgpu_ras.c:434:9-17: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_xgmi.c:220:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_xgmi.c:249:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/df_v3_6.c:208:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_psp.c:2973:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:75:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:112:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:58:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:93:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_vram_mgr.c:125:9-17: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_gtt_mgr.c:52:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_gtt_mgr.c:71:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_device.c:140:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_device.c:164:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_device.c:186:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_device.c:208:8-16: WARNING: use scnprintf or sprintf drivers/gpu//drm/amd/amdgpu/amdgpu_atombios.c:1916:8-16: WARNING: use scnprintf or sprintf Signed-off-by: Tian Tao Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 8 +++---- drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c | 6 ++---- drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 8 +++---- drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 32 +++++++++++++--------------- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/df_v3_6.c | 2 +- 8 files changed, 29 insertions(+), 35 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c index e05648a8a145..ca7bdbdbf302 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c @@ -1913,7 +1913,7 @@ static ssize_t amdgpu_atombios_get_vbios_version(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct atom_context *ctx = adev->mode_info.atom_context; - return snprintf(buf, PAGE_SIZE, "%s\n", ctx->vbios_version); + return sysfs_emit(buf, "%s\n", ctx->vbios_version); } static DEVICE_ATTR(vbios_version, 0444, amdgpu_atombios_get_vbios_version, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 789b02508077..165192b2c336 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -137,7 +137,7 @@ static ssize_t amdgpu_device_get_pcie_replay_count(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); uint64_t cnt = amdgpu_asic_get_pcie_replay_count(adev); - return snprintf(buf, PAGE_SIZE, "%llu\n", cnt); + return sysfs_emit(buf, "%llu\n", cnt); } static DEVICE_ATTR(pcie_replay_count, S_IRUGO, @@ -161,7 +161,7 @@ static ssize_t amdgpu_device_get_product_name(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%s\n", adev->product_name); + return sysfs_emit(buf, "%s\n", adev->product_name); } static DEVICE_ATTR(product_name, S_IRUGO, @@ -183,7 +183,7 @@ static ssize_t amdgpu_device_get_product_number(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%s\n", adev->product_number); + return sysfs_emit(buf, "%s\n", adev->product_number); } static DEVICE_ATTR(product_number, S_IRUGO, @@ -205,7 +205,7 @@ static ssize_t amdgpu_device_get_serial_number(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%s\n", adev->serial); + return sysfs_emit(buf, "%s\n", adev->serial); } static DEVICE_ATTR(serial_number, S_IRUGO, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c index 8980329cded0..540c01052b21 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c @@ -49,8 +49,7 @@ static ssize_t amdgpu_mem_info_gtt_total_show(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT); - return snprintf(buf, PAGE_SIZE, "%llu\n", - man->size * PAGE_SIZE); + return sysfs_emit(buf, "%llu\n", man->size * PAGE_SIZE); } /** @@ -68,8 +67,7 @@ static ssize_t amdgpu_mem_info_gtt_used_show(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_TT); - return snprintf(buf, PAGE_SIZE, "%llu\n", - amdgpu_gtt_mgr_usage(man)); + return sysfs_emit(buf, "%llu\n", amdgpu_gtt_mgr_usage(man)); } static DEVICE_ATTR(mem_info_gtt_total, S_IRUGO, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index c566e56154b0..f5ae3107a1a4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -2993,7 +2993,7 @@ static ssize_t psp_usbc_pd_fw_sysfs_read(struct device *dev, return ret; } - return snprintf(buf, PAGE_SIZE, "%x\n", fw_ver); + return sysfs_emit(buf, "%x\n", fw_ver); } static ssize_t psp_usbc_pd_fw_sysfs_write(struct device *dev, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 17652972fd49..c5a73b5c26d9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -431,15 +431,13 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev, }; if (!amdgpu_ras_get_error_query_ready(obj->adev)) - return snprintf(buf, PAGE_SIZE, - "Query currently inaccessible\n"); + return sysfs_emit(buf, "Query currently inaccessible\n"); if (amdgpu_ras_query_error_status(obj->adev, &info)) return -EINVAL; - return snprintf(buf, PAGE_SIZE, "%s: %lu\n%s: %lu\n", - "ue", info.ue_count, - "ce", info.ce_count); + return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count, + "ce", info.ce_count); } /* obj begin */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index b2fc475ce6f7..592a2dd16493 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -52,7 +52,7 @@ static ssize_t amdgpu_mem_info_vram_total_show(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%llu\n", adev->gmc.real_vram_size); + return sysfs_emit(buf, "%llu\n", adev->gmc.real_vram_size); } /** @@ -69,7 +69,7 @@ static ssize_t amdgpu_mem_info_vis_vram_total_show(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%llu\n", adev->gmc.visible_vram_size); + return sysfs_emit(buf, "%llu\n", adev->gmc.visible_vram_size); } /** @@ -87,8 +87,7 @@ static ssize_t amdgpu_mem_info_vram_used_show(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM); - return snprintf(buf, PAGE_SIZE, "%llu\n", - amdgpu_vram_mgr_usage(man)); + return sysfs_emit(buf, "%llu\n", amdgpu_vram_mgr_usage(man)); } /** @@ -106,8 +105,7 @@ static ssize_t amdgpu_mem_info_vis_vram_used_show(struct device *dev, struct amdgpu_device *adev = drm_to_adev(ddev); struct ttm_resource_manager *man = ttm_manager_type(&adev->mman.bdev, TTM_PL_VRAM); - return snprintf(buf, PAGE_SIZE, "%llu\n", - amdgpu_vram_mgr_vis_usage(man)); + return sysfs_emit(buf, "%llu\n", amdgpu_vram_mgr_vis_usage(man)); } static ssize_t amdgpu_mem_info_vram_vendor(struct device *dev, @@ -119,27 +117,27 @@ static ssize_t amdgpu_mem_info_vram_vendor(struct device *dev, switch (adev->gmc.vram_vendor) { case SAMSUNG: - return snprintf(buf, PAGE_SIZE, "samsung\n"); + return sysfs_emit(buf, "samsung\n"); case INFINEON: - return snprintf(buf, PAGE_SIZE, "infineon\n"); + return sysfs_emit(buf, "infineon\n"); case ELPIDA: - return snprintf(buf, PAGE_SIZE, "elpida\n"); + return sysfs_emit(buf, "elpida\n"); case ETRON: - return snprintf(buf, PAGE_SIZE, "etron\n"); + return sysfs_emit(buf, "etron\n"); case NANYA: - return snprintf(buf, PAGE_SIZE, "nanya\n"); + return sysfs_emit(buf, "nanya\n"); case HYNIX: - return snprintf(buf, PAGE_SIZE, "hynix\n"); + return sysfs_emit(buf, "hynix\n"); case MOSEL: - return snprintf(buf, PAGE_SIZE, "mosel\n"); + return sysfs_emit(buf, "mosel\n"); case WINBOND: - return snprintf(buf, PAGE_SIZE, "winbond\n"); + return sysfs_emit(buf, "winbond\n"); case ESMT: - return snprintf(buf, PAGE_SIZE, "esmt\n"); + return sysfs_emit(buf, "esmt\n"); case MICRON: - return snprintf(buf, PAGE_SIZE, "micron\n"); + return sysfs_emit(buf, "micron\n"); default: - return snprintf(buf, PAGE_SIZE, "unknown\n"); + return sysfs_emit(buf, "unknown\n"); } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 1429c6897d3f..5a1b598f4d30 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -216,7 +216,7 @@ static ssize_t amdgpu_xgmi_show_device_id(struct device *dev, struct drm_device *ddev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(ddev); - return snprintf(buf, PAGE_SIZE, "%llu\n", adev->gmc.xgmi.node_id); + return sysfs_emit(buf, "%llu\n", adev->gmc.xgmi.node_id); } @@ -245,7 +245,7 @@ static ssize_t amdgpu_xgmi_show_error(struct device *dev, adev->df.funcs->set_fica(adev, ficaa_pie_status_in, 0, 0); - return snprintf(buf, PAGE_SIZE, "%u\n", error_count); + return sysfs_emit(buf, "%u\n", error_count); } diff --git a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c index 44109a6b8f44..0d8459d63bac 100644 --- a/drivers/gpu/drm/amd/amdgpu/df_v3_6.c +++ b/drivers/gpu/drm/amd/amdgpu/df_v3_6.c @@ -205,7 +205,7 @@ static ssize_t df_v3_6_get_df_cntr_avail(struct device *dev, count++; } - return snprintf(buf, PAGE_SIZE, "%i\n", count); + return sysfs_emit(buf, "%i\n", count); } /* device attr for available perfmon counters */ -- cgit v1.2.3 From 5a4345270474c886dceee48c5b54b5c2c07aa877 Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Thu, 1 Apr 2021 19:10:54 +0800 Subject: drm/amdgpu: support sdma error injection Signed-off-by: Stanley.Yang Reivewed-by: Dennis Li Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index c5a73b5c26d9..32be6dd1fcd7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -925,6 +925,7 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, ret = -EINVAL; break; case AMDGPU_RAS_BLOCK__UMC: + case AMDGPU_RAS_BLOCK__SDMA: case AMDGPU_RAS_BLOCK__MMHUB: case AMDGPU_RAS_BLOCK__PCIE_BIF: ret = psp_ras_trigger_error(&adev->psp, &block_info); -- cgit v1.2.3 From f08726868c7543e0754212dcadae0d6911f721bd Mon Sep 17 00:00:00 2001 From: Bernard Zhao Date: Wed, 31 Mar 2021 06:12:03 -0700 Subject: drm/amd: cleanup coding style a bit Fix patch check warning: WARNING: suspect code indent for conditional statements (8, 17) + if (obj && obj->use < 0) { + DRM_ERROR("RAS ERROR: Unbalance obj(%s) use\n", obj->head.name); WARNING: braces {} are not necessary for single statement blocks + if (obj && obj->use < 0) { + DRM_ERROR("RAS ERROR: Unbalance obj(%s) use\n", obj->head.name); + } Signed-off-by: Bernard Zhao Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 32be6dd1fcd7..26458946145c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -447,11 +447,10 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev, static inline void put_obj(struct ras_manager *obj) { - if (obj && --obj->use == 0) + if (obj && (--obj->use == 0)) list_del(&obj->node); - if (obj && obj->use < 0) { - DRM_ERROR("RAS ERROR: Unbalance obj(%s) use\n", obj->head.name); - } + if (obj && (obj->use < 0)) + DRM_ERROR("RAS ERROR: Unbalance obj(%s) use\n", obj->head.name); } /* make one obj and return it. */ -- cgit v1.2.3 From 75f06251c921baf99c003662c529c25ba9937b2d Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Mon, 8 Mar 2021 16:40:07 +0800 Subject: drm/amdgpu: initialze ras caps per paltform config Driver only manages GFX/SDMA/MMHUB RAS in platforms that gpu node is connected to cpu through XGMI, other than that, it queries VBIOS for RAS capabilities. Signed-off-by: Hawking Zhang Acked-by: Alex Deucher Reviewed-by: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 35 ++++++++++++++++++++++----------- 1 file changed, 23 insertions(+), 12 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 26458946145c..1708045e2a0d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1936,6 +1936,7 @@ static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev) return adev->asic_type == CHIP_VEGA10 || adev->asic_type == CHIP_VEGA20 || adev->asic_type == CHIP_ARCTURUS || + adev->asic_type == CHIP_ALDEBARAN || adev->asic_type == CHIP_SIENNA_CICHLID; } @@ -1958,19 +1959,29 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev, !amdgpu_ras_asic_supported(adev)) return; - if (amdgpu_atomfirmware_mem_ecc_supported(adev)) { - dev_info(adev->dev, "MEM ECC is active.\n"); - *hw_supported |= (1 << AMDGPU_RAS_BLOCK__UMC | - 1 << AMDGPU_RAS_BLOCK__DF); - } else - dev_info(adev->dev, "MEM ECC is not presented.\n"); + if (!adev->gmc.xgmi.connected_to_cpu) { + if (amdgpu_atomfirmware_mem_ecc_supported(adev)) { + dev_info(adev->dev, "MEM ECC is active.\n"); + *hw_supported |= (1 << AMDGPU_RAS_BLOCK__UMC | + 1 << AMDGPU_RAS_BLOCK__DF); + } else { + dev_info(adev->dev, "MEM ECC is not presented.\n"); + } - if (amdgpu_atomfirmware_sram_ecc_supported(adev)) { - dev_info(adev->dev, "SRAM ECC is active.\n"); - *hw_supported |= ~(1 << AMDGPU_RAS_BLOCK__UMC | - 1 << AMDGPU_RAS_BLOCK__DF); - } else - dev_info(adev->dev, "SRAM ECC is not presented.\n"); + if (amdgpu_atomfirmware_sram_ecc_supported(adev)) { + dev_info(adev->dev, "SRAM ECC is active.\n"); + *hw_supported |= ~(1 << AMDGPU_RAS_BLOCK__UMC | + 1 << AMDGPU_RAS_BLOCK__DF); + } else { + dev_info(adev->dev, "SRAM ECC is not presented.\n"); + } + } else { + /* driver only manages a few IP blocks RAS feature + * when GPU is connected cpu through XGMI */ + *hw_supported |= (1 << AMDGPU_RAS_BLOCK__GFX | + 1 << AMDGPU_RAS_BLOCK__SDMA | + 1 << AMDGPU_RAS_BLOCK__MMHUB); + } /* hw_supported needs to be aligned with RAS block mask. */ *hw_supported &= AMDGPU_RAS_BLOCK_MASK; -- cgit v1.2.3 From 6e36f23193cc870856a41e87281f62fb2b04bd1f Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Fri, 2 Apr 2021 14:39:36 +0800 Subject: drm/amdgpu: split nbio callbacks into ras and non-ras ones nbio ras is not managed by gpu driver when gpu is connected to cpu through xgmi. split nbio callbacks into ras and non-ras ones so gpu driver only initializes nbio ras callbacks when it manages nbio ras. Signed-off-by: Hawking Zhang Reviewed-by: Dennis Li Reviewed-by: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h | 19 ++++++++++++------- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 30 ++++++++++++++++++++++++------ drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c | 16 ++++++++++------ drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h | 1 + drivers/gpu/drm/amd/amdgpu/soc15.c | 15 ++++++++++----- 6 files changed, 63 insertions(+), 30 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index 03412543427a..90f50561b43a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -199,13 +199,13 @@ irqreturn_t amdgpu_irq_handler(int irq, void *arg) * ack the interrupt if it is there */ if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__PCIE_BIF)) { - if (adev->nbio.funcs && - adev->nbio.funcs->handle_ras_controller_intr_no_bifring) - adev->nbio.funcs->handle_ras_controller_intr_no_bifring(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->handle_ras_controller_intr_no_bifring) + adev->nbio.ras_funcs->handle_ras_controller_intr_no_bifring(adev); - if (adev->nbio.funcs && - adev->nbio.funcs->handle_ras_err_event_athub_intr_no_bifring) - adev->nbio.funcs->handle_ras_err_event_athub_intr_no_bifring(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->handle_ras_err_event_athub_intr_no_bifring) + adev->nbio.ras_funcs->handle_ras_err_event_athub_intr_no_bifring(adev); } return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h index 7c11bce4514b..25ee53545837 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h @@ -47,6 +47,17 @@ struct nbio_hdp_flush_reg { u32 ref_and_mask_sdma7; }; +struct amdgpu_nbio_ras_funcs { + void (*handle_ras_controller_intr_no_bifring)(struct amdgpu_device *adev); + void (*handle_ras_err_event_athub_intr_no_bifring)(struct amdgpu_device *adev); + int (*init_ras_controller_interrupt)(struct amdgpu_device *adev); + int (*init_ras_err_event_athub_interrupt)(struct amdgpu_device *adev); + void (*query_ras_error_count)(struct amdgpu_device *adev, + void *ras_error_status); + int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); +}; + struct amdgpu_nbio_funcs { const struct nbio_hdp_flush_reg *hdp_flush_reg; u32 (*get_hdp_flush_req_offset)(struct amdgpu_device *adev); @@ -79,13 +90,6 @@ struct amdgpu_nbio_funcs { void (*ih_control)(struct amdgpu_device *adev); void (*init_registers)(struct amdgpu_device *adev); void (*remap_hdp_registers)(struct amdgpu_device *adev); - void (*handle_ras_controller_intr_no_bifring)(struct amdgpu_device *adev); - void (*handle_ras_err_event_athub_intr_no_bifring)(struct amdgpu_device *adev); - int (*init_ras_controller_interrupt)(struct amdgpu_device *adev); - int (*init_ras_err_event_athub_interrupt)(struct amdgpu_device *adev); - void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); - int (*ras_late_init)(struct amdgpu_device *adev); void (*enable_aspm)(struct amdgpu_device *adev, bool enable); void (*program_aspm)(struct amdgpu_device *adev); @@ -97,6 +101,7 @@ struct amdgpu_nbio { struct amdgpu_irq_src ras_err_event_athub_irq; struct ras_common_if *ras_if; const struct amdgpu_nbio_funcs *funcs; + const struct amdgpu_nbio_ras_funcs *ras_funcs; }; int amdgpu_nbio_ras_late_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 1708045e2a0d..ac3f4c3266bc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -804,8 +804,9 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, adev->mmhub.funcs->query_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__PCIE_BIF: - if (adev->nbio.funcs->query_ras_error_count) - adev->nbio.funcs->query_ras_error_count(adev, &err_data); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->query_ras_error_count) + adev->nbio.ras_funcs->query_ras_error_count(adev, &err_data); break; case AMDGPU_RAS_BLOCK__XGMI_WAFL: amdgpu_xgmi_query_ras_error_count(adev, &err_data); @@ -2030,14 +2031,31 @@ int amdgpu_ras_init(struct amdgpu_device *adev) /* Might need get this flag from vbios. */ con->flags = RAS_DEFAULT_FLAGS; - if (adev->nbio.funcs->init_ras_controller_interrupt) { - r = adev->nbio.funcs->init_ras_controller_interrupt(adev); + /* initialize nbio ras function ahead of any other + * ras functions so hardware fatal error interrupt + * can be enabled as early as possible */ + switch (adev->asic_type) { + case CHIP_VEGA20: + case CHIP_ARCTURUS: + case CHIP_ALDEBARAN: + if (!adev->gmc.xgmi.connected_to_cpu) + adev->nbio.ras_funcs = &nbio_v7_4_ras_funcs; + break; + default: + /* nbio ras is not available */ + break; + } + + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->init_ras_controller_interrupt) { + r = adev->nbio.ras_funcs->init_ras_controller_interrupt(adev); if (r) goto release_con; } - if (adev->nbio.funcs->init_ras_err_event_athub_interrupt) { - r = adev->nbio.funcs->init_ras_err_event_athub_interrupt(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt) { + r = adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt(adev); if (r) goto release_con; } diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c index c477f8972d5d..af44aad78171 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c @@ -557,6 +557,16 @@ static void nbio_v7_4_enable_doorbell_interrupt(struct amdgpu_device *adev, DOORBELL_INTERRUPT_DISABLE, enable ? 0 : 1); } +const struct amdgpu_nbio_ras_funcs nbio_v7_4_ras_funcs = { + .handle_ras_controller_intr_no_bifring = nbio_v7_4_handle_ras_controller_intr_no_bifring, + .handle_ras_err_event_athub_intr_no_bifring = nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring, + .init_ras_controller_interrupt = nbio_v7_4_init_ras_controller_interrupt, + .init_ras_err_event_athub_interrupt = nbio_v7_4_init_ras_err_event_athub_interrupt, + .query_ras_error_count = nbio_v7_4_query_ras_error_count, + .ras_late_init = amdgpu_nbio_ras_late_init, + .ras_fini = amdgpu_nbio_ras_fini, +}; + const struct amdgpu_nbio_funcs nbio_v7_4_funcs = { .get_hdp_flush_req_offset = nbio_v7_4_get_hdp_flush_req_offset, .get_hdp_flush_done_offset = nbio_v7_4_get_hdp_flush_done_offset, @@ -577,10 +587,4 @@ const struct amdgpu_nbio_funcs nbio_v7_4_funcs = { .ih_control = nbio_v7_4_ih_control, .init_registers = nbio_v7_4_init_registers, .remap_hdp_registers = nbio_v7_4_remap_hdp_registers, - .handle_ras_controller_intr_no_bifring = nbio_v7_4_handle_ras_controller_intr_no_bifring, - .handle_ras_err_event_athub_intr_no_bifring = nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring, - .init_ras_controller_interrupt = nbio_v7_4_init_ras_controller_interrupt, - .init_ras_err_event_athub_interrupt = nbio_v7_4_init_ras_err_event_athub_interrupt, - .query_ras_error_count = nbio_v7_4_query_ras_error_count, - .ras_late_init = amdgpu_nbio_ras_late_init, }; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h index b1ac82872752..b8216581ec8d 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.h @@ -28,5 +28,6 @@ extern const struct nbio_hdp_flush_reg nbio_v7_4_hdp_flush_reg; extern const struct amdgpu_nbio_funcs nbio_v7_4_funcs; +extern const struct amdgpu_nbio_ras_funcs nbio_v7_4_ras_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index 64c98b9c7fe8..5c5eb3aed1b3 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -1523,8 +1523,9 @@ static int soc15_common_late_init(void *handle) if (adev->hdp.funcs->reset_ras_error_count) adev->hdp.funcs->reset_ras_error_count(adev); - if (adev->nbio.funcs->ras_late_init) - r = adev->nbio.funcs->ras_late_init(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->ras_late_init) + r = adev->nbio.ras_funcs->ras_late_init(adev); return r; } @@ -1545,7 +1546,9 @@ static int soc15_common_sw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - amdgpu_nbio_ras_fini(adev); + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->ras_fini) + adev->nbio.ras_funcs->ras_fini(adev); adev->df.funcs->sw_fini(adev); return 0; } @@ -1609,9 +1612,11 @@ static int soc15_common_hw_fini(void *handle) if (adev->nbio.ras_if && amdgpu_ras_is_supported(adev, adev->nbio.ras_if->block)) { - if (adev->nbio.funcs->init_ras_controller_interrupt) + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->init_ras_controller_interrupt) amdgpu_irq_put(adev, &adev->nbio.ras_controller_irq, 0); - if (adev->nbio.funcs->init_ras_err_event_athub_interrupt) + if (adev->nbio.ras_funcs && + adev->nbio.ras_funcs->init_ras_err_event_athub_interrupt) amdgpu_irq_put(adev, &adev->nbio.ras_err_event_athub_irq, 0); } -- cgit v1.2.3 From 52137ca8526cdeceab7651c314ce68ac49963512 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Thu, 18 Mar 2021 20:18:19 +0800 Subject: drm/amdgpu: move xgmi ras functions to xgmi_ras_funcs xgmi ras is not managed by gpu driver when gpu is connected to cpu through xgmi. move all xgmi ras functions to xgmi_ras_funcs so gpu driver only initializes xgmi ras functions when it manages xgmi ras. Signed-off-by: Hawking Zhang Reviewed-by: Dennis Li Reviewed-by: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c | 16 ++++++++++++++-- drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h | 9 +++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 4 +++- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 21 ++++++++++++++------- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h | 7 +------ drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 2 +- 6 files changed, 42 insertions(+), 17 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index b9d68fd2610c..082f9d03a94a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -403,14 +403,26 @@ int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev) return r; } - return amdgpu_xgmi_ras_late_init(adev); + if (!adev->gmc.xgmi.connected_to_cpu) + adev->gmc.xgmi.ras_funcs = &xgmi_ras_funcs; + + if (adev->gmc.xgmi.ras_funcs && + adev->gmc.xgmi.ras_funcs->ras_late_init) { + r = adev->gmc.xgmi.ras_funcs->ras_late_init(adev); + if (r) + return r; + } + + return 0; } void amdgpu_gmc_ras_fini(struct amdgpu_device *adev) { amdgpu_umc_ras_fini(adev); amdgpu_mmhub_ras_fini(adev); - amdgpu_xgmi_ras_fini(adev); + if (adev->gmc.xgmi.ras_funcs && + adev->gmc.xgmi.ras_funcs->ras_fini) + adev->gmc.xgmi.ras_funcs->ras_fini(adev); } /* diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h index 7e248a4e2fa3..cbb7735c6988 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h @@ -135,6 +135,14 @@ struct amdgpu_gmc_funcs { unsigned int (*get_vbios_fb_size)(struct amdgpu_device *adev); }; +struct amdgpu_xgmi_ras_funcs { + int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); + int (*query_ras_error_count)(struct amdgpu_device *adev, + void *ras_error_status); + void (*reset_ras_error_count)(struct amdgpu_device *adev); +}; + struct amdgpu_xgmi { /* from psp */ u64 node_id; @@ -151,6 +159,7 @@ struct amdgpu_xgmi { struct ras_common_if *ras_if; bool connected_to_cpu; bool pending_reset; + const struct amdgpu_xgmi_ras_funcs *ras_funcs; }; struct amdgpu_gmc { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index ac3f4c3266bc..172738cc99db 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -809,7 +809,9 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, adev->nbio.ras_funcs->query_ras_error_count(adev, &err_data); break; case AMDGPU_RAS_BLOCK__XGMI_WAFL: - amdgpu_xgmi_query_ras_error_count(adev, &err_data); + if (adev->gmc.xgmi.ras_funcs && + adev->gmc.xgmi.ras_funcs->query_ras_error_count) + adev->gmc.xgmi.ras_funcs->query_ras_error_count(adev, &err_data); break; default: break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 5a1b598f4d30..8567d5d77346 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -628,7 +628,7 @@ int amdgpu_xgmi_remove_device(struct amdgpu_device *adev) return psp_xgmi_terminate(&adev->psp); } -int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) +static int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) { int r; struct ras_ih_if ih_info = { @@ -642,7 +642,7 @@ int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) adev->gmc.xgmi.num_physical_nodes == 0) return 0; - amdgpu_xgmi_reset_ras_error_count(adev); + adev->gmc.xgmi.ras_funcs->reset_ras_error_count(adev); if (!adev->gmc.xgmi.ras_if) { adev->gmc.xgmi.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); @@ -664,7 +664,7 @@ int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev) return r; } -void amdgpu_xgmi_ras_fini(struct amdgpu_device *adev) +static void amdgpu_xgmi_ras_fini(struct amdgpu_device *adev) { if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL) && adev->gmc.xgmi.ras_if) { @@ -691,7 +691,7 @@ static void pcs_clear_status(struct amdgpu_device *adev, uint32_t pcs_status_reg WREG32_PCIE(pcs_status_reg, 0); } -void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev) +static void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev) { uint32_t i; @@ -751,8 +751,8 @@ static int amdgpu_xgmi_query_pcs_error_status(struct amdgpu_device *adev, return 0; } -int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, - void *ras_error_status) +static int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, + void *ras_error_status) { struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; int i; @@ -801,10 +801,17 @@ int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, break; } - amdgpu_xgmi_reset_ras_error_count(adev); + adev->gmc.xgmi.ras_funcs->reset_ras_error_count(adev); err_data->ue_count += ue_cnt; err_data->ce_count += ce_cnt; return 0; } + +const struct amdgpu_xgmi_ras_funcs xgmi_ras_funcs = { + .ras_late_init = amdgpu_xgmi_ras_late_init, + .ras_fini = amdgpu_xgmi_ras_fini, + .query_ras_error_count = amdgpu_xgmi_query_ras_error_count, + .reset_ras_error_count = amdgpu_xgmi_reset_ras_error_count, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h index 148560d63554..12969c0830d5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h @@ -50,6 +50,7 @@ struct amdgpu_pcs_ras_field { uint32_t pcs_err_shift; }; +extern const struct amdgpu_xgmi_ras_funcs xgmi_ras_funcs; struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev); void amdgpu_put_xgmi_hive(struct amdgpu_hive_info *hive); int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev); @@ -58,14 +59,8 @@ int amdgpu_xgmi_remove_device(struct amdgpu_device *adev); int amdgpu_xgmi_set_pstate(struct amdgpu_device *adev, int pstate); int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev, struct amdgpu_device *peer_adev); -int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev); -void amdgpu_xgmi_ras_fini(struct amdgpu_device *adev); uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev, uint64_t addr); -int amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev, - void *ras_error_status); -void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev); - static inline bool amdgpu_xgmi_same_hive(struct amdgpu_device *adev, struct amdgpu_device *bo_adev) { diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 468acc06b681..6d784c6098bd 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -1208,7 +1208,7 @@ static int gmc_v9_0_early_init(void *handle) adev->gmc.xgmi.supported = true; adev->gmc.xgmi.connected_to_cpu = adev->smuio.funcs->is_host_gpu_xgmi_supported(adev); - } + } adev->gmc.shared_aperture_start = 0x2000000000000000ULL; adev->gmc.shared_aperture_end = -- cgit v1.2.3 From 49070c4ea3d97b76c5666466efb35dcc42c6c8fd Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Wed, 17 Mar 2021 19:17:52 +0800 Subject: drm/amdgpu: split umc callbacks to ras and non-ras ones umc ras is not managed by gpu driver when gpu is connected to cpu through xgmi. split umc callbacks into ras and non-ras ones so gpu driver only initializes umc ras callbacks when it manages umc ras. Signed-off-by: Hawking Zhang Reviewed-by: Dennis Li Reviewed-by: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c | 11 ++++++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 10 ++++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 17 +++++++++-------- drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h | 9 +++++++-- drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 16 ++++++++-------- drivers/gpu/drm/amd/amdgpu/umc_v6_1.c | 4 +++- drivers/gpu/drm/amd/amdgpu/umc_v6_1.h | 2 +- drivers/gpu/drm/amd/amdgpu/umc_v6_7.c | 4 +++- drivers/gpu/drm/amd/amdgpu/umc_v6_7.h | 2 +- drivers/gpu/drm/amd/amdgpu/umc_v8_7.c | 4 +++- drivers/gpu/drm/amd/amdgpu/umc_v8_7.h | 2 +- 12 files changed, 51 insertions(+), 32 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 082f9d03a94a..3411c6577a18 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -391,8 +391,9 @@ int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev) { int r; - if (adev->umc.funcs && adev->umc.funcs->ras_late_init) { - r = adev->umc.funcs->ras_late_init(adev); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->ras_late_init) { + r = adev->umc.ras_funcs->ras_late_init(adev); if (r) return r; } @@ -418,8 +419,12 @@ int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev) void amdgpu_gmc_ras_fini(struct amdgpu_device *adev) { - amdgpu_umc_ras_fini(adev); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->ras_fini) + adev->umc.ras_funcs->ras_fini(adev); + amdgpu_mmhub_ras_fini(adev); + if (adev->gmc.xgmi.ras_funcs && adev->gmc.xgmi.ras_funcs->ras_fini) adev->gmc.xgmi.ras_funcs->ras_fini(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 172738cc99db..459a470744f4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -774,13 +774,15 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, switch (info->head.block) { case AMDGPU_RAS_BLOCK__UMC: - if (adev->umc.funcs->query_ras_error_count) - adev->umc.funcs->query_ras_error_count(adev, &err_data); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_count) + adev->umc.ras_funcs->query_ras_error_count(adev, &err_data); /* umc query_ras_error_address is also responsible for clearing * error status */ - if (adev->umc.funcs->query_ras_error_address) - adev->umc.funcs->query_ras_error_address(adev, &err_data); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_address) + adev->umc.ras_funcs->query_ras_error_address(adev, &err_data); break; case AMDGPU_RAS_BLOCK__SDMA: if (adev->sdma.funcs->query_ras_error_count) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index a2975c8092a9..ea6f99be070b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -60,8 +60,9 @@ int amdgpu_umc_ras_late_init(struct amdgpu_device *adev) } /* ras init of specific umc version */ - if (adev->umc.funcs && adev->umc.funcs->err_cnt_init) - adev->umc.funcs->err_cnt_init(adev); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->err_cnt_init) + adev->umc.ras_funcs->err_cnt_init(adev); return 0; @@ -95,12 +96,12 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); - if (adev->umc.funcs && - adev->umc.funcs->query_ras_error_count) - adev->umc.funcs->query_ras_error_count(adev, ras_error_status); + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_count) + adev->umc.ras_funcs->query_ras_error_count(adev, ras_error_status); - if (adev->umc.funcs && - adev->umc.funcs->query_ras_error_address && + if (adev->umc.ras_funcs && + adev->umc.ras_funcs->query_ras_error_address && adev->umc.max_ras_err_cnt_per_query) { err_data->err_addr = kcalloc(adev->umc.max_ras_err_cnt_per_query, @@ -116,7 +117,7 @@ int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, /* umc query_ras_error_address is also responsible for clearing * error status */ - adev->umc.funcs->query_ras_error_address(adev, ras_error_status); + adev->umc.ras_funcs->query_ras_error_address(adev, ras_error_status); } /* only uncorrectable error needs gpu reset */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index 183814493658..bbcccf53080d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -35,13 +35,17 @@ #define LOOP_UMC_CH_INST(ch_inst) for ((ch_inst) = 0; (ch_inst) < adev->umc.channel_inst_num; (ch_inst)++) #define LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) LOOP_UMC_INST((umc_inst)) LOOP_UMC_CH_INST((ch_inst)) -struct amdgpu_umc_funcs { +struct amdgpu_umc_ras_funcs { void (*err_cnt_init)(struct amdgpu_device *adev); int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); + void *ras_error_status); void (*query_ras_error_address)(struct amdgpu_device *adev, void *ras_error_status); +}; + +struct amdgpu_umc_funcs { void (*init_registers)(struct amdgpu_device *adev); }; @@ -59,6 +63,7 @@ struct amdgpu_umc { struct ras_common_if *ras_if; const struct amdgpu_umc_funcs *funcs; + const struct amdgpu_umc_ras_funcs *ras_funcs; }; int amdgpu_umc_ras_late_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index 33e54eed2eec..2bfd620576f2 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -655,7 +655,7 @@ static void gmc_v10_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V8_7_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V8_7_PER_CHANNEL_OFFSET_SIENNA; adev->umc.channel_idx_tbl = &umc_v8_7_channel_idx_tbl[0][0]; - adev->umc.funcs = &umc_v8_7_funcs; + adev->umc.ras_funcs = &umc_v8_7_ras_funcs; break; default: break; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 6d784c6098bd..1970ae9d5376 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -1155,7 +1155,7 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V6_1_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V6_1_PER_CHANNEL_OFFSET_VG20; adev->umc.channel_idx_tbl = &umc_v6_1_channel_idx_tbl[0][0]; - adev->umc.funcs = &umc_v6_1_funcs; + adev->umc.ras_funcs = &umc_v6_1_ras_funcs; break; case CHIP_ARCTURUS: adev->umc.max_ras_err_cnt_per_query = UMC_V6_1_TOTAL_CHANNEL_NUM; @@ -1163,7 +1163,7 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V6_1_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V6_1_PER_CHANNEL_OFFSET_ARCT; adev->umc.channel_idx_tbl = &umc_v6_1_channel_idx_tbl[0][0]; - adev->umc.funcs = &umc_v6_1_funcs; + adev->umc.ras_funcs = &umc_v6_1_ras_funcs; break; default: break; @@ -1194,12 +1194,6 @@ static int gmc_v9_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - gmc_v9_0_set_gmc_funcs(adev); - gmc_v9_0_set_irq_funcs(adev); - gmc_v9_0_set_umc_funcs(adev); - gmc_v9_0_set_mmhub_funcs(adev); - gmc_v9_0_set_gfxhub_funcs(adev); - if (adev->asic_type == CHIP_VEGA20 || adev->asic_type == CHIP_ARCTURUS) adev->gmc.xgmi.supported = true; @@ -1210,6 +1204,12 @@ static int gmc_v9_0_early_init(void *handle) adev->smuio.funcs->is_host_gpu_xgmi_supported(adev); } + gmc_v9_0_set_gmc_funcs(adev); + gmc_v9_0_set_irq_funcs(adev); + gmc_v9_0_set_umc_funcs(adev); + gmc_v9_0_set_mmhub_funcs(adev); + gmc_v9_0_set_gfxhub_funcs(adev); + adev->gmc.shared_aperture_start = 0x2000000000000000ULL; adev->gmc.shared_aperture_end = adev->gmc.shared_aperture_start + (4ULL << 30) - 1; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c index 96d7769609f4..20b44983ac94 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.c @@ -22,6 +22,7 @@ */ #include "umc_v6_1.h" #include "amdgpu_ras.h" +#include "amdgpu_umc.h" #include "amdgpu.h" #include "rsmu/rsmu_0_0_2_offset.h" @@ -464,9 +465,10 @@ static void umc_v6_1_err_cnt_init(struct amdgpu_device *adev) umc_v6_1_enable_umc_index_mode(adev); } -const struct amdgpu_umc_funcs umc_v6_1_funcs = { +const struct amdgpu_umc_ras_funcs umc_v6_1_ras_funcs = { .err_cnt_init = umc_v6_1_err_cnt_init, .ras_late_init = amdgpu_umc_ras_late_init, + .ras_fini = amdgpu_umc_ras_fini, .query_ras_error_count = umc_v6_1_query_ras_error_count, .query_ras_error_address = umc_v6_1_query_ras_error_address, }; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h index 0ce1d323cfdd..5dc36c730bb2 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_1.h @@ -45,7 +45,7 @@ /* umc ce count initial value */ #define UMC_V6_1_CE_CNT_INIT (UMC_V6_1_CE_CNT_MAX - UMC_V6_1_CE_INT_THRESHOLD) -extern const struct amdgpu_umc_funcs umc_v6_1_funcs; +extern const struct amdgpu_umc_ras_funcs umc_v6_1_ras_funcs; extern const uint32_t umc_v6_1_channel_idx_tbl[UMC_V6_1_UMC_INSTANCE_NUM][UMC_V6_1_CHANNEL_INSTANCE_NUM]; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c index 3c3fb01b7c20..3a8f787374c0 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c @@ -22,6 +22,7 @@ */ #include "umc_v6_7.h" #include "amdgpu_ras.h" +#include "amdgpu_umc.h" #include "amdgpu.h" #include "umc/umc_6_7_0_offset.h" @@ -272,8 +273,9 @@ static void umc_v6_7_query_ras_error_address(struct amdgpu_device *adev, } } -const struct amdgpu_umc_funcs umc_v6_7_funcs = { +const struct amdgpu_umc_ras_funcs umc_v6_7_ras_funcs = { .ras_late_init = amdgpu_umc_ras_late_init, + .ras_fini = amdgpu_umc_ras_fini, .query_ras_error_count = umc_v6_7_query_ras_error_count, .query_ras_error_address = umc_v6_7_query_ras_error_address, }; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h index e59dbdb6ef9b..4eb85f247e96 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h @@ -32,6 +32,6 @@ #define UMC_V6_7_INST_DIST 0x40000 -extern const struct amdgpu_umc_funcs umc_v6_7_funcs; +extern const struct amdgpu_umc_ras_funcs umc_v6_7_ras_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c index a064c097690c..89d20adfa001 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.c @@ -22,6 +22,7 @@ */ #include "umc_v8_7.h" #include "amdgpu_ras.h" +#include "amdgpu_umc.h" #include "amdgpu.h" #include "rsmu/rsmu_0_0_2_offset.h" @@ -323,9 +324,10 @@ static void umc_v8_7_err_cnt_init(struct amdgpu_device *adev) } } -const struct amdgpu_umc_funcs umc_v8_7_funcs = { +const struct amdgpu_umc_ras_funcs umc_v8_7_ras_funcs = { .err_cnt_init = umc_v8_7_err_cnt_init, .ras_late_init = amdgpu_umc_ras_late_init, + .ras_fini = amdgpu_umc_ras_fini, .query_ras_error_count = umc_v8_7_query_ras_error_count, .query_ras_error_address = umc_v8_7_query_ras_error_address, }; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h index d4d0468e3df5..37e6dc7c28e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_7.h @@ -44,7 +44,7 @@ /* umc ce count initial value */ #define UMC_V8_7_CE_CNT_INIT (UMC_V8_7_CE_CNT_MAX - UMC_V8_7_CE_INT_THRESHOLD) -extern const struct amdgpu_umc_funcs umc_v8_7_funcs; +extern const struct amdgpu_umc_ras_funcs umc_v8_7_ras_funcs; extern const uint32_t umc_v8_7_channel_idx_tbl[UMC_V8_7_UMC_INSTANCE_NUM][UMC_V8_7_CHANNEL_INSTANCE_NUM]; -- cgit v1.2.3 From 8bc7b360ad4b0a090380d7548dbf24a627f0b035 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Fri, 19 Mar 2021 15:50:14 +0800 Subject: drm/amdgpu: split mmhub callbacks into ras and non-ras ones mmhub ras is only avaiable in cerntain mmhub ip generation. Signed-off-by: Hawking Zhang Reviewed-by: Dennis Li Reviewed-by: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 11 ++++++----- drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c | 28 +++++++++++++++++++++++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h | 12 ++++++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 20 ++++++++++++-------- drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 5 +++-- drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c | 6 +++++- drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h | 1 + drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c | 8 ++++++-- drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h | 1 + drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c | 1 - drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c | 1 - drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c | 8 ++++++-- drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h | 1 + 13 files changed, 74 insertions(+), 29 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 438e2f732377..b4ad1c055c70 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -3142,8 +3142,9 @@ static void amdgpu_device_xgmi_reset_func(struct work_struct *__work) if (adev->asic_reset_res) goto fail; - if (adev->mmhub.funcs && adev->mmhub.funcs->reset_ras_error_count) - adev->mmhub.funcs->reset_ras_error_count(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->reset_ras_error_count) + adev->mmhub.ras_funcs->reset_ras_error_count(adev); } else { task_barrier_full(&hive->tb); @@ -4378,9 +4379,9 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle, if (!r && amdgpu_ras_intr_triggered()) { list_for_each_entry(tmp_adev, device_list_handle, reset_list) { - if (tmp_adev->mmhub.funcs && - tmp_adev->mmhub.funcs->reset_ras_error_count) - tmp_adev->mmhub.funcs->reset_ras_error_count(tmp_adev); + if (tmp_adev->mmhub.ras_funcs && + tmp_adev->mmhub.ras_funcs->reset_ras_error_count) + tmp_adev->mmhub.ras_funcs->reset_ras_error_count(tmp_adev); } amdgpu_ras_intr_cleared(); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 3411c6577a18..013efc746821 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -30,6 +30,9 @@ #include "amdgpu_gmc.h" #include "amdgpu_ras.h" #include "amdgpu_xgmi.h" +#include "mmhub_v1_0.h" +#include "mmhub_v9_4.h" +#include "mmhub_v1_7.h" /** * amdgpu_gmc_pdb0_alloc - allocate vram for pdb0 @@ -398,8 +401,25 @@ int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev) return r; } - if (adev->mmhub.funcs && adev->mmhub.funcs->ras_late_init) { - r = adev->mmhub.funcs->ras_late_init(adev); + /* initialize mmhub ras funcs */ + switch (adev->asic_type) { + case CHIP_VEGA20: + adev->mmhub.ras_funcs = &mmhub_v1_0_ras_funcs; + break; + case CHIP_ARCTURUS: + adev->mmhub.ras_funcs = &mmhub_v9_4_ras_funcs; + break; + case CHIP_ALDEBARAN: + adev->mmhub.ras_funcs = &mmhub_v1_7_ras_funcs; + break; + default: + /* mmhub ras is not available */ + break; + } + + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->ras_late_init) { + r = adev->mmhub.ras_funcs->ras_late_init(adev); if (r) return r; } @@ -423,7 +443,9 @@ void amdgpu_gmc_ras_fini(struct amdgpu_device *adev) adev->umc.ras_funcs->ras_fini) adev->umc.ras_funcs->ras_fini(adev); - amdgpu_mmhub_ras_fini(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->ras_fini) + amdgpu_mmhub_ras_fini(adev); if (adev->gmc.xgmi.ras_funcs && adev->gmc.xgmi.ras_funcs->ras_fini) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h index 1ae9bdae7311..11aa29933c1f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mmhub.h @@ -21,12 +21,16 @@ #ifndef __AMDGPU_MMHUB_H__ #define __AMDGPU_MMHUB_H__ -struct amdgpu_mmhub_funcs { - void (*ras_init)(struct amdgpu_device *adev); +struct amdgpu_mmhub_ras_funcs { int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); void (*query_ras_error_count)(struct amdgpu_device *adev, - void *ras_error_status); + void *ras_error_status); + void (*query_ras_error_status)(struct amdgpu_device *adev); void (*reset_ras_error_count)(struct amdgpu_device *adev); +}; + +struct amdgpu_mmhub_funcs { u64 (*get_fb_location)(struct amdgpu_device *adev); void (*init)(struct amdgpu_device *adev); int (*gart_enable)(struct amdgpu_device *adev); @@ -40,12 +44,12 @@ struct amdgpu_mmhub_funcs { uint64_t page_table_base); void (*update_power_gating)(struct amdgpu_device *adev, bool enable); - void (*query_ras_error_status)(struct amdgpu_device *adev); }; struct amdgpu_mmhub { struct ras_common_if *ras_if; const struct amdgpu_mmhub_funcs *funcs; + const struct amdgpu_mmhub_ras_funcs *ras_funcs; }; int amdgpu_mmhub_ras_late_init(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 459a470744f4..b55f470eb747 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -799,11 +799,13 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, adev->gfx.funcs->query_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__MMHUB: - if (adev->mmhub.funcs->query_ras_error_count) - adev->mmhub.funcs->query_ras_error_count(adev, &err_data); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->query_ras_error_count) + adev->mmhub.ras_funcs->query_ras_error_count(adev, &err_data); - if (adev->mmhub.funcs->query_ras_error_status) - adev->mmhub.funcs->query_ras_error_status(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->query_ras_error_status) + adev->mmhub.ras_funcs->query_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__PCIE_BIF: if (adev->nbio.ras_funcs && @@ -857,8 +859,9 @@ int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, adev->gfx.funcs->reset_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__MMHUB: - if (adev->mmhub.funcs->reset_ras_error_count) - adev->mmhub.funcs->reset_ras_error_count(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->reset_ras_error_count) + adev->mmhub.ras_funcs->reset_ras_error_count(adev); break; case AMDGPU_RAS_BLOCK__SDMA: if (adev->sdma.funcs->reset_ras_error_count) @@ -1515,8 +1518,9 @@ static void amdgpu_ras_error_status_query(struct amdgpu_device *adev, adev->gfx.funcs->query_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__MMHUB: - if (adev->mmhub.funcs->query_ras_error_status) - adev->mmhub.funcs->query_ras_error_status(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->query_ras_error_status) + adev->mmhub.ras_funcs->query_ras_error_status(adev); break; default: break; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index d139b5f737e1..e1f9cc96e567 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -1241,8 +1241,9 @@ static int gmc_v9_0_late_init(void *handle) } } - if (adev->mmhub.funcs && adev->mmhub.funcs->reset_ras_error_count) - adev->mmhub.funcs->reset_ras_error_count(adev); + if (adev->mmhub.ras_funcs && + adev->mmhub.ras_funcs->reset_ras_error_count) + adev->mmhub.ras_funcs->reset_ras_error_count(adev); r = amdgpu_gmc_ras_late_init(adev); if (r) diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c index d7b39c07de20..aa9be5612c89 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c @@ -776,10 +776,14 @@ static void mmhub_v1_0_reset_ras_error_count(struct amdgpu_device *adev) } } -const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs = { +const struct amdgpu_mmhub_ras_funcs mmhub_v1_0_ras_funcs = { .ras_late_init = amdgpu_mmhub_ras_late_init, + .ras_fini = amdgpu_mmhub_ras_fini, .query_ras_error_count = mmhub_v1_0_query_ras_error_count, .reset_ras_error_count = mmhub_v1_0_reset_ras_error_count, +}; + +const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs = { .get_fb_location = mmhub_v1_0_get_fb_location, .init = mmhub_v1_0_init, .gart_enable = mmhub_v1_0_gart_enable, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h index d77f5b65a618..4661b094e007 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h @@ -24,5 +24,6 @@ #define __MMHUB_V1_0_H__ extern const struct amdgpu_mmhub_funcs mmhub_v1_0_funcs; +extern const struct amdgpu_mmhub_ras_funcs mmhub_v1_0_ras_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c index ae7d8a1738a3..7977a7879b32 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c @@ -1313,10 +1313,15 @@ static void mmhub_v1_7_query_ras_error_status(struct amdgpu_device *adev) } } -const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs = { +const struct amdgpu_mmhub_ras_funcs mmhub_v1_7_ras_funcs = { .ras_late_init = amdgpu_mmhub_ras_late_init, + .ras_fini = amdgpu_mmhub_ras_fini, .query_ras_error_count = mmhub_v1_7_query_ras_error_count, .reset_ras_error_count = mmhub_v1_7_reset_ras_error_count, + .query_ras_error_status = mmhub_v1_7_query_ras_error_status, +}; + +const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs = { .get_fb_location = mmhub_v1_7_get_fb_location, .init = mmhub_v1_7_init, .gart_enable = mmhub_v1_7_gart_enable, @@ -1325,5 +1330,4 @@ const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs = { .set_clockgating = mmhub_v1_7_set_clockgating, .get_clockgating = mmhub_v1_7_get_clockgating, .setup_vm_pt_regs = mmhub_v1_7_setup_vm_pt_regs, - .query_ras_error_status = mmhub_v1_7_query_ras_error_status, }; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h index bf2fbeb172d1..a7f9dfc24697 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.h @@ -24,5 +24,6 @@ #define __MMHUB_V1_7_H__ extern const struct amdgpu_mmhub_funcs mmhub_v1_7_funcs; +extern const struct amdgpu_mmhub_ras_funcs mmhub_v1_7_ras_funcs; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c index f107385faba2..da7edd1ed6b2 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c @@ -689,7 +689,6 @@ static void mmhub_v2_0_get_clockgating(struct amdgpu_device *adev, u32 *flags) } const struct amdgpu_mmhub_funcs mmhub_v2_0_funcs = { - .ras_late_init = amdgpu_mmhub_ras_late_init, .init = mmhub_v2_0_init, .gart_enable = mmhub_v2_0_gart_enable, .set_fault_enable_default = mmhub_v2_0_set_fault_enable_default, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c index ab9be5ad5a5f..1141c37432f0 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c @@ -616,7 +616,6 @@ static void mmhub_v2_3_get_clockgating(struct amdgpu_device *adev, u32 *flags) } const struct amdgpu_mmhub_funcs mmhub_v2_3_funcs = { - .ras_late_init = amdgpu_mmhub_ras_late_init, .init = mmhub_v2_3_init, .gart_enable = mmhub_v2_3_gart_enable, .set_fault_enable_default = mmhub_v2_3_set_fault_enable_default, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c index 4a31737b6bb0..0cffa820ea6e 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c @@ -1652,10 +1652,15 @@ static void mmhub_v9_4_query_ras_error_status(struct amdgpu_device *adev) } } -const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = { +const struct amdgpu_mmhub_ras_funcs mmhub_v9_4_ras_funcs = { .ras_late_init = amdgpu_mmhub_ras_late_init, + .ras_fini = amdgpu_mmhub_ras_fini, .query_ras_error_count = mmhub_v9_4_query_ras_error_count, .reset_ras_error_count = mmhub_v9_4_reset_ras_error_count, + .query_ras_error_status = mmhub_v9_4_query_ras_error_status, +}; + +const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = { .get_fb_location = mmhub_v9_4_get_fb_location, .init = mmhub_v9_4_init, .gart_enable = mmhub_v9_4_gart_enable, @@ -1664,5 +1669,4 @@ const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs = { .set_clockgating = mmhub_v9_4_set_clockgating, .get_clockgating = mmhub_v9_4_get_clockgating, .setup_vm_pt_regs = mmhub_v9_4_setup_vm_pt_regs, - .query_ras_error_status = mmhub_v9_4_query_ras_error_status, }; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h index 92404a8f66f3..90436efa92ef 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.h @@ -24,5 +24,6 @@ #define __MMHUB_V9_4_H__ extern const struct amdgpu_mmhub_funcs mmhub_v9_4_funcs; +extern const struct amdgpu_mmhub_ras_funcs mmhub_v9_4_ras_funcs; #endif -- cgit v1.2.3 From 719a9b332305b8c4b91805c4bedee27ce82ee916 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Fri, 19 Mar 2021 16:59:09 +0800 Subject: drm/amdgpu: split gfx callbacks into ras and non-ras ones gfx ras is only available in cerntain ip generations. Signed-off-by: Hawking Zhang Reviewed-by: Dennis Li Reviewed-by: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 5 ++- drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h | 22 +++++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 30 +++++++++------ drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 68 +++++++++++++-------------------- drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c | 20 +++++++--- drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h | 12 +----- drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c | 13 ++++++- drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h | 9 +---- 8 files changed, 92 insertions(+), 87 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 12e8b527776b..95d4f43a03df 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -677,8 +677,9 @@ int amdgpu_gfx_process_ras_data_cb(struct amdgpu_device *adev, */ if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) { kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); - if (adev->gfx.funcs->query_ras_error_count) - adev->gfx.funcs->query_ras_error_count(adev, err_data); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->query_ras_error_count) + adev->gfx.ras_funcs->query_ras_error_count(adev, err_data); amdgpu_ras_reset_gpu(adev); } return AMDGPU_RAS_SUCCESS; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index 38af93f501e1..d43fe2ed8116 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h @@ -205,6 +205,19 @@ struct amdgpu_cu_info { uint32_t bitmap[4][4]; }; +struct amdgpu_gfx_ras_funcs { + int (*ras_late_init)(struct amdgpu_device *adev); + void (*ras_fini)(struct amdgpu_device *adev); + int (*ras_error_inject)(struct amdgpu_device *adev, + void *inject_if); + int (*query_ras_error_count)(struct amdgpu_device *adev, + void *ras_error_status); + void (*reset_ras_error_count)(struct amdgpu_device *adev); + void (*query_ras_error_status)(struct amdgpu_device *adev); + void (*reset_ras_error_status)(struct amdgpu_device *adev); + void (*enable_watchdog_timer)(struct amdgpu_device *adev); +}; + struct amdgpu_gfx_funcs { /* get the gpu clock counter */ uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev); @@ -220,14 +233,8 @@ struct amdgpu_gfx_funcs { uint32_t *dst); void (*select_me_pipe_q)(struct amdgpu_device *adev, u32 me, u32 pipe, u32 queue, u32 vmid); - int (*ras_error_inject)(struct amdgpu_device *adev, void *inject_if); - int (*query_ras_error_count) (struct amdgpu_device *adev, void *ras_error_status); - void (*reset_ras_error_count) (struct amdgpu_device *adev); void (*init_spm_golden)(struct amdgpu_device *adev); - void (*query_ras_error_status) (struct amdgpu_device *adev); - void (*reset_ras_error_status) (struct amdgpu_device *adev); void (*update_perfmon_mgcg)(struct amdgpu_device *adev, bool enable); - void (*enable_watchdog_timer)(struct amdgpu_device *adev); }; struct sq_work { @@ -330,7 +337,8 @@ struct amdgpu_gfx { DECLARE_BITMAP (pipe_reserve_bitmap, AMDGPU_MAX_COMPUTE_QUEUES); /*ras */ - struct ras_common_if *ras_if; + struct ras_common_if *ras_if; + const struct amdgpu_gfx_ras_funcs *ras_funcs; }; #define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev)) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index b55f470eb747..1d905bcbc1ac 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -792,11 +792,13 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, } break; case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.funcs->query_ras_error_count) - adev->gfx.funcs->query_ras_error_count(adev, &err_data); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->query_ras_error_count) + adev->gfx.ras_funcs->query_ras_error_count(adev, &err_data); - if (adev->gfx.funcs->query_ras_error_status) - adev->gfx.funcs->query_ras_error_status(adev); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->query_ras_error_status) + adev->gfx.ras_funcs->query_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__MMHUB: if (adev->mmhub.ras_funcs && @@ -852,11 +854,13 @@ int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, switch (block) { case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.funcs->reset_ras_error_count) - adev->gfx.funcs->reset_ras_error_count(adev); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->reset_ras_error_count) + adev->gfx.ras_funcs->reset_ras_error_count(adev); - if (adev->gfx.funcs->reset_ras_error_status) - adev->gfx.funcs->reset_ras_error_status(adev); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->reset_ras_error_status) + adev->gfx.ras_funcs->reset_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__MMHUB: if (adev->mmhub.ras_funcs && @@ -926,8 +930,9 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, switch (info->head.block) { case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.funcs->ras_error_inject) - ret = adev->gfx.funcs->ras_error_inject(adev, info); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->ras_error_inject) + ret = adev->gfx.ras_funcs->ras_error_inject(adev, info); else ret = -EINVAL; break; @@ -1514,8 +1519,9 @@ static void amdgpu_ras_error_status_query(struct amdgpu_device *adev, */ switch (info->head.block) { case AMDGPU_RAS_BLOCK__GFX: - if (adev->gfx.funcs->query_ras_error_status) - adev->gfx.funcs->query_ras_error_status(adev); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->query_ras_error_status) + adev->gfx.ras_funcs->query_ras_error_status(adev); break; case AMDGPU_RAS_BLOCK__MMHUB: if (adev->mmhub.ras_funcs && diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 2b7ab8f41cd9..06811a1f4625 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -2103,45 +2103,22 @@ static void gfx_v9_0_select_me_pipe_q(struct amdgpu_device *adev, } static const struct amdgpu_gfx_funcs gfx_v9_0_gfx_funcs = { - .get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter, - .select_se_sh = &gfx_v9_0_select_se_sh, - .read_wave_data = &gfx_v9_0_read_wave_data, - .read_wave_sgprs = &gfx_v9_0_read_wave_sgprs, - .read_wave_vgprs = &gfx_v9_0_read_wave_vgprs, - .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q, + .get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter, + .select_se_sh = &gfx_v9_0_select_se_sh, + .read_wave_data = &gfx_v9_0_read_wave_data, + .read_wave_sgprs = &gfx_v9_0_read_wave_sgprs, + .read_wave_vgprs = &gfx_v9_0_read_wave_vgprs, + .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q, +}; + +static const struct amdgpu_gfx_ras_funcs gfx_v9_0_ras_funcs = { + .ras_late_init = amdgpu_gfx_ras_late_init, + .ras_fini = amdgpu_gfx_ras_fini, .ras_error_inject = &gfx_v9_0_ras_error_inject, .query_ras_error_count = &gfx_v9_0_query_ras_error_count, .reset_ras_error_count = &gfx_v9_0_reset_ras_error_count, }; -static const struct amdgpu_gfx_funcs gfx_v9_4_gfx_funcs = { - .get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter, - .select_se_sh = &gfx_v9_0_select_se_sh, - .read_wave_data = &gfx_v9_0_read_wave_data, - .read_wave_sgprs = &gfx_v9_0_read_wave_sgprs, - .read_wave_vgprs = &gfx_v9_0_read_wave_vgprs, - .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q, - .ras_error_inject = &gfx_v9_4_ras_error_inject, - .query_ras_error_count = &gfx_v9_4_query_ras_error_count, - .reset_ras_error_count = &gfx_v9_4_reset_ras_error_count, - .query_ras_error_status = &gfx_v9_4_query_ras_error_status, -}; - -static const struct amdgpu_gfx_funcs gfx_v9_4_2_gfx_funcs = { - .get_gpu_clock_counter = &gfx_v9_0_get_gpu_clock_counter, - .select_se_sh = &gfx_v9_0_select_se_sh, - .read_wave_data = &gfx_v9_0_read_wave_data, - .read_wave_sgprs = &gfx_v9_0_read_wave_sgprs, - .read_wave_vgprs = &gfx_v9_0_read_wave_vgprs, - .select_me_pipe_q = &gfx_v9_0_select_me_pipe_q, - .ras_error_inject = &gfx_v9_4_2_ras_error_inject, - .query_ras_error_count = &gfx_v9_4_2_query_ras_error_count, - .reset_ras_error_count = &gfx_v9_4_2_reset_ras_error_count, - .query_ras_error_status = &gfx_v9_4_2_query_ras_error_status, - .reset_ras_error_status = &gfx_v9_4_2_reset_ras_error_status, - .enable_watchdog_timer = &gfx_v9_4_2_enable_watchdog_timer, -}; - static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) { u32 gb_addr_config; @@ -2168,6 +2145,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) DRM_INFO("fix gfx.config for vega12\n"); break; case CHIP_VEGA20: + adev->gfx.ras_funcs = &gfx_v9_0_ras_funcs; adev->gfx.config.max_hw_contexts = 8; adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; adev->gfx.config.sc_prim_fifo_size_backend = 0x100; @@ -2193,7 +2171,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) gb_addr_config = RAVEN_GB_ADDR_CONFIG_GOLDEN; break; case CHIP_ARCTURUS: - adev->gfx.funcs = &gfx_v9_4_gfx_funcs; + adev->gfx.ras_funcs = &gfx_v9_4_ras_funcs; adev->gfx.config.max_hw_contexts = 8; adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; adev->gfx.config.sc_prim_fifo_size_backend = 0x100; @@ -2214,7 +2192,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) gb_addr_config |= 0x22010042; break; case CHIP_ALDEBARAN: - adev->gfx.funcs = &gfx_v9_4_2_gfx_funcs; + adev->gfx.ras_funcs = &gfx_v9_4_2_ras_funcs; adev->gfx.config.max_hw_contexts = 8; adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; adev->gfx.config.sc_prim_fifo_size_backend = 0x100; @@ -2447,7 +2425,9 @@ static int gfx_v9_0_sw_fini(void *handle) int i; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - amdgpu_gfx_ras_fini(adev); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->ras_fini) + adev->gfx.ras_funcs->ras_fini(adev); for (i = 0; i < adev->gfx.num_gfx_rings; i++) amdgpu_ring_fini(&adev->gfx.gfx_ring[i]); @@ -4766,12 +4746,16 @@ static int gfx_v9_0_ecc_late_init(void *handle) if (r) return r; - r = amdgpu_gfx_ras_late_init(adev); - if (r) - return r; + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->ras_late_init) { + r = adev->gfx.ras_funcs->ras_late_init(adev); + if (r) + return r; + } - if (adev->gfx.funcs->enable_watchdog_timer) - adev->gfx.funcs->enable_watchdog_timer(adev); + if (adev->gfx.ras_funcs && + adev->gfx.ras_funcs->enable_watchdog_timer) + adev->gfx.ras_funcs->enable_watchdog_timer(adev); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c index bc699d680ce8..830080ff90d8 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c @@ -863,8 +863,8 @@ static int gfx_v9_4_ras_error_count(struct amdgpu_device *adev, return 0; } -int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, - void *ras_error_status) +static int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, + void *ras_error_status) { struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; uint32_t sec_count = 0, ded_count = 0; @@ -906,7 +906,7 @@ int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, return 0; } -void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev) +static void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev) { int i, j, k; @@ -971,7 +971,8 @@ void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev) WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255); } -int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, void *inject_if) +static int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, + void *inject_if) { struct ras_inject_if *info = (struct ras_inject_if *)inject_if; int ret; @@ -996,7 +997,7 @@ int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, void *inject_if) static const struct soc15_reg_entry gfx_v9_4_rdrsp_status_regs = { SOC15_REG_ENTRY(GC, 0, mmGCEA_ERR_STATUS), 0, 1, 32 }; -void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev) +static void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev) { uint32_t i, j; uint32_t reg_value; @@ -1021,3 +1022,12 @@ void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev) gfx_v9_4_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); mutex_unlock(&adev->grbm_idx_mutex); } + +const struct amdgpu_gfx_ras_funcs gfx_v9_4_ras_funcs = { + .ras_late_init = amdgpu_gfx_ras_late_init, + .ras_fini = amdgpu_gfx_ras_fini, + .ras_error_inject = &gfx_v9_4_ras_error_inject, + .query_ras_error_count = &gfx_v9_4_query_ras_error_count, + .reset_ras_error_count = &gfx_v9_4_reset_ras_error_count, + .query_ras_error_status = &gfx_v9_4_query_ras_error_status, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h index 875f18473a98..bdd16b568021 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.h @@ -24,16 +24,6 @@ #ifndef __GFX_V9_4_H__ #define __GFX_V9_4_H__ -void gfx_v9_4_clear_ras_edc_counter(struct amdgpu_device *adev); - -int gfx_v9_4_query_ras_error_count(struct amdgpu_device *adev, - void *ras_error_status); - -int gfx_v9_4_ras_error_inject(struct amdgpu_device *adev, - void *inject_if); - -void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev); - -void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev); +extern const struct amdgpu_gfx_ras_funcs gfx_v9_4_ras_funcs; #endif /* __GFX_V9_4_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c index 2e94998c9812..9ca76a3ac38c 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c @@ -1283,4 +1283,15 @@ static void gfx_v9_4_2_reset_sq_timeout_status(struct amdgpu_device *adev) } gfx_v9_4_2_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); mutex_unlock(&adev->grbm_idx_mutex); -} \ No newline at end of file +} + +const struct amdgpu_gfx_ras_funcs gfx_v9_4_2_ras_funcs = { + .ras_late_init = amdgpu_gfx_ras_late_init, + .ras_fini = amdgpu_gfx_ras_fini, + .ras_error_inject = &gfx_v9_4_2_ras_error_inject, + .query_ras_error_count = &gfx_v9_4_2_query_ras_error_count, + .reset_ras_error_count = &gfx_v9_4_2_reset_ras_error_count, + .query_ras_error_status = &gfx_v9_4_2_query_ras_error_status, + .reset_ras_error_status = &gfx_v9_4_2_reset_ras_error_status, + .enable_watchdog_timer = &gfx_v9_4_2_enable_watchdog_timer, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h index c143d178ef98..81c5833b6b9f 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.h @@ -30,11 +30,6 @@ void gfx_v9_4_2_init_golden_registers(struct amdgpu_device *adev, uint32_t die_id); void gfx_v9_4_2_set_power_brake_sequence(struct amdgpu_device *adev); -void gfx_v9_4_2_reset_ras_error_count(struct amdgpu_device *adev); -int gfx_v9_4_2_ras_error_inject(struct amdgpu_device *adev, void *inject_if); -void gfx_v9_4_2_query_ras_error_status(struct amdgpu_device *adev); -int gfx_v9_4_2_query_ras_error_count(struct amdgpu_device *adev, - void *ras_error_status); -void gfx_v9_4_2_reset_ras_error_status(struct amdgpu_device *adev); -void gfx_v9_4_2_enable_watchdog_timer(struct amdgpu_device *adev); +extern const struct amdgpu_gfx_ras_funcs gfx_v9_4_2_ras_funcs; + #endif /* __GFX_V9_4_2_H__ */ -- cgit v1.2.3 From 134d16d50f0948f00e7172b509e869b6eaecf437 Mon Sep 17 00:00:00 2001 From: John Clements Date: Thu, 25 Mar 2021 17:10:10 +0800 Subject: drm/amdgpu: RAS harvest on driver load In event of RAS UE + warm reset, error counters shall be harvested and cleared on driver load Reviewed-by: Hawking Zhang Signed-off-by: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 1d905bcbc1ac..b0fe5885e4c6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2090,6 +2090,32 @@ release_con: return r; } +static int amdgpu_persistent_edc_harvesting_supported(struct amdgpu_device *adev) +{ + if (adev->gmc.xgmi.connected_to_cpu) + return 1; + return 0; +} + +static int amdgpu_persistent_edc_harvesting(struct amdgpu_device *adev, + struct ras_common_if *ras_block) +{ + struct ras_query_if info = { + .head = *ras_block, + }; + + if (!amdgpu_persistent_edc_harvesting_supported(adev)) + return 0; + + if (amdgpu_ras_query_error_status(adev, &info) != 0) + DRM_WARN("RAS init harvest failure"); + + if (amdgpu_ras_reset_error_status(adev, ras_block->block) != 0) + DRM_WARN("RAS init harvest reset failure"); + + return 0; +} + /* helper function to handle common stuff in ip late init phase */ int amdgpu_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block, @@ -2119,6 +2145,9 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev, return r; } + /* check for errors on warm reset edc persisant supported ASIC */ + amdgpu_persistent_edc_harvesting(adev, ras_block); + /* in resume phase, no need to create ras fs node */ if (adev->in_suspend || amdgpu_in_reset(adev)) return 0; -- cgit v1.2.3 From cbb8f989d5a07cb3e39e9c149a6f89d6c83432aa Mon Sep 17 00:00:00 2001 From: John Clements Date: Fri, 9 Apr 2021 17:25:29 +0800 Subject: drm/amdgpu: page retire over debugfs mechanism added support in RAS debugfs to add bad page for isolated page retirement testing Reviewed-by: Hawking Zhang Signed-off-by: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 67 +++++++++++++++++++++++++++++++++ 1 file changed, 67 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index b0fe5885e4c6..0541196ae1ed 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -99,6 +99,49 @@ static bool amdgpu_ras_get_error_query_ready(struct amdgpu_device *adev) return false; } +static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t address) +{ + struct ras_err_data err_data = {0, 0, 0, NULL}; + struct eeprom_table_record err_rec; + + if ((address >= adev->gmc.mc_vram_size) || + (address >= RAS_UMC_INJECT_ADDR_LIMIT)) { + dev_warn(adev->dev, + "RAS WARN: input address 0x%llx is invalid.\n", + address); + return -EINVAL; + } + + if (amdgpu_ras_check_bad_page(adev, address)) { + dev_warn(adev->dev, + "RAS WARN: 0x%llx has been marked as bad page!\n", + address); + return 0; + } + + memset(&err_rec, 0x0, sizeof(struct eeprom_table_record)); + + err_rec.address = address; + err_rec.retired_page = address >> AMDGPU_GPU_PAGE_SHIFT; + err_rec.ts = (uint64_t)ktime_get_real_seconds(); + err_rec.err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; + + err_data.err_addr = &err_rec; + err_data.err_addr_cnt = 1; + + if (amdgpu_bad_page_threshold != 0) { + amdgpu_ras_add_bad_pages(adev, err_data.err_addr, + err_data.err_addr_cnt); + amdgpu_ras_save_bad_pages(adev); + } + + dev_warn(adev->dev, "WARNING: THIS IS ONLY FOR TEST PURPOSES AND WILL CORRUPT RAS EEPROM\n"); + dev_warn(adev->dev, "Clear EEPROM:\n"); + dev_warn(adev->dev, " echo 1 > /sys/kernel/debug/dri/0/ras/ras_eeprom_reset\n"); + + return 0; +} + static ssize_t amdgpu_ras_debugfs_read(struct file *f, char __user *buf, size_t size, loff_t *pos) { @@ -178,11 +221,25 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, op = 1; else if (sscanf(str, "inject %32s %8s", block_name, err) == 2) op = 2; + else if (sscanf(str, "retire_page") == 0) + op = 3; else if (str[0] && str[1] && str[2] && str[3]) /* ascii string, but commands are not matched. */ return -EINVAL; if (op != -1) { + + if (op == 3) { + if (sscanf(str, "%*s %llu", &address) != 1) + if (sscanf(str, "%*s 0x%llx", &address) != 1) + return -EINVAL; + + data->op = op; + data->inject.address = address; + + return 0; + } + if (amdgpu_ras_find_block_id_by_name(block_name, &block_id)) return -EINVAL; @@ -310,6 +367,16 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, const char __user * if (ret) return -EINVAL; + if (data.op == 3) + { + ret = amdgpu_reserve_page_direct(adev, data.inject.address); + + if (ret) + return size; + else + return ret; + } + if (!amdgpu_ras_is_supported(adev, data.head.block)) return -EINVAL; -- cgit v1.2.3 From 6cb7a1d40acb2425b9dabb99cdaf522bf47295cb Mon Sep 17 00:00:00 2001 From: Luben Tuikov Date: Tue, 13 Apr 2021 07:49:11 -0400 Subject: drm/amdgpu: Fix a bug for input with double sscanf Remove double-sscanf to scan for %llu and 0x%llx, as that is not going to work! The %llu will consume the "0" in "0x" of your input, and the hex value you think you're entering will always be 0. That is, a valid hex value can never be consumed. On the other hand, just entering a hex number without leading 0x will either be scanned as a string and not match, for instance FAB123, or the leading decimal portion is scanned as the %llu, for instance 123FAB will be scanned as 123, which is not correct. Thus remove the first %llu scan and leave only the %llx scan, removing the leading 0x since %llx can scan either. Addresses are usually always hex values, so this suffices. Cc: Alexander Deucher Cc: Xinhui Pan Cc: Hawking Zhang Signed-off-by: Luben Tuikov Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 13 +++++-------- 1 file changed, 5 insertions(+), 8 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 0541196ae1ed..9041453465f1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -230,9 +230,8 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, if (op != -1) { if (op == 3) { - if (sscanf(str, "%*s %llu", &address) != 1) - if (sscanf(str, "%*s 0x%llx", &address) != 1) - return -EINVAL; + if (sscanf(str, "%*s %llx", &address) != 1) + return -EINVAL; data->op = op; data->inject.address = address; @@ -255,11 +254,9 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, data->op = op; if (op == 2) { - if (sscanf(str, "%*s %*s %*s %u %llu %llu", - &sub_block, &address, &value) != 3) - if (sscanf(str, "%*s %*s %*s 0x%x 0x%llx 0x%llx", - &sub_block, &address, &value) != 3) - return -EINVAL; + if (sscanf(str, "%*s %*s %*s %x %llx %llx", + &sub_block, &address, &value) != 3) + return -EINVAL; data->head.sub_block_index = sub_block; data->inject.address = address; data->inject.value = value; -- cgit v1.2.3 From 80b0cd0fb97ec3c02b0f2c4dc26fe3e8be94dc01 Mon Sep 17 00:00:00 2001 From: Luben Tuikov Date: Tue, 13 Apr 2021 08:31:59 -0400 Subject: drm/amdgpu: Fix a bug in checking the result of reserve page Fix if (ret) --> if (!ret), a bug, for "retire_page", which caused the kernel to recall the method with *pos == end of file, and that bounced back with error. On the first run, we advanced *pos, but returned 0 back to fs layer, also a bug. Fix the logic of the check of the result of amdgpu_reserve_page_direct()--it is 0 on success, and non-zero on error, not the other way around. This patch fixes this bug. Cc: Alexander Deucher Cc: John Clements Cc: Hawking Zhang Signed-off-by: Luben Tuikov Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 9041453465f1..3ca6b51f0c9c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -114,7 +114,7 @@ static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t addre if (amdgpu_ras_check_bad_page(adev, address)) { dev_warn(adev->dev, - "RAS WARN: 0x%llx has been marked as bad page!\n", + "RAS WARN: 0x%llx has already been marked as bad page!\n", address); return 0; } @@ -228,7 +228,6 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, return -EINVAL; if (op != -1) { - if (op == 3) { if (sscanf(str, "%*s %llx", &address) != 1) return -EINVAL; @@ -364,11 +363,9 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, const char __user * if (ret) return -EINVAL; - if (data.op == 3) - { + if (data.op == 3) { ret = amdgpu_reserve_page_direct(adev, data.inject.address); - - if (ret) + if (!ret) return size; else return ret; -- cgit v1.2.3 From 7fb6407145479dad41deecb5919d248539e38667 Mon Sep 17 00:00:00 2001 From: Luben Tuikov Date: Tue, 13 Apr 2021 08:40:18 -0400 Subject: drm/amdgpu: Add bad_page_cnt_threshold to debugfs Add bad_page_cnt_threshold to debugfs, an optional file system used for debugging, for reporting purposes only--it usually matches the size of EEPROM but may be different depending on the "bad_page_threshold" kernel module option. The "bad_page_cnt_threshold" is a dynamically computed value. It depends on three things: the VRAM size; the size of the EEPROM (or the size allocated to the RAS table therein); and the "bad_page_threshold" module parameter. It is a dynamically computed value, when the amdgpu module is run, on which further parameters and logic depend, and as such it is helpful to see the dynamically computed value in debugfs. Cc: Alexander Deucher Cc: Hawking Zhang Signed-off-by: Luben Tuikov Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 3ca6b51f0c9c..30cda4b8a461 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1263,6 +1263,8 @@ static struct dentry *amdgpu_ras_debugfs_create_ctrl_node(struct amdgpu_device * &amdgpu_ras_debugfs_ctrl_ops); debugfs_create_file("ras_eeprom_reset", S_IWUGO | S_IRUGO, dir, adev, &amdgpu_ras_debugfs_eeprom_ops); + debugfs_create_u32("bad_page_cnt_threshold", 0444, dir, + &con->bad_page_cnt_threshold); /* * After one uncorrectable error happens, usually GPU recovery will -- cgit v1.2.3 From 737c375b88eb33102164ab38fae7d22a501a47d8 Mon Sep 17 00:00:00 2001 From: Luben Tuikov Date: Tue, 13 Apr 2021 08:48:59 -0400 Subject: drm/amdgpu: Fix kernel-doc for the RAS sysfs interface Imporve the kernel-doc for the RAS sysfs interface. Fix the grammar, fix the context. Cc: Alexander Deucher Cc: Hawking Zhang Signed-off-by: Luben Tuikov Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 47 +++++++++++++++++---------------- 1 file changed, 24 insertions(+), 23 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 30cda4b8a461..44dfb3613e37 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -274,7 +274,7 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, /** * DOC: AMDGPU RAS debugfs control interface * - * It accepts struct ras_debug_if who has two members. + * The control interface accepts struct ras_debug_if which has two members. * * First member: ras_debug_if::head or ras_debug_if::inject. * @@ -299,32 +299,33 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, * * How to use the interface? * - * Programs + * In a program * - * Copy the struct ras_debug_if in your codes and initialize it. - * Write the struct to the control node. + * Copy the struct ras_debug_if in your code and initialize it. + * Write the struct to the control interface. * - * Shells + * From shell * * .. code-block:: bash * - * echo op block [error [sub_block address value]] > .../ras/ras_ctrl + * echo "disable " > /sys/kernel/debug/dri//ras/ras_ctrl + * echo "enable " > /sys/kernel/debug/dri//ras/ras_ctrl + * echo "inject
> /sys/kernel/debug/dri//ras/ras_ctrl * - * Parameters: + * Where N, is the card which you want to affect. * - * op: disable, enable, inject - * disable: only block is needed - * enable: block and error are needed - * inject: error, address, value are needed - * block: umc, sdma, gfx, ......... + * "disable" requires only the block. + * "enable" requires the block and error type. + * "inject" requires the block, error type, address, and value. + * The block is one of: umc, sdma, gfx, etc. * see ras_block_string[] for details - * error: ue, ce - * ue: multi_uncorrectable - * ce: single_correctable - * sub_block: - * sub block index, pass 0 if there is no sub block + * The error type is one of: ue, ce, where, + * ue is multi-uncorrectable + * ce is single-correctable + * The sub-block is a the sub-block index, pass 0 if there is no sub-block. + * The address and value are hexadecimal numbers, leading 0x is optional. * - * here are some examples for bash commands: + * For instance, * * .. code-block:: bash * @@ -332,17 +333,17 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, * echo inject umc ce 0 0 0 > /sys/kernel/debug/dri/0/ras/ras_ctrl * echo disable umc > /sys/kernel/debug/dri/0/ras/ras_ctrl * - * How to check the result? + * How to check the result of the operation? * - * For disable/enable, please check ras features at + * To check disable/enable, see "ras" features at, * /sys/class/drm/card[0/1/2...]/device/ras/features * - * For inject, please check corresponding err count at - * /sys/class/drm/card[0/1/2...]/device/ras/[gfx/sdma/...]_err_count + * To check inject, see the corresponding error count at, + * /sys/class/drm/card[0/1/2...]/device/ras/[gfx|sdma|umc|...]_err_count * * .. note:: * Operations are only allowed on blocks which are supported. - * Please check ras mask at /sys/module/amdgpu/parameters/ras_mask + * Check the "ras" mask at /sys/module/amdgpu/parameters/ras_mask * to see which blocks support RAS on a particular asic. * */ -- cgit v1.2.3 From 546aa546b05aef582956f44ad917c0eee2cb8027 Mon Sep 17 00:00:00 2001 From: Luben Tuikov Date: Wed, 14 Apr 2021 11:17:01 -0400 Subject: drm/amdgpu: Add double-sscanf but invert Add back the double-sscanf so that both decimal and hexadecimal values could be read in, but this time invert the scan so that hexadecimal format with a leading 0x is tried first, and if that fails, then try decimal format. Also use a logical-AND instead of nesting double if-conditional. See commit "drm/amdgpu: Fix a bug for input with double sscanf" Cc: Alexander Deucher Cc: Hawking Zhang Signed-off-by: Luben Tuikov Reviewed-by: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 44dfb3613e37..38a691a3b600 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -229,7 +229,8 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, if (op != -1) { if (op == 3) { - if (sscanf(str, "%*s %llx", &address) != 1) + if (sscanf(str, "%*s 0x%llx", &address) != 1 && + sscanf(str, "%*s %llu", &address) != 1) return -EINVAL; data->op = op; @@ -253,7 +254,9 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, data->op = op; if (op == 2) { - if (sscanf(str, "%*s %*s %*s %x %llx %llx", + if (sscanf(str, "%*s %*s %*s 0x%x 0x%llx 0x%llx", + &sub_block, &address, &value) != 3 && + sscanf(str, "%*s %*s %*s %u %llu %llu", &sub_block, &address, &value) != 3) return -EINVAL; data->head.sub_block_index = sub_block; -- cgit v1.2.3 From 6df23f4c5cef5b6d8879720ef7c842adffc565c9 Mon Sep 17 00:00:00 2001 From: Dennis Li Date: Fri, 16 Apr 2021 22:41:11 +0800 Subject: drm/amdgpu: fix a error injection failed issue because "sscanf(str, "retire_page")" always return 0, if application use the raw data for error injection, it always wrongly falls into "op == 3". Change to use strstr instead. Signed-off-by: Dennis Li Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 38a691a3b600..7438d4e84776 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -221,7 +221,7 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, op = 1; else if (sscanf(str, "inject %32s %8s", block_name, err) == 2) op = 2; - else if (sscanf(str, "retire_page") == 0) + else if (strstr(str, "retire_page") != NULL) op = 3; else if (str[0] && str[1] && str[2] && str[3]) /* ascii string, but commands are not matched. */ -- cgit v1.2.3 From 1f0d8e3781f40c6cae7cb68a4cccfc54dd4ad3a1 Mon Sep 17 00:00:00 2001 From: Mukul Joshi Date: Wed, 24 Mar 2021 11:36:33 -0400 Subject: drm/amdgpu: Reset RAS error count and status regs Reset the RAS error count and error status registers after reading to prevent over reporting error counts on Aldebaran. Signed-off-by: Mukul Joshi Reviewed-By: John Clements Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 7438d4e84776..b0d2fc9454ca 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -501,6 +501,12 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev, if (amdgpu_ras_query_error_status(obj->adev, &info)) return -EINVAL; + + if (obj->adev->asic_type == CHIP_ALDEBARAN) { + if (amdgpu_ras_reset_error_status(obj->adev, info.head.block)) + DRM_WARN("Failed to reset error counter and error status"); + } + return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count, "ce", info.ce_count); } -- cgit v1.2.3