From a80fe1a698dcc7f5a493af9a16d516f8cbb20564 Mon Sep 17 00:00:00 2001 From: Tao Zhou Date: Thu, 29 Jun 2023 17:48:38 +0800 Subject: drm/amdgpu: skip address adjustment for GFX RAS injection The address parameter of GFX RAS injection isn't related to XGMI node number, keep it unchanged. Signed-off-by: Tao Zhou Reviewed-by: Candice Li Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 3 ++- 1 file changed, 2 insertions(+), 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 8aaa427f8c0f..06aec5f2dd1a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1159,7 +1159,8 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, } /* Calculate XGMI relative offset */ - if (adev->gmc.xgmi.num_physical_nodes > 1) { + if (adev->gmc.xgmi.num_physical_nodes > 1 && + info->head.block != AMDGPU_RAS_BLOCK__GFX) { block_info.address = amdgpu_xgmi_get_relative_phy_addr(adev, block_info.address); -- cgit From cb906ce32b468099da17ef4181d60eaa77ee89d7 Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Sun, 25 Jun 2023 18:19:08 +0800 Subject: drm/amdgpu: Enable aqua vanjaram RAS Enable RAS for aqua vanjaram. Changed from V1: Split the change in amdgpu_ras_asic_supported into a separated patch. Changed from V2: Avoid to modify global variable amdgpu_ras_enable. Signed-off-by: Stanley.Yang Reviewed-by: Hawking Zhang Reviewed-by: Tao Zhou 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 06aec5f2dd1a..51731ccaa0a0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2415,6 +2415,7 @@ static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev) if (adev->asic_type == CHIP_IP_DISCOVERY) { switch (adev->ip_versions[MP0_HWIP][0]) { case IP_VERSION(13, 0, 0): + case IP_VERSION(13, 0, 6): case IP_VERSION(13, 0, 10): return true; default: -- cgit From 276f6e8cb769b0b9782d9daae307f6a182af3368 Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Thu, 13 Jul 2023 21:07:05 +0800 Subject: drm/amdgpu: Disable RAS by default on APU flatform Disable RAS feature by default for aqua vanjaram on APU platform. Changed from V1: Splite Disable RAS by default on APU platform into a separated patch. Changed from V2: Avoid to modify global variable amdgpu_ras_enable. Signed-off-by: Stanley.Yang Reviewed-by: Hawking Zhang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 13 +++++++++++-- 1 file changed, 11 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 51731ccaa0a0..fc317b005943 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2517,8 +2517,17 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) /* hw_supported needs to be aligned with RAS block mask. */ adev->ras_hw_enabled &= AMDGPU_RAS_BLOCK_MASK; - adev->ras_enabled = amdgpu_ras_enable == 0 ? 0 : - adev->ras_hw_enabled & amdgpu_ras_mask; + + /* + * Disable ras feature for aqua vanjaram + * by default on apu platform. + */ + if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 6)) + adev->ras_enabled = amdgpu_ras_enable != 1 ? 0 : + adev->ras_hw_enabled & amdgpu_ras_mask; + else + adev->ras_enabled = amdgpu_ras_enable == 0 ? 0 : + adev->ras_hw_enabled & amdgpu_ras_mask; } static void amdgpu_ras_counte_dw(struct work_struct *work) -- cgit From adf64e214280a0230af0638c1825b3812421c958 Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Tue, 18 Jul 2023 13:02:58 -0500 Subject: drm/amd: Avoid reading the VBIOS part number twice The VBIOS part number is read both in amdgpu_atom_parse() as well as in atom_get_vbios_pn() and stored twice in the `struct atom_context` structure. Remove the first unnecessary read and move the `pr_info` line from that read into the second. v2: squash in unused variable removal Signed-off-by: Mario Limonciello Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c | 20 ++++++++++---------- drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 8 ++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c | 8 ++++---- drivers/gpu/drm/amd/amdgpu/atom.c | 14 ++------------ drivers/gpu/drm/amd/amdgpu/atom.h | 2 -- 7 files changed, 22 insertions(+), 34 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 f4e3c133a16c..dce9e7d5e4ec 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c @@ -1776,7 +1776,7 @@ 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 sysfs_emit(buf, "%s\n", ctx->vbios_version); + return sysfs_emit(buf, "%s\n", ctx->vbios_pn); } static DEVICE_ATTR(vbios_version, 0444, amdgpu_atombios_get_vbios_version, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c index 4620c4712ce3..c9f16eab0f3d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c @@ -60,10 +60,10 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev, u32 *fru_addr) switch (adev->asic_type) { case CHIP_VEGA20: /* D161 and D163 are the VG20 server SKUs */ - if (strnstr(atom_ctx->vbios_version, "D161", - sizeof(atom_ctx->vbios_version)) || - strnstr(atom_ctx->vbios_version, "D163", - sizeof(atom_ctx->vbios_version))) { + if (strnstr(atom_ctx->vbios_pn, "D161", + sizeof(atom_ctx->vbios_pn)) || + strnstr(atom_ctx->vbios_pn, "D163", + sizeof(atom_ctx->vbios_pn))) { if (fru_addr) *fru_addr = FRU_EEPROM_MADDR_6; return true; @@ -72,16 +72,16 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev, u32 *fru_addr) } case CHIP_ALDEBARAN: /* All Aldebaran SKUs have an FRU */ - if (!strnstr(atom_ctx->vbios_version, "D673", - sizeof(atom_ctx->vbios_version))) + if (!strnstr(atom_ctx->vbios_pn, "D673", + sizeof(atom_ctx->vbios_pn))) if (fru_addr) *fru_addr = FRU_EEPROM_MADDR_6; return true; case CHIP_SIENNA_CICHLID: - if (strnstr(atom_ctx->vbios_version, "D603", - sizeof(atom_ctx->vbios_version))) { - if (strnstr(atom_ctx->vbios_version, "D603GLXE", - sizeof(atom_ctx->vbios_version))) { + if (strnstr(atom_ctx->vbios_pn, "D603", + sizeof(atom_ctx->vbios_pn))) { + if (strnstr(atom_ctx->vbios_pn, "D603GLXE", + sizeof(atom_ctx->vbios_pn))) { return false; } else { if (fru_addr) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index cab2fdd5b76a..12f3c7ad7d04 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -1722,7 +1722,7 @@ static int amdgpu_debugfs_firmware_info_show(struct seq_file *m, void *unused) seq_printf(m, "MES feature version: %u, firmware version: 0x%08x\n", fw_info.feature, fw_info.ver); - seq_printf(m, "VBIOS version: %s\n", ctx->vbios_version); + seq_printf(m, "VBIOS version: %s\n", ctx->vbios_pn); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index fc317b005943..0ba86cc88481 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2442,10 +2442,10 @@ static void amdgpu_ras_get_quirks(struct amdgpu_device *adev) if (!ctx) return; - if (strnstr(ctx->vbios_version, "D16406", - sizeof(ctx->vbios_version)) || - strnstr(ctx->vbios_version, "D36002", - sizeof(ctx->vbios_version))) + if (strnstr(ctx->vbios_pn, "D16406", + sizeof(ctx->vbios_pn)) || + strnstr(ctx->vbios_pn, "D36002", + sizeof(ctx->vbios_pn))) adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__GFX); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index 0648dfe559af..4287743e1212 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -194,9 +194,9 @@ static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev, /* VEGA20 and ARCTURUS */ if (adev->asic_type == CHIP_VEGA20) control->i2c_address = EEPROM_I2C_MADDR_0; - else if (strnstr(atom_ctx->vbios_version, + else if (strnstr(atom_ctx->vbios_pn, "D342", - sizeof(atom_ctx->vbios_version))) + sizeof(atom_ctx->vbios_pn))) control->i2c_address = EEPROM_I2C_MADDR_0; else control->i2c_address = EEPROM_I2C_MADDR_4; @@ -205,8 +205,8 @@ static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev, control->i2c_address = EEPROM_I2C_MADDR_0; return true; case IP_VERSION(13, 0, 2): - if (strnstr(atom_ctx->vbios_version, "D673", - sizeof(atom_ctx->vbios_version))) + if (strnstr(atom_ctx->vbios_pn, "D673", + sizeof(atom_ctx->vbios_pn))) control->i2c_address = EEPROM_I2C_MADDR_4; else control->i2c_address = EEPROM_I2C_MADDR_0; diff --git a/drivers/gpu/drm/amd/amdgpu/atom.c b/drivers/gpu/drm/amd/amdgpu/atom.c index 1c5d9388ad0b..9f63ddb89b75 100644 --- a/drivers/gpu/drm/amd/amdgpu/atom.c +++ b/drivers/gpu/drm/amd/amdgpu/atom.c @@ -1438,6 +1438,8 @@ static void atom_get_vbios_pn(struct atom_context *ctx) ctx->vbios_pn[count] = 0; } + + pr_info("ATOM BIOS: %s\n", ctx->vbios_pn); } static void atom_get_vbios_version(struct atom_context *ctx) @@ -1460,11 +1462,9 @@ struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios) int base; struct atom_context *ctx = kzalloc(sizeof(struct atom_context), GFP_KERNEL); - char *str; struct _ATOM_ROM_HEADER *atom_rom_header; struct _ATOM_MASTER_DATA_TABLE *master_table; struct _ATOM_FIRMWARE_INFO *atom_fw_info; - u16 idx; if (!ctx) return NULL; @@ -1502,16 +1502,6 @@ struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios) return NULL; } - idx = CU16(ATOM_ROM_PART_NUMBER_PTR); - if (idx == 0) - idx = 0x80; - - str = CSTR(idx); - if (*str != '\0') { - pr_info("ATOM BIOS: %s\n", str); - strlcpy(ctx->vbios_version, str, sizeof(ctx->vbios_version)); - } - atom_rom_header = (struct _ATOM_ROM_HEADER *)CSTR(base); if (atom_rom_header->usMasterDataTableOffset != 0) { master_table = (struct _ATOM_MASTER_DATA_TABLE *) diff --git a/drivers/gpu/drm/amd/amdgpu/atom.h b/drivers/gpu/drm/amd/amdgpu/atom.h index 0c1839824520..55bf99d5288d 100644 --- a/drivers/gpu/drm/amd/amdgpu/atom.h +++ b/drivers/gpu/drm/amd/amdgpu/atom.h @@ -33,7 +33,6 @@ struct drm_device; #define ATOM_ATI_MAGIC_PTR 0x30 #define ATOM_ATI_MAGIC " 761295520" #define ATOM_ROM_TABLE_PTR 0x48 -#define ATOM_ROM_PART_NUMBER_PTR 0x6E #define ATOM_ROM_MAGIC "ATOM" #define ATOM_ROM_MAGIC_PTR 4 @@ -143,7 +142,6 @@ struct atom_context { int io_mode; uint32_t *scratch; int scratch_size_bytes; - char vbios_version[20]; uint8_t name[STRLEN_LONG]; uint8_t vbios_pn[STRLEN_LONG]; -- cgit From fcb7a1849aea4a31b7155d0816ad5070983da3d0 Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Fri, 21 Jul 2023 16:38:02 +0800 Subject: drm/amdgpu: Check APU flag to disable RAS Only disable RAS by default for aqua vanjaram on APU platform. Signed-off-by: Stanley.Yang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 3 ++- 1 file changed, 2 insertions(+), 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 0ba86cc88481..62011a521833 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2522,7 +2522,8 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) * Disable ras feature for aqua vanjaram * by default on apu platform. */ - if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 6)) + if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 6) && + adev->gmc.is_app_apu) adev->ras_enabled = amdgpu_ras_enable != 1 ? 0 : adev->ras_hw_enabled & amdgpu_ras_mask; else -- cgit From f957138cc30ae904346f117013fc7b428700c1f7 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Mon, 31 Jul 2023 23:43:12 +0800 Subject: drm/amdgpu: Only create err_count sysfs when hw_op is supported Some IP blocks only support partial ras feature and don't have ras counter and/or ras error status register at all. Driver should not create err_count sysfs node for those IP blocks. Signed-off-by: Hawking Zhang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 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 62011a521833..fa53a51db69f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2777,23 +2777,28 @@ int amdgpu_ras_block_late_init(struct amdgpu_device *adev, goto cleanup; } - r = amdgpu_ras_sysfs_create(adev, ras_block); - if (r) - goto interrupt; + if (ras_obj->hw_ops && + (ras_obj->hw_ops->query_ras_error_count || + ras_obj->hw_ops->query_ras_error_status)) { + r = amdgpu_ras_sysfs_create(adev, ras_block); + if (r) + goto interrupt; - /* Those are the cached values at init. - */ - query_info = kzalloc(sizeof(struct ras_query_if), GFP_KERNEL); - if (!query_info) - return -ENOMEM; - memcpy(&query_info->head, ras_block, sizeof(struct ras_common_if)); + /* Those are the cached values at init. + */ + query_info = kzalloc(sizeof(*query_info), GFP_KERNEL); + if (!query_info) + return -ENOMEM; + memcpy(&query_info->head, ras_block, sizeof(struct ras_common_if)); - if (amdgpu_ras_query_error_count(adev, &ce_count, &ue_count, query_info) == 0) { - atomic_set(&con->ras_ce_count, ce_count); - atomic_set(&con->ras_ue_count, ue_count); + if (amdgpu_ras_query_error_count(adev, &ce_count, &ue_count, query_info) == 0) { + atomic_set(&con->ras_ce_count, ce_count); + atomic_set(&con->ras_ue_count, ue_count); + } + + kfree(query_info); } - kfree(query_info); return 0; interrupt: -- cgit From 62c4b772bdd91ed1745c4a55b681e94aa9e48071 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Mon, 3 Jul 2023 14:38:45 +0800 Subject: drm/amdgpu: Apply poison mode check to GFX IP only For GFX IP that only supports poison consumption, GFX RAS won't be marked as enabled. i.e., hardware doesn't support gfx sram ecc. But driver still needs to issue firmware to enable poison consumption mode for GFX IP. In such case, check poison mode and treat GFX IP as RAS capable IP block. Signed-off-by: Hawking Zhang Reviewed-by: Tao Zhou 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 fa53a51db69f..664b841f5cec 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3153,6 +3153,7 @@ int amdgpu_ras_is_supported(struct amdgpu_device *adev, * that the ras block supports ras function. */ if (!ret && + block == AMDGPU_RAS_BLOCK__GFX && amdgpu_ras_is_poison_mode_supported(adev) && amdgpu_ras_get_ras_block(adev, block, 0)) ret = 1; -- cgit From 6fc9d92c3d27f572eb1d884662f199c0d7f90d16 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Mon, 3 Jul 2023 16:17:46 +0800 Subject: drm/amdgpu: Issue ras enable_feature for gfx ip only For non-GFX IP blocks, set up ras obj if ras feature is allowed. For GFX IP blocks, force issue ras enable_feature command to firmware and only set up ras obj if ras feature is allowed Signed-off-by: Hawking Zhang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 30 ++++++++++-------------------- 1 file changed, 10 insertions(+), 20 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 664b841f5cec..bb29cb57add5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -757,16 +757,6 @@ static int __amdgpu_ras_feature_enable(struct amdgpu_device *adev, return 0; } -static int amdgpu_ras_check_feature_allowed(struct amdgpu_device *adev, - struct ras_common_if *head) -{ - if (amdgpu_ras_is_feature_allowed(adev, head) || - amdgpu_ras_is_poison_mode_supported(adev)) - return 1; - else - return 0; -} - /* wrapper of psp_ras_enable_features */ int amdgpu_ras_feature_enable(struct amdgpu_device *adev, struct ras_common_if *head, bool enable) @@ -778,7 +768,16 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev, if (!con) return -EINVAL; - if (head->block == AMDGPU_RAS_BLOCK__GFX) { + /* Do not enable ras feature if it is not allowed */ + if (enable && + head->block != AMDGPU_RAS_BLOCK__GFX && + !amdgpu_ras_is_feature_allowed(adev, head)) + goto out; + + /* Only enable gfx ras feature from host side */ + if (head->block == AMDGPU_RAS_BLOCK__GFX && + !amdgpu_sriov_vf(adev) && + !amdgpu_ras_intr_triggered()) { info = kzalloc(sizeof(union ta_ras_cmd_input), GFP_KERNEL); if (!info) return -ENOMEM; @@ -794,16 +793,7 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev, .error_type = amdgpu_ras_error_to_ta(head->type), }; } - } - /* Do not enable if it is not allowed. */ - if (enable && !amdgpu_ras_check_feature_allowed(adev, head)) - goto out; - - /* Only enable ras feature operation handle on host side */ - if (head->block == AMDGPU_RAS_BLOCK__GFX && - !amdgpu_sriov_vf(adev) && - !amdgpu_ras_intr_triggered()) { ret = psp_ras_enable_features(&adev->psp, info, enable); if (ret) { dev_err(adev->dev, "ras %s %s failed poison:%d ret:%d\n", -- cgit From 7692e1ee2446fd1940b5caa6e09779504a58881a Mon Sep 17 00:00:00 2001 From: Tao Zhou Date: Thu, 8 Jun 2023 16:25:42 +0800 Subject: drm/amdgpu: add RAS fatal error handler for NBIO v7.9 Register RAS fatal error interrupt and add handler. v2: only register NBIO RAS for dGPU platform. change nbio_v7_9_set_ras_controller_irq_state and nbio_v7_9_set_ras_err_event_athub_irq_state to dummy functions. Signed-off-by: Tao Zhou Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 5 + drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c | 187 ++++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/nbio_v7_9.h | 1 + 3 files changed, 193 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 bb29cb57add5..00658c2816dc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -35,6 +35,7 @@ #include "amdgpu_xgmi.h" #include "ivsrcid/nbio/irqsrcs_nbif_7_4.h" #include "nbio_v4_3.h" +#include "nbio_v7_9.h" #include "atom.h" #include "amdgpu_reset.h" @@ -2644,6 +2645,10 @@ int amdgpu_ras_init(struct amdgpu_device *adev) * check DF RAS */ adev->nbio.ras = &nbio_v4_3_ras; break; + case IP_VERSION(7, 9, 0): + if (!adev->gmc.is_app_apu) + adev->nbio.ras = &nbio_v7_9_ras; + break; default: /* nbio ras is not available */ break; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c index 1d1ab188ef15..781f98655567 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c @@ -471,3 +471,190 @@ const struct amdgpu_nbio_funcs nbio_v7_9_funcs = { .init_registers = nbio_v7_9_init_registers, .get_pcie_replay_count = nbio_v7_9_get_pcie_replay_count, }; + +static void nbio_v7_9_query_ras_error_count(struct amdgpu_device *adev, + void *ras_error_status) +{ + return; +} + +static void nbio_v7_9_handle_ras_controller_intr_no_bifring(struct amdgpu_device *adev) +{ + uint32_t bif_doorbell_intr_cntl; + struct ras_manager *obj = amdgpu_ras_find_obj(adev, adev->nbio.ras_if); + struct ras_err_data err_data = {0, 0, 0, NULL}; + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + + bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL); + + if (REG_GET_FIELD(bif_doorbell_intr_cntl, + BIF_BX0_BIF_DOORBELL_INT_CNTL, RAS_CNTLR_INTERRUPT_STATUS)) { + /* driver has to clear the interrupt status when bif ring is disabled */ + bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl, + BIF_BX0_BIF_DOORBELL_INT_CNTL, + RAS_CNTLR_INTERRUPT_CLEAR, 1); + WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl); + + if (!ras->disable_ras_err_cnt_harvest) { + /* + * clear error status after ras_controller_intr + * according to hw team and count ue number + * for query + */ + nbio_v7_9_query_ras_error_count(adev, &err_data); + + /* logging on error cnt and printing for awareness */ + obj->err_data.ue_count += err_data.ue_count; + obj->err_data.ce_count += err_data.ce_count; + + if (err_data.ce_count) + dev_info(adev->dev, "%ld correctable hardware " + "errors detected in %s block, " + "no user action is needed.\n", + obj->err_data.ce_count, + get_ras_block_str(adev->nbio.ras_if)); + + if (err_data.ue_count) + dev_info(adev->dev, "%ld uncorrectable hardware " + "errors detected in %s block\n", + obj->err_data.ue_count, + get_ras_block_str(adev->nbio.ras_if)); + } + + dev_info(adev->dev, "RAS controller interrupt triggered " + "by NBIF error\n"); + + /* ras_controller_int is dedicated for nbif ras error, + * not the global interrupt for sync flood + */ + amdgpu_ras_reset_gpu(adev); + } +} + +static void nbio_v7_9_handle_ras_err_event_athub_intr_no_bifring(struct amdgpu_device *adev) +{ + uint32_t bif_doorbell_intr_cntl; + + bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL); + + if (REG_GET_FIELD(bif_doorbell_intr_cntl, + BIF_BX0_BIF_DOORBELL_INT_CNTL, RAS_ATHUB_ERR_EVENT_INTERRUPT_STATUS)) { + /* driver has to clear the interrupt status when bif ring is disabled */ + bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl, + BIF_BX0_BIF_DOORBELL_INT_CNTL, + RAS_ATHUB_ERR_EVENT_INTERRUPT_CLEAR, 1); + + WREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl); + + amdgpu_ras_global_ras_isr(adev); + } +} + +static int nbio_v7_9_set_ras_controller_irq_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *src, + unsigned type, + enum amdgpu_interrupt_state state) +{ + /* Dummy function, there is no initialization operation in driver */ + + return 0; +} + +static int nbio_v7_9_process_ras_controller_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + /* By design, the ih cookie for ras_controller_irq should be written + * to BIFring instead of general iv ring. However, due to known bif ring + * hw bug, it has to be disabled. There is no chance the process function + * will be involked. Just left it as a dummy one. + */ + return 0; +} + +static int nbio_v7_9_set_ras_err_event_athub_irq_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *src, + unsigned type, + enum amdgpu_interrupt_state state) +{ + /* Dummy function, there is no initialization operation in driver */ + + return 0; +} + +static int nbio_v7_9_process_err_event_athub_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + /* By design, the ih cookie for err_event_athub_irq should be written + * to BIFring instead of general iv ring. However, due to known bif ring + * hw bug, it has to be disabled. There is no chance the process function + * will be involked. Just left it as a dummy one. + */ + return 0; +} + +static const struct amdgpu_irq_src_funcs nbio_v7_9_ras_controller_irq_funcs = { + .set = nbio_v7_9_set_ras_controller_irq_state, + .process = nbio_v7_9_process_ras_controller_irq, +}; + +static const struct amdgpu_irq_src_funcs nbio_v7_9_ras_err_event_athub_irq_funcs = { + .set = nbio_v7_9_set_ras_err_event_athub_irq_state, + .process = nbio_v7_9_process_err_event_athub_irq, +}; + +static int nbio_v7_9_init_ras_controller_interrupt (struct amdgpu_device *adev) +{ + int r; + + /* init the irq funcs */ + adev->nbio.ras_controller_irq.funcs = + &nbio_v7_9_ras_controller_irq_funcs; + adev->nbio.ras_controller_irq.num_types = 1; + + /* register ras controller interrupt */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, + NBIF_7_4__SRCID__RAS_CONTROLLER_INTERRUPT, + &adev->nbio.ras_controller_irq); + + return r; +} + +static int nbio_v7_9_init_ras_err_event_athub_interrupt (struct amdgpu_device *adev) +{ + + int r; + + /* init the irq funcs */ + adev->nbio.ras_err_event_athub_irq.funcs = + &nbio_v7_9_ras_err_event_athub_irq_funcs; + adev->nbio.ras_err_event_athub_irq.num_types = 1; + + /* register ras err event athub interrupt */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, + NBIF_7_4__SRCID__ERREVENT_ATHUB_INTERRUPT, + &adev->nbio.ras_err_event_athub_irq); + + return r; +} + +const struct amdgpu_ras_block_hw_ops nbio_v7_9_ras_hw_ops = { + .query_ras_error_count = nbio_v7_9_query_ras_error_count, +}; + +struct amdgpu_nbio_ras nbio_v7_9_ras = { + .ras_block = { + .ras_comm = { + .name = "pcie_bif", + .block = AMDGPU_RAS_BLOCK__PCIE_BIF, + .type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE, + }, + .hw_ops = &nbio_v7_9_ras_hw_ops, + .ras_late_init = amdgpu_nbio_ras_late_init, + }, + .handle_ras_controller_intr_no_bifring = nbio_v7_9_handle_ras_controller_intr_no_bifring, + .handle_ras_err_event_athub_intr_no_bifring = nbio_v7_9_handle_ras_err_event_athub_intr_no_bifring, + .init_ras_controller_interrupt = nbio_v7_9_init_ras_controller_interrupt, + .init_ras_err_event_athub_interrupt = nbio_v7_9_init_ras_err_event_athub_interrupt, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.h b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.h index 8e04eb484328..73709771950d 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.h +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.h @@ -28,5 +28,6 @@ extern const struct nbio_hdp_flush_reg nbio_v7_9_hdp_flush_reg; extern const struct amdgpu_nbio_funcs nbio_v7_9_funcs; +extern struct amdgpu_nbio_ras nbio_v7_9_ras; #endif -- cgit From bc0f80802d735950738167e2cc0b51b0dd41e68d Mon Sep 17 00:00:00 2001 From: Candice Li Date: Tue, 8 Aug 2023 16:59:18 +0800 Subject: drm/amdgpu: Extend poison mode check to SDMA/VCN/JPEG Treat SDMA/VCN/JPEG as RAS capable IP blocks in poison mode. Signed-off-by: Candice Li Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 5 ++++- 1 file changed, 4 insertions(+), 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 00658c2816dc..886c7c6f1c3b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3148,7 +3148,10 @@ int amdgpu_ras_is_supported(struct amdgpu_device *adev, * that the ras block supports ras function. */ if (!ret && - block == AMDGPU_RAS_BLOCK__GFX && + (block == AMDGPU_RAS_BLOCK__GFX || + block == AMDGPU_RAS_BLOCK__SDMA || + block == AMDGPU_RAS_BLOCK__VCN || + block == AMDGPU_RAS_BLOCK__JPEG) && amdgpu_ras_is_poison_mode_supported(adev) && amdgpu_ras_get_ras_block(adev, block, 0)) ret = 1; -- cgit From 8b3a7a707c6c5f7ccde47cf2427a560675cc5202 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Wed, 9 Aug 2023 19:07:50 +0800 Subject: drm/amdgpu: Remove unnecessary ras cap check RAS global isr will only be invoked by hardware interrupt. Don't need to query ras capability in isr In addition, amdgpu_ras_interrupt_fatal_error_handler ensures the isr won't be called from guest linux side by accident. The RAS cap check in isr that introduced to fix sriov crash is not needed any more Signed-off-by: Hawking Zhang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 4 ---- 1 file changed, 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 886c7c6f1c3b..da52d7c6d210 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2970,10 +2970,6 @@ int amdgpu_ras_fini(struct amdgpu_device *adev) void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev) { - amdgpu_ras_check_supported(adev); - if (!adev->ras_hw_enabled) - return; - if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) { struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); -- cgit From 1b98a5f8e04b944ba93444a8004690391a60fcf1 Mon Sep 17 00:00:00 2001 From: YiPeng Chai Date: Tue, 8 Aug 2023 10:02:30 +0800 Subject: drm/amdgpu: mode1 reset needs to recover mp1 for mp0 v13_0_10 Mode1 reset needs to recover mp1 in fatal error case for mp0 v13_0_10. v2: Define a macro to wrap psp function calls. Signed-off-by: YiPeng Chai Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h | 5 +++++ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 2 ++ drivers/gpu/drm/amd/amdgpu/psp_v13_0.c | 24 +++++++++++++++++++++++- 3 files changed, 30 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index 3c7d9051b8c6..3384eb94fde0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -132,6 +132,7 @@ struct psp_funcs { int (*read_usbc_pd_fw)(struct psp_context *psp, uint32_t *fw_ver); int (*update_spirom)(struct psp_context *psp, uint64_t fw_pri_mc_addr); int (*vbflash_stat)(struct psp_context *psp); + int (*fatal_error_recovery_quirk)(struct psp_context *psp); }; struct ta_funcs { @@ -445,6 +446,10 @@ struct amdgpu_psp_funcs { ((psp)->funcs->vbflash_stat ? \ (psp)->funcs->vbflash_stat((psp)) : -EINVAL) +#define psp_fatal_error_recovery_quirk(psp) \ + ((psp)->funcs->fatal_error_recovery_quirk ? \ + (psp)->funcs->fatal_error_recovery_quirk((psp)) : 0) + extern const struct amd_ip_funcs psp_ip_funcs; extern const struct amdgpu_ip_block_version psp_v3_1_ip_block; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index da52d7c6d210..7689395e44fd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2064,6 +2064,8 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) if (ras->gpu_reset_flags & AMDGPU_RAS_GPU_RESET_MODE1_RESET) { ras->gpu_reset_flags &= ~AMDGPU_RAS_GPU_RESET_MODE1_RESET; set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); + + psp_fatal_error_recovery_quirk(&adev->psp); } } diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c index 58db1ee631b3..10b17bd5aebe 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c @@ -691,6 +691,27 @@ static int psp_v13_0_vbflash_status(struct psp_context *psp) return RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_115); } +static int psp_v13_0_fatal_error_recovery_quirk(struct psp_context *psp) +{ + struct amdgpu_device *adev = psp->adev; + + if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 10)) { + uint32_t reg_data; + /* MP1 fatal error: trigger PSP dram read to unhalt PSP + * during MP1 triggered sync flood. + */ + reg_data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67); + WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_67, reg_data + 0x10); + + /* delay 1000ms for the mode1 reset for fatal error + * to be recovered back. + */ + msleep(1000); + } + + return 0; +} + static const struct psp_funcs psp_v13_0_funcs = { .init_microcode = psp_v13_0_init_microcode, .bootloader_load_kdb = psp_v13_0_bootloader_load_kdb, @@ -710,7 +731,8 @@ static const struct psp_funcs psp_v13_0_funcs = { .load_usbc_pd_fw = psp_v13_0_load_usbc_pd_fw, .read_usbc_pd_fw = psp_v13_0_read_usbc_pd_fw, .update_spirom = psp_v13_0_update_spirom, - .vbflash_stat = psp_v13_0_vbflash_status + .vbflash_stat = psp_v13_0_vbflash_status, + .fatal_error_recovery_quirk = psp_v13_0_fatal_error_recovery_quirk, }; void psp_v13_0_set_psp_funcs(struct psp_context *psp) -- cgit From 80578f1641aaba95c70adc40d84d1b619e92f2d4 Mon Sep 17 00:00:00 2001 From: YiPeng Chai Date: Tue, 15 Aug 2023 17:39:30 +0800 Subject: drm/amdgpu: Enable ras for mp0 v13_0_6 sriov Enable ras for mp0 v13_0_6 sriov Signed-off-by: YiPeng Chai Reviewed-by: Stanley.Yang 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 7689395e44fd..378478cf9c21 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2399,6 +2399,7 @@ static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) { switch (adev->ip_versions[MP0_HWIP][0]) { case IP_VERSION(13, 0, 2): + case IP_VERSION(13, 0, 6): return true; default: return false; -- cgit From ec70578c83eaf52b78488c887bc2d40a4c3bffe2 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Thu, 24 Aug 2023 16:18:59 +0800 Subject: drm/amdgpu: Allow issue disable gfx ras cmd to firmware Disable gfx ras command is needed in some use cases Signed-off-by: Hawking Zhang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 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 378478cf9c21..e47600a8e88e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -769,9 +769,10 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev, if (!con) return -EINVAL; - /* Do not enable ras feature if it is not allowed */ - if (enable && - head->block != AMDGPU_RAS_BLOCK__GFX && + /* For non-gfx ip, do not enable ras feature if it is not allowed */ + /* For gfx ip, regardless of feature support status, */ + /* Force issue enable or disable ras feature commands */ + if (head->block != AMDGPU_RAS_BLOCK__GFX && !amdgpu_ras_is_feature_allowed(adev, head)) goto out; -- cgit From bf7aa8bea9cbe26dfc7fe5cf9c0ed05a419dbb64 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Tue, 29 Aug 2023 23:20:27 +0800 Subject: drm/amdgpu: Free ras cmd input buffer properly Do not access the pointer for ras input cmd buffer if it is even not allocated. Signed-off-by: Hawking Zhang Reviewed-by: Stanley Yang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 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 e47600a8e88e..8eb6f6943778 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -764,7 +764,7 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev, { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); union ta_ras_cmd_input *info; - int ret = 0; + int ret; if (!con) return -EINVAL; @@ -774,7 +774,7 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev, /* Force issue enable or disable ras feature commands */ if (head->block != AMDGPU_RAS_BLOCK__GFX && !amdgpu_ras_is_feature_allowed(adev, head)) - goto out; + return 0; /* Only enable gfx ras feature from host side */ if (head->block == AMDGPU_RAS_BLOCK__GFX && @@ -802,16 +802,16 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev, enable ? "enable":"disable", get_ras_block_str(head), amdgpu_ras_is_poison_mode_supported(adev), ret); - goto out; + return ret; } + + kfree(info); } /* setup the obj */ __amdgpu_ras_feature_enable(adev, head, enable); -out: - if (head->block == AMDGPU_RAS_BLOCK__GFX) - kfree(info); - return ret; + + return 0; } /* Only used in device probe stage and called only once. */ -- cgit From 9f9d4651f70e43df76614634670c02b99cfee57e Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Fri, 8 Sep 2023 21:21:55 +0800 Subject: drm/amdgpu: fallback to old RAS error message for aqua_vanjaram So driver doesn't generate incorrect message until the new format is settled down for aqua_vanjaram Signed-off-by: Hawking Zhang Reviewed-by: Yang Wang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 6 ++++-- 1 file changed, 4 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 8eb6f6943778..632478874f7d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1053,7 +1053,8 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, info->ce_count = obj->err_data.ce_count; if (err_data.ce_count) { - if (adev->smuio.funcs && + if (!adev->aid_mask && + adev->smuio.funcs && adev->smuio.funcs->get_socket_id && adev->smuio.funcs->get_die_id) { dev_info(adev->dev, "socket: %d, die: %d " @@ -1073,7 +1074,8 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, } } if (err_data.ue_count) { - if (adev->smuio.funcs && + if (!adev->aid_mask && + adev->smuio.funcs && adev->smuio.funcs->get_socket_id && adev->smuio.funcs->get_die_id) { dev_info(adev->dev, "socket: %d, die: %d " -- cgit From 4e8303cf2c4dd27374a16a8881ec1a1cd5baf86f Mon Sep 17 00:00:00 2001 From: Lijo Lazar Date: Mon, 11 Sep 2023 13:48:11 +0530 Subject: drm/amdgpu: Use function for IP version check Use an inline function for version check. Gives more flexibility to handle any format changes. Signed-off-by: Lijo Lazar Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/aldebaran.c | 6 +- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 6 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 29 ++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 92 ++++++++++---------- drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 28 ++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c | 3 +- drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 4 +- drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c | 4 +- drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 12 ++- drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c | 11 ++- drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c | 48 ++++++----- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 38 +++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c | 4 +- drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c | 4 +- drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c | 7 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 3 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c | 15 ++-- drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c | 4 +- drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c | 11 +-- drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c | 4 +- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 4 +- drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 3 +- drivers/gpu/drm/amd/amdgpu/athub_v1_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/athub_v2_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/athub_v2_1.c | 2 +- drivers/gpu/drm/amd/amdgpu/athub_v3_0.c | 6 +- drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 98 ++++++++++++---------- drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c | 27 +++--- drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 83 +++++++++--------- drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c | 8 +- drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c | 13 +-- drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c | 4 +- drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c | 28 ++++--- drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c | 8 +- drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 96 +++++++++++---------- drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c | 18 ++-- drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c | 6 +- drivers/gpu/drm/amd/amdgpu/imu_v11_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c | 4 +- drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/mes_v10_1.c | 10 +-- drivers/gpu/drm/amd/amdgpu/mes_v11_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c | 12 +-- drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c | 2 +- drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.c | 2 +- drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c | 2 +- drivers/gpu/drm/amd/amdgpu/navi10_ih.c | 4 +- drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c | 2 +- drivers/gpu/drm/amd/amdgpu/nbio_v4_3.c | 6 +- drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c | 10 +-- drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c | 4 +- drivers/gpu/drm/amd/amdgpu/nv.c | 8 +- drivers/gpu/drm/amd/amdgpu/psp_v10_0.c | 7 +- drivers/gpu/drm/amd/amdgpu/psp_v11_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/psp_v13_0.c | 6 +- drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c | 2 +- drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 48 ++++++----- drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c | 7 +- drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c | 4 +- drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c | 9 +- drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c | 3 +- drivers/gpu/drm/amd/amdgpu/sienna_cichlid.c | 2 +- drivers/gpu/drm/amd/amdgpu/soc15.c | 27 +++--- drivers/gpu/drm/amd/amdgpu/soc21.c | 12 +-- drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c | 9 +- drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c | 9 +- drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c | 14 ++-- drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c | 7 +- drivers/gpu/drm/amd/amdgpu/vega20_ih.c | 14 ++-- drivers/gpu/drm/amd/amdkfd/kfd_device.c | 12 +-- drivers/gpu/drm/amd/amdkfd/kfd_migrate.c | 2 +- drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c | 3 +- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 2 +- drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 2 +- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 34 ++++---- .../drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c | 10 +-- drivers/gpu/drm/amd/pm/amdgpu_pm.c | 12 +-- drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c | 24 +++--- drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c | 46 ++++++---- .../drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c | 63 ++++++++------ drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c | 32 +++---- drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c | 8 +- drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c | 18 ++-- .../gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c | 8 +- .../gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c | 2 +- .../gpu/drm/amd/pm/swsmu/smu13/yellow_carp_ppt.c | 18 ++-- 94 files changed, 714 insertions(+), 593 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran.c b/drivers/gpu/drm/amd/amdgpu/aldebaran.c index 5d2516210a3a..02f4c6f9d4f6 100644 --- a/drivers/gpu/drm/amd/amdgpu/aldebaran.c +++ b/drivers/gpu/drm/amd/amdgpu/aldebaran.c @@ -35,7 +35,7 @@ static bool aldebaran_is_mode2_default(struct amdgpu_reset_control *reset_ctl) { struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle; - if ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2) && + if ((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 2) && adev->gmc.xgmi.connected_to_cpu)) return true; @@ -154,7 +154,7 @@ aldebaran_mode2_perform_reset(struct amdgpu_reset_control *reset_ctl, if (reset_device_list == NULL) return -EINVAL; - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2) && + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 2) && reset_context->hive == NULL) { /* Wrong context, return error */ return -EINVAL; @@ -335,7 +335,7 @@ aldebaran_mode2_restore_hwcontext(struct amdgpu_reset_control *reset_ctl, if (reset_device_list == NULL) return -EINVAL; - if (reset_context->reset_req_dev->ip_versions[MP1_HWIP][0] == + if (amdgpu_ip_version(reset_context->reset_req_dev, MP1_HWIP, 0) == IP_VERSION(13, 0, 2) && reset_context->hive == NULL) { /* Wrong context, return error */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 62bbfdd502af..96ea0a43cdd2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1105,6 +1105,12 @@ struct amdgpu_device { bool debug_disable_soft_recovery; }; +static inline uint32_t amdgpu_ip_version(const struct amdgpu_device *adev, + uint8_t ip, uint8_t inst) +{ + return adev->ip_versions[ip][inst]; +} + static inline struct amdgpu_device *drm_to_adev(struct drm_device *ddev) { return container_of(ddev, struct amdgpu_device, ddev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index 26ff5f8d9795..38b5457baded 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -707,7 +707,7 @@ void amdgpu_amdkfd_set_compute_idle(struct amdgpu_device *adev, bool idle) /* Temporary workaround to fix issues observed in some * compute applications when GFXOFF is enabled on GFX11. */ - if (IP_VERSION_MAJ(adev->ip_versions[GC_HWIP][0]) == 11) { + if (IP_VERSION_MAJ(amdgpu_ip_version(adev, GC_HWIP, 0)) == 11) { pr_debug("GFXOFF is %s\n", idle ? "enabled" : "disabled"); amdgpu_gfx_off_ctrl(adev, idle); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c index d67d003bada2..b61a32d6af4b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c @@ -658,7 +658,7 @@ static int kgd_gfx_v11_validate_trap_override_request(struct amdgpu_device *adev KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH | KFD_DBG_TRAP_MASK_DBG_MEMORY_VIOLATION; - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(11, 0, 4)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 4)) *trap_mask_supported |= KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START | KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c index 04b8c7dacd30..51011e8ee90d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c @@ -677,7 +677,7 @@ void kgd_gfx_v9_set_wave_launch_stall(struct amdgpu_device *adev, int i; uint32_t data = RREG32(SOC15_REG_OFFSET(GC, 0, mmSPI_GDBG_WAVE_CNTL)); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1)) data = REG_SET_FIELD(data, SPI_GDBG_WAVE_CNTL, STALL_VMID, stall ? 1 << vmid : 0); else diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 3d540b0cf0e1..ca56b5a543b4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -1003,8 +1003,8 @@ static int amdgpu_device_asic_init(struct amdgpu_device *adev) amdgpu_asic_pre_asic_init(adev); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3) || - adev->ip_versions[GC_HWIP][0] >= IP_VERSION(11, 0, 0)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) || + amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0)) { amdgpu_psp_wait_for_bootloader(adev); ret = amdgpu_atomfirmware_asic_init(adev, true); return ret; @@ -2845,7 +2845,7 @@ static void amdgpu_device_smu_fini_early(struct amdgpu_device *adev) { int i, r; - if (adev->ip_versions[GC_HWIP][0] > IP_VERSION(9, 0, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) > IP_VERSION(9, 0, 0)) return; for (i = 0; i < adev->num_ip_blocks; i++) { @@ -3098,8 +3098,10 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) /* SDMA 5.x+ is part of GFX power domain so it's covered by GFXOFF */ if (adev->in_s0ix && - (adev->ip_versions[SDMA0_HWIP][0] >= IP_VERSION(5, 0, 0)) && - (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SDMA)) + (amdgpu_ip_version(adev, SDMA0_HWIP, 0) >= + IP_VERSION(5, 0, 0)) && + (adev->ip_blocks[i].version->type == + AMD_IP_BLOCK_TYPE_SDMA)) continue; /* Once swPSP provides the IMU, RLC FW binaries to TOS during cold-boot. @@ -3590,8 +3592,8 @@ static void amdgpu_device_set_mcbp(struct amdgpu_device *adev) adev->gfx.mcbp = true; else if (amdgpu_mcbp == 0) adev->gfx.mcbp = false; - else if ((adev->ip_versions[GC_HWIP][0] >= IP_VERSION(9, 0, 0)) && - (adev->ip_versions[GC_HWIP][0] < IP_VERSION(10, 0, 0)) && + else if ((amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(9, 0, 0)) && + (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(10, 0, 0)) && adev->gfx.num_gfx_rings) adev->gfx.mcbp = true; @@ -3811,7 +3813,8 @@ int amdgpu_device_init(struct amdgpu_device *adev, * internal path natively support atomics, set have_atomics_support to true. */ } else if ((adev->flags & AMD_IS_APU) && - (adev->ip_versions[GC_HWIP][0] > IP_VERSION(9, 0, 0))) { + (amdgpu_ip_version(adev, GC_HWIP, 0) > + IP_VERSION(9, 0, 0))) { adev->have_atomics_support = true; } else { adev->have_atomics_support = @@ -5444,8 +5447,9 @@ retry: /* Rest of adevs pre asic reset from XGMI hive. */ adev->asic_reset_res = r; /* Aldebaran and gfx_11_0_3 support ras in SRIOV, so need resume ras during reset */ - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(11, 0, 3)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(9, 4, 2) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3)) amdgpu_ras_resume(adev); } else { r = amdgpu_do_asic_reset(device_list_handle, reset_context); @@ -5470,7 +5474,8 @@ skip_hw_reset: drm_sched_start(&ring->sched, true); } - if (adev->enable_mes && adev->ip_versions[GC_HWIP][0] != IP_VERSION(11, 0, 3)) + if (adev->enable_mes && + amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(11, 0, 3)) amdgpu_mes_self_test(tmp_adev); if (!drm_drv_uses_atomic_modeset(adev_to_drm(tmp_adev)) && !job_signaled) @@ -6147,7 +6152,7 @@ bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev) return true; default: /* IP discovery */ - if (!adev->ip_versions[DCE_HWIP][0] || + if (!amdgpu_ip_version(adev, DCE_HWIP, 0) || (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK)) return false; return true; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index 961a642f5bde..40a059feb8c5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -311,8 +311,8 @@ static void amdgpu_discovery_harvest_config_quirk(struct amdgpu_device *adev) * So far, apply this quirk only on those Navy Flounder boards which * have a bad harvest table of VCN config. */ - if ((adev->ip_versions[UVD_HWIP][1] == IP_VERSION(3, 0, 1)) && - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 2))) { + if ((amdgpu_ip_version(adev, UVD_HWIP, 1) == IP_VERSION(3, 0, 1)) && + (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 3, 2))) { switch (adev->pdev->revision) { case 0xC1: case 0xC2: @@ -1363,8 +1363,8 @@ static void amdgpu_discovery_harvest_ip(struct amdgpu_device *adev) * so read harvest bit per IP data structure to set * harvest configuration. */ - if (adev->ip_versions[GC_HWIP][0] < IP_VERSION(10, 2, 0) && - adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 4, 3)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(10, 2, 0) && + amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 3)) { if ((adev->pdev->device == 0x731E && (adev->pdev->revision == 0xC6 || adev->pdev->revision == 0xC7)) || @@ -1607,7 +1607,7 @@ static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev) static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev) { /* what IP to use for this? */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 1, 0): case IP_VERSION(9, 2, 1): @@ -1645,7 +1645,7 @@ static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add common ip block(GC_HWIP:0x%x)\n", - adev->ip_versions[GC_HWIP][0]); + amdgpu_ip_version(adev, GC_HWIP, 0)); return -EINVAL; } return 0; @@ -1654,7 +1654,7 @@ static int amdgpu_discovery_set_common_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_gmc_ip_blocks(struct amdgpu_device *adev) { /* use GC or MMHUB IP version */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 1, 0): case IP_VERSION(9, 2, 1): @@ -1690,9 +1690,8 @@ static int amdgpu_discovery_set_gmc_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &gmc_v11_0_ip_block); break; default: - dev_err(adev->dev, - "Failed to add gmc ip block(GC_HWIP:0x%x)\n", - adev->ip_versions[GC_HWIP][0]); + dev_err(adev->dev, "Failed to add gmc ip block(GC_HWIP:0x%x)\n", + amdgpu_ip_version(adev, GC_HWIP, 0)); return -EINVAL; } return 0; @@ -1700,7 +1699,7 @@ static int amdgpu_discovery_set_gmc_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_ih_ip_blocks(struct amdgpu_device *adev) { - switch (adev->ip_versions[OSSSYS_HWIP][0]) { + switch (amdgpu_ip_version(adev, OSSSYS_HWIP, 0)) { case IP_VERSION(4, 0, 0): case IP_VERSION(4, 0, 1): case IP_VERSION(4, 1, 0): @@ -1733,7 +1732,7 @@ static int amdgpu_discovery_set_ih_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add ih ip block(OSSSYS_HWIP:0x%x)\n", - adev->ip_versions[OSSSYS_HWIP][0]); + amdgpu_ip_version(adev, OSSSYS_HWIP, 0)); return -EINVAL; } return 0; @@ -1741,7 +1740,7 @@ static int amdgpu_discovery_set_ih_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev) { - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(9, 0, 0): amdgpu_device_ip_block_add(adev, &psp_v3_1_ip_block); break; @@ -1787,7 +1786,7 @@ static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add psp ip block(MP0_HWIP:0x%x)\n", - adev->ip_versions[MP0_HWIP][0]); + amdgpu_ip_version(adev, MP0_HWIP, 0)); return -EINVAL; } return 0; @@ -1795,7 +1794,7 @@ static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev) { - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(9, 0, 0): case IP_VERSION(10, 0, 0): case IP_VERSION(10, 0, 1): @@ -1836,7 +1835,7 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add smu ip block(MP1_HWIP:0x%x)\n", - adev->ip_versions[MP1_HWIP][0]); + amdgpu_ip_version(adev, MP1_HWIP, 0)); return -EINVAL; } return 0; @@ -1861,8 +1860,8 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) return 0; #if defined(CONFIG_DRM_AMD_DC) - if (adev->ip_versions[DCE_HWIP][0]) { - switch (adev->ip_versions[DCE_HWIP][0]) { + if (amdgpu_ip_version(adev, DCE_HWIP, 0)) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(1, 0, 0): case IP_VERSION(1, 0, 1): case IP_VERSION(2, 0, 2): @@ -1888,11 +1887,11 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add dm ip block(DCE_HWIP:0x%x)\n", - adev->ip_versions[DCE_HWIP][0]); + amdgpu_ip_version(adev, DCE_HWIP, 0)); return -EINVAL; } - } else if (adev->ip_versions[DCI_HWIP][0]) { - switch (adev->ip_versions[DCI_HWIP][0]) { + } else if (amdgpu_ip_version(adev, DCI_HWIP, 0)) { + switch (amdgpu_ip_version(adev, DCI_HWIP, 0)) { case IP_VERSION(12, 0, 0): case IP_VERSION(12, 0, 1): case IP_VERSION(12, 1, 0): @@ -1904,7 +1903,7 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add dm ip block(DCI_HWIP:0x%x)\n", - adev->ip_versions[DCI_HWIP][0]); + amdgpu_ip_version(adev, DCI_HWIP, 0)); return -EINVAL; } } @@ -1914,7 +1913,7 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_gc_ip_blocks(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 1, 0): case IP_VERSION(9, 2, 1): @@ -1954,9 +1953,8 @@ static int amdgpu_discovery_set_gc_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &gfx_v11_0_ip_block); break; default: - dev_err(adev->dev, - "Failed to add gfx ip block(GC_HWIP:0x%x)\n", - adev->ip_versions[GC_HWIP][0]); + dev_err(adev->dev, "Failed to add gfx ip block(GC_HWIP:0x%x)\n", + amdgpu_ip_version(adev, GC_HWIP, 0)); return -EINVAL; } return 0; @@ -1964,7 +1962,7 @@ static int amdgpu_discovery_set_gc_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev) { - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(4, 0, 0): case IP_VERSION(4, 0, 1): case IP_VERSION(4, 1, 0): @@ -2004,7 +2002,7 @@ static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add sdma ip block(SDMA0_HWIP:0x%x)\n", - adev->ip_versions[SDMA0_HWIP][0]); + amdgpu_ip_version(adev, SDMA0_HWIP, 0)); return -EINVAL; } return 0; @@ -2012,8 +2010,8 @@ static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev) { - if (adev->ip_versions[VCE_HWIP][0]) { - switch (adev->ip_versions[UVD_HWIP][0]) { + if (amdgpu_ip_version(adev, VCE_HWIP, 0)) { + switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) { case IP_VERSION(7, 0, 0): case IP_VERSION(7, 2, 0): /* UVD is not supported on vega20 SR-IOV */ @@ -2023,10 +2021,10 @@ static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add uvd v7 ip block(UVD_HWIP:0x%x)\n", - adev->ip_versions[UVD_HWIP][0]); + amdgpu_ip_version(adev, UVD_HWIP, 0)); return -EINVAL; } - switch (adev->ip_versions[VCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, VCE_HWIP, 0)) { case IP_VERSION(4, 0, 0): case IP_VERSION(4, 1, 0): /* VCE is not supported on vega20 SR-IOV */ @@ -2036,11 +2034,11 @@ static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add VCE v4 ip block(VCE_HWIP:0x%x)\n", - adev->ip_versions[VCE_HWIP][0]); + amdgpu_ip_version(adev, VCE_HWIP, 0)); return -EINVAL; } } else { - switch (adev->ip_versions[UVD_HWIP][0]) { + switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) { case IP_VERSION(1, 0, 0): case IP_VERSION(1, 0, 1): amdgpu_device_ip_block_add(adev, &vcn_v1_0_ip_block); @@ -2091,7 +2089,7 @@ static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev) default: dev_err(adev->dev, "Failed to add vcn/jpeg ip block(UVD_HWIP:0x%x)\n", - adev->ip_versions[UVD_HWIP][0]); + amdgpu_ip_version(adev, UVD_HWIP, 0)); return -EINVAL; } } @@ -2100,7 +2098,7 @@ static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_mes_ip_blocks(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): @@ -2138,7 +2136,7 @@ static int amdgpu_discovery_set_mes_ip_blocks(struct amdgpu_device *adev) static void amdgpu_discovery_init_soc_config(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 3): aqua_vanjaram_init_soc_config(adev); break; @@ -2149,7 +2147,7 @@ static void amdgpu_discovery_init_soc_config(struct amdgpu_device *adev) static int amdgpu_discovery_set_vpe_ip_blocks(struct amdgpu_device *adev) { - switch (adev->ip_versions[VPE_HWIP][0]) { + switch (amdgpu_ip_version(adev, VPE_HWIP, 0)) { case IP_VERSION(6, 1, 0): amdgpu_device_ip_block_add(adev, &vpe_v6_1_ip_block); break; @@ -2162,7 +2160,7 @@ static int amdgpu_discovery_set_vpe_ip_blocks(struct amdgpu_device *adev) static int amdgpu_discovery_set_umsch_mm_ip_blocks(struct amdgpu_device *adev) { - switch (adev->ip_versions[VCN_HWIP][0]) { + switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) { case IP_VERSION(4, 0, 5): if (amdgpu_umsch_mm & 0x1) { amdgpu_device_ip_block_add(adev, &umsch_mm_v4_0_ip_block); @@ -2356,7 +2354,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) amdgpu_discovery_init_soc_config(adev); amdgpu_discovery_sysfs_init(adev); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 2, 1): case IP_VERSION(9, 4, 0): @@ -2410,7 +2408,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) return -EINVAL; } - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 1, 0): case IP_VERSION(9, 2, 2): case IP_VERSION(9, 3, 0): @@ -2429,11 +2427,11 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) break; } - if (adev->ip_versions[XGMI_HWIP][0] == IP_VERSION(4, 8, 0)) + if (amdgpu_ip_version(adev, XGMI_HWIP, 0) == IP_VERSION(4, 8, 0)) adev->gmc.xgmi.supported = true; /* set NBIO version */ - switch (adev->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(6, 1, 0): case IP_VERSION(6, 2, 0): adev->nbio.funcs = &nbio_v6_1_funcs; @@ -2495,7 +2493,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) break; } - switch (adev->ip_versions[HDP_HWIP][0]) { + switch (amdgpu_ip_version(adev, HDP_HWIP, 0)) { case IP_VERSION(4, 0, 0): case IP_VERSION(4, 0, 1): case IP_VERSION(4, 1, 0): @@ -2527,7 +2525,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) break; } - switch (adev->ip_versions[DF_HWIP][0]) { + switch (amdgpu_ip_version(adev, DF_HWIP, 0)) { case IP_VERSION(3, 6, 0): case IP_VERSION(3, 6, 1): case IP_VERSION(3, 6, 2): @@ -2547,7 +2545,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) break; } - switch (adev->ip_versions[SMUIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, SMUIO_HWIP, 0)) { case IP_VERSION(9, 0, 0): case IP_VERSION(9, 0, 1): case IP_VERSION(10, 0, 0): @@ -2590,7 +2588,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) break; } - switch (adev->ip_versions[LSDMA_HWIP][0]) { + switch (amdgpu_ip_version(adev, LSDMA_HWIP, 0)) { case IP_VERSION(6, 0, 0): case IP_VERSION(6, 0, 1): case IP_VERSION(6, 0, 2): diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index 363e6a2cad8c..0cacd0b9f8be 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -766,11 +766,13 @@ static int convert_tiling_flags_to_modifier(struct amdgpu_framebuffer *afb) return -EINVAL; } - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(11, 0, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0)) version = AMD_FMT_MOD_TILE_VER_GFX11; - else if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) + else if (amdgpu_ip_version(adev, GC_HWIP, 0) >= + IP_VERSION(10, 3, 0)) version = AMD_FMT_MOD_TILE_VER_GFX10_RBPLUS; - else if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 0, 0)) + else if (amdgpu_ip_version(adev, GC_HWIP, 0) >= + IP_VERSION(10, 0, 0)) version = AMD_FMT_MOD_TILE_VER_GFX10; else version = AMD_FMT_MOD_TILE_VER_GFX9; @@ -779,13 +781,15 @@ static int convert_tiling_flags_to_modifier(struct amdgpu_framebuffer *afb) case 0: /* Z microtiling */ return -EINVAL; case 1: /* S microtiling */ - if (adev->ip_versions[GC_HWIP][0] < IP_VERSION(11, 0, 0)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) < + IP_VERSION(11, 0, 0)) { if (!has_xor) version = AMD_FMT_MOD_TILE_VER_GFX9; } break; case 2: - if (adev->ip_versions[GC_HWIP][0] < IP_VERSION(11, 0, 0)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) < + IP_VERSION(11, 0, 0)) { if (!has_xor && afb->base.format->cpp[0] != 4) version = AMD_FMT_MOD_TILE_VER_GFX9; } @@ -838,10 +842,12 @@ static int convert_tiling_flags_to_modifier(struct amdgpu_framebuffer *afb) u64 render_dcc_offset; /* Enable constant encode on RAVEN2 and later. */ - bool dcc_constant_encode = (adev->asic_type > CHIP_RAVEN || - (adev->asic_type == CHIP_RAVEN && - adev->external_rev_id >= 0x81)) && - adev->ip_versions[GC_HWIP][0] < IP_VERSION(11, 0, 0); + bool dcc_constant_encode = + (adev->asic_type > CHIP_RAVEN || + (adev->asic_type == CHIP_RAVEN && + adev->external_rev_id >= 0x81)) && + amdgpu_ip_version(adev, GC_HWIP, 0) < + IP_VERSION(11, 0, 0); int max_cblock_size = dcc_i64b ? AMD_FMT_MOD_DCC_BLOCK_64B : dcc_i128b ? AMD_FMT_MOD_DCC_BLOCK_128B : @@ -878,7 +884,9 @@ static int convert_tiling_flags_to_modifier(struct amdgpu_framebuffer *afb) if (adev->family >= AMDGPU_FAMILY_NV) { int extra_pipe = 0; - if ((adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) && + if ((amdgpu_ip_version(adev, GC_HWIP, + 0) >= + IP_VERSION(10, 3, 0)) && pipes == packers && pipes > 1) extra_pipe = 1; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index f735e91eef5c..4605803fc92f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -2274,7 +2274,7 @@ amdgpu_pci_remove(struct pci_dev *pdev) pm_runtime_forbid(dev->dev); } - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2) && + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 2) && !amdgpu_sriov_vf(adev)) { bool need_to_reset_gpu = false; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index 7537f5aa76f0..dc230212746a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -570,7 +570,8 @@ static bool amdgpu_fence_need_ring_interrupt_restore(struct amdgpu_ring *ring) switch (ring->funcs->type) { case AMDGPU_RING_TYPE_SDMA: /* SDMA 5.x+ is part of GFX power domain so it's covered by GFXOFF */ - if (adev->ip_versions[SDMA0_HWIP][0] >= IP_VERSION(5, 0, 0)) + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) >= + IP_VERSION(5, 0, 0)) is_gfx_power_domain = true; break; case AMDGPU_RING_TYPE_GFX: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 2382921710ec..9a158018ae16 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -158,7 +158,7 @@ static bool amdgpu_gfx_is_compute_multipipe_capable(struct amdgpu_device *adev) return amdgpu_compute_multipipe == 1; } - if (adev->ip_versions[GC_HWIP][0] > IP_VERSION(9, 0, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) > IP_VERSION(9, 0, 0)) return true; /* FIXME: spreading the queues across pipes causes perf regressions @@ -385,7 +385,7 @@ int amdgpu_gfx_mqd_sw_init(struct amdgpu_device *adev, u32 domain = AMDGPU_GEM_DOMAIN_GTT; /* Only enable on gfx10 and 11 for now to avoid changing behavior on older chips */ - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 0, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 0, 0)) domain |= AMDGPU_GEM_DOMAIN_VRAM; /* create MQD for KIQ */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 5f7641d9f346..de7b379a9cc8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -588,7 +588,7 @@ int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev) */ void amdgpu_gmc_tmz_set(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { /* RAVEN */ case IP_VERSION(9, 2, 2): case IP_VERSION(9, 1, 0): @@ -652,7 +652,7 @@ void amdgpu_gmc_tmz_set(struct amdgpu_device *adev) void amdgpu_gmc_noretry_set(struct amdgpu_device *adev) { struct amdgpu_gmc *gmc = &adev->gmc; - uint32_t gc_ver = adev->ip_versions[GC_HWIP][0]; + uint32_t gc_ver = amdgpu_ip_version(adev, GC_HWIP, 0); bool noretry_default = (gc_ver == IP_VERSION(9, 0, 1) || gc_ver == IP_VERSION(9, 3, 0) || gc_ver == IP_VERSION(9, 4, 0) || diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index a74ea53899f6..dfd24a582391 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -501,18 +501,22 @@ static int amdgpu_hw_ip_info(struct amdgpu_device *adev, if (adev->asic_type >= CHIP_VEGA10) { switch (type) { case AMD_IP_BLOCK_TYPE_GFX: - result->ip_discovery_version = adev->ip_versions[GC_HWIP][0]; + result->ip_discovery_version = + amdgpu_ip_version(adev, GC_HWIP, 0); break; case AMD_IP_BLOCK_TYPE_SDMA: - result->ip_discovery_version = adev->ip_versions[SDMA0_HWIP][0]; + result->ip_discovery_version = + amdgpu_ip_version(adev, SDMA0_HWIP, 0); break; case AMD_IP_BLOCK_TYPE_UVD: case AMD_IP_BLOCK_TYPE_VCN: case AMD_IP_BLOCK_TYPE_JPEG: - result->ip_discovery_version = adev->ip_versions[UVD_HWIP][0]; + result->ip_discovery_version = + amdgpu_ip_version(adev, UVD_HWIP, 0); break; case AMD_IP_BLOCK_TYPE_VCE: - result->ip_discovery_version = adev->ip_versions[VCE_HWIP][0]; + result->ip_discovery_version = + amdgpu_ip_version(adev, VCE_HWIP, 0); break; default: result->ip_discovery_version = 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c index b6015157763a..59f10b353b3a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c @@ -132,7 +132,8 @@ int amdgpu_mes_init(struct amdgpu_device *adev) adev->mes.gfx_hqd_mask[i] = i ? 0 : 0xfffffffe; for (i = 0; i < AMDGPU_MES_MAX_SDMA_PIPES; i++) { - if (adev->ip_versions[SDMA0_HWIP][0] < IP_VERSION(6, 0, 0)) + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) < + IP_VERSION(6, 0, 0)) adev->mes.sdma_hqd_mask[i] = i ? 0 : 0x3fc; /* zero sdma_hqd_mask for non-existent engine */ else if (adev->sdma.num_instances == 1) @@ -1335,8 +1336,10 @@ int amdgpu_mes_self_test(struct amdgpu_device *adev) for (i = 0; i < ARRAY_SIZE(queue_types); i++) { /* On GFX v10.3, fw hasn't supported to map sdma queue. */ - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0) && - adev->ip_versions[GC_HWIP][0] < IP_VERSION(11, 0, 0) && + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= + IP_VERSION(10, 3, 0) && + amdgpu_ip_version(adev, GC_HWIP, 0) < + IP_VERSION(11, 0, 0) && queue_types[i][0] == AMDGPU_RING_TYPE_SDMA) continue; @@ -1397,7 +1400,7 @@ int amdgpu_mes_init_microcode(struct amdgpu_device *adev, int pipe) amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix)); - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(11, 0, 0)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0)) { snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mes%s.bin", ucode_prefix, pipe == AMDGPU_MES_SCHED_PIPE ? "_2" : "1"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index 60bc9d4f0761..a82913320c56 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -100,7 +100,7 @@ static void psp_check_pmfw_centralized_cstate_management(struct psp_context *psp return; } - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 4): case IP_VERSION(11, 0, 5): @@ -128,7 +128,7 @@ static int psp_init_sriov_microcode(struct psp_context *psp) amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix)); - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(9, 0, 0): case IP_VERSION(11, 0, 7): case IP_VERSION(11, 0, 9): @@ -162,7 +162,7 @@ static int psp_early_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; struct psp_context *psp = &adev->psp; - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(9, 0, 0): psp_v3_1_set_psp_funcs(psp); psp->autoload_supported = false; @@ -334,7 +334,7 @@ static bool psp_get_runtime_db_entry(struct amdgpu_device *adev, bool ret = false; int i; - if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 6)) + if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6)) return false; db_header_pos = adev->gmc.mc_vram_size - PSP_RUNTIME_DB_OFFSET; @@ -413,7 +413,7 @@ static int psp_sw_init(void *handle) adev->psp.xgmi_context.supports_extended_data = !adev->gmc.xgmi.connected_to_cpu && - adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 2); + amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 2); memset(&scpm_entry, 0, sizeof(scpm_entry)); if ((psp_get_runtime_db_entry(adev, @@ -773,7 +773,7 @@ static int psp_load_toc(struct psp_context *psp, static bool psp_boottime_tmr(struct psp_context *psp) { - switch (psp->adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(psp->adev, MP0_HWIP, 0)) { case IP_VERSION(13, 0, 6): return true; default: @@ -828,7 +828,7 @@ static int psp_tmr_init(struct psp_context *psp) static bool psp_skip_tmr(struct psp_context *psp) { - switch (psp->adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(psp->adev, MP0_HWIP, 0)) { case IP_VERSION(11, 0, 9): case IP_VERSION(11, 0, 7): case IP_VERSION(13, 0, 2): @@ -1215,8 +1215,8 @@ int psp_xgmi_terminate(struct psp_context *psp) struct amdgpu_device *adev = psp->adev; /* XGMI TA unload currently is not supported on Arcturus/Aldebaran A+A */ - if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(11, 0, 4) || - (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 2) && + if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(11, 0, 4) || + (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 2) && adev->gmc.xgmi.connected_to_cpu)) return 0; @@ -1313,9 +1313,11 @@ int psp_xgmi_get_node_id(struct psp_context *psp, uint64_t *node_id) static bool psp_xgmi_peer_link_info_supported(struct psp_context *psp) { - return (psp->adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 2) && + return (amdgpu_ip_version(psp->adev, MP0_HWIP, 0) == + IP_VERSION(13, 0, 2) && psp->xgmi_context.context.bin_desc.fw_version >= 0x2000000b) || - psp->adev->ip_versions[MP0_HWIP][0] >= IP_VERSION(13, 0, 6); + amdgpu_ip_version(psp->adev, MP0_HWIP, 0) >= + IP_VERSION(13, 0, 6); } /* @@ -1424,8 +1426,10 @@ int psp_xgmi_get_topology_info(struct psp_context *psp, if (psp_xgmi_peer_link_info_supported(psp)) { struct ta_xgmi_cmd_get_peer_link_info_output *link_info_output; bool requires_reflection = - (psp->xgmi_context.supports_extended_data && get_extended_data) || - psp->adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 6); + (psp->xgmi_context.supports_extended_data && + get_extended_data) || + amdgpu_ip_version(psp->adev, MP0_HWIP, 0) == + IP_VERSION(13, 0, 6); xgmi_cmd->cmd_id = TA_COMMAND_XGMI__GET_PEER_LINKS; @@ -2517,10 +2521,9 @@ static int psp_load_smu_fw(struct psp_context *psp) if (!ucode->fw || amdgpu_sriov_vf(psp->adev)) return 0; - if ((amdgpu_in_reset(adev) && - ras && adev->ras_enabled && - (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(11, 0, 4) || - adev->ip_versions[MP0_HWIP][0] == IP_VERSION(11, 0, 2)))) { + if ((amdgpu_in_reset(adev) && ras && adev->ras_enabled && + (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(11, 0, 4) || + amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(11, 0, 2)))) { ret = amdgpu_dpm_set_mp1_state(adev, PP_MP1_STATE_UNLOAD); if (ret) DRM_WARN("Failed to set MP1 state prepare for reload\n"); @@ -2603,9 +2606,12 @@ static int psp_load_non_psp_fw(struct psp_context *psp) continue; if (psp->autoload_supported && - (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(11, 0, 7) || - adev->ip_versions[MP0_HWIP][0] == IP_VERSION(11, 0, 11) || - adev->ip_versions[MP0_HWIP][0] == IP_VERSION(11, 0, 12)) && + (amdgpu_ip_version(adev, MP0_HWIP, 0) == + IP_VERSION(11, 0, 7) || + amdgpu_ip_version(adev, MP0_HWIP, 0) == + IP_VERSION(11, 0, 11) || + amdgpu_ip_version(adev, MP0_HWIP, 0) == + IP_VERSION(11, 0, 12)) && (ucode->ucode_id == AMDGPU_UCODE_ID_SDMA1 || ucode->ucode_id == AMDGPU_UCODE_ID_SDMA2 || ucode->ucode_id == AMDGPU_UCODE_ID_SDMA3)) @@ -3146,7 +3152,7 @@ static int psp_init_sos_base_fw(struct amdgpu_device *adev) le32_to_cpu(sos_hdr->header.ucode_array_offset_bytes); if (adev->gmc.xgmi.connected_to_cpu || - (adev->ip_versions[MP0_HWIP][0] != IP_VERSION(13, 0, 2))) { + (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 2))) { adev->psp.sos.fw_version = le32_to_cpu(sos_hdr->header.ucode_version); adev->psp.sos.feature_version = le32_to_cpu(sos_hdr->sos.fw_version); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 632478874f7d..eeb695f9ff1c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -201,8 +201,8 @@ static ssize_t amdgpu_ras_debugfs_read(struct file *f, char __user *buf, return -EINVAL; /* Hardware counter will be reset automatically after the query on Vega20 and Arcturus */ - if (obj->adev->ip_versions[MP0_HWIP][0] != IP_VERSION(11, 0, 2) && - obj->adev->ip_versions[MP0_HWIP][0] != IP_VERSION(11, 0, 4)) { + if (amdgpu_ip_version(obj->adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 2) && + amdgpu_ip_version(obj->adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 4)) { if (amdgpu_ras_reset_error_status(obj->adev, info.head.block)) dev_warn(obj->adev->dev, "Failed to reset error counter and error status"); } @@ -611,8 +611,8 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev, if (amdgpu_ras_query_error_status(obj->adev, &info)) return -EINVAL; - if (obj->adev->ip_versions[MP0_HWIP][0] != IP_VERSION(11, 0, 2) && - obj->adev->ip_versions[MP0_HWIP][0] != IP_VERSION(11, 0, 4)) { + if (amdgpu_ip_version(obj->adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 2) && + amdgpu_ip_version(obj->adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 4)) { if (amdgpu_ras_reset_error_status(obj->adev, info.head.block)) dev_warn(obj->adev->dev, "Failed to reset error counter and error status"); } @@ -1208,8 +1208,8 @@ static int amdgpu_ras_query_error_count_helper(struct amdgpu_device *adev, /* some hardware/IP supports read to clear * no need to explictly reset the err status after the query call */ - if (adev->ip_versions[MP0_HWIP][0] != IP_VERSION(11, 0, 2) && - adev->ip_versions[MP0_HWIP][0] != IP_VERSION(11, 0, 4)) { + if (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 2) && + amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(11, 0, 4)) { if (amdgpu_ras_reset_error_status(adev, query_info->head.block)) dev_warn(adev->dev, "Failed to reset error counter and error status\n"); @@ -1905,14 +1905,18 @@ static void amdgpu_ras_log_on_err_counter(struct amdgpu_device *adev) * should be removed until smu fix handle ecc_info table. */ if ((info.head.block == AMDGPU_RAS_BLOCK__UMC) && - (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2))) + (amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(13, 0, 2))) continue; amdgpu_ras_query_error_status(adev, &info); - if (adev->ip_versions[MP0_HWIP][0] != IP_VERSION(11, 0, 2) && - adev->ip_versions[MP0_HWIP][0] != IP_VERSION(11, 0, 4) && - adev->ip_versions[MP0_HWIP][0] != IP_VERSION(13, 0, 0)) { + if (amdgpu_ip_version(adev, MP0_HWIP, 0) != + IP_VERSION(11, 0, 2) && + amdgpu_ip_version(adev, MP0_HWIP, 0) != + IP_VERSION(11, 0, 4) && + amdgpu_ip_version(adev, MP0_HWIP, 0) != + IP_VERSION(13, 0, 0)) { if (amdgpu_ras_reset_error_status(adev, info.head.block)) dev_warn(adev->dev, "Failed to reset error counter and error status"); } @@ -2400,7 +2404,7 @@ static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev) static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev) { if (amdgpu_sriov_vf(adev)) { - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(13, 0, 2): case IP_VERSION(13, 0, 6): return true; @@ -2410,7 +2414,7 @@ static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev) } if (adev->asic_type == CHIP_IP_DISCOVERY) { - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(13, 0, 0): case IP_VERSION(13, 0, 6): case IP_VERSION(13, 0, 10): @@ -2484,8 +2488,10 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) /* VCN/JPEG RAS can be supported on both bare metal and * SRIOV environment */ - if (adev->ip_versions[VCN_HWIP][0] == IP_VERSION(2, 6, 0) || - adev->ip_versions[VCN_HWIP][0] == IP_VERSION(4, 0, 0)) + if (amdgpu_ip_version(adev, VCN_HWIP, 0) == + IP_VERSION(2, 6, 0) || + amdgpu_ip_version(adev, VCN_HWIP, 0) == + IP_VERSION(4, 0, 0)) adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN | 1 << AMDGPU_RAS_BLOCK__JPEG); else @@ -2519,7 +2525,7 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) * Disable ras feature for aqua vanjaram * by default on apu platform. */ - if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 6) && + if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) && adev->gmc.is_app_apu) adev->ras_enabled = amdgpu_ras_enable != 1 ? 0 : adev->ras_hw_enabled & amdgpu_ras_mask; @@ -2634,7 +2640,7 @@ int amdgpu_ras_init(struct amdgpu_device *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->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(7, 4, 0): case IP_VERSION(7, 4, 1): case IP_VERSION(7, 4, 4): diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index 595d5e535aca..8ced4be784e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -153,7 +153,7 @@ static bool __is_ras_eeprom_supported(struct amdgpu_device *adev) { - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 2): /* VEGA20 and ARCTURUS */ case IP_VERSION(11, 0, 7): /* Sienna cichlid */ case IP_VERSION(13, 0, 0): @@ -191,7 +191,7 @@ static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev, return true; } - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 2): /* VEGA20 and ARCTURUS */ if (adev->asic_type == CHIP_VEGA20) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c index 02d874799c16..970bfece775c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.c @@ -30,7 +30,7 @@ int amdgpu_reset_init(struct amdgpu_device *adev) { int ret = 0; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 2): case IP_VERSION(13, 0, 6): ret = aldebaran_reset_init(adev); @@ -52,7 +52,7 @@ int amdgpu_reset_fini(struct amdgpu_device *adev) { int ret = 0; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 2): case IP_VERSION(13, 0, 6): ret = aldebaran_reset_fini(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c index 572f861e3f70..e8cbc4142d80 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c @@ -251,8 +251,11 @@ int amdgpu_sdma_init_microcode(struct amdgpu_device *adev, else { /* Use a single copy per SDMA firmware type. PSP uses the same instance for all * groups of SDMAs */ - if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 4, 2) && - adev->firmware.load_type == AMDGPU_FW_LOAD_PSP && + if (amdgpu_ip_version(adev, SDMA0_HWIP, + 0) == + IP_VERSION(4, 4, 2) && + adev->firmware.load_type == + AMDGPU_FW_LOAD_PSP && adev->sdma.num_inst_per_aid == i) { break; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 4e51dce3aab5..05991c5c8ddb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -1727,7 +1727,8 @@ static int amdgpu_ttm_reserve_tmr(struct amdgpu_device *adev) reserve_size = amdgpu_atomfirmware_get_fw_reserved_fb_size(adev); - if (!adev->bios && adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) + if (!adev->bios && + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) reserve_size = max(reserve_size, (uint32_t)280 << 20); else if (!reserve_size) reserve_size = DISCOVERY_TMR_OFFSET; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c index 5aff383473f2..771ef8017a98 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c @@ -1119,7 +1119,7 @@ int amdgpu_ucode_init_bo(struct amdgpu_device *adev) static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int block_type) { if (block_type == MP0_HWIP) { - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(9, 0, 0): switch (adev->asic_type) { case CHIP_VEGA10: @@ -1170,7 +1170,7 @@ static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int bl return "yellow_carp"; } } else if (block_type == MP1_HWIP) { - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(9, 0, 0): case IP_VERSION(10, 0, 0): case IP_VERSION(10, 0, 1): @@ -1196,7 +1196,7 @@ static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int bl return "aldebaran_smc"; } } else if (block_type == SDMA0_HWIP) { - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(4, 0, 0): return "vega10_sdma"; case IP_VERSION(4, 0, 1): @@ -1240,7 +1240,7 @@ static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int bl return "vangogh_sdma"; } } else if (block_type == UVD_HWIP) { - switch (adev->ip_versions[UVD_HWIP][0]) { + switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) { case IP_VERSION(1, 0, 0): case IP_VERSION(1, 0, 1): if (adev->apu_flags & AMD_APU_IS_RAVEN2) @@ -1265,7 +1265,8 @@ static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int bl case IP_VERSION(3, 0, 0): case IP_VERSION(3, 0, 64): case IP_VERSION(3, 0, 192): - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 3, 0)) return "sienna_cichlid_vcn"; return "navy_flounder_vcn"; case IP_VERSION(3, 0, 2): @@ -1278,7 +1279,7 @@ static const char *amdgpu_ucode_legacy_naming(struct amdgpu_device *adev, int bl return "yellow_carp_vcn"; } } else if (block_type == GC_HWIP) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): return "vega10"; case IP_VERSION(9, 2, 1): @@ -1331,7 +1332,7 @@ void amdgpu_ucode_ip_version_decode(struct amdgpu_device *adev, int block_type, int maj, min, rev; char *ip_name; const char *legacy; - uint32_t version = adev->ip_versions[block_type][0]; + uint32_t version = amdgpu_ip_version(adev, block_type, 0); legacy = amdgpu_ucode_legacy_naming(adev, block_type); if (legacy) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index db0d94ca4ffc..24fcc9a2e422 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -28,7 +28,7 @@ static int amdgpu_umc_convert_error_address(struct amdgpu_device *adev, struct ras_err_data *err_data, uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst) { - switch (adev->ip_versions[UMC_HWIP][0]) { + switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) { case IP_VERSION(6, 7, 0): umc_v6_7_convert_error_address(adev, err_data, err_addr, ch_inst, umc_inst); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c index 562ed4e937ba..4825c3353db4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c @@ -580,7 +580,7 @@ int amdgpu_umsch_mm_init_microcode(struct amdgpu_umsch_mm *umsch) const char *fw_name = NULL; int r; - switch (adev->ip_versions[VCN_HWIP][0]) { + switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) { case IP_VERSION(4, 0, 5): fw_name = "amdgpu/umsch_mm_4_0_0.bin"; break; @@ -757,7 +757,7 @@ static int umsch_mm_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - switch (adev->ip_versions[VCN_HWIP][0]) { + switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) { case IP_VERSION(4, 0, 5): umsch_mm_v4_0_set_funcs(&adev->umsch_mm); break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index 80bcbe744e58..c93f3a4c0e31 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -126,7 +126,7 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) * Hence, check for these versions here - notice this is * restricted to Vangogh (Deck's APU). */ - if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(3, 0, 2)) { + if (amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(3, 0, 2)) { const char *bios_ver = dmi_get_system_info(DMI_BIOS_VERSION); if (bios_ver && (!strncmp("F7A0113", bios_ver, 7) || @@ -171,7 +171,7 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev) if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8); - if (adev->ip_versions[UVD_HWIP][0] >= IP_VERSION(4, 0, 0)) { + if (amdgpu_ip_version(adev, UVD_HWIP, 0) >= IP_VERSION(4, 0, 0)) { fw_shared_size = AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)); log_offset = offsetof(struct amdgpu_vcn4_fw_shared, fw_log); } else { @@ -267,7 +267,7 @@ static bool amdgpu_vcn_using_unified_queue(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; bool ret = false; - if (adev->ip_versions[UVD_HWIP][0] >= IP_VERSION(4, 0, 0)) + if (amdgpu_ip_version(adev, UVD_HWIP, 0) >= IP_VERSION(4, 0, 0)) ret = true; return ret; @@ -998,7 +998,7 @@ int amdgpu_vcn_unified_ring_test_ib(struct amdgpu_ring *ring, long timeout) struct amdgpu_device *adev = ring->adev; long r; - if (adev->ip_versions[UVD_HWIP][0] != IP_VERSION(4, 0, 3)) { + if (amdgpu_ip_version(adev, UVD_HWIP, 0) != IP_VERSION(4, 0, 3)) { r = amdgpu_vcn_enc_ring_test_ib(ring, timeout); if (r) goto error; @@ -1048,7 +1048,8 @@ void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev) adev->firmware.fw_size += ALIGN(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE); - if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(4, 0, 3)) + if (amdgpu_ip_version(adev, UVD_HWIP, 0) == + IP_VERSION(4, 0, 3)) break; } dev_info(adev->dev, "Will use PSP to load VCN firmware\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 96857ae7fb5b..a0aa624f5a92 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -837,7 +837,7 @@ enum amdgpu_sriov_vf_mode amdgpu_virt_get_sriov_vf_mode(struct amdgpu_device *ad void amdgpu_virt_post_reset(struct amdgpu_device *adev) { - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(11, 0, 3)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3)) { /* force set to GFXOFF state after reset, * to avoid some invalid operation before GC enable */ @@ -847,7 +847,7 @@ void amdgpu_virt_post_reset(struct amdgpu_device *adev) bool amdgpu_virt_fw_load_skip_check(struct amdgpu_device *adev, uint32_t ucode_id) { - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(13, 0, 0): /* no vf autoload, white list */ if (ucode_id == AMDGPU_UCODE_ID_VCN1 || diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 231f5d9e5791..8ce91f69bbeb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -885,12 +885,12 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm, * heavy-weight flush TLB unconditionally. */ flush_tlb |= adev->gmc.xgmi.num_physical_nodes && - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 0); + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 0); /* * On GFX8 and older any 8 PTE block with a valid bit set enters the TLB */ - flush_tlb |= adev->ip_versions[GC_HWIP][0] < IP_VERSION(9, 0, 0); + flush_tlb |= amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(9, 0, 0); memset(¶ms, 0, sizeof(params)); params.adev = adev; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c index ae070072705a..a68354d03a49 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c @@ -123,7 +123,7 @@ static int vpe_early_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; struct amdgpu_vpe *vpe = &adev->vpe; - switch (adev->ip_versions[VPE_HWIP][0]) { + switch (amdgpu_ip_version(adev, VPE_HWIP, 0)) { case IP_VERSION(6, 1, 0): vpe_v6_1_set_funcs(vpe); break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 7e91b24784e5..061534e845a7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -948,7 +948,8 @@ static int amdgpu_xgmi_query_pcs_error_status(struct amdgpu_device *adev, uint32_t field_array_size = 0; if (is_xgmi_pcs) { - if (adev->ip_versions[XGMI_HWIP][0] == IP_VERSION(6, 1, 0)) { + if (amdgpu_ip_version(adev, XGMI_HWIP, 0) == + IP_VERSION(6, 1, 0)) { pcs_ras_fields = &xgmi3x16_pcs_ras_fields[0]; field_array_size = ARRAY_SIZE(xgmi3x16_pcs_ras_fields); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c index a13c443ea10f..42f4e163e251 100644 --- a/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/athub_v1_0.c @@ -68,7 +68,7 @@ int athub_v1_0_set_clockgating(struct amdgpu_device *adev, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[ATHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) { case IP_VERSION(9, 0, 0): case IP_VERSION(9, 1, 0): case IP_VERSION(9, 2, 0): diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c index a9521c98e7f7..5a122f50a6e7 100644 --- a/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/athub_v2_0.c @@ -77,7 +77,7 @@ int athub_v2_0_set_clockgating(struct amdgpu_device *adev, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[ATHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) { case IP_VERSION(1, 3, 1): case IP_VERSION(2, 0, 0): case IP_VERSION(2, 0, 2): diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c index 78508ae6a670..e143fcc46148 100644 --- a/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c +++ b/drivers/gpu/drm/amd/amdgpu/athub_v2_1.c @@ -70,7 +70,7 @@ int athub_v2_1_set_clockgating(struct amdgpu_device *adev, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[ATHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) { case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v3_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v3_0.c index f0e235f98afb..5a318bc03d23 100644 --- a/drivers/gpu/drm/amd/amdgpu/athub_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/athub_v3_0.c @@ -36,7 +36,7 @@ static uint32_t athub_v3_0_get_cg_cntl(struct amdgpu_device *adev) { uint32_t data; - switch (adev->ip_versions[ATHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) { case IP_VERSION(3, 0, 1): data = RREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL_V3_0_1); break; @@ -49,7 +49,7 @@ static uint32_t athub_v3_0_get_cg_cntl(struct amdgpu_device *adev) static void athub_v3_0_set_cg_cntl(struct amdgpu_device *adev, uint32_t data) { - switch (adev->ip_versions[ATHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) { case IP_VERSION(3, 0, 1): WREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL_V3_0_1, data); break; @@ -99,7 +99,7 @@ int athub_v3_0_set_clockgating(struct amdgpu_device *adev, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[ATHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) { case IP_VERSION(3, 0, 0): case IP_VERSION(3, 0, 1): case IP_VERSION(3, 0, 2): diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 62329a822022..35357364b5b3 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -3627,7 +3627,7 @@ static void gfx_v10_0_set_kiq_pm4_funcs(struct amdgpu_device *adev) static void gfx_v10_0_init_spm_golden_registers(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): soc15_program_register_sequence(adev, golden_settings_gc_rlc_spm_10_0_nv10, @@ -3650,7 +3650,7 @@ static void gfx_v10_0_init_spm_golden_registers(struct amdgpu_device *adev) static void gfx_v10_0_init_golden_registers(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): soc15_program_register_sequence(adev, golden_settings_gc_10_1, @@ -3891,7 +3891,7 @@ static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev) { adev->gfx.cp_fw_write_wait = false; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 1): @@ -3942,7 +3942,7 @@ static bool gfx_v10_0_navi10_gfxoff_should_enable(struct amdgpu_device *adev) static void gfx_v10_0_check_gfxoff_flag(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): if (!gfx_v10_0_navi10_gfxoff_should_enable(adev)) adev->pm.pp_feature &= ~PP_GFXOFF_MASK; @@ -3964,8 +3964,8 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev) DRM_DEBUG("\n"); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 1) && - (!(adev->pdev->device == 0x7340 && adev->pdev->revision != 0x00))) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 1) && + (!(adev->pdev->device == 0x7340 && adev->pdev->revision != 0x00))) wks = "_wks"; amdgpu_ucode_ip_version_decode(adev, GC_HWIP, ucode_prefix, sizeof(ucode_prefix)); @@ -4144,7 +4144,7 @@ static void gfx_v10_0_init_rlcg_reg_access_ctrl(struct amdgpu_device *adev) reg_access_ctrl->scratch_reg3 = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG3); reg_access_ctrl->grbm_cntl = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_CNTL); reg_access_ctrl->grbm_idx = SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): reg_access_ctrl->spare_int = SOC15_REG_OFFSET(GC, 0, mmRLC_SPARE_INT_0_Sienna_Cichlid); @@ -4358,7 +4358,7 @@ static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev) { u32 gb_addr_config; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): @@ -4491,7 +4491,7 @@ static int gfx_v10_0_sw_init(void *handle) struct amdgpu_kiq *kiq; struct amdgpu_device *adev = (struct amdgpu_device *)handle; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): @@ -4749,9 +4749,12 @@ static void gfx_v10_0_setup_rb(struct amdgpu_device *adev) for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { bitmap = i * adev->gfx.config.max_sh_per_se + j; - if (((adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0)) || - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 3)) || - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 6))) && + if (((amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 3, 0)) || + (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 3, 3)) || + (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 3, 6))) && ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1)) continue; gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff, 0); @@ -4779,7 +4782,7 @@ static u32 gfx_v10_0_init_pa_sc_tile_steering_override(struct amdgpu_device *ade /* for ASICs that integrates GFX v10.3 * pa_sc_tile_steering_override should be set to 0 */ - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0)) return 0; /* init num_sc */ @@ -4960,7 +4963,7 @@ static void gfx_v10_0_get_tcc_info(struct amdgpu_device *adev) /* TCCs are global (not instanced). */ uint32_t tcc_disable; - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0)) { tcc_disable = RREG32_SOC15(GC, 0, mmCGTS_TCC_DISABLE_gc_10_3) | RREG32_SOC15(GC, 0, mmCGTS_USER_TCC_DISABLE_gc_10_3); } else { @@ -5037,7 +5040,7 @@ static int gfx_v10_0_init_csb(struct amdgpu_device *adev) adev->gfx.rlc.funcs->get_csb_buffer(adev, adev->gfx.rlc.cs_ptr); /* csib */ - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 2)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 2)) { WREG32_SOC15_RLC(GC, 0, mmRLC_CSIB_ADDR_HI, adev->gfx.rlc.clear_state_gpu_addr >> 32); WREG32_SOC15_RLC(GC, 0, mmRLC_CSIB_ADDR_LO, @@ -5666,7 +5669,7 @@ static int gfx_v10_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable) tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, enable ? 0 : 1); tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, enable ? 0 : 1); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 2)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 2)) WREG32_SOC15_RLC(GC, 0, mmCP_ME_CNTL, tmp); else WREG32_SOC15(GC, 0, mmCP_ME_CNTL, tmp); @@ -6057,7 +6060,7 @@ static void gfx_v10_0_cp_gfx_set_doorbell(struct amdgpu_device *adev, } WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_CONTROL, tmp); } - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): @@ -6190,7 +6193,7 @@ static int gfx_v10_0_cp_gfx_resume(struct amdgpu_device *adev) static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) { if (enable) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): @@ -6206,7 +6209,7 @@ static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable) break; } } else { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): @@ -6306,7 +6309,7 @@ static void gfx_v10_0_kiq_setting(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; /* tell RLC which is KIQ queue */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): @@ -6917,7 +6920,7 @@ static bool gfx_v10_0_check_grbm_cam_remapping(struct amdgpu_device *adev) * check if mmVGT_ESGS_RING_SIZE_UMD * has been remapped to mmVGT_ESGS_RING_SIZE */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 4): @@ -6966,7 +6969,7 @@ static void gfx_v10_0_setup_grbm_cam_remapping(struct amdgpu_device *adev) */ WREG32_SOC15(GC, 0, mmGRBM_CAM_INDEX, 0); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): @@ -7139,19 +7142,19 @@ static int gfx_v10_0_hw_init(void *handle) * init golden registers and rlc resume may override some registers, * reconfig them here */ - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 10) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 1) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 2)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 10) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 1) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 2)) gfx_v10_0_tcp_harvest(adev); r = gfx_v10_0_cp_resume(adev); if (r) return r; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 3, 0)) gfx_v10_3_program_pbb_mode(adev); - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0)) gfx_v10_3_set_power_brake_sequence(adev); return r; @@ -7255,7 +7258,7 @@ static int gfx_v10_0_soft_reset(void *handle) /* GRBM_STATUS2 */ tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS2); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): @@ -7312,7 +7315,7 @@ static uint64_t gfx_v10_0_get_gpu_clock_counter(struct amdgpu_device *adev) { uint64_t clock, clock_lo, clock_hi, hi_check; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 3): case IP_VERSION(10, 3, 7): @@ -7399,7 +7402,7 @@ static int gfx_v10_0_early_init(void *handle) adev->gfx.funcs = &gfx_v10_0_gfx_funcs; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): @@ -7470,7 +7473,7 @@ static void gfx_v10_0_set_safe_mode(struct amdgpu_device *adev, int xcc_id) data = RLC_SAFE_MODE__CMD_MASK; data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): @@ -7508,7 +7511,7 @@ static void gfx_v10_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id) uint32_t data; data = RLC_SAFE_MODE__CMD_MASK; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): @@ -7819,7 +7822,7 @@ static void gfx_v10_0_apply_medium_grain_clock_gating_workaround(struct amdgpu_d mmCGTS_SA1_QUAD1_SM_CTRL_REG }; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 2)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 2)) { for (i = 0; i < ARRAY_SIZE(tcp_ctrl_regs_nv12); i++) { reg_idx = adev->reg_offset[GC_HWIP][0][mmCGTS_SA0_WGP00_CU0_TCP_CTRL_REG_BASE_IDX] + tcp_ctrl_regs_nv12[i]; @@ -7864,9 +7867,12 @@ static int gfx_v10_0_update_gfx_clock_gating(struct amdgpu_device *adev, /* === CGCG + CGLS === */ gfx_v10_0_update_coarse_grain_clock_gating(adev, enable); - if ((adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 10)) || - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 1)) || - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 2))) + if ((amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 1, 10)) || + (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 1, 1)) || + (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 1, 2))) gfx_v10_0_apply_medium_grain_clock_gating_workaround(adev); } else { /* CGCG/CGLS should be disabled before MGCG/MGLS @@ -7966,7 +7972,7 @@ static void gfx_v10_cntl_power_gating(struct amdgpu_device *adev, bool enable) * Power/performance team will optimize it and might give a new value later. */ if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 3): case IP_VERSION(10, 3, 6): @@ -8027,7 +8033,7 @@ static int gfx_v10_0_set_powergating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): @@ -8064,7 +8070,7 @@ static int gfx_v10_0_set_clockgating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): @@ -9311,7 +9317,7 @@ static void gfx_v10_0_set_irq_funcs(struct amdgpu_device *adev) static void gfx_v10_0_set_rlc_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 3): @@ -9428,10 +9434,14 @@ static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev, for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { bitmap = i * adev->gfx.config.max_sh_per_se + j; - if (((adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0)) || - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 3)) || - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 6)) || - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 7))) && + if (((amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 3, 0)) || + (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 3, 3)) || + (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 3, 6)) || + (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(10, 3, 7))) && ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1)) continue; mask = 1; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index 39c434ca0dad..42fc0cc13fdd 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c @@ -288,7 +288,7 @@ static void gfx_v11_0_set_kiq_pm4_funcs(struct amdgpu_device *adev) static void gfx_v11_0_init_golden_registers(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 1): case IP_VERSION(11, 0, 4): soc15_program_register_sequence(adev, @@ -493,7 +493,7 @@ out: static void gfx_v11_0_check_fw_cp_gfx_shadow(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 2): case IP_VERSION(11, 0, 3): @@ -884,8 +884,7 @@ static const struct amdgpu_gfx_funcs gfx_v11_0_gfx_funcs = { static int gfx_v11_0_gpu_early_init(struct amdgpu_device *adev) { - - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 2): adev->gfx.config.max_hw_contexts = 8; @@ -1332,7 +1331,7 @@ static int gfx_v11_0_sw_init(void *handle) adev->gfxhub.funcs->init(adev); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 2): case IP_VERSION(11, 0, 3): @@ -1364,8 +1363,8 @@ static int gfx_v11_0_sw_init(void *handle) } /* Enable CG flag in one VF mode for enabling RLC safe mode enter/exit */ - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(11, 0, 3) && - amdgpu_sriov_is_pp_one_vf(adev)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3) && + amdgpu_sriov_is_pp_one_vf(adev)) adev->cg_flags = AMD_CG_SUPPORT_GFX_CGCG; /* EOP Event */ @@ -2592,9 +2591,11 @@ static int gfx_v11_0_wait_for_rlc_autoload_complete(struct amdgpu_device *adev) for (i = 0; i < adev->usec_timeout; i++) { cp_status = RREG32_SOC15(GC, 0, regCP_STAT); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(11, 0, 1) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(11, 0, 4) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(11, 5, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(11, 0, 1) || + amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(11, 0, 4) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 5, 0)) bootload_status = RREG32_SOC15(GC, 0, regRLC_RLCS_BOOTLOAD_STATUS_gc_11_0_1); else @@ -5025,7 +5026,7 @@ static void gfx_v11_cntl_power_gating(struct amdgpu_device *adev, bool enable) // Program RLC_PG_DELAY3 for CGPG hysteresis if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 1): case IP_VERSION(11, 0, 4): WREG32_SOC15(GC, 0, regRLC_PG_DELAY_3, RLC_PG_DELAY_3_DEFAULT_GC_11_0_1); @@ -5054,7 +5055,7 @@ static int gfx_v11_0_set_powergating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 2): case IP_VERSION(11, 0, 3): @@ -5086,7 +5087,7 @@ static int gfx_v11_0_set_clockgating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 1): case IP_VERSION(11, 0, 2): diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index f99a3a6bfd91..e3ff6e46f3f7 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -895,7 +895,7 @@ static void gfx_v9_0_set_kiq_pm4_funcs(struct amdgpu_device *adev) static void gfx_v9_0_init_golden_registers(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): soc15_program_register_sequence(adev, golden_settings_gc_9_0, @@ -951,8 +951,8 @@ static void gfx_v9_0_init_golden_registers(struct amdgpu_device *adev) break; } - if ((adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 4, 1)) && - (adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 4, 2))) + if ((amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 1)) && + (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 2))) soc15_program_register_sequence(adev, golden_settings_gc_9_x_common, (const u32)ARRAY_SIZE(golden_settings_gc_9_x_common)); } @@ -1095,14 +1095,14 @@ static void gfx_v9_0_check_fw_write_wait(struct amdgpu_device *adev) adev->gfx.me_fw_write_wait = false; adev->gfx.mec_fw_write_wait = false; - if ((adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 4, 1)) && + if ((amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 1)) && ((adev->gfx.mec_fw_version < 0x000001a5) || - (adev->gfx.mec_feature_version < 46) || - (adev->gfx.pfp_fw_version < 0x000000b7) || - (adev->gfx.pfp_feature_version < 46))) + (adev->gfx.mec_feature_version < 46) || + (adev->gfx.pfp_fw_version < 0x000000b7) || + (adev->gfx.pfp_feature_version < 46))) DRM_WARN_ONCE("CP firmware version too old, please update!"); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): if ((adev->gfx.me_fw_version >= 0x0000009c) && (adev->gfx.me_feature_version >= 42) && @@ -1202,7 +1202,7 @@ static bool is_raven_kicker(struct amdgpu_device *adev) static bool check_if_enlarge_doorbell_range(struct amdgpu_device *adev) { - if ((adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 3, 0)) && + if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 3, 0)) && (adev->gfx.me_fw_version >= 0x000000a5) && (adev->gfx.me_feature_version >= 52)) return true; @@ -1215,7 +1215,7 @@ static void gfx_v9_0_check_if_need_gfxoff(struct amdgpu_device *adev) if (gfx_v9_0_should_disable_gfxoff(adev->pdev)) adev->pm.pp_feature &= ~PP_GFXOFF_MASK; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 2, 1): case IP_VERSION(9, 4, 0): @@ -1326,9 +1326,9 @@ out: static bool gfx_v9_0_load_mec2_fw_bin_support(struct amdgpu_device *adev) { - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 3, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 3, 0)) return false; return true; @@ -1485,7 +1485,7 @@ static void gfx_v9_0_init_always_on_cu_mask(struct amdgpu_device *adev) if (adev->flags & AMD_IS_APU) always_on_cu_num = 4; - else if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 2, 1)) + else if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 2, 1)) always_on_cu_num = 8; else always_on_cu_num = 12; @@ -1836,7 +1836,7 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) u32 gb_addr_config; int err; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): adev->gfx.config.max_hw_contexts = 8; adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; @@ -2002,7 +2002,7 @@ static int gfx_v9_0_sw_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; unsigned int hw_prio; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 2, 1): case IP_VERSION(9, 4, 0): @@ -2363,7 +2363,7 @@ static void gfx_v9_0_init_sq_config(struct amdgpu_device *adev) { uint32_t tmp; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 1): tmp = RREG32_SOC15(GC, 0, mmSQ_CONFIG); tmp = REG_SET_FIELD(tmp, SQ_CONFIG, DISABLE_BARRIER_WAITCNT, @@ -2700,7 +2700,7 @@ static void gfx_v9_0_init_gfx_power_gating(struct amdgpu_device *adev) /* program GRBM_REG_SAVE_GFX_IDLE_THRESHOLD to 0x55f0 */ data |= (0x55f0 << RLC_AUTO_PG_CTRL__GRBM_REG_SAVE_GFX_IDLE_THRESHOLD__SHIFT); WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_AUTO_PG_CTRL), data); - if (adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 3, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 3, 0)) pwr_10_0_gfxip_control_over_cgpg(adev, true); } } @@ -2812,7 +2812,8 @@ static void gfx_v9_0_init_pg(struct amdgpu_device *adev) * And it's needed by gfxoff feature. */ if (adev->gfx.rlc.is_rlc_v2_1) { - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 2, 1) || + if (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(9, 2, 1) || (adev->apu_flags & AMD_APU_IS_RAVEN2)) gfx_v9_1_init_rlc_save_restore_list(adev); gfx_v9_0_enable_save_restore_machine(adev); @@ -2925,7 +2926,7 @@ static int gfx_v9_0_rlc_resume(struct amdgpu_device *adev) return r; } - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 2, 2): case IP_VERSION(9, 1, 0): gfx_v9_0_init_lbpw(adev); @@ -3713,8 +3714,8 @@ static void gfx_v9_0_init_tcp_config(struct amdgpu_device *adev) { u32 tmp; - if (adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 4, 1) && - adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 4, 2)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 1) && + amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 2)) return; tmp = RREG32_SOC15(GC, 0, mmTCP_ADDR_CONFIG); @@ -3754,7 +3755,7 @@ static int gfx_v9_0_hw_init(void *handle) if (r) return r; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2)) gfx_v9_4_2_set_power_brake_sequence(adev); return r; @@ -3802,7 +3803,7 @@ static int gfx_v9_0_hw_fini(void *handle) /* Skip stopping RLC with A+A reset or when RLC controls GFX clock */ if ((adev->gmc.xgmi.connected_to_cpu && amdgpu_in_reset(adev)) || - (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(9, 4, 2))) { + (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(9, 4, 2))) { dev_dbg(adev->dev, "Skipping RLC halt\n"); return 0; } @@ -3986,7 +3987,7 @@ static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev) { uint64_t clock, clock_lo, clock_hi, hi_check; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 3, 0): preempt_disable(); clock_hi = RREG32_SOC15_NO_KIQ(SMUIO, 0, mmGOLDEN_TSC_COUNT_UPPER_Renoir); @@ -4005,7 +4006,9 @@ static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev) default: amdgpu_gfx_off_ctrl(adev, false); mutex_lock(&adev->gfx.gpu_clock_mutex); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 0, 1) && amdgpu_sriov_runtime(adev)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(9, 0, 1) && + amdgpu_sriov_runtime(adev)) { clock = gfx_v9_0_kiq_read_clock(adev); } else { WREG32_SOC15(GC, 0, mmRLC_CAPTURE_GPU_CLOCK_COUNT, 1); @@ -4357,7 +4360,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) if (!ring->sched.ready) return 0; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1)) { vgpr_init_shader_ptr = vgpr_init_compute_shader_arcturus; vgpr_init_shader_size = sizeof(vgpr_init_compute_shader_arcturus); vgpr_init_regs_ptr = vgpr_init_regs_arcturus; @@ -4509,8 +4512,8 @@ static int gfx_v9_0_early_init(void *handle) adev->gfx.funcs = &gfx_v9_0_gfx_funcs; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2)) adev->gfx.num_gfx_rings = 0; else adev->gfx.num_gfx_rings = GFX9_NUM_GFX_RINGS; @@ -4548,7 +4551,7 @@ static int gfx_v9_0_ecc_late_init(void *handle) } /* requires IBs so do in late init after IB pool is initialized */ - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2)) r = gfx_v9_4_2_do_edc_gpr_workarounds(adev); else r = gfx_v9_0_do_edc_gpr_workarounds(adev); @@ -4580,7 +4583,7 @@ static int gfx_v9_0_late_init(void *handle) if (r) return r; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2)) gfx_v9_4_2_debug_trap_config_init(adev, adev->vm_manager.first_kfd_vmid, AMDGPU_NUM_VMID); else @@ -4676,7 +4679,7 @@ static void gfx_v9_0_update_medium_grain_clock_gating(struct amdgpu_device *adev /* 1 - RLC_CGTT_MGCG_OVERRIDE */ def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE); - if (adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 2, 1)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 2, 1)) data &= ~RLC_CGTT_MGCG_OVERRIDE__CPF_CGTT_SCLK_OVERRIDE_MASK; data &= ~(RLC_CGTT_MGCG_OVERRIDE__GRBM_CGTT_SCLK_OVERRIDE_MASK | @@ -4710,7 +4713,7 @@ static void gfx_v9_0_update_medium_grain_clock_gating(struct amdgpu_device *adev /* 1 - MGCG_OVERRIDE */ def = data = RREG32_SOC15(GC, 0, mmRLC_CGTT_MGCG_OVERRIDE); - if (adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 2, 1)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 2, 1)) data |= RLC_CGTT_MGCG_OVERRIDE__CPF_CGTT_SCLK_OVERRIDE_MASK; data |= (RLC_CGTT_MGCG_OVERRIDE__RLC_CGTT_SCLK_OVERRIDE_MASK | @@ -4816,7 +4819,7 @@ static void gfx_v9_0_update_coarse_grain_clock_gating(struct amdgpu_device *adev /* enable cgcg FSM(0x0000363F) */ def = RREG32_SOC15(GC, 0, mmRLC_CGCG_CGLS_CTRL); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1)) data = (0x2000 << RLC_CGCG_CGLS_CTRL__CGCG_GFX_IDLE_THRESHOLD__SHIFT) | RLC_CGCG_CGLS_CTRL__CGCG_EN_MASK; else @@ -4951,7 +4954,7 @@ static int gfx_v9_0_set_powergating_state(void *handle, struct amdgpu_device *adev = (struct amdgpu_device *)handle; bool enable = (state == AMD_PG_STATE_GATE); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 2, 2): case IP_VERSION(9, 1, 0): case IP_VERSION(9, 3, 0): @@ -4998,7 +5001,7 @@ static int gfx_v9_0_set_clockgating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 2, 1): case IP_VERSION(9, 4, 0): @@ -5048,7 +5051,7 @@ static void gfx_v9_0_get_clockgating_state(void *handle, u64 *flags) if (data & CP_MEM_SLP_CNTL__CP_MEM_LS_EN_MASK) *flags |= AMD_CG_SUPPORT_GFX_CP_LS | AMD_CG_SUPPORT_GFX_MGLS; - if (adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 4, 1)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 1)) { /* AMD_CG_SUPPORT_GFX_3D_CGCG */ data = RREG32_KIQ(SOC15_REG_OFFSET(GC, 0, mmRLC_CGCG_CGLS_CTRL_3D)); if (data & RLC_CGCG_CGLS_CTRL_3D__CGCG_EN_MASK) @@ -7087,7 +7090,7 @@ static void gfx_v9_0_set_irq_funcs(struct amdgpu_device *adev) static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 2, 1): case IP_VERSION(9, 4, 0): @@ -7106,7 +7109,7 @@ static void gfx_v9_0_set_rlc_funcs(struct amdgpu_device *adev) static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev) { /* init asci gds info */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 2, 1): case IP_VERSION(9, 4, 0): @@ -7128,7 +7131,7 @@ static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev) break; } - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): case IP_VERSION(9, 4, 0): adev->gds.gds_compute_max_wave_id = 0x7ff; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c index 32a740104868..fbfe0a1c4b19 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c @@ -682,7 +682,7 @@ static int gfx_v9_4_3_gpu_early_init(struct amdgpu_device *adev) adev->gfx.funcs = &gfx_v9_4_3_gfx_funcs; adev->gfx.ras = &gfx_v9_4_3_ras; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 3): adev->gfx.config.max_hw_contexts = 8; adev->gfx.config.sc_prim_fifo_size_frontend = 0x20; @@ -2430,7 +2430,7 @@ static int gfx_v9_4_3_set_clockgating_state(void *handle, return 0; num_xcc = NUM_XCC(adev->gfx.xcc_mask); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 3): for (i = 0; i < num_xcc; i++) gfx_v9_4_3_xcc_update_gfx_clock_gating( @@ -4231,7 +4231,7 @@ static void gfx_v9_4_3_set_rlc_funcs(struct amdgpu_device *adev) static void gfx_v9_4_3_set_gds_init(struct amdgpu_device *adev) { /* init asci gds info */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 3): /* 9.4.3 removed all the GDS internal memory, * only support GWS opcode in kernel, like barrier @@ -4243,7 +4243,7 @@ static void gfx_v9_4_3_set_gds_init(struct amdgpu_device *adev) break; } - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 3): /* deprecated for 9.4.3, no usage at all */ adev->gds.gds_compute_max_wave_id = 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c index 0834af771549..ff60670b8464 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_2.c @@ -356,11 +356,14 @@ static void gfxhub_v1_2_xcc_setup_vmid_config(struct amdgpu_device *adev, * the SQ per-process. * Retry faults need to be enabled for that to work. */ - tmp = REG_SET_FIELD(tmp, VM_CONTEXT1_CNTL, - RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, - !adev->gmc.noretry || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)); + tmp = REG_SET_FIELD( + tmp, VM_CONTEXT1_CNTL, + RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, + !adev->gmc.noretry || + amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(9, 4, 2) || + amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(9, 4, 3)); WREG32_SOC15_OFFSET(GC, GET_INST(GC, j), regVM_CONTEXT1_CNTL, i * hub->ctx_distance, tmp); WREG32_SOC15_OFFSET(GC, GET_INST(GC, j), diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c index 7708d5ded7b8..f829c441640a 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c @@ -510,7 +510,7 @@ static int gfxhub_v2_1_get_xgmi_info(struct amdgpu_device *adev) u32 max_num_physical_nodes = 0; u32 max_physical_node_id = 0; - switch (adev->ip_versions[XGMI_HWIP][0]) { + switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) { case IP_VERSION(4, 8, 0): max_num_physical_nodes = 4; max_physical_node_id = 3; @@ -548,7 +548,7 @@ static void gfxhub_v2_1_utcl2_harvest(struct amdgpu_device *adev) adev->gfx.config.max_sh_per_se * adev->gfx.config.max_shader_engines); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 1): case IP_VERSION(10, 3, 3): /* Get SA disabled bitmap from eFuse setting */ diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index fa87a85e1017..d3da13f4c80e 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -145,7 +145,8 @@ static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev, * the new fast GRBM interface. */ if ((entry->vmid_src == AMDGPU_GFXHUB(0)) && - (adev->ip_versions[GC_HWIP][0] < IP_VERSION(10, 3, 0))) + (amdgpu_ip_version(adev, GC_HWIP, 0) < + IP_VERSION(10, 3, 0))) RREG32(hub->vm_l2_pro_fault_status); status = RREG32(hub->vm_l2_pro_fault_status); @@ -278,7 +279,7 @@ static void gmc_v10_0_flush_vm_hub(struct amdgpu_device *adev, uint32_t vmid, * to avoid a false ACK due to the new fast GRBM interface. */ if ((vmhub == AMDGPU_GFXHUB(0)) && - (adev->ip_versions[GC_HWIP][0] < IP_VERSION(10, 3, 0))) + (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(10, 3, 0))) RREG32_RLC_NO_KIQ(hub->vm_inv_eng0_req + hub->eng_distance * eng, hub_ip); @@ -680,7 +681,7 @@ static void gmc_v10_0_set_gmc_funcs(struct amdgpu_device *adev) static void gmc_v10_0_set_umc_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[UMC_HWIP][0]) { + switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) { case IP_VERSION(8, 7, 0): adev->umc.max_ras_err_cnt_per_query = UMC_V8_7_TOTAL_CHANNEL_NUM; adev->umc.channel_inst_num = UMC_V8_7_CHANNEL_INSTANCE_NUM; @@ -697,7 +698,7 @@ static void gmc_v10_0_set_umc_funcs(struct amdgpu_device *adev) static void gmc_v10_0_set_mmhub_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(2, 3, 0): case IP_VERSION(2, 4, 0): case IP_VERSION(2, 4, 1): @@ -711,7 +712,7 @@ static void gmc_v10_0_set_mmhub_funcs(struct amdgpu_device *adev) static void gmc_v10_0_set_gfxhub_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): @@ -825,7 +826,7 @@ static int gmc_v10_0_mc_init(struct amdgpu_device *adev) /* set the gart size */ if (amdgpu_gart_size == -1) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { default: adev->gmc.gart_size = 512ULL << 20; break; @@ -892,7 +893,7 @@ static int gmc_v10_0_sw_init(void *handle) adev->gmc.vram_vendor = vram_vendor; } - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): adev->gmc.mall_size = 128 * 1024 * 1024; break; @@ -910,7 +911,7 @@ static int gmc_v10_0_sw_init(void *handle) break; } - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): case IP_VERSION(10, 1, 2): @@ -1195,7 +1196,8 @@ static int gmc_v10_0_set_clockgating_state(void *handle, * is a new problem observed at DF 3.0.3, however with the same suspend sequence not * seen any issue on the DF 3.0.2 series platform. */ - if (adev->in_s0ix && adev->ip_versions[DF_HWIP][0] > IP_VERSION(3, 0, 2)) { + if (adev->in_s0ix && + amdgpu_ip_version(adev, DF_HWIP, 0) > IP_VERSION(3, 0, 2)) { dev_dbg(adev->dev, "keep mmhub clock gating being enabled for s0ix\n"); return 0; } @@ -1204,7 +1206,7 @@ static int gmc_v10_0_set_clockgating_state(void *handle, if (r) return r; - if (adev->ip_versions[ATHUB_HWIP][0] >= IP_VERSION(2, 1, 0)) + if (amdgpu_ip_version(adev, ATHUB_HWIP, 0) >= IP_VERSION(2, 1, 0)) return athub_v2_1_set_clockgating(adev, state); else return athub_v2_0_set_clockgating(adev, state); @@ -1214,13 +1216,13 @@ static void gmc_v10_0_get_clockgating_state(void *handle, u64 *flags) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 3) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 4)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 3) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 1, 4)) return; adev->mmhub.funcs->get_clockgating(adev, flags); - if (adev->ip_versions[ATHUB_HWIP][0] >= IP_VERSION(2, 1, 0)) + if (amdgpu_ip_version(adev, ATHUB_HWIP, 0) >= IP_VERSION(2, 1, 0)) athub_v2_1_get_clockgating(adev, flags); else athub_v2_0_get_clockgating(adev, flags); diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c index 671e288c7575..e1f47f9c1881 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c @@ -588,7 +588,7 @@ static void gmc_v11_0_set_gmc_funcs(struct amdgpu_device *adev) static void gmc_v11_0_set_umc_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[UMC_HWIP][0]) { + switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) { case IP_VERSION(8, 10, 0): adev->umc.channel_inst_num = UMC_V8_10_CHANNEL_INSTANCE_NUM; adev->umc.umc_inst_num = UMC_V8_10_UMC_INSTANCE_NUM; @@ -611,7 +611,7 @@ static void gmc_v11_0_set_umc_funcs(struct amdgpu_device *adev) static void gmc_v11_0_set_mmhub_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(3, 0, 1): adev->mmhub.funcs = &mmhub_v3_0_1_funcs; break; @@ -629,7 +629,7 @@ static void gmc_v11_0_set_mmhub_funcs(struct amdgpu_device *adev) static void gmc_v11_0_set_gfxhub_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 3): adev->gfxhub.funcs = &gfxhub_v3_0_3_funcs; break; @@ -782,7 +782,7 @@ static int gmc_v11_0_sw_init(void *handle) adev->gmc.vram_type = vram_type; adev->gmc.vram_vendor = vram_vendor; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 1): case IP_VERSION(11, 0, 2): diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 3d13d0bba7b1..268ee533e7c1 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -640,7 +640,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, addr, entry->client_id, soc15_ih_clientid_name[entry->client_id]); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) dev_err(adev->dev, " cookie node_id %d fault from die %s%d%s\n", node_id, node_id % 4 == 3 ? "RSV" : "AID", node_id / 4, node_id % 4 == 1 ? ".XCD0" : node_id % 4 == 2 ? ".XCD1" : ""); @@ -654,7 +654,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, * the new fast GRBM interface. */ if ((entry->vmid_src == AMDGPU_GFXHUB(0)) && - (adev->ip_versions[GC_HWIP][0] < IP_VERSION(9, 4, 2))) + (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(9, 4, 2))) RREG32(hub->vm_l2_pro_fault_status); status = RREG32(hub->vm_l2_pro_fault_status); @@ -671,7 +671,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, gfxhub_client_ids[cid], cid); } else { - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(9, 0, 0): mmhub_cid = mmhub_client_ids_vega10[cid][rw]; break; @@ -772,8 +772,8 @@ static uint32_t gmc_v9_0_get_invalidate_req(unsigned int vmid, static bool gmc_v9_0_use_invalidate_semaphore(struct amdgpu_device *adev, uint32_t vmhub) { - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) return false; return ((vmhub == AMDGPU_MMHUB0(0) || @@ -824,7 +824,7 @@ static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, hub = &adev->vmhub[vmhub]; if (adev->gmc.xgmi.num_physical_nodes && - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 0)) { + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 0)) { /* Vega20+XGMI caches PTEs in TC and TLB. Add a * heavy-weight TLB flush (type 2), which flushes * both. Due to a race condition with concurrent @@ -834,7 +834,7 @@ static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, inv_req = gmc_v9_0_get_invalidate_req(vmid, 2); inv_req2 = gmc_v9_0_get_invalidate_req(vmid, flush_type); } else if (flush_type == 2 && - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3) && + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) && adev->rev_id == 0) { inv_req = gmc_v9_0_get_invalidate_req(vmid, 0); inv_req2 = gmc_v9_0_get_invalidate_req(vmid, flush_type); @@ -896,7 +896,7 @@ static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, * GRBM interface. */ if ((vmhub == AMDGPU_GFXHUB(0)) && - (adev->ip_versions[GC_HWIP][0] < IP_VERSION(9, 4, 2))) + (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(9, 4, 2))) RREG32_NO_KIQ(hub->vm_inv_eng0_req + hub->eng_distance * eng); @@ -969,7 +969,8 @@ static int gmc_v9_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev, * still need a second TLB flush after this. */ bool vega20_xgmi_wa = (adev->gmc.xgmi.num_physical_nodes && - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 0)); + amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(9, 4, 0)); /* 2 dwords flush + 8 dwords fence */ unsigned int ndw = kiq->pmf->invalidate_tlbs_size + 8; @@ -984,7 +985,8 @@ static int gmc_v9_0_flush_gpu_tlb_pasid(struct amdgpu_device *adev, pasid, 2, all_hub); if (flush_type == 2 && - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3) && + amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(9, 4, 3) && adev->rev_id == 0) kiq->pmf->kiq_invalidate_tlbs(ring, pasid, 0, all_hub); @@ -1192,7 +1194,7 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev, bool snoop = false; bool is_local; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 1): case IP_VERSION(9, 4, 2): if (is_vram) { @@ -1206,8 +1208,10 @@ static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev, /* FIXME: is this still needed? Or does * amdgpu_ttm_tt_pde_flags already handle this? */ - if ((adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) && + if ((amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(9, 4, 2) || + amdgpu_ip_version(adev, GC_HWIP, 0) == + IP_VERSION(9, 4, 3)) && adev->gmc.xgmi.connected_to_cpu) snoop = true; } else { @@ -1316,7 +1320,7 @@ static void gmc_v9_0_override_vm_pte_flags(struct amdgpu_device *adev, /* Only GFX 9.4.3 APUs associate GPUs with NUMA nodes. Local system * memory can use more efficient MTYPEs. */ - if (adev->ip_versions[GC_HWIP][0] != IP_VERSION(9, 4, 3)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(9, 4, 3)) return; /* Only direct-mapped memory allows us to determine the NUMA node from @@ -1385,7 +1389,7 @@ static unsigned int gmc_v9_0_get_vbios_fb_size(struct amdgpu_device *adev) } else { u32 viewport; - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(1, 0, 0): case IP_VERSION(1, 0, 1): viewport = RREG32_SOC15(DCE, 0, mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION); @@ -1456,7 +1460,7 @@ static void gmc_v9_0_set_gmc_funcs(struct amdgpu_device *adev) static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[UMC_HWIP][0]) { + switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) { case IP_VERSION(6, 0, 0): adev->umc.funcs = &umc_v6_0_funcs; break; @@ -1510,7 +1514,7 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) static void gmc_v9_0_set_mmhub_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(9, 4, 1): adev->mmhub.funcs = &mmhub_v9_4_funcs; break; @@ -1528,7 +1532,7 @@ static void gmc_v9_0_set_mmhub_funcs(struct amdgpu_device *adev) static void gmc_v9_0_set_mmhub_ras_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(9, 4, 0): adev->mmhub.ras = &mmhub_v1_0_ras; break; @@ -1549,7 +1553,7 @@ static void gmc_v9_0_set_mmhub_ras_funcs(struct amdgpu_device *adev) static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev) { - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) adev->gfxhub.funcs = &gfxhub_v1_2_funcs; else adev->gfxhub.funcs = &gfxhub_v1_0_funcs; @@ -1565,7 +1569,7 @@ static void gmc_v9_0_set_mca_ras_funcs(struct amdgpu_device *adev) struct amdgpu_mca *mca = &adev->mca; /* is UMC the right IP to check for MCA? Maybe DF? */ - switch (adev->ip_versions[UMC_HWIP][0]) { + switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) { case IP_VERSION(6, 7, 0): if (!adev->gmc.xgmi.connected_to_cpu) { mca->mp0.ras = &mca_v3_0_mp0_ras; @@ -1592,18 +1596,18 @@ static int gmc_v9_0_early_init(void *handle) * 9.4.0, 9.4.1 and 9.4.3 don't have XGMI defined * in their IP discovery tables */ - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 0) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 0) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) adev->gmc.xgmi.supported = true; - if (adev->ip_versions[XGMI_HWIP][0] == IP_VERSION(6, 1, 0)) { + if (amdgpu_ip_version(adev, XGMI_HWIP, 0) == IP_VERSION(6, 1, 0)) { adev->gmc.xgmi.supported = true; adev->gmc.xgmi.connected_to_cpu = adev->smuio.funcs->is_host_gpu_xgmi_supported(adev); } - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) { enum amdgpu_pkg_type pkg_type = adev->smuio.funcs->get_pkg_type(adev); /* On GFXIP 9.4.3. APU, there is no physical VRAM domain present @@ -1652,7 +1656,7 @@ static int gmc_v9_0_late_init(void *handle) * writes, while disables HBM ECC for vega10. */ if (!amdgpu_sriov_vf(adev) && - (adev->ip_versions[UMC_HWIP][0] == IP_VERSION(6, 0, 0))) { + (amdgpu_ip_version(adev, UMC_HWIP, 0) == IP_VERSION(6, 0, 0))) { if (!(adev->ras_enabled & (1 << AMDGPU_RAS_BLOCK__UMC))) { if (adev->df.funcs && adev->df.funcs->enable_ecc_force_par_wr_rmw) @@ -1760,7 +1764,7 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev) /* set the gart size */ if (amdgpu_gart_size == -1) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): /* all engines support GPUVM */ case IP_VERSION(9, 2, 1): /* all engines support GPUVM */ case IP_VERSION(9, 4, 0): @@ -1839,8 +1843,8 @@ static int gmc_v9_0_gart_init(struct amdgpu_device *adev) */ static void gmc_v9_0_save_registers(struct amdgpu_device *adev) { - if ((adev->ip_versions[DCE_HWIP][0] == IP_VERSION(1, 0, 0)) || - (adev->ip_versions[DCE_HWIP][0] == IP_VERSION(1, 0, 1))) + if ((amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 0)) || + (amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 1))) adev->gmc.sdpif_register = RREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0); } @@ -2035,7 +2039,7 @@ static int gmc_v9_0_sw_init(void *handle) spin_lock_init(&adev->gmc.invalidate_lock); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) { gmc_v9_4_3_init_vram_info(adev); } else if (!adev->bios) { if (adev->flags & AMD_IS_APU) { @@ -2075,7 +2079,7 @@ static int gmc_v9_0_sw_init(void *handle) adev->gmc.vram_type = vram_type; adev->gmc.vram_vendor = vram_vendor; } - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 1, 0): case IP_VERSION(9, 2, 2): set_bit(AMDGPU_GFXHUB(0), adev->vmhubs_mask); @@ -2108,7 +2112,7 @@ static int gmc_v9_0_sw_init(void *handle) amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 47); else amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2)) adev->gmc.translate_further = adev->vm_manager.num_level > 1; break; case IP_VERSION(9, 4, 1): @@ -2140,7 +2144,7 @@ static int gmc_v9_0_sw_init(void *handle) if (r) return r; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1)) { r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VMC1, VMC_1_0__SRCID__VM_FAULT, &adev->gmc.vm_fault); if (r) @@ -2169,7 +2173,10 @@ static int gmc_v9_0_sw_init(void *handle) */ adev->gmc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */ - dma_addr_bits = adev->ip_versions[GC_HWIP][0] >= IP_VERSION(9, 4, 2) ? 48:44; + dma_addr_bits = amdgpu_ip_version(adev, GC_HWIP, 0) >= + IP_VERSION(9, 4, 2) ? + 48 : + 44; r = dma_set_mask_and_coherent(adev->dev, DMA_BIT_MASK(dma_addr_bits)); if (r) { dev_warn(adev->dev, "amdgpu: No suitable DMA available.\n"); @@ -2183,7 +2190,7 @@ static int gmc_v9_0_sw_init(void *handle) amdgpu_gmc_get_vbios_allocations(adev); - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) { + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) { r = gmc_v9_0_init_mem_ranges(adev); if (r) return r; @@ -2209,9 +2216,11 @@ static int gmc_v9_0_sw_init(void *handle) * for video processing. */ adev->vm_manager.first_kfd_vmid = - (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) ? 3 : 8; + (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 1) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) || + amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) ? + 3 : + 8; amdgpu_vm_manager_init(adev); @@ -2221,7 +2230,7 @@ static int gmc_v9_0_sw_init(void *handle) if (r) return r; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) amdgpu_gmc_sysfs_init(adev); return 0; @@ -2231,7 +2240,7 @@ static int gmc_v9_0_sw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 3)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) amdgpu_gmc_sysfs_fini(adev); adev->gmc.num_mem_partitions = 0; kfree(adev->gmc.mem_partitions); @@ -2253,8 +2262,7 @@ static int gmc_v9_0_sw_fini(void *handle) static void gmc_v9_0_init_golden_registers(struct amdgpu_device *adev) { - - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(9, 0, 0): if (amdgpu_sriov_vf(adev)) break; @@ -2288,8 +2296,8 @@ static void gmc_v9_0_init_golden_registers(struct amdgpu_device *adev) */ void gmc_v9_0_restore_registers(struct amdgpu_device *adev) { - if ((adev->ip_versions[DCE_HWIP][0] == IP_VERSION(1, 0, 0)) || - (adev->ip_versions[DCE_HWIP][0] == IP_VERSION(1, 0, 1))) { + if ((amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 0)) || + (amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 1))) { WREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0, adev->gmc.sdpif_register); WARN_ON(adev->gmc.sdpif_register != RREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0)); diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c index 71d1a2e3bac9..3f3a6445c006 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c @@ -49,8 +49,8 @@ static void hdp_v4_0_flush_hdp(struct amdgpu_device *adev, static void hdp_v4_0_invalidate_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring) { - if (adev->ip_versions[HDP_HWIP][0] == IP_VERSION(4, 4, 0) || - adev->ip_versions[HDP_HWIP][0] == IP_VERSION(4, 4, 2)) + if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 0) || + amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 2)) return; if (!ring || !ring->funcs->emit_wreg) @@ -80,7 +80,7 @@ static void hdp_v4_0_reset_ras_error_count(struct amdgpu_device *adev) if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__HDP)) return; - if (adev->ip_versions[HDP_HWIP][0] >= IP_VERSION(4, 4, 0)) + if (amdgpu_ip_version(adev, HDP_HWIP, 0) >= IP_VERSION(4, 4, 0)) WREG32_SOC15(HDP, 0, mmHDP_EDC_CNT, 0); else /*read back hdp ras counter to reset it to 0 */ @@ -92,10 +92,10 @@ static void hdp_v4_0_update_clock_gating(struct amdgpu_device *adev, { uint32_t def, data; - if (adev->ip_versions[HDP_HWIP][0] == IP_VERSION(4, 0, 0) || - adev->ip_versions[HDP_HWIP][0] == IP_VERSION(4, 0, 1) || - adev->ip_versions[HDP_HWIP][0] == IP_VERSION(4, 1, 1) || - adev->ip_versions[HDP_HWIP][0] == IP_VERSION(4, 1, 0)) { + if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 0, 0) || + amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 0, 1) || + amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 1, 1) || + amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 1, 0)) { def = data = RREG32(SOC15_REG_OFFSET(HDP, 0, mmHDP_MEM_POWER_LS)); if (enable && (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS)) @@ -137,7 +137,7 @@ static void hdp_v4_0_get_clockgating_state(struct amdgpu_device *adev, static void hdp_v4_0_init_registers(struct amdgpu_device *adev) { - switch (adev->ip_versions[HDP_HWIP][0]) { + switch (amdgpu_ip_version(adev, HDP_HWIP, 0)) { case IP_VERSION(4, 2, 1): WREG32_FIELD15(HDP, 0, HDP_MMHUB_CNTL, HDP_MMHUB_GCC, 1); break; @@ -147,7 +147,7 @@ static void hdp_v4_0_init_registers(struct amdgpu_device *adev) WREG32_FIELD15(HDP, 0, HDP_MISC_CNTL, FLUSH_INVALIDATE_CACHE, 1); - if (adev->ip_versions[HDP_HWIP][0] == IP_VERSION(4, 4, 0)) + if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 0)) WREG32_FIELD15(HDP, 0, HDP_MISC_CNTL, READ_BUFFER_WATERMARK, 2); WREG32_SOC15(HDP, 0, mmHDP_NONSURFACE_BASE, (adev->gmc.vram_start >> 8)); diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c index 6f20f9889a78..ab06c2b4b20b 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c @@ -51,7 +51,7 @@ static void hdp_v6_0_update_clock_gating(struct amdgpu_device *adev, AMD_CG_SUPPORT_HDP_SD))) return; - if (adev->ip_versions[HDP_HWIP][0] == IP_VERSION(6, 1, 0)) + if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(6, 1, 0)) hdp_clk_cntl = RREG32_SOC15(HDP, 0, regHDP_CLK_CNTL_V6_1); else hdp_clk_cntl = RREG32_SOC15(HDP, 0, regHDP_CLK_CNTL); @@ -61,7 +61,7 @@ static void hdp_v6_0_update_clock_gating(struct amdgpu_device *adev, * forced on IPH & RC clock */ hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL, RC_MEM_CLK_SOFT_OVERRIDE, 1); - if (adev->ip_versions[HDP_HWIP][0] == IP_VERSION(6, 1, 0)) + if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(6, 1, 0)) WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL_V6_1, hdp_clk_cntl); else WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl); @@ -126,7 +126,7 @@ static void hdp_v6_0_update_clock_gating(struct amdgpu_device *adev, /* disable IPH & RC clock override after clock/power mode changing */ hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL, RC_MEM_CLK_SOFT_OVERRIDE, 0); - if (adev->ip_versions[HDP_HWIP][0] == IP_VERSION(6, 1, 0)) + if (amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(6, 1, 0)) WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL_V6_1, hdp_clk_cntl); else WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl); diff --git a/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c index 0a8bc6c94fa9..875fb5ac70b5 100644 --- a/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/imu_v11_0.c @@ -353,7 +353,7 @@ static void imu_v11_0_program_rlc_ram(struct amdgpu_device *adev) WREG32_SOC15(GC, 0, regGFX_IMU_RLC_RAM_INDEX, 0x2); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): program_imu_rlc_ram(adev, imu_rlc_ram_golden_11, (const u32)ARRAY_SIZE(imu_rlc_ram_golden_11)); diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c index aadb74de52bc..e67a337457ed 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c @@ -128,7 +128,7 @@ static int jpeg_v2_5_sw_init(void *handle) ring = adev->jpeg.inst[i].ring_dec; ring->use_doorbell = true; - if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(2, 5, 0)) + if (amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(2, 5, 0)) ring->vm_hub = AMDGPU_MMHUB1(0); else ring->vm_hub = AMDGPU_MMHUB0(0); @@ -822,7 +822,7 @@ static struct amdgpu_jpeg_ras jpeg_v2_6_ras = { static void jpeg_v2_5_set_ras_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[JPEG_HWIP][0]) { + switch (amdgpu_ip_version(adev, JPEG_HWIP, 0)) { case IP_VERSION(2, 6, 0): adev->jpeg.ras = &jpeg_v2_6_ras; break; diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c index df4440c21bbf..a92481da60cd 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v3_0.c @@ -52,7 +52,7 @@ static int jpeg_v3_0_early_init(void *handle) u32 harvest; - switch (adev->ip_versions[UVD_HWIP][0]) { + switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) { case IP_VERSION(3, 1, 1): case IP_VERSION(3, 1, 2): break; diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c index 3eb3dcd56b57..98ed49b16e62 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c @@ -831,7 +831,7 @@ static struct amdgpu_jpeg_ras jpeg_v4_0_ras = { static void jpeg_v4_0_set_ras_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[JPEG_HWIP][0]) { + switch (amdgpu_ip_version(adev, JPEG_HWIP, 0)) { case IP_VERSION(4, 0, 0): adev->jpeg.ras = &jpeg_v4_0_ras; break; diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c index eb06d749876f..1e5ad1e08d2a 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c @@ -558,7 +558,7 @@ static int mes_v10_1_load_microcode(struct amdgpu_device *adev, WREG32_SOC15(GC, 0, mmCP_MES_MDBOUND_LO, 0x3FFFF); /* invalidate ICACHE */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): data = RREG32_SOC15(GC, 0, mmCP_MES_IC_OP_CNTL_Sienna_Cichlid); break; @@ -568,7 +568,7 @@ static int mes_v10_1_load_microcode(struct amdgpu_device *adev, } data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, PRIME_ICACHE, 0); data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, INVALIDATE_CACHE, 1); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): WREG32_SOC15(GC, 0, mmCP_MES_IC_OP_CNTL_Sienna_Cichlid, data); break; @@ -578,7 +578,7 @@ static int mes_v10_1_load_microcode(struct amdgpu_device *adev, } /* prime the ICACHE. */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): data = RREG32_SOC15(GC, 0, mmCP_MES_IC_OP_CNTL_Sienna_Cichlid); break; @@ -587,7 +587,7 @@ static int mes_v10_1_load_microcode(struct amdgpu_device *adev, break; } data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, PRIME_ICACHE, 1); - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): WREG32_SOC15(GC, 0, mmCP_MES_IC_OP_CNTL_Sienna_Cichlid, data); break; @@ -995,7 +995,7 @@ static void mes_v10_1_kiq_setting(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; /* tell RLC which is KIQ queue */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 1): diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c index 3fa5bc3ddf92..4a3020b5b30f 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c @@ -1316,7 +1316,7 @@ static int mes_v11_0_late_init(void *handle) /* it's only intended for use in mes_self_test case, not for s0ix and reset */ if (!amdgpu_in_reset(adev) && !adev->in_s0ix && !adev->in_suspend && - (adev->ip_versions[GC_HWIP][0] != IP_VERSION(11, 0, 3))) + (amdgpu_ip_version(adev, GC_HWIP, 0) != IP_VERSION(11, 0, 3))) amdgpu_mes_self_test(adev); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c index 8f76c6ecf50a..37458f906980 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c @@ -151,7 +151,7 @@ mmhub_v2_0_print_l2_protection_fault_status(struct amdgpu_device *adev, dev_err(adev->dev, "MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", status); - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(2, 0, 0): case IP_VERSION(2, 0, 2): mmhub_cid = mmhub_client_ids_navi1x[cid][rw]; @@ -568,7 +568,7 @@ static void mmhub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *ad if (!(adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)) return; - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): @@ -601,7 +601,7 @@ static void mmhub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *ad DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK); } - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): @@ -625,7 +625,7 @@ static void mmhub_v2_0_update_medium_grain_light_sleep(struct amdgpu_device *ade if (!(adev->cg_flags & AMD_CG_SUPPORT_MC_LS)) return; - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): @@ -651,7 +651,7 @@ static int mmhub_v2_0_set_clockgating(struct amdgpu_device *adev, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(2, 0, 0): case IP_VERSION(2, 0, 2): case IP_VERSION(2, 1, 0): @@ -676,7 +676,7 @@ static void mmhub_v2_0_get_clockgating(struct amdgpu_device *adev, u64 *flags) if (amdgpu_sriov_vf(adev)) *flags = 0; - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c index 1dce053a4c4d..4ddd9448e2bc 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_3.c @@ -90,7 +90,7 @@ mmhub_v2_3_print_l2_protection_fault_status(struct amdgpu_device *adev, dev_err(adev->dev, "MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", status); - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(2, 3, 0): case IP_VERSION(2, 4, 0): case IP_VERSION(2, 4, 1): diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.c index 7c9ab5491067..9627df8b194b 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0.c @@ -107,7 +107,7 @@ mmhub_v3_0_print_l2_protection_fault_status(struct amdgpu_device *adev, dev_err(adev->dev, "MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", status); - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(3, 0, 0): case IP_VERSION(3, 0, 1): mmhub_cid = mmhub_client_ids_v3_0_0[cid][rw]; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.c index db79e6f92441..77bff803b452 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_0_1.c @@ -108,7 +108,7 @@ mmhub_v3_0_1_print_l2_protection_fault_status(struct amdgpu_device *adev, "MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", status); - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(3, 0, 1): mmhub_cid = mmhub_client_ids_v3_0_1[cid][rw]; break; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c index 8194ee2b96c4..3d80a184ce6b 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c @@ -96,7 +96,7 @@ mmhub_v3_3_print_l2_protection_fault_status(struct amdgpu_device *adev, "MMVM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", status); - switch (adev->ip_versions[MMHUB_HWIP][0]) { + switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) { case IP_VERSION(3, 3, 0): mmhub_cid = mmhub_client_ids_v3_3[cid][rw]; break; diff --git a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c index b6a8478dabf4..bb1873363d75 100644 --- a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c @@ -107,7 +107,7 @@ force_update_wptr_for_self_int(struct amdgpu_device *adev, { u32 ih_cntl, ih_rb_cntl; - if (adev->ip_versions[OSSSYS_HWIP][0] < IP_VERSION(5, 0, 3)) + if (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) < IP_VERSION(5, 0, 3)) return; ih_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_CNTL2); @@ -330,7 +330,7 @@ static int navi10_ih_irq_init(struct amdgpu_device *adev) if (unlikely(adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT)) { if (ih[0]->use_bus_addr) { - switch (adev->ip_versions[OSSSYS_HWIP][0]) { + switch (amdgpu_ip_version(adev, OSSSYS_HWIP, 0)) { case IP_VERSION(5, 0, 3): case IP_VERSION(5, 2, 0): case IP_VERSION(5, 2, 1): diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c index 4038455d7998..e523627cfe25 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c @@ -536,7 +536,7 @@ static void nbio_v2_3_clear_doorbell_interrupt(struct amdgpu_device *adev) { uint32_t reg, reg_data; - if (adev->ip_versions[NBIO_HWIP][0] != IP_VERSION(3, 3, 0)) + if (amdgpu_ip_version(adev, NBIO_HWIP, 0) != IP_VERSION(3, 3, 0)) return; reg = RREG32_SOC15(NBIO, 0, mmBIF_RB_CNTL); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.c b/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.c index e5b5b0f4940f..a3622897e3fe 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v4_3.c @@ -338,7 +338,7 @@ const struct nbio_hdp_flush_reg nbio_v4_3_hdp_flush_reg = { static void nbio_v4_3_init_registers(struct amdgpu_device *adev) { - if (adev->ip_versions[NBIO_HWIP][0] == IP_VERSION(4, 3, 0)) { + if (amdgpu_ip_version(adev, NBIO_HWIP, 0) == IP_VERSION(4, 3, 0)) { uint32_t data; data = RREG32_SOC15(NBIO, 0, regRCC_DEV0_EPF2_STRAP2); @@ -392,8 +392,8 @@ static void nbio_v4_3_program_aspm(struct amdgpu_device *adev) #ifdef CONFIG_PCIEASPM uint32_t def, data; - if (!(adev->ip_versions[PCIE_HWIP][0] == IP_VERSION(7, 4, 0)) && - !(adev->ip_versions[PCIE_HWIP][0] == IP_VERSION(7, 6, 0))) + if (!(amdgpu_ip_version(adev, PCIE_HWIP, 0) == IP_VERSION(7, 4, 0)) && + !(amdgpu_ip_version(adev, PCIE_HWIP, 0) == IP_VERSION(7, 6, 0))) return; def = data = RREG32_SOC15(NBIO, 0, regPCIE_LC_CNTL); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c index 4ef1fa4603c8..e962821ae6a1 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_2.c @@ -59,7 +59,7 @@ static u32 nbio_v7_2_get_rev_id(struct amdgpu_device *adev) { u32 tmp; - switch (adev->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(7, 2, 1): case IP_VERSION(7, 3, 0): case IP_VERSION(7, 5, 0): @@ -78,7 +78,7 @@ static u32 nbio_v7_2_get_rev_id(struct amdgpu_device *adev) static void nbio_v7_2_mc_access_enable(struct amdgpu_device *adev, bool enable) { - switch (adev->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(7, 2, 1): case IP_VERSION(7, 3, 0): case IP_VERSION(7, 5, 0): @@ -262,7 +262,7 @@ static void nbio_v7_2_update_medium_grain_light_sleep(struct amdgpu_device *adev { uint32_t def, data; - switch (adev->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(7, 2, 1): case IP_VERSION(7, 3, 0): case IP_VERSION(7, 5, 0): @@ -369,7 +369,7 @@ const struct nbio_hdp_flush_reg nbio_v7_2_hdp_flush_reg = { static void nbio_v7_2_init_registers(struct amdgpu_device *adev) { uint32_t def, data; - switch (adev->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(7, 2, 1): case IP_VERSION(7, 3, 0): case IP_VERSION(7, 5, 0): @@ -394,7 +394,7 @@ static void nbio_v7_2_init_registers(struct amdgpu_device *adev) break; } - switch (adev->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(7, 3, 0): case IP_VERSION(7, 5, 1): data = RREG32_SOC15(NBIO, 0, regRCC_DEV2_EPF0_STRAP2); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c index 685abf57ffdd..7d6d7734dbec 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c @@ -347,7 +347,7 @@ static void nbio_v7_4_init_registers(struct amdgpu_device *adev) adev->rmmio_remap.reg_offset = SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_DEV0_EPF0_VF0_HDP_MEM_COHERENCY_FLUSH_CNTL) << 2; - if (adev->ip_versions[NBIO_HWIP][0] == IP_VERSION(7, 4, 4) && + if (amdgpu_ip_version(adev, NBIO_HWIP, 0) == IP_VERSION(7, 4, 4) && !amdgpu_sriov_vf(adev)) { baco_cntl = RREG32_SOC15(NBIO, 0, mmBACO_CNTL); if (baco_cntl & @@ -702,7 +702,7 @@ static void nbio_v7_4_program_aspm(struct amdgpu_device *adev) #ifdef CONFIG_PCIEASPM uint32_t def, data; - if (adev->ip_versions[NBIO_HWIP][0] == IP_VERSION(7, 4, 4)) + if (amdgpu_ip_version(adev, NBIO_HWIP, 0) == IP_VERSION(7, 4, 4)) return; def = data = RREG32_PCIE(smnPCIE_LC_CNTL); diff --git a/drivers/gpu/drm/amd/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c index 13aca808ecab..0535cabe3b16 100644 --- a/drivers/gpu/drm/amd/amdgpu/nv.c +++ b/drivers/gpu/drm/amd/amdgpu/nv.c @@ -214,7 +214,7 @@ static int nv_query_video_codecs(struct amdgpu_device *adev, bool encode, if (adev->vcn.num_vcn_inst == hweight8(adev->vcn.harvest_config)) return -EINVAL; - switch (adev->ip_versions[UVD_HWIP][0]) { + switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) { case IP_VERSION(3, 0, 0): case IP_VERSION(3, 0, 64): case IP_VERSION(3, 0, 192): @@ -453,7 +453,7 @@ nv_asic_reset_method(struct amdgpu_device *adev) dev_warn(adev->dev, "Specified reset method:%d isn't supported, using AUTO instead.\n", amdgpu_reset_method); - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 5, 0): case IP_VERSION(13, 0, 1): case IP_VERSION(13, 0, 3): @@ -669,7 +669,7 @@ static int nv_common_early_init(void *handle) /* TODO: split the GC and PG flags based on the relevant IP version for which * they are relevant. */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(10, 1, 10): adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | AMD_CG_SUPPORT_GFX_CGCG | @@ -1073,7 +1073,7 @@ static int nv_common_set_clockgating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(2, 3, 0): case IP_VERSION(2, 3, 1): case IP_VERSION(2, 3, 2): diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c index 5f10883da6a2..145186a1e48f 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c @@ -58,9 +58,10 @@ static int psp_v10_0_init_microcode(struct psp_context *psp) return err; err = psp_init_ta_microcode(psp, ucode_prefix); - if ((adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 1, 0)) && - (adev->pdev->revision == 0xa1) && - (psp->securedisplay_context.context.bin_desc.fw_version >= 0x27000008)) { + if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 1, 0)) && + (adev->pdev->revision == 0xa1) && + (psp->securedisplay_context.context.bin_desc.fw_version >= + 0x27000008)) { adev->psp.securedisplay_context.context.bin_desc.size_bytes = 0; } return err; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index 8f84fe40abbb..efa37e3b7931 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -95,7 +95,7 @@ static int psp_v11_0_init_microcode(struct psp_context *psp) amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix)); - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(11, 0, 2): case IP_VERSION(11, 0, 4): err = psp_init_sos_microcode(psp, ucode_prefix); diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c index 469eed084976..54008a8991fc 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c @@ -79,7 +79,7 @@ static int psp_v13_0_init_microcode(struct psp_context *psp) amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix)); - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(13, 0, 2): err = psp_init_sos_microcode(psp, ucode_prefix); if (err) @@ -181,7 +181,7 @@ static int psp_v13_0_wait_for_bootloader_steady_state(struct psp_context *psp) { struct amdgpu_device *adev = psp->adev; - if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 6)) { + if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6)) { psp_v13_0_wait_for_vmbx_ready(psp); return psp_v13_0_wait_for_bootloader(psp); @@ -728,7 +728,7 @@ static int psp_v13_0_fatal_error_recovery_quirk(struct psp_context *psp) { struct amdgpu_device *adev = psp->adev; - if (adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 10)) { + if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 10)) { uint32_t reg_data; /* MP1 fatal error: trigger PSP dram read to unhalt PSP * during MP1 triggered sync flood. diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c index d5ba58eba3e2..eaa5512a21da 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c @@ -40,7 +40,7 @@ static int psp_v13_0_4_init_microcode(struct psp_context *psp) amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix)); - switch (adev->ip_versions[MP0_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(13, 0, 4): err = psp_init_toc_microcode(psp, ucode_prefix); if (err) diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index cd37f45e01a1..8562ac7f7ff0 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -469,7 +469,7 @@ static int sdma_v4_0_irq_id_to_seq(unsigned client_id) static void sdma_v4_0_init_golden_registers(struct amdgpu_device *adev) { - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(4, 0, 0): soc15_program_register_sequence(adev, golden_settings_sdma_4, @@ -539,7 +539,7 @@ static void sdma_v4_0_setup_ulv(struct amdgpu_device *adev) * The only chips with SDMAv4 and ULV are VG10 and VG20. * Server SKUs take a different hysteresis setting from other SKUs. */ - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(4, 0, 0): if (adev->pdev->device == 0x6860) break; @@ -578,8 +578,10 @@ static int sdma_v4_0_init_microcode(struct amdgpu_device *adev) int ret, i; for (i = 0; i < adev->sdma.num_instances; i++) { - if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) || - adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 4, 0)) { + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == + IP_VERSION(4, 2, 2) || + amdgpu_ip_version(adev, SDMA0_HWIP, 0) == + IP_VERSION(4, 4, 0)) { /* Acturus & Aldebaran will leverage the same FW memory for every SDMA instance */ ret = amdgpu_sdma_init_microcode(adev, 0, true); @@ -978,7 +980,8 @@ static void sdma_v4_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable) * Arcturus for the moment and firmware version 14 * and above. */ - if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) && + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == + IP_VERSION(4, 2, 2) && adev->sdma.instance[i].fw_version >= 14) WREG32_SDMA(i, mmSDMA0_PUB_DUMMY_REG2, enable); /* Extend page fault timeout to avoid interrupt storm */ @@ -1255,7 +1258,7 @@ static void sdma_v4_0_init_pg(struct amdgpu_device *adev) if (!(adev->pg_flags & AMD_PG_SUPPORT_SDMA)) return; - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(4, 1, 0): case IP_VERSION(4, 1, 1): case IP_VERSION(4, 1, 2): @@ -1698,7 +1701,7 @@ static bool sdma_v4_0_fw_support_paging_queue(struct amdgpu_device *adev) { uint fw_version = adev->sdma.instance[0].fw_version; - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(4, 0, 0): return fw_version >= 430; case IP_VERSION(4, 0, 1): @@ -1723,7 +1726,7 @@ static int sdma_v4_0_early_init(void *handle) } /* TODO: Page queue breaks driver reload under SRIOV */ - if ((adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 0, 0)) && + if ((amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 0, 0)) && amdgpu_sriov_vf((adev))) adev->sdma.has_page_queue = false; else if (sdma_v4_0_fw_support_paging_queue(adev)) @@ -1823,7 +1826,9 @@ static int sdma_v4_0_sw_init(void *handle) * On Arcturus, SDMA instance 5~7 has a different vmhub * type(AMDGPU_MMHUB1). */ - if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) && i >= 5) + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == + IP_VERSION(4, 2, 2) && + i >= 5) ring->vm_hub = AMDGPU_MMHUB1(0); else ring->vm_hub = AMDGPU_MMHUB0(0); @@ -1843,8 +1848,10 @@ static int sdma_v4_0_sw_init(void *handle) /* paging queue use same doorbell index/routing as gfx queue * with 0x400 (4096 dwords) offset on second doorbell page */ - if (adev->ip_versions[SDMA0_HWIP][0] >= IP_VERSION(4, 0, 0) && - adev->ip_versions[SDMA0_HWIP][0] < IP_VERSION(4, 2, 0)) { + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) >= + IP_VERSION(4, 0, 0) && + amdgpu_ip_version(adev, SDMA0_HWIP, 0) < + IP_VERSION(4, 2, 0)) { ring->doorbell_index = adev->doorbell_index.sdma_engine[i] << 1; ring->doorbell_index += 0x400; @@ -1856,7 +1863,9 @@ static int sdma_v4_0_sw_init(void *handle) (adev->doorbell_index.sdma_engine[i] + 1) << 1; } - if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) && i >= 5) + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == + IP_VERSION(4, 2, 2) && + i >= 5) ring->vm_hub = AMDGPU_MMHUB1(0); else ring->vm_hub = AMDGPU_MMHUB0(0); @@ -1890,8 +1899,8 @@ static int sdma_v4_0_sw_fini(void *handle) amdgpu_ring_fini(&adev->sdma.instance[i].page); } - if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 2) || - adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 4, 0)) + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 2, 2) || + amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 0)) amdgpu_sdma_destroy_inst_ctx(adev, true); else amdgpu_sdma_destroy_inst_ctx(adev, false); @@ -2036,14 +2045,16 @@ static int sdma_v4_0_process_trap_irq(struct amdgpu_device *adev, amdgpu_fence_process(&adev->sdma.instance[instance].ring); break; case 1: - if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 2, 0)) + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == + IP_VERSION(4, 2, 0)) amdgpu_fence_process(&adev->sdma.instance[instance].page); break; case 2: /* XXX compute */ break; case 3: - if (adev->ip_versions[SDMA0_HWIP][0] != IP_VERSION(4, 2, 0)) + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) != + IP_VERSION(4, 2, 0)) amdgpu_fence_process(&adev->sdma.instance[instance].page); break; } @@ -2259,7 +2270,7 @@ static int sdma_v4_0_set_powergating_state(void *handle, { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(4, 1, 0): case IP_VERSION(4, 1, 1): case IP_VERSION(4, 1, 2): @@ -2622,7 +2633,7 @@ static struct amdgpu_sdma_ras sdma_v4_0_ras = { static void sdma_v4_0_set_ras_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(4, 2, 0): case IP_VERSION(4, 2, 2): adev->sdma.ras = &sdma_v4_0_ras; @@ -2633,7 +2644,6 @@ static void sdma_v4_0_set_ras_funcs(struct amdgpu_device *adev) default: break; } - } const struct amdgpu_ip_block_version sdma_v4_0_ip_block = { diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c index 267c1b7b8dcd..1cadd3cb26a9 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c @@ -132,7 +132,8 @@ static int sdma_v4_4_2_init_microcode(struct amdgpu_device *adev) int ret, i; for (i = 0; i < adev->sdma.num_instances; i++) { - if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 4, 2)) { + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == + IP_VERSION(4, 4, 2)) { ret = amdgpu_sdma_init_microcode(adev, 0, true); break; } else { @@ -1231,7 +1232,7 @@ static void sdma_v4_4_2_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t re static bool sdma_v4_4_2_fw_support_paging_queue(struct amdgpu_device *adev) { - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(4, 4, 2): return false; default: @@ -1401,7 +1402,7 @@ static int sdma_v4_4_2_sw_fini(void *handle) amdgpu_ring_fini(&adev->sdma.instance[i].page); } - if (adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(4, 4, 2)) + if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) == IP_VERSION(4, 4, 2)) amdgpu_sdma_destroy_inst_ctx(adev, true); else amdgpu_sdma_destroy_inst_ctx(adev, false); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c index 1cc34efb455b..e0527e5ed7d1 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c @@ -184,7 +184,7 @@ static u32 sdma_v5_0_get_reg_offset(struct amdgpu_device *adev, u32 instance, u3 static void sdma_v5_0_init_golden_registers(struct amdgpu_device *adev) { - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(5, 0, 0): soc15_program_register_sequence(adev, golden_settings_sdma_5, @@ -1697,7 +1697,7 @@ static int sdma_v5_0_set_clockgating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(5, 0, 0): case IP_VERSION(5, 0, 2): case IP_VERSION(5, 0, 5): diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c index 2b3ebebc4299..0ccb7523bc55 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c @@ -1510,7 +1510,7 @@ static int sdma_v5_2_process_illegal_inst_irq(struct amdgpu_device *adev, static bool sdma_v5_2_firmware_mgcg_support(struct amdgpu_device *adev, int i) { - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(5, 2, 1): if (adev->sdma.instance[i].fw_version < 70) return false; @@ -1575,8 +1575,9 @@ static void sdma_v5_2_update_medium_grain_light_sleep(struct amdgpu_device *adev int i; for (i = 0; i < adev->sdma.num_instances; i++) { - - if (adev->sdma.instance[i].fw_version < 70 && adev->ip_versions[SDMA0_HWIP][0] == IP_VERSION(5, 2, 1)) + if (adev->sdma.instance[i].fw_version < 70 && + amdgpu_ip_version(adev, SDMA0_HWIP, 0) == + IP_VERSION(5, 2, 1)) adev->cg_flags &= ~AMD_CG_SUPPORT_SDMA_LS; if (enable && (adev->cg_flags & AMD_CG_SUPPORT_SDMA_LS)) { @@ -1605,7 +1606,7 @@ static int sdma_v5_2_set_clockgating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(5, 2, 0): case IP_VERSION(5, 2, 2): case IP_VERSION(5, 2, 1): diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c index 0e25b6fb1340..4d6de77d289e 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c @@ -1246,14 +1246,13 @@ static struct amdgpu_sdma_ras sdma_v6_0_3_ras = { static void sdma_v6_0_set_ras_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[SDMA0_HWIP][0]) { + switch (amdgpu_ip_version(adev, SDMA0_HWIP, 0)) { case IP_VERSION(6, 0, 3): adev->sdma.ras = &sdma_v6_0_3_ras; break; default: break; } - } static int sdma_v6_0_early_init(void *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.c b/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.c index 07ded70f4df9..93f6772d1b24 100644 --- a/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.c +++ b/drivers/gpu/drm/amd/amdgpu/sienna_cichlid.c @@ -36,7 +36,7 @@ static bool sienna_cichlid_is_mode2_default(struct amdgpu_reset_control *reset_c #if 0 struct amdgpu_device *adev = (struct amdgpu_device *)reset_ctl->handle; - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 7) && + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 7) && adev->pm.fw_version >= 0x3a5500 && !amdgpu_sriov_vf(adev)) return true; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index 9c72add6f93d..66ed28136bc8 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -174,8 +174,8 @@ static const struct amdgpu_video_codecs vcn_4_0_3_video_codecs_encode = { static int soc15_query_video_codecs(struct amdgpu_device *adev, bool encode, const struct amdgpu_video_codecs **codecs) { - if (adev->ip_versions[VCE_HWIP][0]) { - switch (adev->ip_versions[VCE_HWIP][0]) { + if (amdgpu_ip_version(adev, VCE_HWIP, 0)) { + switch (amdgpu_ip_version(adev, VCE_HWIP, 0)) { case IP_VERSION(4, 0, 0): case IP_VERSION(4, 1, 0): if (encode) @@ -187,7 +187,7 @@ static int soc15_query_video_codecs(struct amdgpu_device *adev, bool encode, return -EINVAL; } } else { - switch (adev->ip_versions[UVD_HWIP][0]) { + switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) { case IP_VERSION(1, 0, 0): case IP_VERSION(1, 0, 1): if (encode) @@ -324,12 +324,12 @@ static u32 soc15_get_xclk(struct amdgpu_device *adev) { u32 reference_clock = adev->clock.spll.reference_freq; - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(12, 0, 0) || - adev->ip_versions[MP1_HWIP][0] == IP_VERSION(12, 0, 1) || - adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 6)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(12, 0, 0) || + amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(12, 0, 1) || + amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 6)) return 10000; - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(10, 0, 0) || - adev->ip_versions[MP1_HWIP][0] == IP_VERSION(10, 0, 1)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(10, 0, 0) || + amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(10, 0, 1)) return reference_clock / 4; return reference_clock; @@ -523,7 +523,7 @@ soc15_asic_reset_method(struct amdgpu_device *adev) dev_warn(adev->dev, "Specified reset method:%d isn't supported, using AUTO instead.\n", amdgpu_reset_method); - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(10, 0, 0): case IP_VERSION(10, 0, 1): case IP_VERSION(12, 0, 0): @@ -599,7 +599,7 @@ static int soc15_asic_reset(struct amdgpu_device *adev) static bool soc15_supports_baco(struct amdgpu_device *adev) { - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(9, 0, 0): case IP_VERSION(11, 0, 2): if (adev->asic_type == CHIP_VEGA20) { @@ -938,7 +938,7 @@ static int soc15_common_early_init(void *handle) /* TODO: split the GC and PG flags based on the relevant IP version for which * they are relevant. */ - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 0, 1): adev->asic_funcs = &soc15_asic_funcs; adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG | @@ -1367,7 +1367,7 @@ static int soc15_common_set_clockgating_state(void *handle, if (amdgpu_sriov_vf(adev)) return 0; - switch (adev->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(6, 1, 0): case IP_VERSION(6, 2, 0): case IP_VERSION(7, 4, 0): @@ -1423,8 +1423,7 @@ static void soc15_common_get_clockgating_state(void *handle, u64 *flags) adev->hdp.funcs->get_clock_gating_state(adev, flags); - if (adev->ip_versions[MP0_HWIP][0] != IP_VERSION(13, 0, 2)) { - + if (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 2)) { /* AMD_CG_SUPPORT_DRM_MGCG */ data = RREG32(SOC15_REG_OFFSET(MP0, 0, mmMP0_MISC_CGTT_CTRL0)); if (!(data & 0x01000000)) diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.c b/drivers/gpu/drm/amd/amdgpu/soc21.c index 2ecc8c9a078b..92a80780ab72 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc21.c +++ b/drivers/gpu/drm/amd/amdgpu/soc21.c @@ -153,7 +153,7 @@ static int soc21_query_video_codecs(struct amdgpu_device *adev, bool encode, if (adev->vcn.num_vcn_inst == hweight8(adev->vcn.harvest_config)) return -EINVAL; - switch (adev->ip_versions[UVD_HWIP][0]) { + switch (amdgpu_ip_version(adev, UVD_HWIP, 0)) { case IP_VERSION(4, 0, 0): case IP_VERSION(4, 0, 2): case IP_VERSION(4, 0, 4): @@ -374,7 +374,7 @@ soc21_asic_reset_method(struct amdgpu_device *adev) dev_warn(adev->dev, "Specified reset method:%d isn't supported, using AUTO instead.\n", amdgpu_reset_method); - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 0): case IP_VERSION(13, 0, 7): case IP_VERSION(13, 0, 10): @@ -448,7 +448,7 @@ const struct amdgpu_ip_block_version soc21_common_ip_block = { static bool soc21_need_full_reset(struct amdgpu_device *adev) { - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): return amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC); case IP_VERSION(11, 0, 2): @@ -577,7 +577,7 @@ static int soc21_common_early_init(void *handle) adev->rev_id = amdgpu_device_get_rev_id(adev); adev->external_rev_id = 0xff; - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): adev->cg_flags = AMD_CG_SUPPORT_GFX_CGCG | AMD_CG_SUPPORT_GFX_CGLS | @@ -843,7 +843,7 @@ static int soc21_common_set_clockgating_state(void *handle, { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - switch (adev->ip_versions[NBIO_HWIP][0]) { + switch (amdgpu_ip_version(adev, NBIO_HWIP, 0)) { case IP_VERSION(4, 3, 0): case IP_VERSION(4, 3, 1): case IP_VERSION(7, 7, 0): @@ -865,7 +865,7 @@ static int soc21_common_set_powergating_state(void *handle, { struct amdgpu_device *adev = (struct amdgpu_device *)handle; - switch (adev->ip_versions[LSDMA_HWIP][0]) { + switch (amdgpu_ip_version(adev, LSDMA_HWIP, 0)) { case IP_VERSION(6, 0, 0): case IP_VERSION(6, 0, 2): adev->lsdma.funcs->update_memory_power_gating(adev, diff --git a/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c b/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c index 67164991f541..99713949b61f 100644 --- a/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/umsch_mm_v4_0.c @@ -273,14 +273,15 @@ static int umsch_mm_v4_0_set_hw_resources(struct amdgpu_umsch_mm *umsch) memcpy(set_hw_resources.mmhub_base, adev->reg_offset[MMHUB_HWIP][0], sizeof(uint32_t) * 5); - set_hw_resources.mmhub_version = adev->ip_versions[MMHUB_HWIP][0]; + set_hw_resources.mmhub_version = amdgpu_ip_version(adev, MMHUB_HWIP, 0); memcpy(set_hw_resources.osssys_base, adev->reg_offset[OSSSYS_HWIP][0], sizeof(uint32_t) * 5); - set_hw_resources.osssys_version = adev->ip_versions[OSSSYS_HWIP][0]; + set_hw_resources.osssys_version = + amdgpu_ip_version(adev, OSSSYS_HWIP, 0); - set_hw_resources.vcn_version = adev->ip_versions[VCN_HWIP][0]; - set_hw_resources.vpe_version = adev->ip_versions[VPE_HWIP][0]; + set_hw_resources.vcn_version = amdgpu_ip_version(adev, VCN_HWIP, 0); + set_hw_resources.vpe_version = amdgpu_ip_version(adev, VPE_HWIP, 0); set_hw_resources.api_status.api_completion_fence_addr = umsch->ring.fence_drv.gpu_addr; set_hw_resources.api_status.api_completion_fence_value = ++umsch->ring.fence_drv.sync_seq; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index 6fbea38f4d3e..aba403d71806 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -187,7 +187,7 @@ static int vcn_v2_5_sw_init(void *handle) ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + (amdgpu_sriov_vf(adev) ? 2*j : 8*j); - if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(2, 5, 0)) + if (amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(2, 5, 0)) ring->vm_hub = AMDGPU_MMHUB1(0); else ring->vm_hub = AMDGPU_MMHUB0(0); @@ -207,7 +207,8 @@ static int vcn_v2_5_sw_init(void *handle) ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + (amdgpu_sriov_vf(adev) ? (1 + i + 2*j) : (2 + i + 8*j)); - if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(2, 5, 0)) + if (amdgpu_ip_version(adev, UVD_HWIP, 0) == + IP_VERSION(2, 5, 0)) ring->vm_hub = AMDGPU_MMHUB1(0); else ring->vm_hub = AMDGPU_MMHUB0(0); @@ -794,7 +795,7 @@ static void vcn_v2_6_enable_ras(struct amdgpu_device *adev, int inst_idx, { uint32_t tmp; - if (adev->ip_versions[UVD_HWIP][0] != IP_VERSION(2, 6, 0)) + if (amdgpu_ip_version(adev, UVD_HWIP, 0) != IP_VERSION(2, 6, 0)) return; tmp = VCN_RAS_CNTL__VCPU_VCODEC_REARM_MASK | @@ -1985,7 +1986,7 @@ static struct amdgpu_vcn_ras vcn_v2_6_ras = { static void vcn_v2_5_set_ras_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[VCN_HWIP][0]) { + switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) { case IP_VERSION(2, 6, 0): adev->vcn.ras = &vcn_v2_6_ras; break; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c index a61ecefdafc5..e02af4de521c 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c @@ -100,7 +100,8 @@ static int vcn_v3_0_early_init(void *handle) /* both instances are harvested, disable the block */ return -ENOENT; - if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(3, 0, 33)) + if (amdgpu_ip_version(adev, UVD_HWIP, 0) == + IP_VERSION(3, 0, 33)) adev->vcn.num_enc_rings = 0; else adev->vcn.num_enc_rings = 2; @@ -227,9 +228,10 @@ static int vcn_v3_0_sw_init(void *handle) cpu_to_le32(AMDGPU_VCN_FW_SHARED_FLAG_0_RB); fw_shared->sw_ring.is_enabled = cpu_to_le32(DEC_SW_RING_ENABLED); fw_shared->present_flag_0 |= AMDGPU_VCN_SMU_VERSION_INFO_FLAG; - if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(3, 1, 2)) + if (amdgpu_ip_version(adev, UVD_HWIP, 0) == IP_VERSION(3, 1, 2)) fw_shared->smu_interface_info.smu_interface_type = 2; - else if (adev->ip_versions[UVD_HWIP][0] == IP_VERSION(3, 1, 1)) + else if (amdgpu_ip_version(adev, UVD_HWIP, 0) == + IP_VERSION(3, 1, 1)) fw_shared->smu_interface_info.smu_interface_type = 1; if (amdgpu_vcnfw_log) @@ -1255,7 +1257,8 @@ static int vcn_v3_0_start(struct amdgpu_device *adev) fw_shared->rb.wptr = lower_32_bits(ring->wptr); fw_shared->multi_queue.decode_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET); - if (adev->ip_versions[UVD_HWIP][0] != IP_VERSION(3, 0, 33)) { + if (amdgpu_ip_version(adev, UVD_HWIP, 0) != + IP_VERSION(3, 0, 33)) { fw_shared->multi_queue.encode_generalpurpose_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET); ring = &adev->vcn.inst[i].ring_enc[0]; WREG32_SOC15(VCN, i, mmUVD_RB_RPTR, lower_32_bits(ring->wptr)); @@ -1628,7 +1631,8 @@ static int vcn_v3_0_pause_dpg_mode(struct amdgpu_device *adev, UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK); - if (adev->ip_versions[UVD_HWIP][0] != IP_VERSION(3, 0, 33)) { + if (amdgpu_ip_version(adev, UVD_HWIP, 0) != + IP_VERSION(3, 0, 33)) { /* Restore */ fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr; fw_shared->multi_queue.encode_generalpurpose_queue_mode |= cpu_to_le32(FW_QUEUE_RING_RESET); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c index 29164289c5f3..bd0fef2a92a8 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c @@ -169,7 +169,8 @@ static int vcn_v4_0_sw_init(void *handle) fw_shared->smu_dpm_interface.smu_interface_type = (adev->flags & AMD_IS_APU) ? AMDGPU_VCN_SMU_DPM_INTERFACE_APU : AMDGPU_VCN_SMU_DPM_INTERFACE_DGPU; - if (adev->ip_versions[VCN_HWIP][0] == IP_VERSION(4, 0, 2)) { + if (amdgpu_ip_version(adev, VCN_HWIP, 0) == + IP_VERSION(4, 0, 2)) { fw_shared->present_flag_0 |= AMDGPU_FW_SHARED_FLAG_0_DRM_KEY_INJECT; fw_shared->drm_key_wa.method = AMDGPU_DRM_KEY_INJECT_WORKAROUND_VCNFW_ASD_HANDSHAKING; @@ -1849,7 +1850,7 @@ static void vcn_v4_0_set_unified_ring_funcs(struct amdgpu_device *adev) if (adev->vcn.harvest_config & (1 << i)) continue; - if (adev->ip_versions[VCN_HWIP][0] == IP_VERSION(4, 0, 2)) + if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 2)) vcn_v4_0_unified_ring_vm_funcs.secure_submission_supported = true; adev->vcn.inst[i].ring_enc[0].funcs = @@ -2156,7 +2157,7 @@ static struct amdgpu_vcn_ras vcn_v4_0_ras = { static void vcn_v4_0_set_ras_funcs(struct amdgpu_device *adev) { - switch (adev->ip_versions[VCN_HWIP][0]) { + switch (amdgpu_ip_version(adev, VCN_HWIP, 0)) { case IP_VERSION(4, 0, 0): adev->vcn.ras = &vcn_v4_0_ras; break; diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c index dbc99536440f..ddfc6941f9d5 100644 --- a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c @@ -291,7 +291,7 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev) adev->nbio.funcs->ih_control(adev); - if ((adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 2, 1)) && + if ((amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 2, 1)) && adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) { ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN); if (adev->irq.ih.use_bus_addr) { @@ -304,8 +304,8 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev) /* psp firmware won't program IH_CHICKEN for aldebaran * driver needs to program it properly according to * MC_SPACE type in IH_RB_CNTL */ - if ((adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 4, 0)) || - (adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 4, 2))) { + if ((amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 0)) || + (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 2))) { ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN_ALDEBARAN); if (adev->irq.ih.use_bus_addr) { ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN, @@ -334,8 +334,8 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev) vega20_setup_retry_doorbell(adev->irq.retry_cam_doorbell_index)); /* Enable IH Retry CAM */ - if (adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 4, 0) || - adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 4, 2)) + if (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 0) || + amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 2)) WREG32_FIELD15(OSSSYS, 0, IH_RETRY_INT_CAM_CNTL_ALDEBARAN, ENABLE, 1); else @@ -537,7 +537,7 @@ static int vega20_ih_sw_init(void *handle) return r; if ((adev->flags & AMD_IS_APU) && - (adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 4, 2))) + (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) == IP_VERSION(4, 4, 2))) use_bus_addr = false; r = amdgpu_ih_ring_init(adev, &adev->irq.ih, IH_RING_SIZE, use_bus_addr); @@ -554,7 +554,7 @@ static int vega20_ih_sw_init(void *handle) adev->irq.ih1.use_doorbell = true; adev->irq.ih1.doorbell_index = (adev->doorbell_index.ih + 1) << 1; - if (adev->ip_versions[OSSSYS_HWIP][0] != IP_VERSION(4, 4, 2)) { + if (amdgpu_ip_version(adev, OSSSYS_HWIP, 0) != IP_VERSION(4, 4, 2)) { r = amdgpu_ih_ring_init(adev, &adev->irq.ih2, PAGE_SIZE, true); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c index ab4a63bb3e3e..0a9cf9dfc224 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c @@ -65,7 +65,7 @@ static int kfd_resume(struct kfd_node *kfd); static void kfd_device_info_set_sdma_info(struct kfd_dev *kfd) { - uint32_t sdma_version = kfd->adev->ip_versions[SDMA0_HWIP][0]; + uint32_t sdma_version = amdgpu_ip_version(kfd->adev, SDMA0_HWIP, 0); switch (sdma_version) { case IP_VERSION(4, 0, 0):/* VEGA10 */ @@ -282,7 +282,7 @@ struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf) f2g = &gfx_v8_kfd2kgd; break; default: - switch (adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { /* Vega 10 */ case IP_VERSION(9, 0, 1): gfx_target_version = 90000; @@ -427,9 +427,11 @@ struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf) } if (!f2g) { - if (adev->ip_versions[GC_HWIP][0]) - dev_err(kfd_device, "GC IP %06x %s not supported in kfd\n", - adev->ip_versions[GC_HWIP][0], vf ? "VF" : ""); + if (amdgpu_ip_version(adev, GC_HWIP, 0)) + dev_err(kfd_device, + "GC IP %06x %s not supported in kfd\n", + amdgpu_ip_version(adev, GC_HWIP, 0), + vf ? "VF" : ""); else dev_err(kfd_device, "%s %s not supported in kfd\n", amdgpu_asic_name[adev->asic_type], vf ? "VF" : ""); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c index 7d82c7da223a..192b0d106413 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c @@ -1001,7 +1001,7 @@ int kgd2kfd_init_zone_device(struct amdgpu_device *adev) void *r; /* Page migration works on gfx9 or newer */ - if (adev->ip_versions[GC_HWIP][0] < IP_VERSION(9, 0, 1)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(9, 0, 1)) return -EINVAL; if (adev->gmc.is_app_apu) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c index 1a03173e2313..8ee2bedd301a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c @@ -205,7 +205,8 @@ static int pm_set_resources_v9(struct packet_manager *pm, uint32_t *buffer, static inline bool pm_use_ext_eng(struct kfd_dev *dev) { - return dev->adev->ip_versions[SDMA0_HWIP][0] >= IP_VERSION(5, 2, 0); + return amdgpu_ip_version(dev->adev, SDMA0_HWIP, 0) >= + IP_VERSION(5, 2, 0); } static int pm_map_queues_v9(struct packet_manager *pm, uint32_t *buffer, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index b9950074aee0..9cc32f577e38 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -202,7 +202,7 @@ enum cache_policy { cache_policy_noncoherent }; -#define KFD_GC_VERSION(dev) ((dev)->adev->ip_versions[GC_HWIP][0]) +#define KFD_GC_VERSION(dev) (amdgpu_ip_version((dev)->adev, GC_HWIP, 0)) #define KFD_IS_SOC15(dev) ((KFD_GC_VERSION(dev)) >= (IP_VERSION(9, 0, 1))) #define KFD_SUPPORT_XNACK_PER_PROCESS(dev)\ ((KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 2)) || \ diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 49b62c2172cc..ae5660861133 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -1196,7 +1196,7 @@ svm_range_get_pte_flags(struct kfd_node *node, if (domain == SVM_RANGE_VRAM_DOMAIN) bo_node = prange->svm_bo->node; - switch (node->adev->ip_versions[GC_HWIP][0]) { + switch (amdgpu_ip_version(node->adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 1): if (domain == SVM_RANGE_VRAM_DOMAIN) { if (bo_node == node) { diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 8e98dda1e084..d9c29d24844a 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -1173,7 +1173,7 @@ static int dm_dmub_hw_init(struct amdgpu_device *adev) for (i = 0; i < fb_info->num_fb; ++i) hw_params.fb[i] = &fb_info->fb[i]; - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(3, 1, 3): case IP_VERSION(3, 1, 4): case IP_VERSION(3, 5, 0): @@ -1606,7 +1606,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) init_data.dce_environment = DCE_ENV_PRODUCTION_DRV; - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(2, 1, 0): switch (adev->dm.dmcub_fw_version) { case 0: /* development */ @@ -1631,7 +1631,7 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) init_data.flags.gpu_vm_support = true; break; default: - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(1, 0, 0): case IP_VERSION(1, 0, 1): /* enable S/G on PCO and RV2 */ @@ -2015,7 +2015,7 @@ static int load_dmcu_fw(struct amdgpu_device *adev) return 0; break; default: - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(2, 0, 2): case IP_VERSION(2, 0, 3): case IP_VERSION(2, 0, 0): @@ -2105,7 +2105,7 @@ static int dm_dmub_sw_init(struct amdgpu_device *adev) enum dmub_status status; int r; - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(2, 1, 0): dmub_asic = DMUB_ASIC_DCN21; break; @@ -2477,7 +2477,7 @@ static int amdgpu_dm_smu_write_watermarks_table(struct amdgpu_device *adev) * therefore, this function apply to navi10/12/14 but not Renoir * * */ - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(2, 0, 2): case IP_VERSION(2, 0, 0): break; @@ -4429,7 +4429,7 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev) } /* Use Outbox interrupt */ - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(3, 0, 0): case IP_VERSION(3, 1, 2): case IP_VERSION(3, 1, 3): @@ -4447,12 +4447,12 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev) break; default: DRM_DEBUG_KMS("Unsupported DCN IP version for outbox: 0x%X\n", - adev->ip_versions[DCE_HWIP][0]); + amdgpu_ip_version(adev, DCE_HWIP, 0)); } /* Determine whether to enable PSR support by default. */ if (!(amdgpu_dc_debug_mask & DC_DISABLE_PSR)) { - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(3, 1, 2): case IP_VERSION(3, 1, 3): case IP_VERSION(3, 1, 4): @@ -4470,7 +4470,7 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev) } if (!(amdgpu_dc_debug_mask & DC_DISABLE_REPLAY)) { - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(3, 1, 4): case IP_VERSION(3, 1, 5): case IP_VERSION(3, 1, 6): @@ -4585,7 +4585,7 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev) } break; default: - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(1, 0, 0): case IP_VERSION(1, 0, 1): case IP_VERSION(2, 0, 2): @@ -4611,7 +4611,7 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev) break; default: DRM_ERROR("Unsupported DCE IP versions: 0x%X\n", - adev->ip_versions[DCE_HWIP][0]); + amdgpu_ip_version(adev, DCE_HWIP, 0)); goto fail; } break; @@ -4694,14 +4694,14 @@ static int dm_init_microcode(struct amdgpu_device *adev) char *fw_name_dmub; int r; - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(2, 1, 0): fw_name_dmub = FIRMWARE_RENOIR_DMUB; if (ASICREV_IS_GREEN_SARDINE(adev->external_rev_id)) fw_name_dmub = FIRMWARE_GREEN_SARDINE_DMUB; break; case IP_VERSION(3, 0, 0): - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(10, 3, 0)) fw_name_dmub = FIRMWARE_SIENNA_CICHLID_DMUB; else fw_name_dmub = FIRMWARE_NAVY_FLOUNDER_DMUB; @@ -4831,7 +4831,7 @@ static int dm_early_init(void *handle) break; default: - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(2, 0, 2): case IP_VERSION(3, 0, 0): adev->mode_info.num_crtc = 6; @@ -4868,7 +4868,7 @@ static int dm_early_init(void *handle) break; default: DRM_ERROR("Unsupported DCE IP versions: 0x%x\n", - adev->ip_versions[DCE_HWIP][0]); + amdgpu_ip_version(adev, DCE_HWIP, 0)); return -EINVAL; } break; @@ -11002,7 +11002,7 @@ int amdgpu_dm_process_dmub_set_config_sync( */ bool check_seamless_boot_capability(struct amdgpu_device *adev) { - switch (adev->ip_versions[DCE_HWIP][0]) { + switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(3, 0, 1): if (!adev->mman.keep_stolen_vga_memory) return true; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c index b97cbc4e5477..8038fe3d193e 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c @@ -226,7 +226,7 @@ static void fill_gfx9_tiling_info_from_device(const struct amdgpu_device *adev, tiling_info->gfx9.num_rb_per_se = adev->gfx.config.gb_addr_config_fields.num_rb_per_se; tiling_info->gfx9.shaderEnable = 1; - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0)) tiling_info->gfx9.num_pkrs = adev->gfx.config.gb_addr_config_fields.num_pkrs; } @@ -669,7 +669,7 @@ static int get_plane_modifiers(struct amdgpu_device *adev, unsigned int plane_ty case AMDGPU_FAMILY_YC: case AMDGPU_FAMILY_GC_10_3_6: case AMDGPU_FAMILY_GC_10_3_7: - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(10, 3, 0)) add_gfx10_3_modifiers(adev, mods, &size, &capacity); else add_gfx10_1_modifiers(adev, mods, &size, &capacity); @@ -1069,8 +1069,8 @@ int amdgpu_dm_plane_fill_dc_scaling_info(struct amdgpu_device *adev, * is to gesture the YouTube Android app into full screen * on ChromeOS. */ - if (((adev->ip_versions[DCE_HWIP][0] == IP_VERSION(1, 0, 0)) || - (adev->ip_versions[DCE_HWIP][0] == IP_VERSION(1, 0, 1))) && + if (((amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 0)) || + (amdgpu_ip_version(adev, DCE_HWIP, 0) == IP_VERSION(1, 0, 1))) && (state->fb && state->fb->format->format == DRM_FORMAT_NV12 && (scaling_info->src_rect.x != 0 || scaling_info->src_rect.y != 0))) return -EINVAL; @@ -1509,7 +1509,7 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm, drm_plane_create_rotation_property(plane, DRM_MODE_ROTATE_0, supported_rotations); - if (dm->adev->ip_versions[DCE_HWIP][0] > IP_VERSION(3, 0, 1) && + if (amdgpu_ip_version(dm->adev, DCE_HWIP, 0) > IP_VERSION(3, 0, 1) && plane->type != DRM_PLANE_TYPE_CURSOR) drm_plane_enable_fb_damage_clips(plane); diff --git a/drivers/gpu/drm/amd/pm/amdgpu_pm.c b/drivers/gpu/drm/amd/pm/amdgpu_pm.c index 84e1af6a6ce7..e789a48089ad 100644 --- a/drivers/gpu/drm/amd/pm/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/pm/amdgpu_pm.c @@ -2024,8 +2024,8 @@ static int default_attr_update(struct amdgpu_device *adev, struct amdgpu_device_ uint32_t mask, enum amdgpu_device_attr_states *states) { struct device_attribute *dev_attr = &attr->dev_attr; - uint32_t mp1_ver = adev->ip_versions[MP1_HWIP][0]; - uint32_t gc_ver = adev->ip_versions[GC_HWIP][0]; + uint32_t mp1_ver = amdgpu_ip_version(adev, MP1_HWIP, 0); + uint32_t gc_ver = amdgpu_ip_version(adev, GC_HWIP, 0); const char *attr_name = dev_attr->attr.name; if (!(attr->flags & mask)) { @@ -2917,7 +2917,7 @@ static ssize_t amdgpu_hwmon_show_power_label(struct device *dev, char *buf) { struct amdgpu_device *adev = dev_get_drvdata(dev); - uint32_t gc_ver = adev->ip_versions[GC_HWIP][0]; + uint32_t gc_ver = amdgpu_ip_version(adev, GC_HWIP, 0); if (gc_ver == IP_VERSION(10, 3, 1)) return sysfs_emit(buf, "%s\n", @@ -3205,7 +3205,7 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj, struct device *dev = kobj_to_dev(kobj); struct amdgpu_device *adev = dev_get_drvdata(dev); umode_t effective_mode = attr->mode; - uint32_t gc_ver = adev->ip_versions[GC_HWIP][0]; + uint32_t gc_ver = amdgpu_ip_version(adev, GC_HWIP, 0); uint32_t tmp; /* under multi-vf mode, the hwmon attributes are all not supported */ @@ -4158,8 +4158,8 @@ static void amdgpu_debugfs_prints_cpu_info(struct seq_file *m, static int amdgpu_debugfs_pm_info_pp(struct seq_file *m, struct amdgpu_device *adev) { - uint32_t mp1_ver = adev->ip_versions[MP1_HWIP][0]; - uint32_t gc_ver = adev->ip_versions[GC_HWIP][0]; + uint32_t mp1_ver = amdgpu_ip_version(adev, MP1_HWIP, 0); + uint32_t gc_ver = amdgpu_ip_version(adev, GC_HWIP, 0); uint32_t value; uint64_t value64 = 0; uint32_t query = 0; diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c index ed23d7de3f28..e6f1620acdd4 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c @@ -485,7 +485,7 @@ bool is_support_sw_smu(struct amdgpu_device *adev) if (adev->asic_type == CHIP_VEGA20) return false; - if (adev->ip_versions[MP1_HWIP][0] >= IP_VERSION(11, 0, 0)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) >= IP_VERSION(11, 0, 0)) return true; return false; @@ -603,7 +603,7 @@ static int smu_set_funcs(struct amdgpu_device *adev) if (adev->pm.pp_feature & PP_OVERDRIVE_MASK) smu->od_enabled = true; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 5): case IP_VERSION(11, 0, 9): @@ -775,8 +775,8 @@ static int smu_late_init(void *handle) } } - if ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 1)) || - (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 3))) + if ((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 1)) || + (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 3))) return 0; if (!amdgpu_sriov_vf(adev) || smu->od_enabled) { @@ -1259,7 +1259,7 @@ static int smu_smc_hw_setup(struct smu_context *smu) uint64_t features_supported; int ret = 0; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 7): case IP_VERSION(11, 0, 11): case IP_VERSION(11, 5, 0): @@ -1449,7 +1449,7 @@ static int smu_start_smc_engine(struct smu_context *smu) int ret = 0; if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) { - if (adev->ip_versions[MP1_HWIP][0] < IP_VERSION(11, 0, 0)) { + if (amdgpu_ip_version(adev, MP1_HWIP, 0) < IP_VERSION(11, 0, 0)) { if (smu->ppt_funcs->load_microcode) { ret = smu->ppt_funcs->load_microcode(smu); if (ret) @@ -1549,7 +1549,7 @@ static int smu_disable_dpms(struct smu_context *smu) * For SMU 13.0.0 and 13.0.7, PMFW will handle the DPM features(disablement or others) * properly on suspend/reset/unload. Driver involvement may cause some unexpected issues. */ - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 0): case IP_VERSION(13, 0, 7): case IP_VERSION(13, 0, 10): @@ -1570,7 +1570,7 @@ static int smu_disable_dpms(struct smu_context *smu) * properly. */ if (smu->uploading_custom_pp_table) { - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 5): case IP_VERSION(11, 0, 9): @@ -1590,7 +1590,7 @@ static int smu_disable_dpms(struct smu_context *smu) * on BACO in. Driver involvement is unnecessary. */ if (use_baco) { - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 7): case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 5): @@ -1607,7 +1607,7 @@ static int smu_disable_dpms(struct smu_context *smu) * for gpu reset and S0i3 cases. Driver involvement is unnecessary. */ if (amdgpu_in_reset(adev) || adev->in_s0ix) { - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 4): case IP_VERSION(13, 0, 11): return 0; @@ -1634,7 +1634,7 @@ static int smu_disable_dpms(struct smu_context *smu) } } - if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(9, 4, 2) && + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(9, 4, 2) && !amdgpu_sriov_vf(adev) && adev->gfx.rlc.funcs->stop) adev->gfx.rlc.funcs->stop(adev); @@ -2391,7 +2391,7 @@ int smu_get_power_limit(void *handle, } else { switch (limit_level) { case SMU_PPT_LIMIT_CURRENT: - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 2): case IP_VERSION(11, 0, 7): case IP_VERSION(11, 0, 11): diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c index 18487ae10bcf..650482cedd1f 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c @@ -345,8 +345,8 @@ navi10_get_allowed_feature_mask(struct smu_context *smu, /* DPM UCLK enablement should be skipped for navi10 A0 secure board */ if (!(is_asic_secure(smu) && - (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 0)) && - (adev->rev_id == 0)) && + (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 0)) && + (adev->rev_id == 0)) && (adev->pm.pp_feature & PP_MCLK_DPM_MASK)) *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_DPM_UCLK_BIT) | FEATURE_MASK(FEATURE_MEM_VDDCI_SCALING_BIT) @@ -354,7 +354,7 @@ navi10_get_allowed_feature_mask(struct smu_context *smu, /* DS SOCCLK enablement should be skipped for navi10 A0 secure board */ if (is_asic_secure(smu) && - (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 0)) && + (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 0)) && (adev->rev_id == 0)) *(uint64_t *)feature_mask &= ~FEATURE_MASK(FEATURE_DS_SOCCLK_BIT); @@ -916,7 +916,7 @@ static int navi1x_get_smu_metrics_data(struct smu_context *smu, return ret; } - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 9): if (smu_version > 0x00341C00) ret = navi12_get_smu_metrics_data(smu, member, value); @@ -926,8 +926,12 @@ static int navi1x_get_smu_metrics_data(struct smu_context *smu, case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 5): default: - if (((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 5)) && smu_version > 0x00351F00) || - ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 0)) && smu_version > 0x002A3B00)) + if (((amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 5)) && + smu_version > 0x00351F00) || + ((amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 0)) && + smu_version > 0x002A3B00)) ret = navi10_get_smu_metrics_data(smu, member, value); else ret = navi10_get_legacy_smu_metrics_data(smu, member, value); @@ -1712,7 +1716,7 @@ static int navi10_populate_umd_state_clk(struct smu_context *smu) uint32_t sclk_freq; pstate_table->gfxclk_pstate.min = gfx_table->min; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 0): switch (adev->pdev->revision) { case 0xf0: /* XTX */ @@ -2754,8 +2758,8 @@ static bool navi10_need_umc_cdr_workaround(struct smu_context *smu) if (!smu_cmn_feature_is_enabled(smu, SMU_FEATURE_DPM_UCLK_BIT)) return false; - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 0) || - adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 5)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 0) || + amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 5)) return true; return false; @@ -2863,8 +2867,10 @@ static int navi10_run_umc_cdr_workaround(struct smu_context *smu) * - PPSMC_MSG_SetDriverDummyTableDramAddrLow * - PPSMC_MSG_GetUMCFWWA */ - if (((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 0)) && (pmfw_version >= 0x2a3500)) || - ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 5)) && (pmfw_version >= 0x351D00))) { + if (((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 0)) && + (pmfw_version >= 0x2a3500)) || + ((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 5)) && + (pmfw_version >= 0x351D00))) { ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_GET_UMC_FW_WA, 0, @@ -2883,13 +2889,15 @@ static int navi10_run_umc_cdr_workaround(struct smu_context *smu) return 0; if (umc_fw_disable_cdr) { - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 0)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 0)) return navi10_umc_hybrid_cdr_workaround(smu); } else { return navi10_set_dummy_pstates_table_location(smu); } } else { - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 0)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 0)) return navi10_umc_hybrid_cdr_workaround(smu); } @@ -3356,7 +3364,7 @@ static ssize_t navi1x_get_gpu_metrics(struct smu_context *smu, return ret; } - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 9): if (smu_version > 0x00341C00) ret = navi12_get_gpu_metrics(smu, table); @@ -3366,8 +3374,12 @@ static ssize_t navi1x_get_gpu_metrics(struct smu_context *smu, case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 5): default: - if (((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 5)) && smu_version > 0x00351F00) || - ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 0)) && smu_version > 0x002A3B00)) + if (((amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 5)) && + smu_version > 0x00351F00) || + ((amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 0)) && + smu_version > 0x002A3B00)) ret = navi10_get_gpu_metrics(smu, table); else ret = navi10_get_legacy_gpu_metrics(smu, table); @@ -3385,7 +3397,7 @@ static int navi10_enable_mgpu_fan_boost(struct smu_context *smu) uint32_t param = 0; /* Navi12 does not support this */ - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 9)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 9)) return 0; /* diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c index 4bb289f9b4b8..164c2264027d 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c @@ -73,12 +73,16 @@ #define SMU_11_0_7_GFX_BUSY_THRESHOLD 15 -#define GET_PPTABLE_MEMBER(field, member) do {\ - if (smu->adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 13))\ - (*member) = (smu->smu_table.driver_pptable + offsetof(PPTable_beige_goby_t, field));\ - else\ - (*member) = (smu->smu_table.driver_pptable + offsetof(PPTable_t, field));\ -} while(0) +#define GET_PPTABLE_MEMBER(field, member) \ + do { \ + if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == \ + IP_VERSION(11, 0, 13)) \ + (*member) = (smu->smu_table.driver_pptable + \ + offsetof(PPTable_beige_goby_t, field)); \ + else \ + (*member) = (smu->smu_table.driver_pptable + \ + offsetof(PPTable_t, field)); \ + } while (0) /* STB FIFO depth is in 64bit units */ #define SIENNA_CICHLID_STB_DEPTH_UNIT_BYTES 8 @@ -91,7 +95,7 @@ static int get_table_size(struct smu_context *smu) { - if (smu->adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 13)) + if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 13)) return sizeof(PPTable_beige_goby_t); else return sizeof(PPTable_t); @@ -309,7 +313,7 @@ sienna_cichlid_get_allowed_feature_mask(struct smu_context *smu, } if ((adev->pm.pp_feature & PP_GFX_DCS_MASK) && - (adev->ip_versions[MP1_HWIP][0] > IP_VERSION(11, 0, 7)) && + (amdgpu_ip_version(adev, MP1_HWIP, 0) > IP_VERSION(11, 0, 7)) && !(adev->flags & AMD_IS_APU)) *(uint64_t *)feature_mask |= FEATURE_MASK(FEATURE_GFX_DCS_BIT); @@ -434,7 +438,7 @@ static int sienna_cichlid_append_powerplay_table(struct smu_context *smu) PPTable_beige_goby_t *ppt_beige_goby; PPTable_t *ppt; - if (smu->adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 13)) + if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 13)) ppt_beige_goby = smu->smu_table.driver_pptable; else ppt = smu->smu_table.driver_pptable; @@ -447,7 +451,7 @@ static int sienna_cichlid_append_powerplay_table(struct smu_context *smu) if (ret) return ret; - if (smu->adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 13)) + if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 13)) smu_memcpy_trailing(ppt_beige_goby, I2cControllers, BoardReserved, smc_dpm_table, I2cControllers); else @@ -725,7 +729,7 @@ static int sienna_cichlid_get_smu_metrics_data(struct smu_context *smu, uint32_t apu_percent = 0; uint32_t dgpu_percent = 0; - switch (smu->adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(smu->adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 7): if (smu->smc_fw_version >= 0x3A4900) use_metrics_v3 = true; @@ -1385,8 +1389,9 @@ static int sienna_cichlid_print_clk_levels(struct smu_context *smu, * and onwards SMU firmwares. */ smu_cmn_get_smc_version(smu, NULL, &smu_version); - if ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 7)) && - (smu_version < 0x003a2900)) + if ((amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 7)) && + (smu_version < 0x003a2900)) break; size += sysfs_emit_at(buf, size, "OD_VDDGFX_OFFSET:\n"); @@ -1494,7 +1499,7 @@ static int sienna_cichlid_populate_umd_state_clk(struct smu_context *smu) pstate_table->socclk_pstate.min = soc_table->min; pstate_table->socclk_pstate.peak = soc_table->max; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 7): case IP_VERSION(11, 0, 11): pstate_table->gfxclk_pstate.standard = SIENNA_CICHLID_UMD_PSTATE_PROFILING_GFXCLK; @@ -1945,7 +1950,8 @@ static int sienna_cichlid_read_sensor(struct smu_context *smu, *size = 4; break; case AMDGPU_PP_SENSOR_SS_APU_SHARE: - if (adev->ip_versions[MP1_HWIP][0] != IP_VERSION(11, 0, 7)) { + if (amdgpu_ip_version(adev, MP1_HWIP, 0) != + IP_VERSION(11, 0, 7)) { ret = sienna_cichlid_get_smu_metrics_data(smu, METRICS_SS_APU_SHARE, (uint32_t *)data); *size = 4; @@ -1954,7 +1960,8 @@ static int sienna_cichlid_read_sensor(struct smu_context *smu, } break; case AMDGPU_PP_SENSOR_SS_DGPU_SHARE: - if (adev->ip_versions[MP1_HWIP][0] != IP_VERSION(11, 0, 7)) { + if (amdgpu_ip_version(adev, MP1_HWIP, 0) != + IP_VERSION(11, 0, 7)) { ret = sienna_cichlid_get_smu_metrics_data(smu, METRICS_SS_DGPU_SHARE, (uint32_t *)data); *size = 4; @@ -1978,7 +1985,7 @@ static void sienna_cichlid_get_unique_id(struct smu_context *smu) /* Only supported as of version 0.58.83.0 and only on Sienna Cichlid */ if (smu->smc_fw_version < 0x3A5300 || - smu->adev->ip_versions[MP1_HWIP][0] != IP_VERSION(11, 0, 7)) + amdgpu_ip_version(smu->adev, MP1_HWIP, 0) != IP_VERSION(11, 0, 7)) return; if (sienna_cichlid_get_smu_metrics_data(smu, METRICS_UNIQUE_ID_UPPER32, &upper32)) @@ -2148,8 +2155,8 @@ static void sienna_cichlid_dump_od_table(struct smu_context *smu, od_table->UclkFmax); smu_cmn_get_smc_version(smu, NULL, &smu_version); - if (!((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 7)) && - (smu_version < 0x003a2900))) + if (!((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 7)) && + (smu_version < 0x003a2900))) dev_dbg(smu->adev->dev, "OD: VddGfxOffset: %d\n", od_table->VddGfxOffset); } @@ -2381,8 +2388,9 @@ static int sienna_cichlid_od_edit_dpm_table(struct smu_context *smu, * and onwards SMU firmwares. */ smu_cmn_get_smc_version(smu, NULL, &smu_version); - if ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 7)) && - (smu_version < 0x003a2900)) { + if ((amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 7)) && + (smu_version < 0x003a2900)) { dev_err(smu->adev->dev, "OD GFX Voltage offset functionality is supported " "only by 58.41.0 and onwards SMU firmwares!\n"); return -EOPNOTSUPP; @@ -3105,7 +3113,8 @@ static void sienna_cichlid_dump_pptable(struct smu_context *smu) PPTable_t *pptable = table_context->driver_pptable; int i; - if (smu->adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 13)) { + if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 13)) { beige_goby_dump_pptable(smu); return; } @@ -3910,7 +3919,7 @@ static ssize_t sienna_cichlid_get_gpu_metrics(struct smu_context *smu, uint16_t average_gfx_activity; int ret = 0; - switch (smu->adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(smu->adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 7): if (smu->smc_fw_version >= 0x3A4900) use_metrics_v3 = true; @@ -4026,8 +4035,10 @@ static ssize_t sienna_cichlid_get_gpu_metrics(struct smu_context *smu, gpu_metrics->current_fan_speed = use_metrics_v3 ? metrics_v3->CurrFanSpeed : use_metrics_v2 ? metrics_v2->CurrFanSpeed : metrics->CurrFanSpeed; - if (((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 7)) && smu->smc_fw_version > 0x003A1E00) || - ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 11)) && smu->smc_fw_version > 0x00410400)) { + if (((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 7)) && + smu->smc_fw_version > 0x003A1E00) || + ((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 11)) && + smu->smc_fw_version > 0x00410400)) { gpu_metrics->pcie_link_width = use_metrics_v3 ? metrics_v3->PcieWidth : use_metrics_v2 ? metrics_v2->PcieWidth : metrics->PcieWidth; gpu_metrics->pcie_link_speed = link_speed[use_metrics_v3 ? metrics_v3->PcieRate : @@ -4253,7 +4264,7 @@ static int sienna_cichlid_get_default_config_table_settings(struct smu_context * table->gfx_activity_average_tau = 10; table->mem_activity_average_tau = 10; table->socket_power_average_tau = 100; - if (adev->ip_versions[MP1_HWIP][0] != IP_VERSION(11, 0, 7)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) != IP_VERSION(11, 0, 7)) table->apu_socket_power_average_tau = 100; return 0; diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c index aa4a5498a12f..ece43b41141c 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c @@ -101,8 +101,8 @@ int smu_v11_0_init_microcode(struct smu_context *smu) struct amdgpu_firmware_info *ucode = NULL; if (amdgpu_sriov_vf(adev) && - ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 9)) || - (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 7)))) + ((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 9)) || + (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 7)))) return 0; amdgpu_ucode_ip_version_decode(adev, MP1_HWIP, ucode_prefix, sizeof(ucode_prefix)); @@ -213,7 +213,7 @@ int smu_v11_0_check_fw_version(struct smu_context *smu) if (smu->is_apu) adev->pm.fw_version = smu_version; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 0): smu->smc_driver_if_version = SMU11_DRIVER_IF_VERSION_NV10; break; @@ -246,7 +246,7 @@ int smu_v11_0_check_fw_version(struct smu_context *smu) break; default: dev_err(smu->adev->dev, "smu unsupported IP version: 0x%x.\n", - adev->ip_versions[MP1_HWIP][0]); + amdgpu_ip_version(adev, MP1_HWIP, 0)); smu->smc_driver_if_version = SMU11_DRIVER_IF_VERSION_INV; break; } @@ -474,9 +474,10 @@ int smu_v11_0_init_power(struct smu_context *smu) { struct amdgpu_device *adev = smu->adev; struct smu_power_context *smu_power = &smu->smu_power; - size_t size = adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 5, 0) ? - sizeof(struct smu_11_5_power_context) : - sizeof(struct smu_11_0_power_context); + size_t size = amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 5, 0) ? + sizeof(struct smu_11_5_power_context) : + sizeof(struct smu_11_0_power_context); smu_power->power_context = kzalloc(size, GFP_KERNEL); if (!smu_power->power_context) @@ -731,10 +732,10 @@ int smu_v11_0_init_display_count(struct smu_context *smu, uint32_t count) /* Navy_Flounder/Dimgrey_Cavefish do not support to change * display num currently */ - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 11) || - adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 5, 0) || - adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 12) || - adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 13)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 11) || + amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 5, 0) || + amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 12) || + amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 13)) return 0; return smu_cmn_send_smc_msg_with_param(smu, @@ -1103,7 +1104,7 @@ int smu_v11_0_gfx_off_control(struct smu_context *smu, bool enable) int ret = 0; struct amdgpu_device *adev = smu->adev; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 5): case IP_VERSION(11, 0, 9): @@ -1591,7 +1592,7 @@ int smu_v11_0_baco_set_state(struct smu_context *smu, enum smu_baco_state state) return 0; if (state == SMU_BACO_STATE_ENTER) { - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(11, 0, 7): case IP_VERSION(11, 0, 11): case IP_VERSION(11, 0, 12): @@ -1610,7 +1611,8 @@ int smu_v11_0_baco_set_state(struct smu_context *smu, enum smu_baco_state state) default: if (!ras || !adev->ras_enabled || adev->gmc.xgmi.pending_reset) { - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 2)) { + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(11, 0, 2)) { data = RREG32_SOC15(THM, 0, mmTHM_BACO_CNTL_ARCT); data |= 0x80000000; WREG32_SOC15(THM, 0, mmTHM_BACO_CNTL_ARCT, data); @@ -1894,7 +1896,7 @@ int smu_v11_0_set_performance_level(struct smu_context *smu, * Separate MCLK and SOCCLK soft min/max settings are not allowed * on Arcturus. */ - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 2)) { + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 0, 2)) { mclk_min = mclk_max = 0; socclk_min = socclk_max = 0; } diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c index c8119491c516..8908bbb3ff1f 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c @@ -1198,8 +1198,12 @@ static int renoir_get_smu_metrics_data(struct smu_context *smu, *value = metrics->AverageUvdActivity / 100; break; case METRICS_CURR_SOCKETPOWER: - if (((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(12, 0, 1)) && (adev->pm.fw_version >= 0x40000f)) || - ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(12, 0, 0)) && (adev->pm.fw_version >= 0x373200))) + if (((amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(12, 0, 1)) && + (adev->pm.fw_version >= 0x40000f)) || + ((amdgpu_ip_version(adev, MP1_HWIP, 0) == + IP_VERSION(12, 0, 0)) && + (adev->pm.fw_version >= 0x373200))) *value = metrics->CurrentSocketPower << 8; else *value = (metrics->CurrentSocketPower << 8) / 1000; diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c index 0232adb95df3..8d1e39589057 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c @@ -196,9 +196,9 @@ int smu_v13_0_init_pptable_microcode(struct smu_context *smu) if (!adev->scpm_enabled) return 0; - if ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 7)) || - (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 0)) || - (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 10))) + if ((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 7)) || + (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 0)) || + (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 10))) return 0; /* override pptable_id from driver parameter */ @@ -234,7 +234,7 @@ int smu_v13_0_check_fw_status(struct smu_context *smu) struct amdgpu_device *adev = smu->adev; uint32_t mp1_fw_flags; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 4): case IP_VERSION(13, 0, 11): mp1_fw_flags = RREG32_PCIE(MP1_Public | @@ -269,7 +269,7 @@ int smu_v13_0_check_fw_version(struct smu_context *smu) smu_minor = (smu_version >> 8) & 0xff; smu_debug = (smu_version >> 0) & 0xff; if (smu->is_apu || - adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 6)) + amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 6)) adev->pm.fw_version = smu_version; /* only for dGPU w/ SMU13*/ @@ -802,7 +802,7 @@ int smu_v13_0_gfx_off_control(struct smu_context *smu, bool enable) int ret = 0; struct amdgpu_device *adev = smu->adev; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 0): case IP_VERSION(13, 0, 1): case IP_VERSION(13, 0, 3): @@ -1780,7 +1780,7 @@ int smu_v13_0_set_performance_level(struct smu_context *smu, * Unset those settings for SMU 13.0.2. As soft limits settings * for those clock domains are not supported. */ - if (smu->adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2)) { + if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 2)) { mclk_min = mclk_max = 0; socclk_min = socclk_max = 0; vclk_min = vclk_max = 0; @@ -1927,7 +1927,7 @@ static int smu_v13_0_get_dpm_level_count(struct smu_context *smu, ret = smu_v13_0_get_dpm_freq_by_index(smu, clk_type, 0xff, value); /* SMU v13.0.2 FW returns 0 based max level, increment by one for it */ - if ((smu->adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2)) && (!ret && value)) + if ((amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 2)) && (!ret && value)) ++(*value); return ret; @@ -1987,7 +1987,7 @@ int smu_v13_0_set_single_dpm_table(struct smu_context *smu, return ret; } - if (smu->adev->ip_versions[MP1_HWIP][0] != IP_VERSION(13, 0, 2)) { + if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) != IP_VERSION(13, 0, 2)) { ret = smu_v13_0_get_fine_grained_status(smu, clk_type, &single_dpm_table->is_fine_grained); diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c index 9ac2be5627d9..0bb9f1426254 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c @@ -2754,7 +2754,7 @@ static int smu_v13_0_0_mode1_reset(struct smu_context *smu) uint32_t param; struct amdgpu_device *adev = smu->adev; - switch (adev->ip_versions[MP1_HWIP][0]) { + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 0): /* SMU 13_0_0 PMFW supports RAS fatal error reset from 78.77 */ smu_v13_0_0_set_mode1_reset_param(smu, 0x004e4d00, ¶m); @@ -2787,7 +2787,7 @@ static int smu_v13_0_0_mode2_reset(struct smu_context *smu) int ret; struct amdgpu_device *adev = smu->adev; - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 10)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 10)) ret = smu_cmn_send_smc_msg(smu, SMU_MSG_Mode2Reset, NULL); else return -EOPNOTSUPP; @@ -2799,7 +2799,7 @@ static int smu_v13_0_0_enable_gfx_features(struct smu_context *smu) { struct amdgpu_device *adev = smu->adev; - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 10)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 10)) return smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_EnableAllSmuFeatures, FEATURE_PWR_GFX, NULL); else @@ -2863,7 +2863,7 @@ static int smu_v13_0_0_check_ecc_table_support(struct smu_context *smu) if (ret) return -EOPNOTSUPP; - if ((adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 10)) && + if ((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 10)) && (smu_version >= SUPPORT_ECCTABLE_SMU_13_0_10_VERSION)) return ret; else diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c index 626591f54bc4..bb98156b2fa1 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c @@ -1144,7 +1144,7 @@ void smu_v13_0_4_set_ppt_funcs(struct smu_context *smu) smu->smc_driver_if_version = SMU13_0_4_DRIVER_IF_VERSION; smu->is_apu = true; - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 4)) + if (amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 4)) smu_v13_0_4_set_smu_mailbox_registers(smu); else smu_v13_0_set_smu_mailbox_registers(smu); diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/yellow_carp_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/yellow_carp_ppt.c index 2e74d749efdd..2d1736234b4a 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/yellow_carp_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/yellow_carp_ppt.c @@ -1024,24 +1024,24 @@ static uint32_t yellow_carp_get_umd_pstate_clk_default(struct smu_context *smu, switch (clk_type) { case SMU_GFXCLK: case SMU_SCLK: - if ((adev->ip_versions[MP1_HWIP][0]) == IP_VERSION(13, 0, 8)) + if ((amdgpu_ip_version(adev, MP1_HWIP, 0)) == IP_VERSION(13, 0, 8)) clk_limit = SMU_13_0_8_UMD_PSTATE_GFXCLK; - if ((adev->ip_versions[MP1_HWIP][0]) == IP_VERSION(13, 0, 1) || - (adev->ip_versions[MP1_HWIP][0]) == IP_VERSION(13, 0, 3)) + if ((amdgpu_ip_version(adev, MP1_HWIP, 0)) == IP_VERSION(13, 0, 1) || + (amdgpu_ip_version(adev, MP1_HWIP, 0)) == IP_VERSION(13, 0, 3)) clk_limit = SMU_13_0_1_UMD_PSTATE_GFXCLK; break; case SMU_SOCCLK: - if ((adev->ip_versions[MP1_HWIP][0]) == IP_VERSION(13, 0, 8)) + if ((amdgpu_ip_version(adev, MP1_HWIP, 0)) == IP_VERSION(13, 0, 8)) clk_limit = SMU_13_0_8_UMD_PSTATE_SOCCLK; - if ((adev->ip_versions[MP1_HWIP][0]) == IP_VERSION(13, 0, 1) || - (adev->ip_versions[MP1_HWIP][0]) == IP_VERSION(13, 0, 3)) + if ((amdgpu_ip_version(adev, MP1_HWIP, 0)) == IP_VERSION(13, 0, 1) || + (amdgpu_ip_version(adev, MP1_HWIP, 0)) == IP_VERSION(13, 0, 3)) clk_limit = SMU_13_0_1_UMD_PSTATE_SOCCLK; break; case SMU_FCLK: - if ((adev->ip_versions[MP1_HWIP][0]) == IP_VERSION(13, 0, 8)) + if ((amdgpu_ip_version(adev, MP1_HWIP, 0)) == IP_VERSION(13, 0, 8)) clk_limit = SMU_13_0_8_UMD_PSTATE_FCLK; - if ((adev->ip_versions[MP1_HWIP][0]) == IP_VERSION(13, 0, 1) || - (adev->ip_versions[MP1_HWIP][0]) == IP_VERSION(13, 0, 3)) + if ((amdgpu_ip_version(adev, MP1_HWIP, 0)) == IP_VERSION(13, 0, 1) || + (amdgpu_ip_version(adev, MP1_HWIP, 0)) == IP_VERSION(13, 0, 3)) clk_limit = SMU_13_0_1_UMD_PSTATE_FCLK; break; default: -- cgit From 4051844c6616c0bf33dd16342967ea5b931746ae Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Wed, 6 Sep 2023 12:37:14 +0800 Subject: drm/amdgpu: add amdgpu mca debug sysfs support add amdgpu mca debug sysfs support. Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c | 116 ++++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h | 2 + drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 2 + 3 files changed, 120 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c index a27cd57b35e9..5a828c175e3a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c @@ -212,3 +212,119 @@ int amdgpu_mca_smu_get_mca_entry(struct amdgpu_device *adev, enum amdgpu_mca_err return -EOPNOTSUPP; } +#if defined(CONFIG_DEBUG_FS) +static int amdgpu_mca_smu_debug_mode_set(void *data, u64 val) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)data; + int ret; + + ret = amdgpu_mca_smu_set_debug_mode(adev, val ? true : false); + if (ret) + return ret; + + dev_info(adev->dev, "amdgpu set smu mca debug mode %s success\n", val ? "on" : "off"); + + return 0; +} + +static void mca_dump_entry(struct seq_file *m, struct mca_bank_entry *entry) +{ + int i, idx = entry->idx; + + seq_printf(m, "mca entry[%d].type: %s\n", idx, entry->type == AMDGPU_MCA_ERROR_TYPE_UE ? "UE" : "CE"); + seq_printf(m, "mca entry[%d].ip: %d\n", idx, entry->ip); + seq_printf(m, "mca entry[%d].info: socketid:%d aid:%d hwid:0x%03x mcatype:0x%04x\n", + idx, entry->info.socket_id, entry->info.aid, entry->info.hwid, entry->info.mcatype); + + for (i = 0; i < ARRAY_SIZE(entry->regs); i++) + seq_printf(m, "mca entry[%d].regs[%d]: 0x%016llx\n", idx, i, entry->regs[i]); +} + +static int mca_dump_show(struct seq_file *m, enum amdgpu_mca_error_type type) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct mca_bank_entry *entry; + uint32_t count = 0; + int i, ret; + + ret = amdgpu_mca_smu_get_valid_mca_count(adev, type, &count); + if (ret) + return ret; + + seq_printf(m, "amdgpu smu %s valid mca count: %d\n", + type == AMDGPU_MCA_ERROR_TYPE_UE ? "UE" : "CE", count); + + if (!count) + return 0; + + entry = kmalloc(sizeof(*entry), GFP_KERNEL); + if (!entry) + return -ENOMEM; + + for (i = 0; i < count; i++) { + memset(entry, 0, sizeof(*entry)); + + ret = amdgpu_mca_smu_get_mca_entry(adev, type, i, entry); + if (ret) + goto err_free_entry; + + mca_dump_entry(m, entry); + } + +err_free_entry: + kfree(entry); + + return ret; +} + +static int mca_dump_ce_show(struct seq_file *m, void *unused) +{ + return mca_dump_show(m, AMDGPU_MCA_ERROR_TYPE_CE); +} + +static int mca_dump_ce_open(struct inode *inode, struct file *file) +{ + return single_open(file, mca_dump_ce_show, inode->i_private); +} + +static const struct file_operations mca_ce_dump_debug_fops = { + .owner = THIS_MODULE, + .open = mca_dump_ce_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static int mca_dump_ue_show(struct seq_file *m, void *unused) +{ + return mca_dump_show(m, AMDGPU_MCA_ERROR_TYPE_UE); +} + +static int mca_dump_ue_open(struct inode *inode, struct file *file) +{ + return single_open(file, mca_dump_ue_show, inode->i_private); +} + +static const struct file_operations mca_ue_dump_debug_fops = { + .owner = THIS_MODULE, + .open = mca_dump_ue_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +DEFINE_DEBUGFS_ATTRIBUTE(mca_debug_mode_fops, NULL, amdgpu_mca_smu_debug_mode_set, "%llu\n"); +#endif + +void amdgpu_mca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root) +{ +#if defined(CONFIG_DEBUG_FS) + if (!root || adev->ip_versions[MP1_HWIP][0] != IP_VERSION(13, 0, 6)) + return; + + debugfs_create_file("mca_debug_mode", 0200, root, adev, &mca_debug_mode_fops); + debugfs_create_file("mca_ue_dump", 0400, root, adev, &mca_ue_dump_debug_fops); + debugfs_create_file("mca_ce_dump", 0400, root, adev, &mca_ce_dump_debug_fops); +#endif +} + diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h index be3189bad97f..28ad463cf5c9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h @@ -112,4 +112,6 @@ int amdgpu_mca_smu_get_error_count(struct amdgpu_device *adev, enum amdgpu_ras_b int amdgpu_mca_smu_get_mca_entry(struct amdgpu_device *adev, enum amdgpu_mca_error_type type, int idx, struct mca_bank_entry *entry); +void amdgpu_mca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index eeb695f9ff1c..4e4ba2149595 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1570,6 +1570,8 @@ void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev) amdgpu_ras_debugfs_create(adev, &fs_info, dir); } } + + amdgpu_mca_smu_debugfs_init(adev, dir); } /* debugfs end */ -- cgit From 5838f74c29ecc4eac397a2374cf208efcf095563 Mon Sep 17 00:00:00 2001 From: Cong Liu Date: Thu, 14 Sep 2023 17:45:33 +0800 Subject: drm/amdgpu: fix a memory leak in amdgpu_ras_feature_enable This patch fixes a memory leak in the amdgpu_ras_feature_enable() function. The leak occurs when the function sends a command to the firmware to enable or disable a RAS feature for a GFX block. If the command fails, the kfree() function is not called to free the info memory. Fixes: 9f051d6ff13f ("drm/amdgpu: Free ras cmd input buffer properly") Reviewed-by: Hawking Zhang Signed-off-by: Cong Liu 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 4e4ba2149595..62dca5d80fb6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -802,6 +802,7 @@ int amdgpu_ras_feature_enable(struct amdgpu_device *adev, enable ? "enable":"disable", get_ras_block_str(head), amdgpu_ras_is_poison_mode_supported(adev), ret); + kfree(info); return ret; } -- cgit From a83f2bf1f4d95582cadb0136f0e1bd5b398b5c90 Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Fri, 15 Sep 2023 18:44:17 +0800 Subject: drm/amdgpu: Fix false positive error log It should first check block ras obj whether be set, it should return 0 directly if block ras obj or hw_ops is not set. If block doesn't support RAS just return 0 is fine. Changed from V1: return 0 directly if block ras obj or hw ops is not set Signed-off-by: Stanley.Yang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 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 62dca5d80fb6..9c8203e87859 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1102,15 +1102,15 @@ int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, { struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0); - if (!amdgpu_ras_is_supported(adev, block)) - return -EINVAL; - - if (!block_obj || !block_obj->hw_ops) { + if (!block_obj || !block_obj->hw_ops) { dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", ras_block_str(block)); - return -EINVAL; + return 0; } + if (!amdgpu_ras_is_supported(adev, block)) + return 0; + if (block_obj->hw_ops->reset_ras_error_count) block_obj->hw_ops->reset_ras_error_count(adev); -- cgit From 625e5f3851b78ddcf1bbcfb74ea39edcdaf07d3d Mon Sep 17 00:00:00 2001 From: Asad Kamal Date: Mon, 2 Oct 2023 14:41:00 +0800 Subject: drm/amdgpu: Expose ras version & schema info Expose ras table version & schema info to sysfs v2: Updated schema to get poison support info from ras context, removed asic specific checks Signed-off-by: Asad Kamal Reviewed-by: Lijo Lazar Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 51 +++++++++++++++++++++++++++++++-- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 3 ++ 2 files changed, 51 insertions(+), 3 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 9c8203e87859..5fb57419ef77 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1370,6 +1370,22 @@ static ssize_t amdgpu_ras_sysfs_features_read(struct device *dev, return sysfs_emit(buf, "feature mask: 0x%x\n", con->features); } +static ssize_t amdgpu_ras_sysfs_version_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct amdgpu_ras *con = + container_of(attr, struct amdgpu_ras, version_attr); + return sysfs_emit(buf, "table version: 0x%x\n", con->eeprom_control.tbl_hdr.version); +} + +static ssize_t amdgpu_ras_sysfs_schema_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct amdgpu_ras *con = + container_of(attr, struct amdgpu_ras, schema_attr); + return sysfs_emit(buf, "schema: 0x%x\n", con->schema); +} + static void amdgpu_ras_sysfs_remove_bad_page_node(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); @@ -1379,11 +1395,13 @@ static void amdgpu_ras_sysfs_remove_bad_page_node(struct amdgpu_device *adev) RAS_FS_NAME); } -static int amdgpu_ras_sysfs_remove_feature_node(struct amdgpu_device *adev) +static int amdgpu_ras_sysfs_remove_dev_attr_node(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); struct attribute *attrs[] = { &con->features_attr.attr, + &con->version_attr.attr, + &con->schema_attr.attr, NULL }; struct attribute_group group = { @@ -1459,7 +1477,7 @@ static int amdgpu_ras_sysfs_remove_all(struct amdgpu_device *adev) if (amdgpu_bad_page_threshold != 0) amdgpu_ras_sysfs_remove_bad_page_node(adev); - amdgpu_ras_sysfs_remove_feature_node(adev); + amdgpu_ras_sysfs_remove_dev_attr_node(adev); return 0; } @@ -1582,6 +1600,10 @@ static BIN_ATTR(gpu_vram_bad_pages, S_IRUGO, amdgpu_ras_sysfs_badpages_read, NULL, 0); static DEVICE_ATTR(features, S_IRUGO, amdgpu_ras_sysfs_features_read, NULL); +static DEVICE_ATTR(version, 0444, + amdgpu_ras_sysfs_version_show, NULL); +static DEVICE_ATTR(schema, 0444, + amdgpu_ras_sysfs_schema_show, NULL); static int amdgpu_ras_fs_init(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); @@ -1590,6 +1612,8 @@ static int amdgpu_ras_fs_init(struct amdgpu_device *adev) }; struct attribute *attrs[] = { &con->features_attr.attr, + &con->version_attr.attr, + &con->schema_attr.attr, NULL }; struct bin_attribute *bin_attrs[] = { @@ -1598,11 +1622,20 @@ static int amdgpu_ras_fs_init(struct amdgpu_device *adev) }; int r; + group.attrs = attrs; + /* add features entry */ con->features_attr = dev_attr_features; - group.attrs = attrs; sysfs_attr_init(attrs[0]); + /* add version entry */ + con->version_attr = dev_attr_version; + sysfs_attr_init(attrs[1]); + + /* add schema entry */ + con->schema_attr = dev_attr_schema; + sysfs_attr_init(attrs[2]); + if (amdgpu_bad_page_threshold != 0) { /* add bad_page_features entry */ bin_attr_gpu_vram_bad_pages.private = NULL; @@ -2594,6 +2627,14 @@ static void amdgpu_ras_query_poison_mode(struct amdgpu_device *adev) } } +static int amdgpu_get_ras_schema(struct amdgpu_device *adev) +{ + return amdgpu_ras_is_poison_mode_supported(adev) ? AMDGPU_RAS_ERROR__POISON : 0 | + AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE | + AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE | + AMDGPU_RAS_ERROR__PARITY; +} + int amdgpu_ras_init(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); @@ -2636,6 +2677,7 @@ int amdgpu_ras_init(struct amdgpu_device *adev) con->update_channel_flag = false; con->features = 0; + con->schema = 0; INIT_LIST_HEAD(&con->head); /* Might need get this flag from vbios. */ con->flags = RAS_DEFAULT_FLAGS; @@ -2691,6 +2733,9 @@ int amdgpu_ras_init(struct amdgpu_device *adev) amdgpu_ras_query_poison_mode(adev); + /* Get RAS schema for particular SOC */ + con->schema = amdgpu_get_ras_schema(adev); + if (amdgpu_ras_fs_init(adev)) { r = -EINVAL; goto release_con; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 7999d202c9bc..728f98c6fc1c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -389,9 +389,12 @@ struct amdgpu_ras { /* ras infrastructure */ /* for ras itself. */ uint32_t features; + uint32_t schema; struct list_head head; /* sysfs */ struct device_attribute features_attr; + struct device_attribute version_attr; + struct device_attribute schema_attr; struct bin_attribute badpages_attr; struct dentry *de_ras_eeprom_table; /* block array */ -- cgit From 5b1270beb3801d328b43577a8bb1152d435bb146 Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Mon, 25 Sep 2023 19:30:26 +0800 Subject: drm/amdgpu: add ras_err_info to identify RAS error source introduced "ras_err_info" to better identify a RAS ERROR source. NOTE: For legacy chips, keep the original RAS error print format. v1: RAS errors may come from different dies during a RAS error query, therefore, need a new data structure to identify the source of RAS ERROR. v2: - use new data structure 'amdgpu_smuio_mcm_config_info' instead of ras_err_id (in v1 patch) - refine ras error dump function name - refine ras error dump log format Signed-off-by: Yang Wang Reviewed-by: Tao Zhou Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 294 +++++++++++++++++++++++++----- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 25 +++ drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h | 5 + drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 27 ++- drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c | 7 +- drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c | 7 +- 6 files changed, 312 insertions(+), 53 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 5fb57419ef77..973073e07b2a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -152,8 +152,9 @@ static bool amdgpu_ras_get_error_query_ready(struct amdgpu_device *adev) static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t address) { - struct ras_err_data err_data = {0, 0, 0, NULL}; + struct ras_err_data err_data; struct eeprom_table_record err_rec; + int ret; if ((address >= adev->gmc.mc_vram_size) || (address >= RAS_UMC_INJECT_ADDR_LIMIT)) { @@ -170,6 +171,10 @@ static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t addre return 0; } + ret = amdgpu_ras_error_data_init(&err_data); + if (ret) + return ret; + memset(&err_rec, 0x0, sizeof(struct eeprom_table_record)); err_data.err_addr = &err_rec; amdgpu_umc_fill_error_record(&err_data, address, address, 0, 0); @@ -180,6 +185,8 @@ static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t addre amdgpu_ras_save_bad_pages(adev, NULL); } + amdgpu_ras_error_data_fini(&err_data); + 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"); @@ -1015,17 +1022,118 @@ static void amdgpu_ras_get_ecc_info(struct amdgpu_device *adev, struct ras_err_d } } +static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, + struct ras_query_if *query_if, + struct ras_err_data *err_data, + bool is_ue) +{ + struct ras_manager *ras_mgr = amdgpu_ras_find_obj(adev, &query_if->head); + const char *blk_name = get_ras_block_str(&query_if->head); + struct amdgpu_smuio_mcm_config_info *mcm_info; + struct ras_err_node *err_node; + struct ras_err_info *err_info; + + if (is_ue) + dev_info(adev->dev, "%ld uncorrectable hardware errors detected in %s block\n", + ras_mgr->err_data.ue_count, blk_name); + else + dev_info(adev->dev, "%ld correctable hardware errors detected in %s block\n", + ras_mgr->err_data.ue_count, blk_name); + + for_each_ras_error(err_node, err_data) { + err_info = &err_node->err_info; + mcm_info = &err_info->mcm_info; + if (is_ue && err_info->ue_count) { + dev_info(adev->dev, "socket: %d, die: %d " + "%lld uncorrectable hardware errors detected in %s block\n", + mcm_info->socket_id, + mcm_info->die_id, + err_info->ue_count, + blk_name); + } else if (!is_ue && err_info->ce_count) { + dev_info(adev->dev, "socket: %d, die: %d " + "%lld correctable hardware errors detected in %s block\n", + mcm_info->socket_id, + mcm_info->die_id, + err_info->ue_count, + blk_name); + } + } +} + +static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, + struct ras_query_if *query_if, + struct ras_err_data *err_data) +{ + struct ras_manager *ras_mgr = amdgpu_ras_find_obj(adev, &query_if->head); + const char *blk_name = get_ras_block_str(&query_if->head); + + if (err_data->ce_count) { + if (!list_empty(&err_data->err_node_list)) { + amdgpu_ras_error_print_error_data(adev, query_if, + err_data, false); + } else if (!adev->aid_mask && + adev->smuio.funcs && + adev->smuio.funcs->get_socket_id && + adev->smuio.funcs->get_die_id) { + dev_info(adev->dev, "socket: %d, die: %d " + "%ld correctable hardware errors " + "detected in %s block, no user " + "action is needed.\n", + adev->smuio.funcs->get_socket_id(adev), + adev->smuio.funcs->get_die_id(adev), + ras_mgr->err_data.ce_count, + blk_name); + } else { + dev_info(adev->dev, "%ld correctable hardware errors " + "detected in %s block, no user " + "action is needed.\n", + ras_mgr->err_data.ce_count, + blk_name); + } + } + + if (err_data->ue_count) { + if (!list_empty(&err_data->err_node_list)) { + amdgpu_ras_error_print_error_data(adev, query_if, + err_data, true); + } else if (!adev->aid_mask && + adev->smuio.funcs && + adev->smuio.funcs->get_socket_id && + adev->smuio.funcs->get_die_id) { + dev_info(adev->dev, "socket: %d, die: %d " + "%ld uncorrectable hardware errors " + "detected in %s block\n", + adev->smuio.funcs->get_socket_id(adev), + adev->smuio.funcs->get_die_id(adev), + ras_mgr->err_data.ue_count, + blk_name); + } else { + dev_info(adev->dev, "%ld uncorrectable hardware errors " + "detected in %s block\n", + ras_mgr->err_data.ue_count, + blk_name); + } + } + +} + /* query/inject/cure begin */ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info) { struct amdgpu_ras_block_object *block_obj = NULL; struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head); - struct ras_err_data err_data = {0, 0, 0, NULL}; + struct ras_err_data err_data; + int ret; if (!obj) return -EINVAL; + ret = amdgpu_ras_error_data_init(&err_data); + if (ret) + return ret; + if (info->head.block == AMDGPU_RAS_BLOCK__UMC) { amdgpu_ras_get_ecc_info(adev, &err_data); } else { @@ -1033,7 +1141,8 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, if (!block_obj || !block_obj->hw_ops) { dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", get_ras_block_str(&info->head)); - return -EINVAL; + ret = -EINVAL; + goto out_fini_err_data; } if (block_obj->hw_ops->query_ras_error_count) @@ -1053,48 +1162,12 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, info->ue_count = obj->err_data.ue_count; info->ce_count = obj->err_data.ce_count; - if (err_data.ce_count) { - if (!adev->aid_mask && - adev->smuio.funcs && - adev->smuio.funcs->get_socket_id && - adev->smuio.funcs->get_die_id) { - dev_info(adev->dev, "socket: %d, die: %d " - "%ld correctable hardware errors " - "detected in %s block, no user " - "action is needed.\n", - adev->smuio.funcs->get_socket_id(adev), - adev->smuio.funcs->get_die_id(adev), - obj->err_data.ce_count, - get_ras_block_str(&info->head)); - } else { - dev_info(adev->dev, "%ld correctable hardware errors " - "detected in %s block, no user " - "action is needed.\n", - obj->err_data.ce_count, - get_ras_block_str(&info->head)); - } - } - if (err_data.ue_count) { - if (!adev->aid_mask && - adev->smuio.funcs && - adev->smuio.funcs->get_socket_id && - adev->smuio.funcs->get_die_id) { - dev_info(adev->dev, "socket: %d, die: %d " - "%ld uncorrectable hardware errors " - "detected in %s block\n", - adev->smuio.funcs->get_socket_id(adev), - adev->smuio.funcs->get_die_id(adev), - obj->err_data.ue_count, - get_ras_block_str(&info->head)); - } else { - dev_info(adev->dev, "%ld uncorrectable hardware errors " - "detected in %s block\n", - obj->err_data.ue_count, - get_ras_block_str(&info->head)); - } - } + amdgpu_ras_error_generate_report(adev, info, &err_data); - return 0; +out_fini_err_data: + amdgpu_ras_error_data_fini(&err_data); + + return ret; } int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, @@ -1744,12 +1817,16 @@ static void amdgpu_ras_interrupt_umc_handler(struct ras_manager *obj, struct amdgpu_iv_entry *entry) { struct ras_ih_data *data = &obj->ih_data; - struct ras_err_data err_data = {0, 0, 0, NULL}; + struct ras_err_data err_data; int ret; if (!data->cb) return; + ret = amdgpu_ras_error_data_init(&err_data); + if (ret) + return; + /* Let IP handle its data, maybe we need get the output * from the callback to update the error type/count, etc */ @@ -1766,6 +1843,8 @@ static void amdgpu_ras_interrupt_umc_handler(struct ras_manager *obj, obj->err_data.ue_count += err_data.ue_count; obj->err_data.ce_count += err_data.ce_count; } + + amdgpu_ras_error_data_fini(&err_data); } static void amdgpu_ras_interrupt_handler(struct ras_manager *obj) @@ -3383,3 +3462,128 @@ void amdgpu_ras_inst_reset_ras_error_count(struct amdgpu_device *adev, WREG32(err_status_hi_offset, 0); } } + +int amdgpu_ras_error_data_init(struct ras_err_data *err_data) +{ + memset(err_data, 0, sizeof(*err_data)); + + INIT_LIST_HEAD(&err_data->err_node_list); + + return 0; +} + +static void amdgpu_ras_error_node_release(struct ras_err_node *err_node) +{ + if (!err_node) + return; + + list_del(&err_node->node); + kvfree(err_node); +} + +void amdgpu_ras_error_data_fini(struct ras_err_data *err_data) +{ + struct ras_err_node *err_node, *tmp; + + list_for_each_entry_safe(err_node, tmp, &err_data->err_node_list, node) { + amdgpu_ras_error_node_release(err_node); + list_del(&err_node->node); + } +} + +static struct ras_err_node *amdgpu_ras_error_find_node_by_id(struct ras_err_data *err_data, + struct amdgpu_smuio_mcm_config_info *mcm_info) +{ + struct ras_err_node *err_node; + struct amdgpu_smuio_mcm_config_info *ref_id; + + if (!err_data || !mcm_info) + return NULL; + + for_each_ras_error(err_node, err_data) { + ref_id = &err_node->err_info.mcm_info; + if ((mcm_info->socket_id >= 0 && mcm_info->socket_id != ref_id->socket_id) || + (mcm_info->die_id >= 0 && mcm_info->die_id != ref_id->die_id)) + continue; + + return err_node; + } + + return NULL; +} + +static struct ras_err_node *amdgpu_ras_error_node_new(void) +{ + struct ras_err_node *err_node; + + err_node = kvzalloc(sizeof(*err_node), GFP_KERNEL); + if (!err_node) + return NULL; + + INIT_LIST_HEAD(&err_node->node); + + return err_node; +} + +static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_data, + struct amdgpu_smuio_mcm_config_info *mcm_info) +{ + struct ras_err_node *err_node; + + err_node = amdgpu_ras_error_find_node_by_id(err_data, mcm_info); + if (err_node) + return &err_node->err_info; + + err_node = amdgpu_ras_error_node_new(); + if (!err_node) + return NULL; + + memcpy(&err_node->err_info.mcm_info, mcm_info, sizeof(*mcm_info)); + + err_data->err_list_count++; + list_add_tail(&err_node->node, &err_data->err_node_list); + + return &err_node->err_info; +} + +int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, + struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count) +{ + struct ras_err_info *err_info; + + if (!err_data || !mcm_info) + return -EINVAL; + + if (!count) + return 0; + + err_info = amdgpu_ras_error_get_info(err_data, mcm_info); + if (!err_info) + return -EINVAL; + + err_info->ue_count += count; + err_data->ue_count += count; + + return 0; +} + +int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, + struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count) +{ + struct ras_err_info *err_info; + + if (!err_data || !mcm_info) + return -EINVAL; + + if (!count) + return 0; + + err_info = amdgpu_ras_error_get_info(err_data, mcm_info); + if (!err_info) + return -EINVAL; + + err_info->ce_count += count; + err_data->ce_count += count; + + return 0; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 46b0dcf846dc..0a5c8a107fb2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -28,6 +28,7 @@ #include #include "ta_ras_if.h" #include "amdgpu_ras_eeprom.h" +#include "amdgpu_smuio.h" struct amdgpu_iv_entry; @@ -443,13 +444,29 @@ struct ras_fs_data { char debugfs_name[32]; }; +struct ras_err_info { + struct amdgpu_smuio_mcm_config_info mcm_info; + u64 ce_count; + u64 ue_count; +}; + +struct ras_err_node { + struct list_head node; + struct ras_err_info err_info; +}; + struct ras_err_data { unsigned long ue_count; unsigned long ce_count; unsigned long err_addr_cnt; struct eeprom_table_record *err_addr; + u32 err_list_count; + struct list_head err_node_list; }; +#define for_each_ras_error(err_node, err_data) \ + list_for_each_entry(err_node, &(err_data)->err_node_list, node) + struct ras_err_handler_data { /* point to bad page records array */ struct eeprom_table_record *bps; @@ -773,4 +790,12 @@ void amdgpu_ras_inst_reset_ras_error_count(struct amdgpu_device *adev, const struct amdgpu_ras_err_status_reg_entry *reg_list, uint32_t reg_list_size, uint32_t instance); + +int amdgpu_ras_error_data_init(struct ras_err_data *err_data); +void amdgpu_ras_error_data_fini(struct ras_err_data *err_data); +int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, + struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count); +int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, + struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count); + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h index 5910d50ac74d..ff4435181055 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_smuio.h @@ -30,6 +30,11 @@ enum amdgpu_pkg_type { AMDGPU_PKG_TYPE_UNKNOWN, }; +struct amdgpu_smuio_mcm_config_info { + int socket_id; + int die_id; +}; + struct amdgpu_smuio_funcs { u32 (*get_rom_index_offset)(struct amdgpu_device *adev); u32 (*get_rom_data_offset)(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index 24fcc9a2e422..f74347cc087a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -45,8 +45,12 @@ static int amdgpu_umc_convert_error_address(struct amdgpu_device *adev, int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev, uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst) { - struct ras_err_data err_data = {0, 0, 0, NULL}; - int ret = AMDGPU_RAS_FAIL; + struct ras_err_data err_data; + int ret; + + ret = amdgpu_ras_error_data_init(&err_data); + if (ret) + return ret; err_data.err_addr = kcalloc(adev->umc.max_ras_err_cnt_per_query, @@ -54,7 +58,8 @@ int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev, if (!err_data.err_addr) { dev_warn(adev->dev, "Failed to alloc memory for umc error record in MCA notifier!\n"); - return AMDGPU_RAS_FAIL; + ret = AMDGPU_RAS_FAIL; + goto out_fini_err_data; } /* @@ -63,7 +68,7 @@ int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev, ret = amdgpu_umc_convert_error_address(adev, &err_data, err_addr, ch_inst, umc_inst); if (ret) - goto out; + goto out_free_err_addr; if (amdgpu_bad_page_threshold != 0) { amdgpu_ras_add_bad_pages(adev, err_data.err_addr, @@ -71,8 +76,12 @@ int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev, amdgpu_ras_save_bad_pages(adev, NULL); } -out: +out_free_err_addr: kfree(err_data.err_addr); + +out_fini_err_data: + amdgpu_ras_error_data_fini(&err_data); + return ret; } @@ -182,18 +191,24 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset) } if (!amdgpu_sriov_vf(adev)) { - struct ras_err_data err_data = {0, 0, 0, NULL}; + struct ras_err_data err_data; struct ras_common_if head = { .block = AMDGPU_RAS_BLOCK__UMC, }; struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head); + ret = amdgpu_ras_error_data_init(&err_data); + if (ret) + return ret; + ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset); if (ret == AMDGPU_RAS_SUCCESS && obj) { obj->err_data.ue_count += err_data.ue_count; obj->err_data.ce_count += err_data.ce_count; } + + amdgpu_ras_error_data_fini(&err_data); } else { if (adev->virt.ops && adev->virt.ops->ras_poison_handler) adev->virt.ops->ras_poison_handler(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c index 7d6d7734dbec..6d24c84924cb 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c @@ -365,9 +365,12 @@ static void nbio_v7_4_handle_ras_controller_intr_no_bifring(struct amdgpu_device { uint32_t bif_doorbell_intr_cntl; struct ras_manager *obj = amdgpu_ras_find_obj(adev, adev->nbio.ras_if); - struct ras_err_data err_data = {0, 0, 0, NULL}; + struct ras_err_data err_data; struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + if (amdgpu_ras_error_data_init(&err_data)) + return; + if (adev->asic_type == CHIP_ALDEBARAN) bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL_ALDE); else @@ -418,6 +421,8 @@ static void nbio_v7_4_handle_ras_controller_intr_no_bifring(struct amdgpu_device */ amdgpu_ras_reset_gpu(adev); } + + amdgpu_ras_error_data_fini(&err_data); } static void nbio_v7_4_handle_ras_err_event_athub_intr_no_bifring(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c index cc034a740605..eccb006e78aa 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c @@ -560,9 +560,12 @@ static void nbio_v7_9_handle_ras_controller_intr_no_bifring(struct amdgpu_device { uint32_t bif_doorbell_intr_cntl; struct ras_manager *obj = amdgpu_ras_find_obj(adev, adev->nbio.ras_if); - struct ras_err_data err_data = {0, 0, 0, NULL}; + struct ras_err_data err_data; struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + if (amdgpu_ras_error_data_init(&err_data)) + return; + bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL); if (REG_GET_FIELD(bif_doorbell_intr_cntl, @@ -607,6 +610,8 @@ static void nbio_v7_9_handle_ras_controller_intr_no_bifring(struct amdgpu_device */ amdgpu_ras_reset_gpu(adev); } + + amdgpu_ras_error_data_fini(&err_data); } static void nbio_v7_9_handle_ras_err_event_athub_intr_no_bifring(struct amdgpu_device *adev) -- cgit From 53dd920c1f471a5763c660a7b94fe0aaf746d357 Mon Sep 17 00:00:00 2001 From: Asad Kamal Date: Thu, 5 Oct 2023 15:40:42 +0800 Subject: drm/amdgpu : Add hive ras recovery check If one of the devices in the hive detects a fatal error, need to send ras recovery reset message to PMFW of all devices in the hive. For that add a flag in hive to indicate that it's undergoing ras recovery Signed-off-by: Asad Kamal Reviewed-by: Hawking Zhang Reviewed-by: Lijo Lazar Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 9 +++++++-- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h | 1 + drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c | 10 +++++++++- 3 files changed, 17 insertions(+), 3 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 973073e07b2a..c7f8dcb3b4d2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2140,9 +2140,11 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) struct amdgpu_device *remote_adev = NULL; struct amdgpu_device *adev = ras->adev; struct list_head device_list, *device_list_handle = NULL; + struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev); + if (hive) + atomic_set(&hive->ras_recovery, 1); if (!ras->disable_ras_err_cnt_harvest) { - struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev); /* Build list of devices to query RAS related errors */ if (hive && adev->gmc.xgmi.num_physical_nodes > 1) { @@ -2159,7 +2161,6 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) amdgpu_ras_log_on_err_counter(remote_adev); } - amdgpu_put_xgmi_hive(hive); } if (amdgpu_device_should_recover_gpu(ras->adev)) { @@ -2194,6 +2195,10 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) amdgpu_device_gpu_recover(ras->adev, NULL, &reset_context); } atomic_set(&ras->in_recovery, 0); + if (hive) { + atomic_set(&hive->ras_recovery, 0); + amdgpu_put_xgmi_hive(hive); + } } /* alloc/realloc bps array */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h index 86fbf56938f4..6cab882e8061 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h @@ -44,6 +44,7 @@ struct amdgpu_hive_info { struct amdgpu_reset_domain *reset_domain; uint32_t device_remove_count; + atomic_t ras_recovery; }; struct amdgpu_pcs_ras_field { diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c index 74fc945a8f9b..c2efb0d24bf8 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c @@ -2238,16 +2238,24 @@ failed: static int smu_v13_0_6_mode1_reset(struct smu_context *smu) { struct amdgpu_device *adev = smu->adev; + struct amdgpu_hive_info *hive = NULL; + u32 hive_ras_recovery = 0; struct amdgpu_ras *ras; u32 fatal_err, param; int ret = 0; + hive = amdgpu_get_xgmi_hive(adev); ras = amdgpu_ras_get_context(adev); fatal_err = 0; param = SMU_RESET_MODE_1; + if (hive) { + hive_ras_recovery = atomic_read(&hive->ras_recovery); + amdgpu_put_xgmi_hive(hive); + } + /* fatal error triggered by ras, PMFW supports the flag */ - if (ras && atomic_read(&ras->in_recovery)) + if (ras && (atomic_read(&ras->in_recovery) || hive_ras_recovery)) fatal_err = 1; param |= (fatal_err << 16); -- cgit From 472c5fb29798695b589fb844f84c6bf4ff07c592 Mon Sep 17 00:00:00 2001 From: Tao Zhou Date: Wed, 11 Oct 2023 18:36:15 +0800 Subject: drm/amdgpu: define ras_reset_error_count function Make the code architecture more simple. v2: reuse ras_reset_error_count in ras_reset_error_status. Signed-off-by: Tao Zhou Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 19 +++++++++++++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 2 ++ 2 files changed, 17 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 c7f8dcb3b4d2..a4384f0eb8bf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1170,23 +1170,34 @@ out_fini_err_data: return ret; } -int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, +int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block block) { struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0); if (!block_obj || !block_obj->hw_ops) { dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", - ras_block_str(block)); - return 0; + ras_block_str(block)); + return -EOPNOTSUPP; } if (!amdgpu_ras_is_supported(adev, block)) - return 0; + return -EOPNOTSUPP; if (block_obj->hw_ops->reset_ras_error_count) block_obj->hw_ops->reset_ras_error_count(adev); + return 0; +} + +int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, + enum amdgpu_ras_block block) +{ + struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0); + + if (amdgpu_ras_reset_error_count(adev, block) == -EOPNOTSUPP) + return 0; + if ((block == AMDGPU_RAS_BLOCK__GFX) || (block == AMDGPU_RAS_BLOCK__MMHUB)) { if (block_obj->hw_ops->reset_ras_error_status) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 0a5c8a107fb2..3f9ac0ab67e6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -714,6 +714,8 @@ void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev); int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info); +int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, + enum amdgpu_ras_block block); int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, enum amdgpu_ras_block block); -- cgit From 9248462d7e0862883df6741ec0e1bb41c3698b22 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Thu, 28 Sep 2023 17:00:31 +0800 Subject: drm/amdgpu: Enable software RAS in vcn v4_0_3 Set VCN/JPEG RAS masks to enable software RAS for VCN and JPEG. Signed-off-by: Hawking Zhang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 4 +++- 1 file changed, 3 insertions(+), 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 a4384f0eb8bf..0df15d856085 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2622,7 +2622,9 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(2, 6, 0) || amdgpu_ip_version(adev, VCN_HWIP, 0) == - IP_VERSION(4, 0, 0)) + IP_VERSION(4, 0, 0) || + amdgpu_ip_version(adev, VCN_HWIP, 0) == + IP_VERSION(4, 0, 3)) adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN | 1 << AMDGPU_RAS_BLOCK__JPEG); else -- cgit From 8a65661114941788a2093193c251e44cf1d6439c Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Thu, 19 Oct 2023 14:55:57 +0800 Subject: drm/amdgpu: Fix delete nodes that have been relesed Fix delete nodes that it has been freed. Fixes: 5b1270beb380 ("drm/amdgpu: add ras_err_info to identify RAS error source") Signed-off-by: Stanley.Yang Reviewed-by: Yang Wang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 4 +--- 1 file changed, 1 insertion(+), 3 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 0df15d856085..b3c32c667e09 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3503,10 +3503,8 @@ void amdgpu_ras_error_data_fini(struct ras_err_data *err_data) { struct ras_err_node *err_node, *tmp; - list_for_each_entry_safe(err_node, tmp, &err_data->err_node_list, node) { + list_for_each_entry_safe(err_node, tmp, &err_data->err_node_list, node) amdgpu_ras_error_node_release(err_node); - list_del(&err_node->node); - } } static struct ras_err_node *amdgpu_ras_error_find_node_by_id(struct ras_err_data *err_data, -- cgit From 49c260bef3ac9fc1bb73acf98036dac64712536d Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Thu, 19 Oct 2023 14:59:19 +0800 Subject: drm/amdgpu: fix typo for amdgpu ras error data print typo fix. Fixes: 5b1270beb380 ("drm/amdgpu: add ras_err_info to identify RAS error source") Signed-off-by: Yang Wang Reviewed-by: Candice Li Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 4 ++-- 1 file changed, 2 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 b3c32c667e09..70dd249f2ba7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1038,7 +1038,7 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, ras_mgr->err_data.ue_count, blk_name); else dev_info(adev->dev, "%ld correctable hardware errors detected in %s block\n", - ras_mgr->err_data.ue_count, blk_name); + ras_mgr->err_data.ce_count, blk_name); for_each_ras_error(err_node, err_data) { err_info = &err_node->err_info; @@ -1055,7 +1055,7 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, "%lld correctable hardware errors detected in %s block\n", mcm_info->socket_id, mcm_info->die_id, - err_info->ue_count, + err_info->ce_count, blk_name); } } -- cgit From 66d64e4e03ef5ecf330075a5f1fc449549ce374a Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Thu, 19 Oct 2023 22:49:32 +0800 Subject: drm/amdgpu: Enable RAS feature by default for APU Enable RAS feature by default for aqua vanjaram on apu platform. Signed-off-by: Stanley.Yang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 14 ++------------ 1 file changed, 2 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 70dd249f2ba7..eb81efacefe2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2653,18 +2653,8 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) /* hw_supported needs to be aligned with RAS block mask. */ adev->ras_hw_enabled &= AMDGPU_RAS_BLOCK_MASK; - - /* - * Disable ras feature for aqua vanjaram - * by default on apu platform. - */ - if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) && - adev->gmc.is_app_apu) - adev->ras_enabled = amdgpu_ras_enable != 1 ? 0 : - adev->ras_hw_enabled & amdgpu_ras_mask; - else - adev->ras_enabled = amdgpu_ras_enable == 0 ? 0 : - adev->ras_hw_enabled & amdgpu_ras_mask; + adev->ras_enabled = amdgpu_ras_enable == 0 ? 0 : + adev->ras_hw_enabled & amdgpu_ras_mask; } static void amdgpu_ras_counte_dw(struct work_struct *work) -- cgit From 8096df766474b54758b268afe900ba9d7ab0cc37 Mon Sep 17 00:00:00 2001 From: Tao Zhou Date: Thu, 12 Oct 2023 11:22:07 +0800 Subject: drm/amdgpu: add set/get mca debug mode operations Record the debug mode status in RAS. Signed-off-by: Tao Zhou Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 21 +++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 5 +++++ 2 files changed, 26 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 eb81efacefe2..3c83a2b8fb2c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3311,6 +3311,27 @@ int amdgpu_ras_reset_gpu(struct amdgpu_device *adev) return 0; } +void amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + + if (con) + con->is_mca_debug_mode = enable; +} + +bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs; + + if (!con) + return false; + + if (mca_funcs && mca_funcs->mca_set_debug_mode) + return con->is_mca_debug_mode; + else + return true; +} /* Register each ip ras block into amdgpu ras */ int amdgpu_ras_register_ras_block(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 3f9ac0ab67e6..2fdfef62ee27 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -434,6 +434,8 @@ struct amdgpu_ras { /* Indicates smu whether need update bad channel info */ bool update_channel_flag; + /* Record status of smu mca debug mode */ + bool is_mca_debug_mode; /* Record special requirements of gpu reset caller */ uint32_t gpu_reset_flags; @@ -768,6 +770,9 @@ struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev); int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con); +void amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable); +bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev); + int amdgpu_ras_register_ras_block(struct amdgpu_device *adev, struct amdgpu_ras_block_object *ras_block_obj); void amdgpu_ras_interrupt_fatal_error_handler(struct amdgpu_device *adev); -- cgit From 53d4d7792757d195979a630a6402f272d3fd2a47 Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Fri, 20 Oct 2023 10:12:07 +0800 Subject: drm/amdgpu: fix find ras error node error the origin function might return the wrong node. Fixes: 5b1270beb380 ("drm/amdgpu: add ras_err_info to identify RAS error source") Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang 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 3c83a2b8fb2c..88e1b47c3d3f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3529,11 +3529,10 @@ static struct ras_err_node *amdgpu_ras_error_find_node_by_id(struct ras_err_data for_each_ras_error(err_node, err_data) { ref_id = &err_node->err_info.mcm_info; - if ((mcm_info->socket_id >= 0 && mcm_info->socket_id != ref_id->socket_id) || - (mcm_info->die_id >= 0 && mcm_info->die_id != ref_id->die_id)) - continue; - return err_node; + if (mcm_info->socket_id == ref_id->socket_id && + mcm_info->die_id == ref_id->die_id) + return err_node; } return NULL; -- cgit From ec3e0a9167e2cc97a9b12d9f2a619afd78b77223 Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Thu, 19 Oct 2023 18:16:14 +0800 Subject: drm/amdgpu: refine ras error kernel log print refine ras error kernel log to avoid user-ridden ambiguity. Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 116 ++++++++++++++++++++++---------- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 5 +- 2 files changed, 82 insertions(+), 39 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 88e1b47c3d3f..fe6b44eb6602 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -635,8 +635,11 @@ 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); + amdgpu_ras_error_data_fini(&obj->err_data); + } + if (obj && (obj->use < 0)) DRM_ERROR("RAS ERROR: Unbalance obj(%s) use\n", get_ras_block_str(&obj->head)); } @@ -666,6 +669,9 @@ static struct ras_manager *amdgpu_ras_create_obj(struct amdgpu_device *adev, if (alive_obj(obj)) return NULL; + if (amdgpu_ras_error_data_init(&obj->err_data)) + return NULL; + obj->head = *head; obj->adev = adev; list_add(&obj->node, &con->head); @@ -1023,44 +1029,68 @@ static void amdgpu_ras_get_ecc_info(struct amdgpu_device *adev, struct ras_err_d } static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, - struct ras_query_if *query_if, + struct ras_manager *ras_mgr, struct ras_err_data *err_data, + const char *blk_name, bool is_ue) { - struct ras_manager *ras_mgr = amdgpu_ras_find_obj(adev, &query_if->head); - const char *blk_name = get_ras_block_str(&query_if->head); struct amdgpu_smuio_mcm_config_info *mcm_info; struct ras_err_node *err_node; struct ras_err_info *err_info; - if (is_ue) - dev_info(adev->dev, "%ld uncorrectable hardware errors detected in %s block\n", - ras_mgr->err_data.ue_count, blk_name); - else - dev_info(adev->dev, "%ld correctable hardware errors detected in %s block\n", - ras_mgr->err_data.ce_count, blk_name); + if (is_ue) { + for_each_ras_error(err_node, err_data) { + err_info = &err_node->err_info; + mcm_info = &err_info->mcm_info; + if (err_info->ue_count) { + dev_info(adev->dev, "socket: %d, die: %d, " + "%lld new uncorrectable hardware errors detected in %s block\n", + mcm_info->socket_id, + mcm_info->die_id, + err_info->ue_count, + blk_name); + } + } - for_each_ras_error(err_node, err_data) { - err_info = &err_node->err_info; - mcm_info = &err_info->mcm_info; - if (is_ue && err_info->ue_count) { - dev_info(adev->dev, "socket: %d, die: %d " - "%lld uncorrectable hardware errors detected in %s block\n", - mcm_info->socket_id, - mcm_info->die_id, - err_info->ue_count, - blk_name); - } else if (!is_ue && err_info->ce_count) { - dev_info(adev->dev, "socket: %d, die: %d " - "%lld correctable hardware errors detected in %s block\n", - mcm_info->socket_id, - mcm_info->die_id, - err_info->ce_count, - blk_name); + for_each_ras_error(err_node, &ras_mgr->err_data) { + err_info = &err_node->err_info; + mcm_info = &err_info->mcm_info; + dev_info(adev->dev, "socket: %d, die: %d, " + "%lld uncorrectable hardware errors detected in total in %s block\n", + mcm_info->socket_id, mcm_info->die_id, err_info->ue_count, blk_name); + } + + } else { + for_each_ras_error(err_node, err_data) { + err_info = &err_node->err_info; + mcm_info = &err_info->mcm_info; + if (err_info->ce_count) { + dev_info(adev->dev, "socket: %d, die: %d, " + "%lld new correctable hardware errors detected in %s block, " + "no user action is needed\n", + mcm_info->socket_id, + mcm_info->die_id, + err_info->ce_count, + blk_name); + } + } + + for_each_ras_error(err_node, &ras_mgr->err_data) { + err_info = &err_node->err_info; + mcm_info = &err_info->mcm_info; + dev_info(adev->dev, "socket: %d, die: %d, " + "%lld correctable hardware errors detected in total in %s block, " + "no user action is needed\n", + mcm_info->socket_id, mcm_info->die_id, err_info->ce_count, blk_name); } } } +static inline bool err_data_has_source_info(struct ras_err_data *data) +{ + return !list_empty(&data->err_node_list); +} + static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, struct ras_query_if *query_if, struct ras_err_data *err_data) @@ -1069,9 +1099,8 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, const char *blk_name = get_ras_block_str(&query_if->head); if (err_data->ce_count) { - if (!list_empty(&err_data->err_node_list)) { - amdgpu_ras_error_print_error_data(adev, query_if, - err_data, false); + if (err_data_has_source_info(err_data)) { + amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, blk_name, false); } else if (!adev->aid_mask && adev->smuio.funcs && adev->smuio.funcs->get_socket_id && @@ -1094,9 +1123,8 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, } if (err_data->ue_count) { - if (!list_empty(&err_data->err_node_list)) { - amdgpu_ras_error_print_error_data(adev, query_if, - err_data, true); + if (err_data_has_source_info(err_data)) { + amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, blk_name, true); } else if (!adev->aid_mask && adev->smuio.funcs && adev->smuio.funcs->get_socket_id && @@ -1118,6 +1146,25 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, } +static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, struct ras_err_data *err_data) +{ + struct ras_err_node *err_node; + struct ras_err_info *err_info; + + if (err_data_has_source_info(err_data)) { + for_each_ras_error(err_node, err_data) { + err_info = &err_node->err_info; + + amdgpu_ras_error_statistic_ce_count(&obj->err_data, &err_info->mcm_info, err_info->ce_count); + amdgpu_ras_error_statistic_ue_count(&obj->err_data, &err_info->mcm_info, err_info->ue_count); + } + } else { + /* for legacy asic path which doesn't has error source info */ + obj->err_data.ue_count += err_data->ue_count; + obj->err_data.ce_count += err_data->ce_count; + } +} + /* query/inject/cure begin */ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info) @@ -1156,8 +1203,7 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, } } - obj->err_data.ue_count += err_data.ue_count; - obj->err_data.ce_count += err_data.ce_count; + amdgpu_rasmgr_error_data_statistic_update(obj, &err_data); info->ue_count = obj->err_data.ue_count; info->ce_count = obj->err_data.ce_count; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 2fdfef62ee27..665414c22ca9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -515,10 +515,7 @@ struct ras_manager { /* IH data */ struct ras_ih_data ih_data; - struct { - unsigned long ue_count; - unsigned long ce_count; - } err_data; + struct ras_err_data err_data; }; struct ras_badpage { -- cgit From f3a3bbf1566c7b6b0f9ac36e8e597c73dc0afdf8 Mon Sep 17 00:00:00 2001 From: Tao Zhou Date: Fri, 20 Oct 2023 17:21:58 +0800 Subject: drm/amdgpu: enable RAS poison mode for APU Enable it by default on APU platform. Signed-off-by: Tao Zhou Reviewed-by: Stanley.Yang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 3 ++- 1 file changed, 2 insertions(+), 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 fe6b44eb6602..f4c2c737b12f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2738,7 +2738,8 @@ static void amdgpu_ras_query_poison_mode(struct amdgpu_device *adev) return; /* Init poison supported flag, the default value is false */ - if (adev->gmc.xgmi.connected_to_cpu) { + if (adev->gmc.xgmi.connected_to_cpu || + adev->gmc.is_app_apu) { /* enabled by default when GPU is connected to CPU */ con->poison_supported = true; } else if (adev->df.funcs && -- cgit From 73582be11ac8f6d6765e185bf48f22efb9d28c3b Mon Sep 17 00:00:00 2001 From: Tao Zhou Date: Thu, 12 Oct 2023 14:33:37 +0800 Subject: drm/amdgpu: bypass RAS error reset in some conditions PMFW is responsible for RAS error reset in some conditions, driver can skip the operation. v2: add check for ras->in_recovery, it's set earlier than amdgpu_in_reset. v3: fix error in gpu reset check. Signed-off-by: Tao Zhou Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 10 +++++++++- 1 file changed, 9 insertions(+), 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 f4c2c737b12f..303fbb6a48b6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1220,6 +1220,8 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, enum amdgpu_ras_block block) { struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0); + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs; if (!block_obj || !block_obj->hw_ops) { dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", @@ -1227,7 +1229,13 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, return -EOPNOTSUPP; } - if (!amdgpu_ras_is_supported(adev, block)) + /* skip ras error reset in gpu reset */ + if ((amdgpu_in_reset(adev) || atomic_read(&ras->in_recovery)) && + mca_funcs && mca_funcs->mca_set_debug_mode) + return -EOPNOTSUPP; + + if (!amdgpu_ras_is_supported(adev, block) || + !amdgpu_ras_get_mca_debug_mode(adev)) return -EOPNOTSUPP; if (block_obj->hw_ops->reset_ras_error_count) -- cgit From d1d4c0b7b65b7fab2bc6f97af9e823b1c42ccdb0 Mon Sep 17 00:00:00 2001 From: Tao Zhou Date: Wed, 25 Oct 2023 11:03:24 +0800 Subject: drm/amdgpu: check RAS supported first in ras_reset_error_count Not all platforms support RAS. Fixes: 73582be11ac8 ("drm/amdgpu: bypass RAS error reset in some conditions") Signed-off-by: Tao Zhou Reviewed-by: Yang Wang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 8 ++++---- 1 file changed, 4 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 303fbb6a48b6..3af50754800d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1229,15 +1229,15 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, return -EOPNOTSUPP; } + if (!amdgpu_ras_is_supported(adev, block) || + !amdgpu_ras_get_mca_debug_mode(adev)) + return -EOPNOTSUPP; + /* skip ras error reset in gpu reset */ if ((amdgpu_in_reset(adev) || atomic_read(&ras->in_recovery)) && mca_funcs && mca_funcs->mca_set_debug_mode) return -EOPNOTSUPP; - if (!amdgpu_ras_is_supported(adev, block) || - !amdgpu_ras_get_mca_debug_mode(adev)) - return -EOPNOTSUPP; - if (block_obj->hw_ops->reset_ras_error_count) block_obj->hw_ops->reset_ras_error_count(adev); -- cgit From 18eae367cb74d05b5e37ce77ef4025b735df012e Mon Sep 17 00:00:00 2001 From: Tao Zhou Date: Mon, 30 Oct 2023 20:44:37 +0800 Subject: drm/amdgpu: check recovery status of xgmi hive in ras_reset_error_count Handle xgmi hive case. Suggested-by: Hawking Zhang Signed-off-by: Tao Zhou Reviewed-by: Stanley.Yang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 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 3af50754800d..b7fe5951b166 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1222,6 +1222,8 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0); struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs; + struct amdgpu_hive_info *hive; + int hive_ras_recovery = 0; if (!block_obj || !block_obj->hw_ops) { dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", @@ -1233,8 +1235,15 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, !amdgpu_ras_get_mca_debug_mode(adev)) return -EOPNOTSUPP; + hive = amdgpu_get_xgmi_hive(adev); + if (hive) { + hive_ras_recovery = atomic_read(&hive->ras_recovery); + amdgpu_put_xgmi_hive(hive); + } + /* skip ras error reset in gpu reset */ - if ((amdgpu_in_reset(adev) || atomic_read(&ras->in_recovery)) && + if ((amdgpu_in_reset(adev) || atomic_read(&ras->in_recovery) || + hive_ras_recovery) && mca_funcs && mca_funcs->mca_set_debug_mode) return -EOPNOTSUPP; -- cgit From 8cc0f5669eb6d4f156c721956da67560c9319317 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Wed, 8 Nov 2023 16:07:45 +0800 Subject: drm/amdgpu: Support multiple error query modes Direct error query mode and firmware error query mode are supported for now. Signed-off-by: Hawking Zhang Reviewed-by: Yang Wang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 93 +++++++++++++++++++++++++-------- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 8 +++ 2 files changed, 78 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 b7fe5951b166..a5c14dbb00b7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1165,13 +1165,53 @@ static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, s } } -/* query/inject/cure begin */ -int amdgpu_ras_query_error_status(struct amdgpu_device *adev, - struct ras_query_if *info) +static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev, + struct ras_query_if *info, + struct ras_err_data *err_data, + unsigned int error_query_mode) { + enum amdgpu_ras_block blk = info ? info->head.block : AMDGPU_RAS_BLOCK_COUNT; struct amdgpu_ras_block_object *block_obj = NULL; + + if (error_query_mode == AMDGPU_RAS_INVALID_ERROR_QUERY) + return -EINVAL; + + if (error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) { + if (info->head.block == AMDGPU_RAS_BLOCK__UMC) { + amdgpu_ras_get_ecc_info(adev, err_data); + } else { + block_obj = amdgpu_ras_get_ras_block(adev, info->head.block, 0); + if (!block_obj || !block_obj->hw_ops) { + dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", + get_ras_block_str(&info->head)); + return -EINVAL; + } + + if (block_obj->hw_ops->query_ras_error_count) + block_obj->hw_ops->query_ras_error_count(adev, &err_data); + + if ((info->head.block == AMDGPU_RAS_BLOCK__SDMA) || + (info->head.block == AMDGPU_RAS_BLOCK__GFX) || + (info->head.block == AMDGPU_RAS_BLOCK__MMHUB)) { + if (block_obj->hw_ops->query_ras_error_status) + block_obj->hw_ops->query_ras_error_status(adev); + } + } + } else { + /* FIXME: add code to check return value later */ + amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_UE, err_data); + amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_CE, err_data); + } + + return 0; +} + +/* query/inject/cure begin */ +int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info) +{ struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head); struct ras_err_data err_data; + unsigned int error_query_mode; int ret; if (!obj) @@ -1181,27 +1221,14 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, if (ret) return ret; - if (info->head.block == AMDGPU_RAS_BLOCK__UMC) { - amdgpu_ras_get_ecc_info(adev, &err_data); - } else { - block_obj = amdgpu_ras_get_ras_block(adev, info->head.block, 0); - if (!block_obj || !block_obj->hw_ops) { - dev_dbg_once(adev->dev, "%s doesn't config RAS function\n", - get_ras_block_str(&info->head)); - ret = -EINVAL; - goto out_fini_err_data; - } - - if (block_obj->hw_ops->query_ras_error_count) - block_obj->hw_ops->query_ras_error_count(adev, &err_data); + if (!amdgpu_ras_get_error_query_mode(adev, &error_query_mode)) + return -EINVAL; - if ((info->head.block == AMDGPU_RAS_BLOCK__SDMA) || - (info->head.block == AMDGPU_RAS_BLOCK__GFX) || - (info->head.block == AMDGPU_RAS_BLOCK__MMHUB)) { - if (block_obj->hw_ops->query_ras_error_status) - block_obj->hw_ops->query_ras_error_status(adev); - } - } + ret = amdgpu_ras_query_error_status_helper(adev, info, + &err_data, + error_query_mode); + if (ret) + goto out_fini_err_data; amdgpu_rasmgr_error_data_statistic_update(obj, &err_data); @@ -3397,6 +3424,26 @@ bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev) return true; } +bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev, + unsigned int *error_query_mode) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs; + + if (!con) { + *error_query_mode = AMDGPU_RAS_INVALID_ERROR_QUERY; + return false; + } + + if (mca_funcs && mca_funcs->mca_set_debug_mode) + *error_query_mode = + (con->is_mca_debug_mode) ? AMDGPU_RAS_DIRECT_ERROR_QUERY : AMDGPU_RAS_FIRMWARE_ERROR_QUERY; + else + *error_query_mode = AMDGPU_RAS_DIRECT_ERROR_QUERY; + + return true; +} + /* Register each ip ras block into amdgpu ras */ int amdgpu_ras_register_ras_block(struct amdgpu_device *adev, struct amdgpu_ras_block_object *ras_block_obj) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 665414c22ca9..19161916ac46 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -320,6 +320,12 @@ enum amdgpu_ras_ret { AMDGPU_RAS_PT, }; +enum amdgpu_ras_error_query_mode { + AMDGPU_RAS_INVALID_ERROR_QUERY = 0, + AMDGPU_RAS_DIRECT_ERROR_QUERY = 1, + AMDGPU_RAS_FIRMWARE_ERROR_QUERY = 2, +}; + /* ras error status reisger fields */ #define ERR_STATUS_LO__ERR_STATUS_VALID_FLAG__SHIFT 0x0 #define ERR_STATUS_LO__ERR_STATUS_VALID_FLAG_MASK 0x00000001L @@ -769,6 +775,8 @@ int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_co void amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable); bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev); +bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev, + unsigned int *mode); int amdgpu_ras_register_ras_block(struct amdgpu_device *adev, struct amdgpu_ras_block_object *ras_block_obj); -- cgit From 4638e0c29a3f2294d5de0d052a4b8c9f33ccb957 Mon Sep 17 00:00:00 2001 From: Vitaly Prosyak Date: Wed, 11 Oct 2023 19:31:48 -0400 Subject: drm/amdgpu: fix software pci_unplug on some chips When software 'pci unplug' using IGT is executed we got a sysfs directory entry is NULL for differant ras blocks like hdp, umc, etc. Before call 'sysfs_remove_file_from_group' and 'sysfs_remove_group' check that 'sd' is not NULL. [ +0.000001] RIP: 0010:sysfs_remove_group+0x83/0x90 [ +0.000002] Code: 31 c0 31 d2 31 f6 31 ff e9 9a a8 b4 00 4c 89 e7 e8 f2 a2 ff ff eb c2 49 8b 55 00 48 8b 33 48 c7 c7 80 65 94 82 e8 cd 82 bb ff <0f> 0b eb cc 66 0f 1f 84 00 00 00 00 00 90 90 90 90 90 90 90 90 90 [ +0.000001] RSP: 0018:ffffc90002067c90 EFLAGS: 00010246 [ +0.000002] RAX: 0000000000000000 RBX: ffffffff824ea180 RCX: 0000000000000000 [ +0.000001] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000 [ +0.000001] RBP: ffffc90002067ca8 R08: 0000000000000000 R09: 0000000000000000 [ +0.000001] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 [ +0.000001] R13: ffff88810a395f48 R14: ffff888101aab0d0 R15: 0000000000000000 [ +0.000001] FS: 00007f5ddaa43a00(0000) GS:ffff88841e800000(0000) knlGS:0000000000000000 [ +0.000002] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ +0.000001] CR2: 00007f8ffa61ba50 CR3: 0000000106432000 CR4: 0000000000350ef0 [ +0.000001] Call Trace: [ +0.000001] [ +0.000001] ? show_regs+0x72/0x90 [ +0.000002] ? sysfs_remove_group+0x83/0x90 [ +0.000002] ? __warn+0x8d/0x160 [ +0.000001] ? sysfs_remove_group+0x83/0x90 [ +0.000001] ? report_bug+0x1bb/0x1d0 [ +0.000003] ? handle_bug+0x46/0x90 [ +0.000001] ? exc_invalid_op+0x19/0x80 [ +0.000002] ? asm_exc_invalid_op+0x1b/0x20 [ +0.000003] ? sysfs_remove_group+0x83/0x90 [ +0.000001] dpm_sysfs_remove+0x61/0x70 [ +0.000002] device_del+0xa3/0x3d0 [ +0.000002] ? ktime_get_mono_fast_ns+0x46/0xb0 [ +0.000002] device_unregister+0x18/0x70 [ +0.000001] i2c_del_adapter+0x26d/0x330 [ +0.000002] arcturus_i2c_control_fini+0x25/0x50 [amdgpu] [ +0.000236] smu_sw_fini+0x38/0x260 [amdgpu] [ +0.000241] amdgpu_device_fini_sw+0x116/0x670 [amdgpu] [ +0.000186] ? mutex_lock+0x13/0x50 [ +0.000003] amdgpu_driver_release_kms+0x16/0x40 [amdgpu] [ +0.000192] drm_minor_release+0x4f/0x80 [drm] [ +0.000025] drm_release+0xfe/0x150 [drm] [ +0.000027] __fput+0x9f/0x290 [ +0.000002] ____fput+0xe/0x20 [ +0.000002] task_work_run+0x61/0xa0 [ +0.000002] exit_to_user_mode_prepare+0x150/0x170 [ +0.000002] syscall_exit_to_user_mode+0x2a/0x50 Cc: Hawking Zhang Cc: Luben Tuikov Cc: Alex Deucher Cc: Christian Koenig Signed-off-by: Vitaly Prosyak Reviewed-by: Luben Tuikov Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 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 a5c14dbb00b7..84e5987b14e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1564,7 +1564,8 @@ static void amdgpu_ras_sysfs_remove_bad_page_node(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - sysfs_remove_file_from_group(&adev->dev->kobj, + if (adev->dev->kobj.sd) + sysfs_remove_file_from_group(&adev->dev->kobj, &con->badpages_attr.attr, RAS_FS_NAME); } @@ -1583,7 +1584,8 @@ static int amdgpu_ras_sysfs_remove_dev_attr_node(struct amdgpu_device *adev) .attrs = attrs, }; - sysfs_remove_group(&adev->dev->kobj, &group); + if (adev->dev->kobj.sd) + sysfs_remove_group(&adev->dev->kobj, &group); return 0; } @@ -1630,7 +1632,8 @@ int amdgpu_ras_sysfs_remove(struct amdgpu_device *adev, if (!obj || !obj->attr_inuse) return -EINVAL; - sysfs_remove_file_from_group(&adev->dev->kobj, + if (adev->dev->kobj.sd) + sysfs_remove_file_from_group(&adev->dev->kobj, &obj->sysfs_attr.attr, RAS_FS_NAME); obj->attr_inuse = 0; -- cgit From 07ee43faeb7eb088e49a7549fcabcae94c443d3b Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Mon, 13 Nov 2023 14:34:55 +0800 Subject: drm/amdgpu: fix ras err_data null pointer issue in amdgpu_ras.c fix ras err_data null pointer issue in amdgpu_ras.c Fixes: 8cc0f5669eb6 ("drm/amdgpu: Support multiple error query modes") Signed-off-by: Yang Wang 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 84e5987b14e0..a3dc68e98910 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1188,7 +1188,7 @@ static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev, } if (block_obj->hw_ops->query_ras_error_count) - block_obj->hw_ops->query_ras_error_count(adev, &err_data); + block_obj->hw_ops->query_ras_error_count(adev, err_data); if ((info->head.block == AMDGPU_RAS_BLOCK__SDMA) || (info->head.block == AMDGPU_RAS_BLOCK__GFX) || -- cgit From 201761b5eb57c3fad810cde555795c3b5721a031 Mon Sep 17 00:00:00 2001 From: Lijo Lazar Date: Thu, 9 Nov 2023 10:50:38 +0530 Subject: drm/amdgpu: Move mca debug mode decision to ras Refactor code such that ras block decides the default mca debug mode, and not swsmu block. By default mca debug mode is set to false. v2: squash in uninitialized value fix (Alex) Signed-off-by: Lijo Lazar Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 14 +++++++++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 2 +- drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c | 12 ------------ 4 files changed, 13 insertions(+), 17 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c index cf33eb219e25..54f2f346579e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c @@ -377,7 +377,7 @@ static int amdgpu_mca_smu_debug_mode_set(void *data, u64 val) struct amdgpu_device *adev = (struct amdgpu_device *)data; int ret; - ret = amdgpu_mca_smu_set_debug_mode(adev, val ? true : false); + ret = amdgpu_ras_set_mca_debug_mode(adev, val ? true : false); if (ret) return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index a3dc68e98910..f6b47ebce9d6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3132,6 +3132,8 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) return 0; + amdgpu_ras_set_mca_debug_mode(adev, false); + list_for_each_entry_safe(node, tmp, &adev->ras_list, node) { if (!node->ras_obj) { dev_warn(adev->dev, "Warning: abnormal ras list node.\n"); @@ -3405,12 +3407,18 @@ int amdgpu_ras_reset_gpu(struct amdgpu_device *adev) return 0; } -void amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable) +int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + int ret = 0; - if (con) - con->is_mca_debug_mode = enable; + if (con) { + ret = amdgpu_mca_smu_set_debug_mode(adev, enable); + if (!ret) + con->is_mca_debug_mode = enable; + } + + return ret; } bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 19161916ac46..6a941eb8fb8f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -773,7 +773,7 @@ struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev); int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con); -void amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable); +int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable); bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev); bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev, unsigned int *mode); diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c index 8ccb8354bb49..dda2249c4994 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c @@ -1516,7 +1516,6 @@ static int smu_v13_0_6_mca_set_debug_mode(struct smu_context *smu, bool enable) if (smu->smc_fw_version < 0x554800) return 0; - amdgpu_ras_set_mca_debug_mode(smu->adev, enable); return smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_ClearMcaOnRead, enable ? 0 : ClearMcaOnRead_UE_FLAG_MASK | ClearMcaOnRead_CE_POLL_MASK, NULL); @@ -2346,16 +2345,6 @@ static int smu_v13_0_6_smu_send_hbm_bad_page_num(struct smu_context *smu, return ret; } -static int smu_v13_0_6_post_init(struct smu_context *smu) -{ - struct amdgpu_device *adev = smu->adev; - - if (!amdgpu_sriov_vf(adev) && adev->ras_enabled) - return smu_v13_0_6_mca_set_debug_mode(smu, false); - - return 0; -} - static int mca_smu_set_debug_mode(struct amdgpu_device *adev, bool enable) { struct smu_context *smu = adev->powerplay.pp_handle; @@ -2920,7 +2909,6 @@ static const struct pptable_funcs smu_v13_0_6_ppt_funcs = { .i2c_init = smu_v13_0_6_i2c_control_init, .i2c_fini = smu_v13_0_6_i2c_control_fini, .send_hbm_bad_pages_num = smu_v13_0_6_smu_send_hbm_bad_page_num, - .post_init = smu_v13_0_6_post_init, }; void smu_v13_0_6_set_ppt_funcs(struct smu_context *smu) -- cgit From f875f61b1fd626a4223a5bdf0339b5372c689e13 Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Thu, 30 Nov 2023 18:39:03 +0800 Subject: drm/amdgpu: enable mca debug mode on APU by default enable MCA debug mode on APU device by default. Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 3 ++- 1 file changed, 2 insertions(+), 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 f6b47ebce9d6..72634d675e27 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3132,7 +3132,8 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) return 0; - amdgpu_ras_set_mca_debug_mode(adev, false); + /* enable MCA debug on APU device */ + amdgpu_ras_set_mca_debug_mode(adev, !!(adev->flags & AMD_IS_APU)); list_for_each_entry_safe(node, tmp, &adev->ras_list, node) { if (!node->ras_obj) { -- cgit From dbf3850d12baf3ba8a80c302f538d1b01940aef7 Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Mon, 4 Dec 2023 10:44:32 +0800 Subject: drm/amdgpu: optimize the printing order of error data sort error data list to optimize the printing order. Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 17 +++++++++++++++++ 1 file changed, 17 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 a3dc68e98910..63fb4cd85e53 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -28,6 +28,7 @@ #include #include #include +#include #include "amdgpu.h" #include "amdgpu_ras.h" @@ -3665,6 +3666,21 @@ static struct ras_err_node *amdgpu_ras_error_node_new(void) return err_node; } +static int ras_err_info_cmp(void *priv, const struct list_head *a, const struct list_head *b) +{ + struct ras_err_node *nodea = container_of(a, struct ras_err_node, node); + struct ras_err_node *nodeb = container_of(b, struct ras_err_node, node); + struct amdgpu_smuio_mcm_config_info *infoa = &nodea->err_info.mcm_info; + struct amdgpu_smuio_mcm_config_info *infob = &nodeb->err_info.mcm_info; + + if (unlikely(infoa->socket_id != infob->socket_id)) + return infoa->socket_id - infob->socket_id; + else + return infoa->die_id - infob->die_id; + + return 0; +} + static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_data, struct amdgpu_smuio_mcm_config_info *mcm_info) { @@ -3682,6 +3698,7 @@ static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_d err_data->err_list_count++; list_add_tail(&err_node->node, &err_data->err_node_list); + list_sort(NULL, &err_data->err_node_list, ras_err_info_cmp); return &err_node->err_info; } -- cgit From 9f91e983ee82d3b6f6d713e1c84ebb8d53180b3d Mon Sep 17 00:00:00 2001 From: YiPeng Chai Date: Tue, 12 Dec 2023 17:26:58 +0800 Subject: drm/amdgpu: MCA supports recording umc address information MCA supports recording umc address information. V2: Move err_addr variable from struct ras_err_node to struct ras_err_info. Signed-off-by: YiPeng Chai Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c | 13 +++++++++++-- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 22 +++++++++++++++------- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 13 +++++++++++-- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c | 2 +- drivers/gpu/drm/amd/amdgpu/umc_v12_0.c | 4 ++-- 8 files changed, 46 insertions(+), 20 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c index 210aea590a52..8911310f98df 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c @@ -218,6 +218,7 @@ static void amdgpu_mca_smu_mca_bank_dump(struct amdgpu_device *adev, int idx, st int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_block blk, enum amdgpu_mca_error_type type, struct ras_err_data *err_data) { struct amdgpu_smuio_mcm_config_info mcm_info; + struct ras_err_addr err_addr = {0}; struct mca_bank_set mca_set; struct mca_bank_node *node; struct mca_bank_entry *entry; @@ -246,10 +247,18 @@ int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_blo mcm_info.socket_id = entry->info.socket_id; mcm_info.die_id = entry->info.aid; + if (blk == AMDGPU_RAS_BLOCK__UMC) { + err_addr.err_status = entry->regs[MCA_REG_IDX_STATUS]; + err_addr.err_ipid = entry->regs[MCA_REG_IDX_IPID]; + err_addr.err_addr = entry->regs[MCA_REG_IDX_ADDR]; + } + if (type == AMDGPU_MCA_ERROR_TYPE_UE) - amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, (uint64_t)count); + amdgpu_ras_error_statistic_ue_count(err_data, + &mcm_info, &err_addr, (uint64_t)count); else - amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, (uint64_t)count); + amdgpu_ras_error_statistic_ce_count(err_data, + &mcm_info, &err_addr, (uint64_t)count); } out_mca_release: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index bacb59d8b701..bad62141f708 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1156,8 +1156,10 @@ static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, s for_each_ras_error(err_node, err_data) { err_info = &err_node->err_info; - amdgpu_ras_error_statistic_ce_count(&obj->err_data, &err_info->mcm_info, err_info->ce_count); - amdgpu_ras_error_statistic_ue_count(&obj->err_data, &err_info->mcm_info, err_info->ue_count); + amdgpu_ras_error_statistic_ce_count(&obj->err_data, + &err_info->mcm_info, NULL, err_info->ce_count); + amdgpu_ras_error_statistic_ue_count(&obj->err_data, + &err_info->mcm_info, NULL, err_info->ue_count); } } else { /* for legacy asic path which doesn't has error source info */ @@ -3691,7 +3693,8 @@ static int ras_err_info_cmp(void *priv, const struct list_head *a, const struct } static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_data, - struct amdgpu_smuio_mcm_config_info *mcm_info) + struct amdgpu_smuio_mcm_config_info *mcm_info, + struct ras_err_addr *err_addr) { struct ras_err_node *err_node; @@ -3705,6 +3708,9 @@ static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_d memcpy(&err_node->err_info.mcm_info, mcm_info, sizeof(*mcm_info)); + if (err_addr) + memcpy(&err_node->err_info.err_addr, err_addr, sizeof(*err_addr)); + err_data->err_list_count++; list_add_tail(&err_node->node, &err_data->err_node_list); list_sort(NULL, &err_data->err_node_list, ras_err_info_cmp); @@ -3713,7 +3719,8 @@ static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_d } int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, - struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count) + struct amdgpu_smuio_mcm_config_info *mcm_info, + struct ras_err_addr *err_addr, u64 count) { struct ras_err_info *err_info; @@ -3723,7 +3730,7 @@ int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, if (!count) return 0; - err_info = amdgpu_ras_error_get_info(err_data, mcm_info); + err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr); if (!err_info) return -EINVAL; @@ -3734,7 +3741,8 @@ int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, } int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, - struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count) + struct amdgpu_smuio_mcm_config_info *mcm_info, + struct ras_err_addr *err_addr, u64 count) { struct ras_err_info *err_info; @@ -3744,7 +3752,7 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, if (!count) return 0; - err_info = amdgpu_ras_error_get_info(err_data, mcm_info); + err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr); if (!err_info) return -EINVAL; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 6a941eb8fb8f..76fb85628716 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -452,10 +452,17 @@ struct ras_fs_data { char debugfs_name[32]; }; +struct ras_err_addr { + uint64_t err_status; + uint64_t err_ipid; + uint64_t err_addr; +}; + struct ras_err_info { struct amdgpu_smuio_mcm_config_info mcm_info; u64 ce_count; u64 ue_count; + struct ras_err_addr err_addr; }; struct ras_err_node { @@ -806,8 +813,10 @@ void amdgpu_ras_inst_reset_ras_error_count(struct amdgpu_device *adev, int amdgpu_ras_error_data_init(struct ras_err_data *err_data); void amdgpu_ras_error_data_fini(struct ras_err_data *err_data); int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, - struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count); + struct amdgpu_smuio_mcm_config_info *mcm_info, + struct ras_err_addr *err_addr, u64 count); int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, - struct amdgpu_smuio_mcm_config_info *mcm_info, u64 count); + struct amdgpu_smuio_mcm_config_info *mcm_info, + struct ras_err_addr *err_addr, u64 count); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 9a95b9f226b8..a6c88f2fe6e5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -1313,10 +1313,10 @@ static void __xgmi_v6_4_0_query_error_count(struct amdgpu_device *adev, struct a switch (xgmi_v6_4_0_pcs_mca_get_error_type(adev, status)) { case AMDGPU_MCA_ERROR_TYPE_UE: - amdgpu_ras_error_statistic_ue_count(err_data, mcm_info, 1ULL); + amdgpu_ras_error_statistic_ue_count(err_data, mcm_info, NULL, 1ULL); break; case AMDGPU_MCA_ERROR_TYPE_CE: - amdgpu_ras_error_statistic_ce_count(err_data, mcm_info, 1ULL); + amdgpu_ras_error_statistic_ce_count(err_data, mcm_info, NULL, 1ULL); break; default: break; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c index 00b21ece081f..131cddbdda0d 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c @@ -3828,8 +3828,8 @@ static void gfx_v9_4_3_inst_query_ras_err_count(struct amdgpu_device *adev, /* the caller should make sure initialize value of * err_data->ue_count and err_data->ce_count */ - amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, ue_count); - amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, ce_count); + amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, NULL, ue_count); + amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, NULL, ce_count); } static void gfx_v9_4_3_inst_reset_ras_err_count(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c index 9b0146732e13..fb53aacdcba2 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c @@ -652,8 +652,8 @@ static void mmhub_v1_8_inst_query_ras_error_count(struct amdgpu_device *adev, AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE, &ue_count); - amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, ce_count); - amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, ue_count); + amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, NULL, ce_count); + amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, NULL, ue_count); } static void mmhub_v1_8_query_ras_error_count(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c index 0f24af6f2810..2d688dca26be 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c @@ -2156,7 +2156,7 @@ static void sdma_v4_4_2_inst_query_ras_error_count(struct amdgpu_device *adev, AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE, &ue_count); - amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, ue_count); + amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, NULL, ue_count); } static void sdma_v4_4_2_query_ras_error_count(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c index e9c2ff74f0bc..8d60c39ae1c5 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c @@ -166,8 +166,8 @@ static int umc_v12_0_query_error_count(struct amdgpu_device *adev, umc_v12_0_query_correctable_error_count(adev, umc_reg_offset, &ce_count); umc_v12_0_query_uncorrectable_error_count(adev, umc_reg_offset, &ue_count); - amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, ue_count); - amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, ce_count); + amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, NULL, ue_count); + amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, NULL, ce_count); return 0; } -- cgit From 091411be7ae899ce23072acf5a83b0b43e9024e1 Mon Sep 17 00:00:00 2001 From: Srinivasan Shanmugam Date: Tue, 19 Dec 2023 19:26:22 +0530 Subject: drm/amdgpu: Use kzalloc instead of kmalloc+__GFP_ZERO in amdgpu_ras.c MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Fixes the below smatch warnings: drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:2543 amdgpu_ras_recovery_init() warn: Please consider using kzalloc instead of kmalloc drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:2830 amdgpu_ras_init() warn: Please consider using kzalloc instead of kmalloc Cc: Christian König Cc: Alex Deucher Signed-off-by: Srinivasan Shanmugam Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 6 +++--- 1 file changed, 3 insertions(+), 3 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 bad62141f708..e541e6925918 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2540,7 +2540,7 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev) return 0; data = &con->eh_data; - *data = kmalloc(sizeof(**data), GFP_KERNEL | __GFP_ZERO); + *data = kzalloc(sizeof(**data), GFP_KERNEL); if (!*data) { ret = -ENOMEM; goto out; @@ -2827,10 +2827,10 @@ int amdgpu_ras_init(struct amdgpu_device *adev) if (con) return 0; - con = kmalloc(sizeof(struct amdgpu_ras) + + con = kzalloc(sizeof(*con) + sizeof(struct ras_manager) * AMDGPU_RAS_BLOCK_COUNT + sizeof(struct ras_manager) * AMDGPU_RAS_MCA_BLOCK_COUNT, - GFP_KERNEL|__GFP_ZERO); + GFP_KERNEL); if (!con) return -ENOMEM; -- cgit From b8d55a90fd55b767c25687747e2b24abd1ef8680 Mon Sep 17 00:00:00 2001 From: Srinivasan Shanmugam Date: Tue, 26 Dec 2023 15:32:19 +0530 Subject: drm/amdgpu: Fix possible NULL dereference in amdgpu_ras_query_error_status_helper() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Return invalid error code -EINVAL for invalid block id. Fixes the below: drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c:1183 amdgpu_ras_query_error_status_helper() error: we previously assumed 'info' could be null (see line 1176) Suggested-by: Hawking Zhang Cc: Tao Zhou Cc: Hawking Zhang Cc: Christian König Cc: Alex Deucher Signed-off-by: Srinivasan Shanmugam Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 3 +++ 1 file changed, 3 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 e541e6925918..39399d0f2ce5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1176,6 +1176,9 @@ static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev, enum amdgpu_ras_block blk = info ? info->head.block : AMDGPU_RAS_BLOCK_COUNT; struct amdgpu_ras_block_object *block_obj = NULL; + if (blk == AMDGPU_RAS_BLOCK_COUNT) + return -EINVAL; + if (error_query_mode == AMDGPU_RAS_INVALID_ERROR_QUERY) return -EINVAL; -- cgit From 6697dbf0afff73fcf2b53e99c4accdab58892e39 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Sun, 31 Dec 2023 17:42:47 +0800 Subject: Revert "drm/amdgpu: enable mca debug mode on APU by default" Not needed any more with firmware fixes Signed-off-by: Hawking Zhang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 3 +-- 1 file changed, 1 insertion(+), 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 39399d0f2ce5..fc42fb6ee191 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3138,8 +3138,7 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) return 0; - /* enable MCA debug on APU device */ - amdgpu_ras_set_mca_debug_mode(adev, !!(adev->flags & AMD_IS_APU)); + amdgpu_ras_set_mca_debug_mode(adev, false); list_for_each_entry_safe(node, tmp, &adev->ras_list, node) { if (!node->ras_obj) { -- cgit From 90bd01471d1c7f2d2db3c69259e247357991fe50 Mon Sep 17 00:00:00 2001 From: Candice Li Date: Thu, 4 Jan 2024 09:34:25 +0800 Subject: drm/amdgpu: Drop unnecessary sentences about CE and deferred error. Remove "no user action is needed" for correctable and deferred error to avoid confusion. Signed-off-by: Candice Li Reviewed-by: Tao Zhou Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 14 +++++--------- drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c | 3 +-- drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c | 3 +-- drivers/gpu/drm/amd/amdgpu/umc_v6_7.c | 2 +- 4 files changed, 8 insertions(+), 14 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 fc42fb6ee191..9a64584318e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1067,8 +1067,7 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, mcm_info = &err_info->mcm_info; if (err_info->ce_count) { dev_info(adev->dev, "socket: %d, die: %d, " - "%lld new correctable hardware errors detected in %s block, " - "no user action is needed\n", + "%lld new correctable hardware errors detected in %s block\n", mcm_info->socket_id, mcm_info->die_id, err_info->ce_count, @@ -1080,8 +1079,7 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, err_info = &err_node->err_info; mcm_info = &err_info->mcm_info; dev_info(adev->dev, "socket: %d, die: %d, " - "%lld correctable hardware errors detected in total in %s block, " - "no user action is needed\n", + "%lld correctable hardware errors detected in total in %s block\n", mcm_info->socket_id, mcm_info->die_id, err_info->ce_count, blk_name); } } @@ -1108,16 +1106,14 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, adev->smuio.funcs->get_die_id) { dev_info(adev->dev, "socket: %d, die: %d " "%ld correctable hardware errors " - "detected in %s block, no user " - "action is needed.\n", + "detected in %s block\n", adev->smuio.funcs->get_socket_id(adev), adev->smuio.funcs->get_die_id(adev), ras_mgr->err_data.ce_count, blk_name); } else { dev_info(adev->dev, "%ld correctable hardware errors " - "detected in %s block, no user " - "action is needed.\n", + "detected in %s block\n", ras_mgr->err_data.ce_count, blk_name); } @@ -1920,7 +1916,7 @@ static void amdgpu_ras_interrupt_poison_creation_handler(struct ras_manager *obj struct amdgpu_iv_entry *entry) { dev_info(obj->adev->dev, - "Poison is created, no user action is needed.\n"); + "Poison is created\n"); } static void amdgpu_ras_interrupt_umc_handler(struct ras_manager *obj, diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c index 6d24c84924cb..19986ff6a48d 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c @@ -401,8 +401,7 @@ static void nbio_v7_4_handle_ras_controller_intr_no_bifring(struct amdgpu_device if (err_data.ce_count) dev_info(adev->dev, "%ld correctable hardware " - "errors detected in %s block, " - "no user action is needed.\n", + "errors detected in %s block\n", obj->err_data.ce_count, get_ras_block_str(adev->nbio.ras_if)); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c index 25a3da83e0fb..e90f33780803 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c @@ -597,8 +597,7 @@ static void nbio_v7_9_handle_ras_controller_intr_no_bifring(struct amdgpu_device if (err_data.ce_count) dev_info(adev->dev, "%ld correctable hardware " - "errors detected in %s block, " - "no user action is needed.\n", + "errors detected in %s block\n", obj->err_data.ce_count, get_ras_block_str(adev->nbio.ras_if)); diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c index 530549314ce4..a3ee3c4c650f 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c @@ -64,7 +64,7 @@ static void umc_v6_7_query_error_status_helper(struct amdgpu_device *adev, uint64_t reg_value; if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1) - dev_info(adev->dev, "Deferred error, no user action is needed.\n"); + dev_info(adev->dev, "Deferred error\n"); if (mc_umc_status) dev_info(adev->dev, "MCA STATUS 0x%llx, umc_reg_offset 0x%x\n", mc_umc_status, umc_reg_offset); -- cgit From fb1e91719983c529f85602fdd08c0b7dbf384b1c Mon Sep 17 00:00:00 2001 From: Candice Li Date: Thu, 4 Jan 2024 10:27:10 +0800 Subject: drm/amdgpu: Support poison error injection via ras_ctrl debugfs Support poison error injection. Signed-off-by: Candice Li Reviewed-by: Tao Zhou 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 9a64584318e0..3f7f5c12961d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -305,11 +305,13 @@ static int amdgpu_ras_debugfs_ctrl_parse_data(struct file *f, return -EINVAL; data->head.block = block_id; - /* only ue and ce errors are supported */ + /* only ue, ce and poison errors are supported */ if (!memcmp("ue", err, 2)) data->head.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; else if (!memcmp("ce", err, 2)) data->head.type = AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE; + else if (!memcmp("poison", err, 6)) + data->head.type = AMDGPU_RAS_ERROR__POISON; else return -EINVAL; @@ -431,9 +433,10 @@ static void amdgpu_ras_instance_mask_check(struct amdgpu_device *adev, * The block is one of: umc, sdma, gfx, etc. * see ras_block_string[] for details * - * The error type is one of: ue, ce, where, + * The error type is one of: ue, ce and poison where, * ue is multi-uncorrectable * ce is single-correctable + * poison is poison * * 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. -- cgit From 73cb81dc548f154547d9205d5b9603ba10e2a402 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Fri, 29 Dec 2023 14:54:35 +0800 Subject: drm/amdgpu: Packed socket_id to ras feature mask Initialize RAS feature mask bit[31:29] with socket_id. Signed-off-by: Hawking Zhang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 5 +++++ 1 file changed, 5 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 3f7f5c12961d..31823a30dea2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2919,6 +2919,11 @@ int amdgpu_ras_init(struct amdgpu_device *adev) amdgpu_ras_query_poison_mode(adev); + /* Packed socket_id to ras feature mask bits[31:29] */ + if (adev->smuio.funcs && + adev->smuio.funcs->get_socket_id) + con->features |= ((adev->smuio.funcs->get_socket_id(adev)) << 29); + /* Get RAS schema for particular SOC */ con->schema = amdgpu_get_ras_schema(adev); -- cgit From cce4febb274cc16655dd4692fbcebc6c7d7953b5 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Tue, 2 Jan 2024 19:32:56 +0800 Subject: drm/amdgpu: Add ras helper to query boot errors v2 Add ras helper function to query boot time gpu errors. v2: use aqua_vanjaram smn addressing pattern Signed-off-by: Hawking Zhang Reviewed-by: Yang Wang Reviewed-by: Le Ma Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 1 + drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 95 +++++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 15 +++++- 3 files changed, 110 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 031dba7582c5..00bf4a364a92 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1333,6 +1333,7 @@ int emu_soc_asic_init(struct amdgpu_device *adev); #define WREG32_FIELD_OFFSET(reg, offset, field, val) \ WREG32(mm##reg + offset, (RREG32(mm##reg + offset) & ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field)) +#define AMDGPU_GET_REG_FIELD(x, h, l) (((x) & GENMASK_ULL(h, l)) >> (l)) /* * BIOS helpers. */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 31823a30dea2..a86f91838301 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3767,3 +3767,98 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, return 0; } + +#define mmMP0_SMN_C2PMSG_92 0x1609C +#define mmMP0_SMN_C2PMSG_126 0x160BE +static void amdgpu_ras_boot_time_error_reporting(struct amdgpu_device *adev, + u32 instance, u32 boot_error) +{ + u32 socket_id, aid_id, hbm_id; + u32 reg_data; + u64 reg_addr; + + socket_id = AMDGPU_RAS_GPU_ERR_SOCKET_ID(boot_error); + aid_id = AMDGPU_RAS_GPU_ERR_AID_ID(boot_error); + hbm_id = AMDGPU_RAS_GPU_ERR_HBM_ID(boot_error); + + /* The pattern for smn addressing in other SOC could be different from + * the one for aqua_vanjaram. We should revisit the code if the pattern + * is changed. In such case, replace the aqua_vanjaram implementation + * with more common helper */ + reg_addr = (mmMP0_SMN_C2PMSG_92 << 2) + + aqua_vanjaram_encode_ext_smn_addressing(instance); + + reg_data = amdgpu_device_indirect_rreg_ext(adev, reg_addr); + dev_err(adev->dev, "socket: %d, aid: %d, firmware boot failed, fw status is 0x%x\n", + socket_id, aid_id, reg_data); + + if (AMDGPU_RAS_GPU_ERR_MEM_TRAINING(boot_error)) + dev_info(adev->dev, "socket: %d, aid: %d, hbm: %d, memory training failed\n", + socket_id, aid_id, hbm_id); + + if (AMDGPU_RAS_GPU_ERR_FW_LOAD(boot_error)) + dev_info(adev->dev, "socket: %d, aid: %d, firmware load failed at boot time\n", + socket_id, aid_id); + + if (AMDGPU_RAS_GPU_ERR_WAFL_LINK_TRAINING(boot_error)) + dev_info(adev->dev, "socket: %d, aid: %d, wafl link training failed\n", + socket_id, aid_id); + + if (AMDGPU_RAS_GPU_ERR_XGMI_LINK_TRAINING(boot_error)) + dev_info(adev->dev, "socket: %d, aid: %d, xgmi link training failed\n", + socket_id, aid_id); + + if (AMDGPU_RAS_GPU_ERR_USR_CP_LINK_TRAINING(boot_error)) + dev_info(adev->dev, "socket: %d, aid: %d, usr cp link training failed\n", + socket_id, aid_id); + + if (AMDGPU_RAS_GPU_ERR_USR_DP_LINK_TRAINING(boot_error)) + dev_info(adev->dev, "socket: %d, aid: %d, usr dp link training failed\n", + socket_id, aid_id); + + if (AMDGPU_RAS_GPU_ERR_HBM_MEM_TEST(boot_error)) + dev_info(adev->dev, "socket: %d, aid: %d, hbm: %d, hbm memory test failed\n", + socket_id, aid_id, hbm_id); + + if (AMDGPU_RAS_GPU_ERR_HBM_BIST_TEST(boot_error)) + dev_info(adev->dev, "socket: %d, aid: %d, hbm: %d, hbm bist test failed\n", + socket_id, aid_id, hbm_id); +} + +static int amdgpu_ras_wait_for_boot_complete(struct amdgpu_device *adev, + u32 instance, u32 *boot_error) +{ + u32 reg_addr; + u32 reg_data; + int retry_loop; + + /* The pattern for smn addressing in other SOC could be different from + * the one for aqua_vanjaram. We should revisit the code if the pattern + * is changed. In such case, replace the aqua_vanjaram implementation + * with more common helper */ + reg_addr = (mmMP0_SMN_C2PMSG_126 << 2) + + aqua_vanjaram_encode_ext_smn_addressing(instance); + + for (retry_loop = 0; retry_loop < 1000; retry_loop++) { + reg_data = amdgpu_device_indirect_rreg_ext(adev, reg_addr); + if (AMDGPU_RAS_GPU_ERR_BOOT_STATUS(reg_data)) { + *boot_error = reg_data; + return 0; + } + msleep(1); + } + + *boot_error = reg_data; + return -ETIME; +} + +void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances) +{ + u32 boot_error = 0; + u32 i; + + for (i = 0; i < num_instances; i++) { + if (amdgpu_ras_wait_for_boot_complete(adev, i, &boot_error)) + amdgpu_ras_boot_time_error_reporting(adev, i, boot_error); + } +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index c36faf353b46..80e574475aa3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -33,6 +33,19 @@ struct amdgpu_iv_entry; +#define AMDGPU_RAS_GPU_ERR_MEM_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 0, 0) +#define AMDGPU_RAS_GPU_ERR_FW_LOAD(x) AMDGPU_GET_REG_FIELD(x, 1, 1) +#define AMDGPU_RAS_GPU_ERR_WAFL_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 2, 2) +#define AMDGPU_RAS_GPU_ERR_XGMI_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 3, 3) +#define AMDGPU_RAS_GPU_ERR_USR_CP_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 4, 4) +#define AMDGPU_RAS_GPU_ERR_USR_DP_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 5, 5) +#define AMDGPU_RAS_GPU_ERR_HBM_MEM_TEST(x) AMDGPU_GET_REG_FIELD(x, 6, 6) +#define AMDGPU_RAS_GPU_ERR_HBM_BIST_TEST(x) AMDGPU_GET_REG_FIELD(x, 7, 7) +#define AMDGPU_RAS_GPU_ERR_SOCKET_ID(x) AMDGPU_GET_REG_FIELD(x, 10, 8) +#define AMDGPU_RAS_GPU_ERR_AID_ID(x) AMDGPU_GET_REG_FIELD(x, 12, 11) +#define AMDGPU_RAS_GPU_ERR_HBM_ID(x) AMDGPU_GET_REG_FIELD(x, 13, 13) +#define AMDGPU_RAS_GPU_ERR_BOOT_STATUS(x) AMDGPU_GET_REG_FIELD(x, 31, 31) + #define AMDGPU_RAS_FLAG_INIT_BY_VBIOS (0x1 << 0) /* position of instance value in sub_block_index of * ta_ras_trigger_error_input, the sub block uses lower 12 bits @@ -819,5 +832,5 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, struct amdgpu_smuio_mcm_config_info *mcm_info, struct ras_err_addr *err_addr, u64 count); - +void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances); #endif -- cgit From 33dcda51e9bd1b904d034fee3332ded8ad254125 Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Fri, 24 Nov 2023 16:07:56 +0800 Subject: drm/amdgpu: add ACA bank dump debugfs support add ACA bank dump debugfs support Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c | 119 ++++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h | 2 + drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 14 ++++ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 1 + 4 files changed, 136 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c index e091e9ac5edd..6074a529caf7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c @@ -701,3 +701,122 @@ int aca_bank_check_error_codes(struct amdgpu_device *adev, struct aca_bank *bank return -EINVAL; } +int amdgpu_aca_smu_set_debug_mode(struct amdgpu_device *adev, bool en) +{ + struct amdgpu_aca *aca = &adev->aca; + const struct aca_smu_funcs *smu_funcs = aca->smu_funcs; + + if (!smu_funcs || !smu_funcs->set_debug_mode) + return -EOPNOTSUPP; + + return smu_funcs->set_debug_mode(adev, en); +} + +#if defined(CONFIG_DEBUG_FS) +static int amdgpu_aca_smu_debug_mode_set(void *data, u64 val) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)data; + int ret; + + ret = amdgpu_ras_set_aca_debug_mode(adev, val ? true : false); + if (ret) + return ret; + + dev_info(adev->dev, "amdgpu set smu aca debug mode %s success\n", val ? "on" : "off"); + + return 0; +} + +static void aca_dump_entry(struct seq_file *m, struct aca_bank *bank, enum aca_error_type type, int idx) +{ + struct aca_bank_info info; + int i, ret; + + ret = aca_bank_info_decode(bank, &info); + if (ret) + return; + + seq_printf(m, "aca entry[%d].type: %s\n", idx, type == ACA_ERROR_TYPE_UE ? "UE" : "CE"); + seq_printf(m, "aca entry[%d].info: socketid:%d aid:%d hwid:0x%03x mcatype:0x%04x\n", + idx, info.socket_id, info.die_id, info.hwid, info.mcatype); + + for (i = 0; i < ARRAY_SIZE(aca_regs); i++) + seq_printf(m, "aca entry[%d].regs[%d]: 0x%016llx\n", idx, aca_regs[i].reg_idx, bank->regs[aca_regs[i].reg_idx]); +} + +struct aca_dump_context { + struct seq_file *m; + int idx; +}; + +static int handler_aca_bank_dump(struct aca_handle *handle, struct aca_bank *bank, + enum aca_error_type type, void *data) +{ + struct aca_dump_context *ctx = (struct aca_dump_context *)data; + + aca_dump_entry(ctx->m, bank, type, ctx->idx++); + + return handler_aca_log_bank_error(handle, bank, type, NULL); +} + +static int aca_dump_show(struct seq_file *m, enum aca_error_type type) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)m->private; + struct aca_dump_context context = { + .m = m, + .idx = 0, + }; + + return aca_banks_update(adev, type, handler_aca_bank_dump, (void *)&context); +} + +static int aca_dump_ce_show(struct seq_file *m, void *unused) +{ + return aca_dump_show(m, ACA_ERROR_TYPE_CE); +} + +static int aca_dump_ce_open(struct inode *inode, struct file *file) +{ + return single_open(file, aca_dump_ce_show, inode->i_private); +} + +static const struct file_operations aca_ce_dump_debug_fops = { + .owner = THIS_MODULE, + .open = aca_dump_ce_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static int aca_dump_ue_show(struct seq_file *m, void *unused) +{ + return aca_dump_show(m, ACA_ERROR_TYPE_UE); +} + +static int aca_dump_ue_open(struct inode *inode, struct file *file) +{ + return single_open(file, aca_dump_ue_show, inode->i_private); +} + +static const struct file_operations aca_ue_dump_debug_fops = { + .owner = THIS_MODULE, + .open = aca_dump_ue_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +DEFINE_DEBUGFS_ATTRIBUTE(aca_debug_mode_fops, NULL, amdgpu_aca_smu_debug_mode_set, "%llu\n"); +#endif + +void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root) +{ +#if defined(CONFIG_DEBUG_FS) + if (!root || adev->ip_versions[MP1_HWIP][0] != IP_VERSION(13, 0, 6)) + return; + + debugfs_create_file("aca_debug_mode", 0200, root, adev, &aca_debug_mode_fops); + debugfs_create_file("aca_ue_dump", 0400, root, adev, &aca_ue_dump_debug_fops); + debugfs_create_file("aca_ce_dump", 0400, root, adev, &aca_ce_dump_debug_fops); +#endif +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h index 8aa3d8c4136c..ac641952ab71 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h @@ -192,4 +192,6 @@ int amdgpu_aca_add_handle(struct amdgpu_device *adev, struct aca_handle *handle, void amdgpu_aca_remove_handle(struct aca_handle *handle); int amdgpu_aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle, enum aca_error_type type, void *data); +int amdgpu_aca_smu_set_debug_mode(struct amdgpu_device *adev, bool en); +void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index a86f91838301..ded19182dc79 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3431,6 +3431,20 @@ int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable) return ret; } +int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + int ret = 0; + + if (con) { + ret = amdgpu_aca_smu_set_debug_mode(adev, enable); + if (!ret) + con->is_mca_debug_mode = enable; + } + + return ret; +} + bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 80e574475aa3..9edc3909f17d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -794,6 +794,7 @@ struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev); int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con); +int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable); int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable); bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev); bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev, -- cgit From 04c4fcd2630d400959f791a598070dab9d2133cd Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Fri, 24 Nov 2023 14:41:06 +0800 Subject: drm/amdgpu: add amdgpu ras aca query interface v1: add ACA error query interface v2: Add a new helper function to determine whether to use ACA or MCA. Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c | 8 +++ drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h | 2 + drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 105 +++++++++++++++++++++++++++----- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 12 +++- 4 files changed, 109 insertions(+), 18 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c index 6074a529caf7..f106a04363ea 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c @@ -577,6 +577,9 @@ int amdgpu_aca_add_handle(struct amdgpu_device *adev, struct aca_handle *handle, { struct amdgpu_aca *aca = &adev->aca; + if (!amdgpu_aca_is_enabled(adev)) + return 0; + return add_aca_handle(adev, &aca->mgr, handle, name, ras_info, data); } @@ -613,6 +616,11 @@ static void aca_manager_fini(struct aca_handle_manager *mgr) remove_aca(handle); } +bool amdgpu_aca_is_enabled(struct amdgpu_device *adev) +{ + return adev->aca.is_enabled; +} + int amdgpu_aca_init(struct amdgpu_device *adev) { struct amdgpu_aca *aca = &adev->aca; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h index ac641952ab71..f6c72371244e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h @@ -172,6 +172,7 @@ struct aca_smu_funcs { struct amdgpu_aca { struct aca_handle_manager mgr; const struct aca_smu_funcs *smu_funcs; + bool is_enabled; }; struct aca_info { @@ -183,6 +184,7 @@ struct aca_info { int amdgpu_aca_init(struct amdgpu_device *adev); void amdgpu_aca_fini(struct amdgpu_device *adev); void amdgpu_aca_set_smu_funcs(struct amdgpu_device *adev, const struct aca_smu_funcs *smu_funcs); +bool amdgpu_aca_is_enabled(struct amdgpu_device *adev); int aca_bank_info_decode(struct aca_bank *bank, struct aca_bank_info *info); int aca_bank_check_error_codes(struct amdgpu_device *adev, struct aca_bank *bank, int *err_codes, int size); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index ded19182dc79..bb93e4e3969f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1167,6 +1167,53 @@ static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, s } } +static struct ras_manager *get_ras_manager(struct amdgpu_device *adev, enum amdgpu_ras_block blk) +{ + struct ras_common_if head; + + memset(&head, 0, sizeof(head)); + head.block = blk; + + return amdgpu_ras_find_obj(adev, &head); +} + +int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk, + const struct aca_info *aca_info, void *data) +{ + struct ras_manager *obj; + + obj = get_ras_manager(adev, blk); + if (!obj) + return -EINVAL; + + return amdgpu_aca_add_handle(adev, &obj->aca_handle, ras_block_str(blk), aca_info, data); +} + +int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk) +{ + struct ras_manager *obj; + + obj = get_ras_manager(adev, blk); + if (!obj) + return -EINVAL; + + amdgpu_aca_remove_handle(&obj->aca_handle); + + return 0; +} + +static int amdgpu_aca_log_ras_error_data(struct amdgpu_device *adev, enum amdgpu_ras_block blk, + enum aca_error_type type, struct ras_err_data *err_data) +{ + struct ras_manager *obj; + + obj = get_ras_manager(adev, blk); + if (!obj) + return -EINVAL; + + return amdgpu_aca_get_error_data(adev, &obj->aca_handle, type, err_data); +} + static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev, struct ras_query_if *info, struct ras_err_data *err_data, @@ -1174,6 +1221,7 @@ static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev, { enum amdgpu_ras_block blk = info ? info->head.block : AMDGPU_RAS_BLOCK_COUNT; struct amdgpu_ras_block_object *block_obj = NULL; + int ret; if (blk == AMDGPU_RAS_BLOCK_COUNT) return -EINVAL; @@ -1203,9 +1251,19 @@ static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev, } } } else { - /* FIXME: add code to check return value later */ - amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_UE, err_data); - amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_CE, err_data); + if (amdgpu_aca_is_enabled(adev)) { + ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_UE, err_data); + if (ret) + return ret; + + ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_CE, err_data); + if (ret) + return ret; + } else { + /* FIXME: add code to check return value later */ + amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_UE, err_data); + amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_CE, err_data); + } } return 0; @@ -1254,6 +1312,7 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0); struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs; + const struct aca_smu_funcs *smu_funcs = adev->aca.smu_funcs; struct amdgpu_hive_info *hive; int hive_ras_recovery = 0; @@ -1264,7 +1323,7 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, } if (!amdgpu_ras_is_supported(adev, block) || - !amdgpu_ras_get_mca_debug_mode(adev)) + !amdgpu_ras_get_aca_debug_mode(adev)) return -EOPNOTSUPP; hive = amdgpu_get_xgmi_hive(adev); @@ -1276,7 +1335,8 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, /* skip ras error reset in gpu reset */ if ((amdgpu_in_reset(adev) || atomic_read(&ras->in_recovery) || hive_ras_recovery) && - mca_funcs && mca_funcs->mca_set_debug_mode) + ((smu_funcs && smu_funcs->set_debug_mode) || + (mca_funcs && mca_funcs->mca_set_debug_mode))) return -EOPNOTSUPP; if (block_obj->hw_ops->reset_ras_error_count) @@ -1772,7 +1832,10 @@ void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev) } } - amdgpu_mca_smu_debugfs_init(adev, dir); + if (amdgpu_aca_is_enabled(adev)) + amdgpu_aca_smu_debugfs_init(adev, dir); + else + amdgpu_mca_smu_debugfs_init(adev, dir); } /* debugfs end */ @@ -2753,6 +2816,9 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) adev->ras_enabled = amdgpu_ras_enable == 0 ? 0 : adev->ras_hw_enabled & amdgpu_ras_mask; + + /* aca is disabled by default */ + adev->aca.is_enabled = false; } static void amdgpu_ras_counte_dw(struct work_struct *work) @@ -3142,7 +3208,10 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) return 0; - amdgpu_ras_set_mca_debug_mode(adev, false); + if (amdgpu_aca_is_enabled(adev)) + amdgpu_ras_set_aca_debug_mode(adev, false); + else + amdgpu_ras_set_mca_debug_mode(adev, false); list_for_each_entry_safe(node, tmp, &adev->ras_list, node) { if (!node->ras_obj) { @@ -3425,7 +3494,7 @@ int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable) if (con) { ret = amdgpu_mca_smu_set_debug_mode(adev, enable); if (!ret) - con->is_mca_debug_mode = enable; + con->is_aca_debug_mode = enable; } return ret; @@ -3437,24 +3506,29 @@ int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable) int ret = 0; if (con) { - ret = amdgpu_aca_smu_set_debug_mode(adev, enable); + if (amdgpu_aca_is_enabled(adev)) + ret = amdgpu_aca_smu_set_debug_mode(adev, enable); + else + ret = amdgpu_mca_smu_set_debug_mode(adev, enable); if (!ret) - con->is_mca_debug_mode = enable; + con->is_aca_debug_mode = enable; } return ret; } -bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev) +bool amdgpu_ras_get_aca_debug_mode(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + const struct aca_smu_funcs *smu_funcs = adev->aca.smu_funcs; const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs; if (!con) return false; - if (mca_funcs && mca_funcs->mca_set_debug_mode) - return con->is_mca_debug_mode; + if ((amdgpu_aca_is_enabled(adev) && smu_funcs && smu_funcs->set_debug_mode) || + (!amdgpu_aca_is_enabled(adev) && mca_funcs && mca_funcs->mca_set_debug_mode)) + return con->is_aca_debug_mode; else return true; } @@ -3464,15 +3538,16 @@ bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev, { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs; + const struct aca_smu_funcs *smu_funcs = adev->aca.smu_funcs; if (!con) { *error_query_mode = AMDGPU_RAS_INVALID_ERROR_QUERY; return false; } - if (mca_funcs && mca_funcs->mca_set_debug_mode) + if ((smu_funcs && smu_funcs->set_debug_mode) || (mca_funcs && mca_funcs->mca_set_debug_mode)) *error_query_mode = - (con->is_mca_debug_mode) ? AMDGPU_RAS_DIRECT_ERROR_QUERY : AMDGPU_RAS_FIRMWARE_ERROR_QUERY; + (con->is_aca_debug_mode) ? AMDGPU_RAS_DIRECT_ERROR_QUERY : AMDGPU_RAS_FIRMWARE_ERROR_QUERY; else *error_query_mode = AMDGPU_RAS_DIRECT_ERROR_QUERY; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 9edc3909f17d..1d568a25f1de 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -455,7 +455,7 @@ struct amdgpu_ras { /* Indicates smu whether need update bad channel info */ bool update_channel_flag; /* Record status of smu mca debug mode */ - bool is_mca_debug_mode; + bool is_aca_debug_mode; /* Record special requirements of gpu reset caller */ uint32_t gpu_reset_flags; @@ -543,6 +543,8 @@ struct ras_manager { struct ras_ih_data ih_data; struct ras_err_data err_data; + + struct aca_handle aca_handle; }; struct ras_badpage { @@ -794,9 +796,9 @@ struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev); int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con); -int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable); int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable); -bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev); +int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable); +bool amdgpu_ras_get_aca_debug_mode(struct amdgpu_device *adev); bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev, unsigned int *mode); @@ -834,4 +836,8 @@ int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, struct amdgpu_smuio_mcm_config_info *mcm_info, struct ras_err_addr *err_addr, u64 count); void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances); +int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk, + const struct aca_info *aca_info, void *data); +int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk); + #endif -- cgit From 37973b69eab43d46dd316372b5739cc4c52a42fd Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Tue, 2 Jan 2024 10:43:23 +0800 Subject: drm/amdgpu: add aca sysfs support add aca sysfs node support Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c | 32 +++++++++++++++++++++++++++++++- drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h | 2 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 15 +++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 3 +++ 4 files changed, 51 insertions(+), 1 deletion(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c index f106a04363ea..8a3c3a49415d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c @@ -572,15 +572,45 @@ static int add_aca_handle(struct amdgpu_device *adev, struct aca_handle_manager return 0; } +static ssize_t aca_sysfs_read(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct aca_handle *handle = container_of(attr, struct aca_handle, aca_attr); + + /* NOTE: the aca cache will be auto cleared once read, + * So the driver should unify the query entry point, forward request to ras query interface directly */ + return amdgpu_ras_aca_sysfs_read(dev, attr, handle, buf, handle->data); +} + +static int add_aca_sysfs(struct amdgpu_device *adev, struct aca_handle *handle) +{ + struct device_attribute *aca_attr = &handle->aca_attr; + + snprintf(handle->attr_name, sizeof(handle->attr_name) - 1, "aca_%s", handle->name); + aca_attr->show = aca_sysfs_read; + aca_attr->attr.name = handle->attr_name; + aca_attr->attr.mode = S_IRUGO; + sysfs_attr_init(&aca_attr->attr); + + return sysfs_add_file_to_group(&adev->dev->kobj, + &aca_attr->attr, + "ras"); +} + int amdgpu_aca_add_handle(struct amdgpu_device *adev, struct aca_handle *handle, const char *name, const struct aca_info *ras_info, void *data) { struct amdgpu_aca *aca = &adev->aca; + int ret; if (!amdgpu_aca_is_enabled(adev)) return 0; - return add_aca_handle(adev, &aca->mgr, handle, name, ras_info, data); + ret = add_aca_handle(adev, &aca->mgr, handle, name, ras_info, data); + if (ret) + return ret; + + return add_aca_sysfs(adev, handle); } static void remove_aca(struct aca_handle *handle) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h index f6c72371244e..6e9a35eda683 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h @@ -149,6 +149,8 @@ struct aca_handle { struct aca_handle_manager *mgr; struct aca_error_cache error_cache; const struct aca_bank_ops *bank_ops; + struct device_attribute aca_attr; + char attr_name[64]; const char *name; u32 mask; void *data; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index bb93e4e3969f..bc459dff42f3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1214,6 +1214,21 @@ static int amdgpu_aca_log_ras_error_data(struct amdgpu_device *adev, enum amdgpu return amdgpu_aca_get_error_data(adev, &obj->aca_handle, type, err_data); } +ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr, + struct aca_handle *handle, char *buf, void *data) +{ + struct ras_manager *obj = container_of(handle, struct ras_manager, aca_handle); + struct ras_query_if info = { + .head = obj->head, + }; + + if (amdgpu_ras_query_error_status(obj->adev, &info)) + return -EINVAL; + + return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count, + "ce", info.ce_count); +} + static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev, struct ras_query_if *info, struct ras_err_data *err_data, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 1d568a25f1de..394394d98bc3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -840,4 +840,7 @@ int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk, const struct aca_info *aca_info, void *data); int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk); +ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr, + struct aca_handle *handle, char *buf, void *data); + #endif -- cgit From 46e2231ce03bc84303fc036562b088849e137a05 Mon Sep 17 00:00:00 2001 From: Candice Li Date: Wed, 3 Jan 2024 15:19:19 +0800 Subject: drm/amdgpu: Log deferred error separately Separate deferred error from UE and CE and log it individually. Signed-off-by: Candice Li Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c | 11 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 116 +++++++++++++++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 6 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 1 + drivers/gpu/drm/amd/amdgpu/umc_v12_0.c | 60 +++++------ drivers/gpu/drm/amd/amdgpu/umc_v12_0.h | 3 + .../gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c | 6 +- 7 files changed, 142 insertions(+), 61 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c index 59fafb8392e0..666fd8fa39ad 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c @@ -256,9 +256,14 @@ int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_blo if (type == AMDGPU_MCA_ERROR_TYPE_UE) amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, &err_addr, (uint64_t)count); - else - amdgpu_ras_error_statistic_ce_count(err_data, - &mcm_info, &err_addr, (uint64_t)count); + else { + if (!!(MCA_REG__STATUS__DEFERRED(entry->regs[MCA_REG_IDX_STATUS]))) + amdgpu_ras_error_statistic_de_count(err_data, + &mcm_info, &err_addr, (uint64_t)count); + else + amdgpu_ras_error_statistic_ce_count(err_data, + &mcm_info, &err_addr, (uint64_t)count); + } } out_mca_release: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index bc459dff42f3..3b1c6b77152e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1036,7 +1036,8 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, struct ras_manager *ras_mgr, struct ras_err_data *err_data, const char *blk_name, - bool is_ue) + bool is_ue, + bool is_de) { struct amdgpu_smuio_mcm_config_info *mcm_info; struct ras_err_node *err_node; @@ -1065,25 +1066,50 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev, } } else { - for_each_ras_error(err_node, err_data) { - err_info = &err_node->err_info; - mcm_info = &err_info->mcm_info; - if (err_info->ce_count) { + if (is_de) { + for_each_ras_error(err_node, err_data) { + err_info = &err_node->err_info; + mcm_info = &err_info->mcm_info; + if (err_info->de_count) { + dev_info(adev->dev, "socket: %d, die: %d, " + "%lld new deferred hardware errors detected in %s block\n", + mcm_info->socket_id, + mcm_info->die_id, + err_info->de_count, + blk_name); + } + } + + for_each_ras_error(err_node, &ras_mgr->err_data) { + err_info = &err_node->err_info; + mcm_info = &err_info->mcm_info; dev_info(adev->dev, "socket: %d, die: %d, " - "%lld new correctable hardware errors detected in %s block\n", - mcm_info->socket_id, - mcm_info->die_id, - err_info->ce_count, - blk_name); + "%lld deferred hardware errors detected in total in %s block\n", + mcm_info->socket_id, mcm_info->die_id, + err_info->de_count, blk_name); + } + } else { + for_each_ras_error(err_node, err_data) { + err_info = &err_node->err_info; + mcm_info = &err_info->mcm_info; + if (err_info->ce_count) { + dev_info(adev->dev, "socket: %d, die: %d, " + "%lld new correctable hardware errors detected in %s block\n", + mcm_info->socket_id, + mcm_info->die_id, + err_info->ce_count, + blk_name); + } } - } - for_each_ras_error(err_node, &ras_mgr->err_data) { - err_info = &err_node->err_info; - mcm_info = &err_info->mcm_info; - dev_info(adev->dev, "socket: %d, die: %d, " - "%lld correctable hardware errors detected in total in %s block\n", - mcm_info->socket_id, mcm_info->die_id, err_info->ce_count, blk_name); + for_each_ras_error(err_node, &ras_mgr->err_data) { + err_info = &err_node->err_info; + mcm_info = &err_info->mcm_info; + dev_info(adev->dev, "socket: %d, die: %d, " + "%lld correctable hardware errors detected in total in %s block\n", + mcm_info->socket_id, mcm_info->die_id, + err_info->ce_count, blk_name); + } } } } @@ -1102,7 +1128,8 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, if (err_data->ce_count) { if (err_data_has_source_info(err_data)) { - amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, blk_name, false); + amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, + blk_name, false, false); } else if (!adev->aid_mask && adev->smuio.funcs && adev->smuio.funcs->get_socket_id && @@ -1124,7 +1151,8 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, if (err_data->ue_count) { if (err_data_has_source_info(err_data)) { - amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, blk_name, true); + amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, + blk_name, true, false); } else if (!adev->aid_mask && adev->smuio.funcs && adev->smuio.funcs->get_socket_id && @@ -1144,6 +1172,28 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev, } } + if (err_data->de_count) { + if (err_data_has_source_info(err_data)) { + amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, + blk_name, false, true); + } else if (!adev->aid_mask && + adev->smuio.funcs && + adev->smuio.funcs->get_socket_id && + adev->smuio.funcs->get_die_id) { + dev_info(adev->dev, "socket: %d, die: %d " + "%ld deferred hardware errors " + "detected in %s block\n", + adev->smuio.funcs->get_socket_id(adev), + adev->smuio.funcs->get_die_id(adev), + ras_mgr->err_data.de_count, + blk_name); + } else { + dev_info(adev->dev, "%ld deferred hardware errors " + "detected in %s block\n", + ras_mgr->err_data.de_count, + blk_name); + } + } } static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, struct ras_err_data *err_data) @@ -1154,7 +1204,8 @@ static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, s if (err_data_has_source_info(err_data)) { for_each_ras_error(err_node, err_data) { err_info = &err_node->err_info; - + amdgpu_ras_error_statistic_de_count(&obj->err_data, + &err_info->mcm_info, NULL, err_info->de_count); amdgpu_ras_error_statistic_ce_count(&obj->err_data, &err_info->mcm_info, NULL, err_info->ce_count); amdgpu_ras_error_statistic_ue_count(&obj->err_data, @@ -1164,6 +1215,7 @@ static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, s /* for legacy asic path which doesn't has error source info */ obj->err_data.ue_count += err_data->ue_count; obj->err_data.ce_count += err_data->ce_count; + obj->err_data.de_count += err_data->de_count; } } @@ -1312,6 +1364,7 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_i info->ue_count = obj->err_data.ue_count; info->ce_count = obj->err_data.ce_count; + info->de_count = obj->err_data.de_count; amdgpu_ras_error_generate_report(adev, info, &err_data); @@ -2029,6 +2082,7 @@ static void amdgpu_ras_interrupt_umc_handler(struct ras_manager *obj, */ obj->err_data.ue_count += err_data.ue_count; obj->err_data.ce_count += err_data.ce_count; + obj->err_data.de_count += err_data.de_count; } amdgpu_ras_error_data_fini(&err_data); @@ -3872,6 +3926,28 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, return 0; } +int amdgpu_ras_error_statistic_de_count(struct ras_err_data *err_data, + struct amdgpu_smuio_mcm_config_info *mcm_info, + struct ras_err_addr *err_addr, u64 count) +{ + struct ras_err_info *err_info; + + if (!err_data || !mcm_info) + return -EINVAL; + + if (!count) + return 0; + + err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr); + if (!err_info) + return -EINVAL; + + err_info->de_count += count; + err_data->de_count += count; + + return 0; +} + #define mmMP0_SMN_C2PMSG_92 0x1609C #define mmMP0_SMN_C2PMSG_126 0x160BE static void amdgpu_ras_boot_time_error_reporting(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 394394d98bc3..d6197bb39c61 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -476,6 +476,7 @@ struct ras_err_info { struct amdgpu_smuio_mcm_config_info mcm_info; u64 ce_count; u64 ue_count; + u64 de_count; struct ras_err_addr err_addr; }; @@ -487,6 +488,7 @@ struct ras_err_node { struct ras_err_data { unsigned long ue_count; unsigned long ce_count; + unsigned long de_count; unsigned long err_addr_cnt; struct eeprom_table_record *err_addr; u32 err_list_count; @@ -564,6 +566,7 @@ struct ras_query_if { struct ras_common_if head; unsigned long ue_count; unsigned long ce_count; + unsigned long de_count; }; struct ras_inject_if { @@ -835,6 +838,9 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, struct amdgpu_smuio_mcm_config_info *mcm_info, struct ras_err_addr *err_addr, u64 count); +int amdgpu_ras_error_statistic_de_count(struct ras_err_data *err_data, + struct amdgpu_smuio_mcm_config_info *mcm_info, + struct ras_err_addr *err_addr, u64 count); void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances); int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk, const struct aca_info *aca_info, void *data); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index 3932ac81a67c..9d1cf41cf483 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -208,6 +208,7 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset) if (ret == AMDGPU_RAS_SUCCESS && obj) { obj->err_data.ue_count += err_data.ue_count; obj->err_data.ce_count += err_data.ce_count; + obj->err_data.de_count += err_data.de_count; } amdgpu_ras_error_data_fini(&err_data); diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c index a5dd73b3fc19..6423dca5b777 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c @@ -89,12 +89,17 @@ static void umc_v12_0_reset_error_count(struct amdgpu_device *adev) umc_v12_0_reset_error_count_per_channel, NULL); } +bool umc_v12_0_is_deferred_error(struct amdgpu_device *adev, uint64_t mc_umc_status) +{ + return (amdgpu_ras_is_poison_mode_supported(adev) && + (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) && + (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1)); +} + bool umc_v12_0_is_uncorrectable_error(struct amdgpu_device *adev, uint64_t mc_umc_status) { - if (amdgpu_ras_is_poison_mode_supported(adev) && - (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) && - (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1)) - return true; + if (umc_v12_0_is_deferred_error(adev, mc_umc_status)) + return false; return ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) && (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 || @@ -104,9 +109,7 @@ bool umc_v12_0_is_uncorrectable_error(struct amdgpu_device *adev, uint64_t mc_um bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_status) { - if (amdgpu_ras_is_poison_mode_supported(adev) && - (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) && - (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1)) + if (umc_v12_0_is_deferred_error(adev, mc_umc_status)) return false; return (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 && @@ -119,29 +122,10 @@ bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_ !(umc_v12_0_is_uncorrectable_error(adev, mc_umc_status))))); } -static void umc_v12_0_query_correctable_error_count(struct amdgpu_device *adev, +static void umc_v12_0_query_error_count_per_type(struct amdgpu_device *adev, uint64_t umc_reg_offset, - unsigned long *error_count) -{ - uint64_t mc_umc_status; - uint64_t mc_umc_status_addr; - - mc_umc_status_addr = - SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0); - - /* Rely on MCUMC_STATUS for correctable error counter - * MCUMC_STATUS is a 64 bit register - */ - mc_umc_status = - RREG64_PCIE_EXT((mc_umc_status_addr + umc_reg_offset) * 4); - - if (umc_v12_0_is_correctable_error(adev, mc_umc_status)) - *error_count += 1; -} - -static void umc_v12_0_query_uncorrectable_error_count(struct amdgpu_device *adev, - uint64_t umc_reg_offset, - unsigned long *error_count) + unsigned long *error_count, + check_error_type_func error_type_func) { uint64_t mc_umc_status; uint64_t mc_umc_status_addr; @@ -149,11 +133,11 @@ static void umc_v12_0_query_uncorrectable_error_count(struct amdgpu_device *adev mc_umc_status_addr = SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0); - /* Check the MCUMC_STATUS. */ + /* Check MCUMC_STATUS */ mc_umc_status = RREG64_PCIE_EXT((mc_umc_status_addr + umc_reg_offset) * 4); - if (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status)) + if (error_type_func(adev, mc_umc_status)) *error_count += 1; } @@ -162,7 +146,7 @@ static int umc_v12_0_query_error_count(struct amdgpu_device *adev, uint32_t ch_inst, void *data) { struct ras_err_data *err_data = (struct ras_err_data *)data; - unsigned long ue_count = 0, ce_count = 0; + unsigned long ue_count = 0, ce_count = 0, de_count = 0; /* NOTE: node_inst is converted by adev->umc.active_mask and the range is [0-3], * which can be used as die ID directly */ @@ -174,11 +158,16 @@ static int umc_v12_0_query_error_count(struct amdgpu_device *adev, uint64_t umc_reg_offset = get_umc_v12_0_reg_offset(adev, node_inst, umc_inst, ch_inst); - umc_v12_0_query_correctable_error_count(adev, umc_reg_offset, &ce_count); - umc_v12_0_query_uncorrectable_error_count(adev, umc_reg_offset, &ue_count); + umc_v12_0_query_error_count_per_type(adev, umc_reg_offset, + &ce_count, umc_v12_0_is_correctable_error); + umc_v12_0_query_error_count_per_type(adev, umc_reg_offset, + &ue_count, umc_v12_0_is_uncorrectable_error); + umc_v12_0_query_error_count_per_type(adev, umc_reg_offset, + &de_count, umc_v12_0_is_deferred_error); amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, NULL, ue_count); amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, NULL, ce_count); + amdgpu_ras_error_statistic_de_count(err_data, &mcm_info, NULL, de_count); return 0; } @@ -392,7 +381,8 @@ static void umc_v12_0_ecc_info_query_ras_error_address(struct amdgpu_device *ade if (!mc_umc_status) continue; - if (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status)) { + if (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status) || + umc_v12_0_is_deferred_error(adev, mc_umc_status)) { uint64_t mca_addr, err_addr, mca_ipid; uint32_t InstanceIdLo; struct amdgpu_smuio_mcm_config_info *mcm_info; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h index e8de3a92251a..5973bfb14fce 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h @@ -121,9 +121,12 @@ (((_ipid_lo) >> 12) & 0xF)) #define MCA_IPID_LO_2_UMC_INST(_ipid_lo) (((_ipid_lo) >> 21) & 0x7) +bool umc_v12_0_is_deferred_error(struct amdgpu_device *adev, uint64_t mc_umc_status); bool umc_v12_0_is_uncorrectable_error(struct amdgpu_device *adev, uint64_t mc_umc_status); bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_status); +typedef bool (*check_error_type_func)(struct amdgpu_device *adev, uint64_t mc_umc_status); + extern const uint32_t umc_v12_0_channel_idx_tbl[] [UMC_V12_0_UMC_INSTANCE_NUM] diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c index c813b3c7940a..952a983da49a 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c @@ -2557,9 +2557,9 @@ static int mca_umc_mca_get_err_count(const struct mca_ras_info *mca_ras, struct return 0; } - if (type == AMDGPU_MCA_ERROR_TYPE_UE && umc_v12_0_is_uncorrectable_error(adev, status0)) - *count = 1; - else if (type == AMDGPU_MCA_ERROR_TYPE_CE && umc_v12_0_is_correctable_error(adev, status0)) + if ((type == AMDGPU_MCA_ERROR_TYPE_UE && umc_v12_0_is_uncorrectable_error(adev, status0)) || + (type == AMDGPU_MCA_ERROR_TYPE_CE && (umc_v12_0_is_correctable_error(adev, status0) || + umc_v12_0_is_deferred_error(adev, status0)))) *count = 1; return 0; -- cgit From 4e2965bd3bdd45c9d514290971f3cfff22c8b484 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Fri, 29 Dec 2023 14:37:45 +0800 Subject: drm/amdgpu: Centralize ras cap query to amdgpu_ras_check_supported Move ras capablity check to amdgpu_ras_check_supported. Driver will query ras capablity through psp interace, or vbios interface, or specific ip callbacks. Signed-off-by: Hawking Zhang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 170 +++++++++++++++++--------------- 1 file changed, 93 insertions(+), 77 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 3b1c6b77152e..5063e8d6dff6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -39,6 +39,7 @@ #include "nbio_v7_9.h" #include "atom.h" #include "amdgpu_reset.h" +#include "amdgpu_psp.h" #ifdef CONFIG_X86_MCE_AMD #include @@ -2811,6 +2812,87 @@ static void amdgpu_ras_get_quirks(struct amdgpu_device *adev) adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__GFX); } +/* Query ras capablity via atomfirmware interface */ +static void amdgpu_ras_query_ras_capablity_from_vbios(struct amdgpu_device *adev) +{ + /* mem_ecc cap */ + if (amdgpu_atomfirmware_mem_ecc_supported(adev)) { + dev_info(adev->dev, "MEM ECC is active.\n"); + adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__UMC | + 1 << AMDGPU_RAS_BLOCK__DF); + } else { + dev_info(adev->dev, "MEM ECC is not presented.\n"); + } + + /* sram_ecc cap */ + if (amdgpu_atomfirmware_sram_ecc_supported(adev)) { + dev_info(adev->dev, "SRAM ECC is active.\n"); + if (!amdgpu_sriov_vf(adev)) + adev->ras_hw_enabled |= ~(1 << AMDGPU_RAS_BLOCK__UMC | + 1 << AMDGPU_RAS_BLOCK__DF); + else + adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__PCIE_BIF | + 1 << AMDGPU_RAS_BLOCK__SDMA | + 1 << AMDGPU_RAS_BLOCK__GFX); + + /* + * VCN/JPEG RAS can be supported on both bare metal and + * SRIOV environment + */ + if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(2, 6, 0) || + amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 0) || + amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3)) + adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN | + 1 << AMDGPU_RAS_BLOCK__JPEG); + else + adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__VCN | + 1 << AMDGPU_RAS_BLOCK__JPEG); + + /* + * XGMI RAS is not supported if xgmi num physical nodes + * is zero + */ + if (!adev->gmc.xgmi.num_physical_nodes) + adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__XGMI_WAFL); + } else { + dev_info(adev->dev, "SRAM ECC is not presented.\n"); + } +} + +/* Query poison mode from umc/df IP callbacks */ +static void amdgpu_ras_query_poison_mode(struct amdgpu_device *adev) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + bool df_poison, umc_poison; + + /* poison setting is useless on SRIOV guest */ + if (amdgpu_sriov_vf(adev) || !con) + return; + + /* Init poison supported flag, the default value is false */ + if (adev->gmc.xgmi.connected_to_cpu || + adev->gmc.is_app_apu) { + /* enabled by default when GPU is connected to CPU */ + con->poison_supported = true; + } else if (adev->df.funcs && + adev->df.funcs->query_ras_poison_mode && + adev->umc.ras && + adev->umc.ras->query_ras_poison_mode) { + df_poison = + adev->df.funcs->query_ras_poison_mode(adev); + umc_poison = + adev->umc.ras->query_ras_poison_mode(adev); + + /* Only poison is set in both DF and UMC, we can support it */ + if (df_poison && umc_poison) + con->poison_supported = true; + else if (df_poison != umc_poison) + dev_warn(adev->dev, + "Poison setting is inconsistent in DF/UMC(%d:%d)!\n", + df_poison, umc_poison); + } +} + /* * check hardware's ras ability which will be saved in hw_supported. * if hardware does not support ras, we can skip some ras initializtion and @@ -2827,49 +2909,13 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) if (!amdgpu_ras_asic_supported(adev)) return; - if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) { - if (amdgpu_atomfirmware_mem_ecc_supported(adev)) { - dev_info(adev->dev, "MEM ECC is active.\n"); - adev->ras_hw_enabled |= (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"); - if (!amdgpu_sriov_vf(adev)) - adev->ras_hw_enabled |= ~(1 << AMDGPU_RAS_BLOCK__UMC | - 1 << AMDGPU_RAS_BLOCK__DF); - else - adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__PCIE_BIF | - 1 << AMDGPU_RAS_BLOCK__SDMA | - 1 << AMDGPU_RAS_BLOCK__GFX); - - /* VCN/JPEG RAS can be supported on both bare metal and - * SRIOV environment - */ - if (amdgpu_ip_version(adev, VCN_HWIP, 0) == - IP_VERSION(2, 6, 0) || - amdgpu_ip_version(adev, VCN_HWIP, 0) == - IP_VERSION(4, 0, 0) || - amdgpu_ip_version(adev, VCN_HWIP, 0) == - IP_VERSION(4, 0, 3)) - adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN | - 1 << AMDGPU_RAS_BLOCK__JPEG); - else - adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__VCN | - 1 << AMDGPU_RAS_BLOCK__JPEG); + /* query ras capability from psp */ + if (amdgpu_psp_get_ras_capability(&adev->psp)) + goto init_ras_enabled_flag; - /* - * XGMI RAS is not supported if xgmi num physical nodes - * is zero - */ - if (!adev->gmc.xgmi.num_physical_nodes) - adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__XGMI_WAFL); - } else { - dev_info(adev->dev, "SRAM ECC is not presented.\n"); - } + /* query ras capablity from bios */ + if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) { + amdgpu_ras_query_ras_capablity_from_vbios(adev); } else { /* driver only manages a few IP blocks RAS feature * when GPU is connected cpu through XGMI */ @@ -2878,8 +2924,13 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) 1 << AMDGPU_RAS_BLOCK__MMHUB); } + /* apply asic specific settings (vega20 only for now) */ amdgpu_ras_get_quirks(adev); + /* query poison mode from umc/df ip callback */ + amdgpu_ras_query_poison_mode(adev); + +init_ras_enabled_flag: /* hw_supported needs to be aligned with RAS block mask. */ adev->ras_hw_enabled &= AMDGPU_RAS_BLOCK_MASK; @@ -2915,39 +2966,6 @@ Out: pm_runtime_put_autosuspend(dev->dev); } -static void amdgpu_ras_query_poison_mode(struct amdgpu_device *adev) -{ - struct amdgpu_ras *con = amdgpu_ras_get_context(adev); - bool df_poison, umc_poison; - - /* poison setting is useless on SRIOV guest */ - if (amdgpu_sriov_vf(adev) || !con) - return; - - /* Init poison supported flag, the default value is false */ - if (adev->gmc.xgmi.connected_to_cpu || - adev->gmc.is_app_apu) { - /* enabled by default when GPU is connected to CPU */ - con->poison_supported = true; - } else if (adev->df.funcs && - adev->df.funcs->query_ras_poison_mode && - adev->umc.ras && - adev->umc.ras->query_ras_poison_mode) { - df_poison = - adev->df.funcs->query_ras_poison_mode(adev); - umc_poison = - adev->umc.ras->query_ras_poison_mode(adev); - - /* Only poison is set in both DF and UMC, we can support it */ - if (df_poison && umc_poison) - con->poison_supported = true; - else if (df_poison != umc_poison) - dev_warn(adev->dev, - "Poison setting is inconsistent in DF/UMC(%d:%d)!\n", - df_poison, umc_poison); - } -} - static int amdgpu_get_ras_schema(struct amdgpu_device *adev) { return amdgpu_ras_is_poison_mode_supported(adev) ? AMDGPU_RAS_ERROR__POISON : 0 | @@ -3052,8 +3070,6 @@ int amdgpu_ras_init(struct amdgpu_device *adev) goto release_con; } - amdgpu_ras_query_poison_mode(adev); - /* Packed socket_id to ras feature mask bits[31:29] */ if (adev->smuio.funcs && adev->smuio.funcs->get_socket_id) -- cgit From 7ed97155b25880a78d39192957a574907210f30d Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Tue, 16 Jan 2024 18:58:39 +0800 Subject: drm/amdgpu: fix UBSAN array-index-out-of-bounds for ras_block_string[] fix array index out of bounds issue for ras_block_string[] array. Fixes: 30df05fb74f6 ("drm/amdgpu: Align ras block enum with firmware") Signed-off-by: Yang Wang Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 5 ++++- 1 file changed, 4 insertions(+), 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 5063e8d6dff6..bf08a7744a4a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -74,6 +74,8 @@ const char *ras_block_string[] = { "mca", "vcn", "jpeg", + "ih", + "mpio", }; const char *ras_mca_block_string[] = { @@ -95,7 +97,8 @@ const char *get_ras_block_str(struct ras_common_if *ras_block) if (!ras_block) return "NULL"; - if (ras_block->block >= AMDGPU_RAS_BLOCK_COUNT) + if (ras_block->block >= AMDGPU_RAS_BLOCK_COUNT || + ras_block->block >= ARRAY_SIZE(ras_block_string)) return "OUT OF RANGE"; if (ras_block->block == AMDGPU_RAS_BLOCK__MCA) -- cgit From 2c7a1560e8a604c37df56fe48f50fb1f8ccbbf22 Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Wed, 17 Jan 2024 11:49:35 +0800 Subject: drm/amdgpu: Show deferred error count for UMC Show deferred error count for UMC syfs node Signed-off-by: Stanley.Yang Reviewed-by: Tao Zhou Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 8 ++++++-- 1 file changed, 6 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 bf08a7744a4a..850cad69e4b7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -632,8 +632,12 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev, dev_warn(obj->adev->dev, "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); + if (info.head.block == AMDGPU_RAS_BLOCK__UMC) + return sysfs_emit(buf, "%s: %lu\n%s: %lu\n%s: %lu\n", "ue", info.ue_count, + "ce", info.ce_count, "de", info.de_count); + else + return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count, + "ce", info.ce_count); } /* obj begin */ -- cgit From 3fdcd0a31d7aed3ef5de104ff8f7e4e4908a0c36 Mon Sep 17 00:00:00 2001 From: YiPeng Chai Date: Thu, 18 Jan 2024 14:57:22 +0800 Subject: drm/amdgpu: Prepare for asynchronous processing of umc page retirement Preparing for asynchronous processing of umc page retirement. Signed-off-by: YiPeng Chai Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 34 +++++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 5 +++++ 2 files changed, 39 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 850cad69e4b7..7e7bb85341af 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2660,6 +2660,25 @@ static void amdgpu_ras_validate_threshold(struct amdgpu_device *adev, } } +static int amdgpu_ras_page_retirement_thread(void *param) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)param; + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + + while (!kthread_should_stop()) { + + wait_event_interruptible(con->page_retirement_wq, + atomic_read(&con->page_retirement_req_cnt)); + + dev_info(adev->dev, "Start processing page retirement. request:%d\n", + atomic_read(&con->page_retirement_req_cnt)); + + atomic_dec(&con->page_retirement_req_cnt); + } + + return 0; +} + int amdgpu_ras_recovery_init(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); @@ -2723,6 +2742,16 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev) } } + mutex_init(&con->page_retirement_lock); + init_waitqueue_head(&con->page_retirement_wq); + atomic_set(&con->page_retirement_req_cnt, 0); + con->page_retirement_thread = + kthread_run(amdgpu_ras_page_retirement_thread, adev, "umc_page_retirement"); + if (IS_ERR(con->page_retirement_thread)) { + con->page_retirement_thread = NULL; + dev_warn(adev->dev, "Failed to create umc_page_retirement thread!!!\n"); + } + #ifdef CONFIG_X86_MCE_AMD if ((adev->asic_type == CHIP_ALDEBARAN) && (adev->gmc.xgmi.connected_to_cpu)) @@ -2758,6 +2787,11 @@ static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev) if (!data) return 0; + if (con->page_retirement_thread) + kthread_stop(con->page_retirement_thread); + + atomic_set(&con->page_retirement_req_cnt, 0); + cancel_work_sync(&con->recovery_work); mutex_lock(&con->recovery_lock); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 2cd89328dc73..9c3df9985fad 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -461,6 +461,11 @@ struct amdgpu_ras { /* Record special requirements of gpu reset caller */ uint32_t gpu_reset_flags; + + struct task_struct *page_retirement_thread; + wait_queue_head_t page_retirement_wq; + struct mutex page_retirement_lock; + atomic_t page_retirement_req_cnt; }; struct ras_fs_data { -- cgit From ee9c3031d027d3a50697ddd9e557e8bb721387db Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Wed, 17 Jan 2024 15:23:41 +0800 Subject: drm/amdgpu: Fix ras features value calltrace The high three bits of ras features mask indicate socket id, it should skip to check high three bits of ras features mask before disable all ras features. Signed-off-by: Stanley.Yang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 11 ++++++----- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 6 ++++++ 2 files changed, 12 insertions(+), 5 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 7e7bb85341af..e7aa274060f8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3114,7 +3114,8 @@ int amdgpu_ras_init(struct amdgpu_device *adev) /* Packed socket_id to ras feature mask bits[31:29] */ if (adev->smuio.funcs && adev->smuio.funcs->get_socket_id) - con->features |= ((adev->smuio.funcs->get_socket_id(adev)) << 29); + con->features |= ((adev->smuio.funcs->get_socket_id(adev)) << + AMDGPU_RAS_FEATURES_SOCKETID_SHIFT); /* Get RAS schema for particular SOC */ con->schema = amdgpu_get_ras_schema(adev); @@ -3320,7 +3321,7 @@ void amdgpu_ras_suspend(struct amdgpu_device *adev) amdgpu_ras_disable_all_features(adev, 0); /* Make sure all ras objects are disabled. */ - if (con->features) + if (AMDGPU_RAS_GET_FEATURES(con->features)) amdgpu_ras_disable_all_features(adev, 1); } @@ -3370,7 +3371,7 @@ int amdgpu_ras_pre_fini(struct amdgpu_device *adev) /* Need disable ras on all IPs here before ip [hw/sw]fini */ - if (con->features) + if (AMDGPU_RAS_GET_FEATURES(con->features)) amdgpu_ras_disable_all_features(adev, 0); amdgpu_ras_recovery_fini(adev); return 0; @@ -3403,9 +3404,9 @@ int amdgpu_ras_fini(struct amdgpu_device *adev) amdgpu_ras_fs_fini(adev); amdgpu_ras_interrupt_remove_all(adev); - WARN(con->features, "Feature mask is not cleared"); + WARN(AMDGPU_RAS_GET_FEATURES(con->features), "Feature mask is not cleared"); - if (con->features) + if (AMDGPU_RAS_GET_FEATURES(con->features)) amdgpu_ras_disable_all_features(adev, 1); cancel_delayed_work_sync(&con->ras_counte_delay_work); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 9c3df9985fad..72022e2c6655 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -53,6 +53,12 @@ struct amdgpu_iv_entry; #define AMDGPU_RAS_INST_MASK 0xfffff000 #define AMDGPU_RAS_INST_SHIFT 0xc +#define AMDGPU_RAS_FEATURES_SOCKETID_SHIFT 29 +#define AMDGPU_RAS_FEATURES_SOCKETID_MASK 0xe0000000 + +/* The high three bits indicates socketid */ +#define AMDGPU_RAS_GET_FEATURES(val) ((val) & ~AMDGPU_RAS_FEATURES_SOCKETID_MASK) + enum amdgpu_ras_block { AMDGPU_RAS_BLOCK__UMC = 0, AMDGPU_RAS_BLOCK__SDMA, -- cgit From 6c23f3d12a92bc044c9373d6099204146178c9f4 Mon Sep 17 00:00:00 2001 From: YiPeng Chai Date: Mon, 15 Jan 2024 10:14:23 +0800 Subject: drm/amdgpu: Use asynchronous polling to handle umc_v12_0 poisoning Use asynchronous polling to handle umc_v12_0 poisoning. v2: 1. Change function name. 2. Change the debugging information content. Signed-off-by: YiPeng Chai Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 5 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 139 +++++++++++++++++++++++++------- drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h | 3 + 3 files changed, 116 insertions(+), 31 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 e7aa274060f8..349c810d2399 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -120,6 +120,8 @@ const char *get_ras_block_str(struct ras_common_if *ras_block) /* typical ECC bad page rate is 1 bad page per 100MB VRAM */ #define RAS_BAD_PAGE_COVER (100 * 1024 * 1024ULL) +#define MAX_UMC_POISON_POLLING_TIME_ASYNC 100 //ms + enum amdgpu_ras_retire_page_reservation { AMDGPU_RAS_RETIRE_PAGE_RESERVED, AMDGPU_RAS_RETIRE_PAGE_PENDING, @@ -2674,6 +2676,9 @@ static int amdgpu_ras_page_retirement_thread(void *param) atomic_read(&con->page_retirement_req_cnt)); atomic_dec(&con->page_retirement_req_cnt); + + amdgpu_umc_bad_page_polling_timeout(adev, + false, MAX_UMC_POISON_POLLING_TIME_ASYNC); } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index 9d1cf41cf483..83983f0e8eb5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -23,6 +23,7 @@ #include "amdgpu.h" #include "umc_v6_7.h" +#define MAX_UMC_POISON_POLLING_TIME_SYNC 20 //ms static int amdgpu_umc_convert_error_address(struct amdgpu_device *adev, struct ras_err_data *err_data, uint64_t err_addr, @@ -85,17 +86,14 @@ out_fini_err_data: return ret; } -static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, - void *ras_error_status, - struct amdgpu_iv_entry *entry, - bool reset) +static void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev, + void *ras_error_status) { struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; struct amdgpu_ras *con = amdgpu_ras_get_context(adev); int ret = 0; unsigned long err_count; - - kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); + mutex_lock(&con->page_retirement_lock); ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc)); if (ret == -EOPNOTSUPP) { if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops && @@ -163,19 +161,85 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, con->update_channel_flag = false; } } - - if (reset) { - /* use mode-2 reset for poison consumption */ - if (!entry) - con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET; - amdgpu_ras_reset_gpu(adev); - } } kfree(err_data->err_addr); + + mutex_unlock(&con->page_retirement_lock); +} + +static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, + void *ras_error_status, + struct amdgpu_iv_entry *entry, + bool reset) +{ + struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + + kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); + amdgpu_umc_handle_bad_pages(adev, ras_error_status); + + if (err_data->ue_count && reset) { + /* use mode-2 reset for poison consumption */ + if (!entry) + con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET; + amdgpu_ras_reset_gpu(adev); + } + return AMDGPU_RAS_SUCCESS; } +int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev, + bool reset, uint32_t timeout_ms) +{ + struct ras_err_data err_data; + struct ras_common_if head = { + .block = AMDGPU_RAS_BLOCK__UMC, + }; + struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head); + uint32_t timeout = timeout_ms; + + memset(&err_data, 0, sizeof(err_data)); + amdgpu_ras_error_data_init(&err_data); + + do { + + amdgpu_umc_handle_bad_pages(adev, &err_data); + + if (timeout && !err_data.de_count) { + msleep(1); + timeout--; + } + + } while (timeout && !err_data.de_count); + + if (!timeout) + dev_warn(adev->dev, "Can't find bad pages\n"); + + if (err_data.de_count) + dev_info(adev->dev, "%ld new deferred hardware errors detected\n", err_data.de_count); + + if (obj) { + obj->err_data.ue_count += err_data.ue_count; + obj->err_data.ce_count += err_data.ce_count; + obj->err_data.de_count += err_data.de_count; + } + + amdgpu_ras_error_data_fini(&err_data); + + kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); + + if (reset) { + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + + /* use mode-2 reset for poison consumption */ + con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET; + amdgpu_ras_reset_gpu(adev); + } + + return 0; +} + int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset) { int ret = AMDGPU_RAS_SUCCESS; @@ -193,25 +257,38 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset) } if (!amdgpu_sriov_vf(adev)) { - struct ras_err_data err_data; - struct ras_common_if head = { - .block = AMDGPU_RAS_BLOCK__UMC, - }; - struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head); - - ret = amdgpu_ras_error_data_init(&err_data); - if (ret) - return ret; - - ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset); - - if (ret == AMDGPU_RAS_SUCCESS && obj) { - obj->err_data.ue_count += err_data.ue_count; - obj->err_data.ce_count += err_data.ce_count; - obj->err_data.de_count += err_data.de_count; - } + if (amdgpu_ip_version(adev, UMC_HWIP, 0) < IP_VERSION(12, 0, 0)) { + struct ras_err_data err_data; + struct ras_common_if head = { + .block = AMDGPU_RAS_BLOCK__UMC, + }; + struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head); + + ret = amdgpu_ras_error_data_init(&err_data); + if (ret) + return ret; - amdgpu_ras_error_data_fini(&err_data); + ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset); + + if (ret == AMDGPU_RAS_SUCCESS && obj) { + obj->err_data.ue_count += err_data.ue_count; + obj->err_data.ce_count += err_data.ce_count; + obj->err_data.de_count += err_data.de_count; + } + + amdgpu_ras_error_data_fini(&err_data); + } else { + if (reset) { + amdgpu_umc_bad_page_polling_timeout(adev, + reset, MAX_UMC_POISON_POLLING_TIME_SYNC); + } else { + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + + atomic_inc(&con->page_retirement_req_cnt); + + wake_up(&con->page_retirement_wq); + } + } } else { if (adev->virt.ops && adev->virt.ops->ras_poison_handler) adev->virt.ops->ras_poison_handler(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index 417a6726c71b..de2dc1853636 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -118,4 +118,7 @@ int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev, int amdgpu_umc_loop_channels(struct amdgpu_device *adev, umc_func func, void *data); + +int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev, + bool reset, uint32_t timeout_ms); #endif -- cgit From 0795b5d234902269fc5182dd8e46f21bdafbcd13 Mon Sep 17 00:00:00 2001 From: YiPeng Chai Date: Mon, 15 Jan 2024 11:02:52 +0800 Subject: drm/amdgpu:Support retiring multiple MCA error address pages Support retiring multiple MCA error address pages in one in-band query for umc v12_0. Signed-off-by: YiPeng Chai Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 43 +++++++++++++++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 8 +++- drivers/gpu/drm/amd/amdgpu/umc_v12_0.c | 66 +++++++++++++++++---------------- 3 files changed, 77 insertions(+), 40 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 349c810d2399..3c1106cf9d80 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3920,8 +3920,7 @@ static int ras_err_info_cmp(void *priv, const struct list_head *a, const struct } static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_data, - struct amdgpu_smuio_mcm_config_info *mcm_info, - struct ras_err_addr *err_addr) + struct amdgpu_smuio_mcm_config_info *mcm_info) { struct ras_err_node *err_node; @@ -3933,10 +3932,9 @@ static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_d if (!err_node) return NULL; - memcpy(&err_node->err_info.mcm_info, mcm_info, sizeof(*mcm_info)); + INIT_LIST_HEAD(&err_node->err_info.err_addr_list); - if (err_addr) - memcpy(&err_node->err_info.err_addr, err_addr, sizeof(*err_addr)); + memcpy(&err_node->err_info.mcm_info, mcm_info, sizeof(*mcm_info)); err_data->err_list_count++; list_add_tail(&err_node->node, &err_data->err_node_list); @@ -3945,6 +3943,29 @@ static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_d return &err_node->err_info; } +void amdgpu_ras_add_mca_err_addr(struct ras_err_info *err_info, struct ras_err_addr *err_addr) +{ + struct ras_err_addr *mca_err_addr; + + mca_err_addr = kzalloc(sizeof(*mca_err_addr), GFP_KERNEL); + if (!mca_err_addr) + return; + + INIT_LIST_HEAD(&mca_err_addr->node); + + mca_err_addr->err_status = err_addr->err_status; + mca_err_addr->err_ipid = err_addr->err_ipid; + mca_err_addr->err_addr = err_addr->err_addr; + + list_add_tail(&mca_err_addr->node, &err_info->err_addr_list); +} + +void amdgpu_ras_del_mca_err_addr(struct ras_err_info *err_info, struct ras_err_addr *mca_err_addr) +{ + list_del(&mca_err_addr->node); + kfree(mca_err_addr); +} + int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, struct amdgpu_smuio_mcm_config_info *mcm_info, struct ras_err_addr *err_addr, u64 count) @@ -3957,10 +3978,13 @@ int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data, if (!count) return 0; - err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr); + err_info = amdgpu_ras_error_get_info(err_data, mcm_info); if (!err_info) return -EINVAL; + if (err_addr && err_addr->err_status) + amdgpu_ras_add_mca_err_addr(err_info, err_addr); + err_info->ue_count += count; err_data->ue_count += count; @@ -3979,7 +4003,7 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data, if (!count) return 0; - err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr); + err_info = amdgpu_ras_error_get_info(err_data, mcm_info); if (!err_info) return -EINVAL; @@ -4001,10 +4025,13 @@ int amdgpu_ras_error_statistic_de_count(struct ras_err_data *err_data, if (!count) return 0; - err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr); + err_info = amdgpu_ras_error_get_info(err_data, mcm_info); if (!err_info) return -EINVAL; + if (err_addr && err_addr->err_status) + amdgpu_ras_add_mca_err_addr(err_info, err_addr); + err_info->de_count += count; err_data->de_count += count; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 72022e2c6655..0b6ffae1e8bb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -480,6 +480,7 @@ struct ras_fs_data { }; struct ras_err_addr { + struct list_head node; uint64_t err_status; uint64_t err_ipid; uint64_t err_addr; @@ -490,7 +491,7 @@ struct ras_err_info { u64 ce_count; u64 ue_count; u64 de_count; - struct ras_err_addr err_addr; + struct list_head err_addr_list; }; struct ras_err_node { @@ -862,4 +863,9 @@ int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk) ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr, struct aca_handle *handle, char *buf, void *data); +void amdgpu_ras_add_mca_err_addr(struct ras_err_info *err_info, + struct ras_err_addr *err_addr); + +void amdgpu_ras_del_mca_err_addr(struct ras_err_info *err_info, + struct ras_err_addr *mca_err_addr); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c index 5ca73fefe358..836a4cc1134e 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c @@ -382,42 +382,46 @@ static void umc_v12_0_ecc_info_query_ras_error_address(struct amdgpu_device *ade { struct ras_err_node *err_node; uint64_t mc_umc_status; + struct ras_err_info *err_info; + struct ras_err_addr *mca_err_addr, *tmp; struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; for_each_ras_error(err_node, err_data) { - mc_umc_status = err_node->err_info.err_addr.err_status; - if (!mc_umc_status) + err_info = &err_node->err_info; + if (list_empty(&err_info->err_addr_list)) continue; - if (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status) || - umc_v12_0_is_deferred_error(adev, mc_umc_status)) { - uint64_t mca_addr, err_addr, mca_ipid; - uint32_t InstanceIdLo; - struct amdgpu_smuio_mcm_config_info *mcm_info; - - mcm_info = &err_node->err_info.mcm_info; - mca_addr = err_node->err_info.err_addr.err_addr; - mca_ipid = err_node->err_info.err_addr.err_ipid; - - err_addr = REG_GET_FIELD(mca_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr); - InstanceIdLo = REG_GET_FIELD(mca_ipid, MCMP1_IPIDT0, InstanceIdLo); - - dev_info(adev->dev, "UMC:IPID:0x%llx, aid:%d, inst:%d, ch:%d, err_addr:0x%llx\n", - mca_ipid, - mcm_info->die_id, - MCA_IPID_LO_2_UMC_INST(InstanceIdLo), - MCA_IPID_LO_2_UMC_CH(InstanceIdLo), - err_addr); - - umc_v12_0_convert_error_address(adev, - err_data, err_addr, - MCA_IPID_LO_2_UMC_CH(InstanceIdLo), - MCA_IPID_LO_2_UMC_INST(InstanceIdLo), - mcm_info->die_id); - - /* Clear umc error address content */ - memset(&err_node->err_info.err_addr, - 0, sizeof(err_node->err_info.err_addr)); + list_for_each_entry_safe(mca_err_addr, tmp, &err_info->err_addr_list, node) { + mc_umc_status = mca_err_addr->err_status; + if (mc_umc_status && + (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status) || + umc_v12_0_is_deferred_error(adev, mc_umc_status))) { + uint64_t mca_addr, err_addr, mca_ipid; + uint32_t InstanceIdLo; + + mca_addr = mca_err_addr->err_addr; + mca_ipid = mca_err_addr->err_ipid; + + err_addr = REG_GET_FIELD(mca_addr, + MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr); + InstanceIdLo = REG_GET_FIELD(mca_ipid, MCMP1_IPIDT0, InstanceIdLo); + + dev_info(adev->dev, "UMC:IPID:0x%llx, aid:%d, inst:%d, ch:%d, err_addr:0x%llx\n", + mca_ipid, + err_info->mcm_info.die_id, + MCA_IPID_LO_2_UMC_INST(InstanceIdLo), + MCA_IPID_LO_2_UMC_CH(InstanceIdLo), + err_addr); + + umc_v12_0_convert_error_address(adev, + err_data, err_addr, + MCA_IPID_LO_2_UMC_CH(InstanceIdLo), + MCA_IPID_LO_2_UMC_INST(InstanceIdLo), + err_info->mcm_info.die_id); + } + + /* Delete error address node from list and free memory */ + amdgpu_ras_del_mca_err_addr(err_info, mca_err_addr); } } } -- cgit From 2866a4549cf434ae8bec1efec439726562d7863c Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Mon, 22 Jan 2024 12:09:31 +0800 Subject: drm/amdgpu: skip call ras_late_init if ras block is not supported skip call ras_late_init callback if ras block is not supported. Signed-off-by: Yang Wang Reviewed-by: Tao Zhou 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 3c1106cf9d80..a32e7eb31354 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3346,12 +3346,15 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev) amdgpu_ras_set_mca_debug_mode(adev, false); list_for_each_entry_safe(node, tmp, &adev->ras_list, node) { - if (!node->ras_obj) { + obj = node->ras_obj; + if (!obj) { dev_warn(adev->dev, "Warning: abnormal ras list node.\n"); continue; } - obj = node->ras_obj; + if (!amdgpu_ras_is_supported(adev, obj->ras_comm.block)) + continue; + if (obj->ras_late_init) { r = obj->ras_late_init(adev, &obj->ras_comm); if (r) { -- cgit From c84a7e21db79fa899b9ad2d56464779f182789cb Mon Sep 17 00:00:00 2001 From: Mukul Joshi Date: Wed, 24 Jan 2024 10:14:51 +0800 Subject: drm/amdgpu: Fix module unload hang with RAS enabled The driver unload hangs because the page retirement kthread cannot be stopped as it is sleeping and waiting on page retirement event to occur. Add kthread_should_stop() to the event condition to wake up the kthread when kthread stop is called during driver unload. Fixes: 3fdcd0a31d7a ("drm/amdgpu: Prepare for asynchronous processing of umc page retirement") Signed-off-by: Mukul Joshi Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 4 ++++ 1 file changed, 4 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 a32e7eb31354..80816c4ec1f1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2670,8 +2670,12 @@ static int amdgpu_ras_page_retirement_thread(void *param) while (!kthread_should_stop()) { wait_event_interruptible(con->page_retirement_wq, + kthread_should_stop() || atomic_read(&con->page_retirement_req_cnt)); + if (kthread_should_stop()) + break; + dev_info(adev->dev, "Start processing page retirement. request:%d\n", atomic_read(&con->page_retirement_req_cnt)); -- cgit From c0c48f0d61ff94864947aa12074f4614b7e989e5 Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Wed, 24 Jan 2024 10:15:10 +0800 Subject: drm/amdgpu: adjust aca init/fini sequence to match gpu reset - move aca init/fini function into ras init/fini to adapt gpu reset sequence. - add new function amdgpu_aca_reset() Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c | 7 +++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h | 1 + drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 6 ------ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 15 +++++++++++++-- 4 files changed, 21 insertions(+), 8 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c index bf1ca6839406..bb9e0612928e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c @@ -682,6 +682,13 @@ void amdgpu_aca_fini(struct amdgpu_device *adev) aca_manager_fini(&aca->mgr); } +int amdgpu_aca_reset(struct amdgpu_device *adev) +{ + amdgpu_aca_fini(adev); + + return amdgpu_aca_init(adev); +} + void amdgpu_aca_set_smu_funcs(struct amdgpu_device *adev, const struct aca_smu_funcs *smu_funcs) { struct amdgpu_aca *aca = &adev->aca; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h index 6e9a35eda683..2da50e095883 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h @@ -185,6 +185,7 @@ struct aca_info { int amdgpu_aca_init(struct amdgpu_device *adev); void amdgpu_aca_fini(struct amdgpu_device *adev); +int amdgpu_aca_reset(struct amdgpu_device *adev); void amdgpu_aca_set_smu_funcs(struct amdgpu_device *adev, const struct aca_smu_funcs *smu_funcs); bool amdgpu_aca_is_enabled(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 1a04ccba9542..9689756bf9f5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -4048,10 +4048,6 @@ int amdgpu_device_init(struct amdgpu_device *adev, amdgpu_device_get_pcie_info(adev); - r = amdgpu_aca_init(adev); - if (r) - return r; - r = amdgpu_device_get_job_timeout_settings(adev); if (r) { dev_err(adev->dev, "invalid lockup_timeout parameter syntax\n"); @@ -4448,8 +4444,6 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev) amdgpu_reset_fini(adev); - amdgpu_aca_fini(adev); - /* free i2c buses */ if (!amdgpu_device_has_dc_support(adev)) amdgpu_i2c_fini(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 80816c4ec1f1..ebcd1cb60052 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3344,10 +3344,18 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) return 0; - if (amdgpu_aca_is_enabled(adev)) + if (amdgpu_aca_is_enabled(adev)) { + if (amdgpu_in_reset(adev)) + r = amdgpu_aca_reset(adev); + else + r = amdgpu_aca_init(adev); + if (r) + return r; + amdgpu_ras_set_aca_debug_mode(adev, false); - else + } else { amdgpu_ras_set_mca_debug_mode(adev, false); + } list_for_each_entry_safe(node, tmp, &adev->ras_list, node) { obj = node->ras_obj; @@ -3416,6 +3424,9 @@ int amdgpu_ras_fini(struct amdgpu_device *adev) amdgpu_ras_fs_fini(adev); amdgpu_ras_interrupt_remove_all(adev); + if (amdgpu_aca_is_enabled(adev)) + amdgpu_aca_fini(adev); + WARN(AMDGPU_RAS_GET_FEATURES(con->features), "Feature mask is not cleared"); if (AMDGPU_RAS_GET_FEATURES(con->features)) -- cgit From ed1e1e42fd68b073fc47aefe94d70364f3a43e97 Mon Sep 17 00:00:00 2001 From: YiPeng Chai Date: Tue, 23 Jan 2024 16:08:11 +0800 Subject: drm/amdgpu: Support passing poison consumption ras block to SRIOV Support passing poison consumption ras blocks to SRIOV. Signed-off-by: YiPeng Chai Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 5 +++-- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c | 5 +++-- drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h | 3 ++- drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h | 3 ++- drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c | 2 +- drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c | 3 ++- drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c | 23 ++++++++++++++++++----- drivers/gpu/drm/amd/amdkfd/kfd_int_process_v10.c | 7 +++++-- drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c | 7 +++++-- drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c | 7 +++++-- 13 files changed, 49 insertions(+), 22 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index 77e263660288..dfb93664e866 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -732,9 +732,10 @@ void amdgpu_amdkfd_debug_mem_fence(struct amdgpu_device *adev) amdgpu_device_flush_hdp(adev, NULL); } -void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev, bool reset) +void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block, bool reset) { - amdgpu_umc_poison_handler(adev, reset); + amdgpu_umc_poison_handler(adev, block, reset); } int amdgpu_amdkfd_send_close_event_drain_irq(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index 584a0cea5572..50d3e0149032 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -334,7 +334,7 @@ void amdgpu_amdkfd_debug_mem_fence(struct amdgpu_device *adev); int amdgpu_amdkfd_get_tile_config(struct amdgpu_device *adev, struct tile_config *config); void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev, - bool reset); + enum amdgpu_ras_block block, bool reset); bool amdgpu_amdkfd_bo_mapped_to_dev(struct amdgpu_device *adev, struct kgd_mem *mem); void amdgpu_amdkfd_block_mmu_notifications(void *p); int amdgpu_amdkfd_criu_resume(void *p); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index ebcd1cb60052..79bf6bd428a5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2041,7 +2041,7 @@ static void amdgpu_ras_interrupt_poison_consumption_handler(struct ras_manager * } } - amdgpu_umc_poison_handler(adev, false); + amdgpu_umc_poison_handler(adev, obj->head.block, false); if (block_obj->hw_ops && block_obj->hw_ops->handle_poison_consumption) poison_stat = block_obj->hw_ops->handle_poison_consumption(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index a6cdb69897f2..20436f81856a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -246,7 +246,8 @@ int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev, return 0; } -int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset) +int amdgpu_umc_poison_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block, bool reset) { int ret = AMDGPU_RAS_SUCCESS; @@ -297,7 +298,7 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset) } } else { if (adev->virt.ops && adev->virt.ops->ras_poison_handler) - adev->virt.ops->ras_poison_handler(adev); + adev->virt.ops->ras_poison_handler(adev, block); else dev_warn(adev->dev, "No ras_poison_handler interface in SRIOV!\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index 83199296ed10..26d2ae498daf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -102,7 +102,8 @@ struct amdgpu_umc { int amdgpu_umc_ras_sw_init(struct amdgpu_device *adev); int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block); -int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset); +int amdgpu_umc_poison_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block, bool reset); int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index f4963330c772..f300d4a4457d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -1189,7 +1189,7 @@ int amdgpu_vcn_process_poison_irq(struct amdgpu_device *adev, amdgpu_ras_interrupt_dispatch(adev, &ih_data); } else { if (adev->virt.ops && adev->virt.ops->ras_poison_handler) - adev->virt.ops->ras_poison_handler(adev); + adev->virt.ops->ras_poison_handler(adev, ras_if->block); else dev_warn(adev->dev, "No ras_poison_handler interface in SRIOV for VCN!\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index 1b49c007ff62..fa7be5f277b9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -88,7 +88,8 @@ struct amdgpu_virt_ops { int (*wait_reset)(struct amdgpu_device *adev); void (*trans_msg)(struct amdgpu_device *adev, enum idh_request req, u32 data1, u32 data2, u32 data3); - void (*ras_poison_handler)(struct amdgpu_device *adev); + void (*ras_poison_handler)(struct amdgpu_device *adev, + enum amdgpu_ras_block block); }; /* diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c index 26d6286d86c9..9e7ce1e6bc06 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c @@ -69,7 +69,7 @@ static int gfx_v11_0_3_rlc_gc_fed_irq(struct amdgpu_device *adev, amdgpu_ras_interrupt_dispatch(adev, &ih_data); } else { if (adev->virt.ops && adev->virt.ops->ras_poison_handler) - adev->virt.ops->ras_poison_handler(adev); + adev->virt.ops->ras_poison_handler(adev, ras_if->block); else dev_warn(adev->dev, "No ras_poison_handler interface in SRIOV for %s!\n", ras_if->name); diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c index 63725b2ebc03..a2bd2c3b1ef9 100644 --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c @@ -404,7 +404,8 @@ static int xgpu_ai_request_init_data(struct amdgpu_device *adev) return xgpu_ai_send_access_requests(adev, IDH_REQ_GPU_INIT_DATA); } -static void xgpu_ai_ras_poison_handler(struct amdgpu_device *adev) +static void xgpu_ai_ras_poison_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block) { xgpu_ai_send_access_requests(adev, IDH_RAS_POISON); } diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c index 6a68ee946f1c..d0a018da3c7a 100644 --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c @@ -152,14 +152,14 @@ static void xgpu_nv_mailbox_trans_msg (struct amdgpu_device *adev, xgpu_nv_mailbox_set_valid(adev, false); } -static int xgpu_nv_send_access_requests(struct amdgpu_device *adev, - enum idh_request req) +static int xgpu_nv_send_access_requests_with_param(struct amdgpu_device *adev, + enum idh_request req, u32 data1, u32 data2, u32 data3) { int r, retry = 1; enum idh_event event = -1; send_request: - xgpu_nv_mailbox_trans_msg(adev, req, 0, 0, 0); + xgpu_nv_mailbox_trans_msg(adev, req, data1, data2, data3); switch (req) { case IDH_REQ_GPU_INIT_ACCESS: @@ -206,6 +206,13 @@ send_request: return 0; } +static int xgpu_nv_send_access_requests(struct amdgpu_device *adev, + enum idh_request req) +{ + return xgpu_nv_send_access_requests_with_param(adev, + req, 0, 0, 0); +} + static int xgpu_nv_request_reset(struct amdgpu_device *adev) { int ret, i = 0; @@ -424,9 +431,15 @@ void xgpu_nv_mailbox_put_irq(struct amdgpu_device *adev) amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0); } -static void xgpu_nv_ras_poison_handler(struct amdgpu_device *adev) +static void xgpu_nv_ras_poison_handler(struct amdgpu_device *adev, + enum amdgpu_ras_block block) { - xgpu_nv_send_access_requests(adev, IDH_RAS_POISON); + if (amdgpu_ip_version(adev, UMC_HWIP, 0) < IP_VERSION(12, 0, 0)) { + xgpu_nv_send_access_requests(adev, IDH_RAS_POISON); + } else { + xgpu_nv_send_access_requests_with_param(adev, + IDH_RAS_POISON, block, 0, 0); + } } const struct amdgpu_virt_ops xgpu_nv_virt_ops = { diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v10.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v10.c index a7697ec8188e..9a06c6fb6605 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v10.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v10.c @@ -132,6 +132,7 @@ enum SQ_INTERRUPT_ERROR_TYPE { static void event_interrupt_poison_consumption(struct kfd_node *dev, uint16_t pasid, uint16_t client_id) { + enum amdgpu_ras_block block = 0; int old_poison, ret = -EINVAL; struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); @@ -151,12 +152,14 @@ static void event_interrupt_poison_consumption(struct kfd_node *dev, case SOC15_IH_CLIENTID_SE3SH: case SOC15_IH_CLIENTID_UTCL2: ret = kfd_dqm_evict_pasid(dev->dqm, pasid); + block = AMDGPU_RAS_BLOCK__GFX; break; case SOC15_IH_CLIENTID_SDMA0: case SOC15_IH_CLIENTID_SDMA1: case SOC15_IH_CLIENTID_SDMA2: case SOC15_IH_CLIENTID_SDMA3: case SOC15_IH_CLIENTID_SDMA4: + block = AMDGPU_RAS_BLOCK__SDMA; break; default: break; @@ -171,12 +174,12 @@ static void event_interrupt_poison_consumption(struct kfd_node *dev, dev_warn(dev->adev->dev, "RAS poison consumption, unmap queue flow succeeded: client id %d\n", client_id); - amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, false); + amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, block, false); } else { dev_warn(dev->adev->dev, "RAS poison consumption, fall back to gpu reset flow: client id %d\n", client_id); - amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, true); + amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, block, true); } } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c index 2a65792fd116..7e2859736a55 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c @@ -191,6 +191,7 @@ static void print_sq_intr_info_error(uint32_t context_id0, uint32_t context_id1) static void event_interrupt_poison_consumption_v11(struct kfd_node *dev, uint16_t pasid, uint16_t source_id) { + enum amdgpu_ras_block block = 0; int ret = -EINVAL; struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); @@ -210,9 +211,11 @@ static void event_interrupt_poison_consumption_v11(struct kfd_node *dev, case SOC15_INTSRC_SQ_INTERRUPT_MSG: if (dev->dqm->ops.reset_queues) ret = dev->dqm->ops.reset_queues(dev->dqm, pasid); + block = AMDGPU_RAS_BLOCK__GFX; break; case SOC21_INTSRC_SDMA_ECC: default: + block = AMDGPU_RAS_BLOCK__GFX; break; } @@ -221,9 +224,9 @@ static void event_interrupt_poison_consumption_v11(struct kfd_node *dev, /* resetting queue passes, do page retirement without gpu reset resetting queue fails, fallback to gpu reset solution */ if (!ret) - amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, false); + amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, block, false); else - amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, true); + amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, block, true); } static bool event_interrupt_isr_v11(struct kfd_node *dev, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c index 27cdaea40501..91dd5e045b51 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c @@ -143,6 +143,7 @@ enum SQ_INTERRUPT_ERROR_TYPE { static void event_interrupt_poison_consumption_v9(struct kfd_node *dev, uint16_t pasid, uint16_t client_id) { + enum amdgpu_ras_block block = 0; int old_poison, ret = -EINVAL; struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); @@ -162,12 +163,14 @@ static void event_interrupt_poison_consumption_v9(struct kfd_node *dev, case SOC15_IH_CLIENTID_SE3SH: case SOC15_IH_CLIENTID_UTCL2: ret = kfd_dqm_evict_pasid(dev->dqm, pasid); + block = AMDGPU_RAS_BLOCK__GFX; break; case SOC15_IH_CLIENTID_SDMA0: case SOC15_IH_CLIENTID_SDMA1: case SOC15_IH_CLIENTID_SDMA2: case SOC15_IH_CLIENTID_SDMA3: case SOC15_IH_CLIENTID_SDMA4: + block = AMDGPU_RAS_BLOCK__SDMA; break; default: break; @@ -182,12 +185,12 @@ static void event_interrupt_poison_consumption_v9(struct kfd_node *dev, dev_warn(dev->adev->dev, "RAS poison consumption, unmap queue flow succeeded: client id %d\n", client_id); - amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, false); + amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, block, false); } else { dev_warn(dev->adev->dev, "RAS poison consumption, fall back to gpu reset flow: client id %d\n", client_id); - amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, true); + amdgpu_amdkfd_ras_poison_consumption_handler(dev->adev, block, true); } } -- cgit From 1731ba9b64f72c7c5632fca9bf4c124613425971 Mon Sep 17 00:00:00 2001 From: Hawking Zhang Date: Mon, 29 Jan 2024 20:29:08 +0800 Subject: drm/amdgpu: Update boot time errors polling sequence Update boot time errors polling sequence to align with the latest firmware change. Signed-off-by: Hawking Zhang Reviewed-by: Frank Min Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 14 +++++++++++++- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 5 +++++ 2 files changed, 18 insertions(+), 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 79bf6bd428a5..084289db23f9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -4120,6 +4120,18 @@ static int amdgpu_ras_wait_for_boot_complete(struct amdgpu_device *adev, u32 reg_data; int retry_loop; + reg_addr = (mmMP0_SMN_C2PMSG_92 << 2) + + aqua_vanjaram_encode_ext_smn_addressing(instance); + + for (retry_loop = 0; retry_loop < AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT; retry_loop++) { + reg_data = amdgpu_device_indirect_rreg_ext(adev, reg_addr); + if ((reg_data & AMDGPU_RAS_BOOT_STATUS_MASK) == AMDGPU_RAS_BOOT_STEADY_STATUS) { + *boot_error = AMDGPU_RAS_BOOT_SUCEESS; + return 0; + } + msleep(1); + } + /* The pattern for smn addressing in other SOC could be different from * the one for aqua_vanjaram. We should revisit the code if the pattern * is changed. In such case, replace the aqua_vanjaram implementation @@ -4127,7 +4139,7 @@ static int amdgpu_ras_wait_for_boot_complete(struct amdgpu_device *adev, reg_addr = (mmMP0_SMN_C2PMSG_126 << 2) + aqua_vanjaram_encode_ext_smn_addressing(instance); - for (retry_loop = 0; retry_loop < 1000; retry_loop++) { + for (retry_loop = 0; retry_loop < AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT; retry_loop++) { reg_data = amdgpu_device_indirect_rreg_ext(adev, reg_addr); if (AMDGPU_RAS_GPU_ERR_BOOT_STATUS(reg_data)) { *boot_error = reg_data; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 0b6ffae1e8bb..d10e5bb0e52f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -46,6 +46,11 @@ struct amdgpu_iv_entry; #define AMDGPU_RAS_GPU_ERR_HBM_ID(x) AMDGPU_GET_REG_FIELD(x, 13, 13) #define AMDGPU_RAS_GPU_ERR_BOOT_STATUS(x) AMDGPU_GET_REG_FIELD(x, 31, 31) +#define AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT 1000 +#define AMDGPU_RAS_BOOT_STEADY_STATUS 0xBA +#define AMDGPU_RAS_BOOT_STATUS_MASK 0xFF +#define AMDGPU_RAS_BOOT_SUCEESS 0x80000000 + #define AMDGPU_RAS_FLAG_INIT_BY_VBIOS (0x1 << 0) /* position of instance value in sub_block_index of * ta_ras_trigger_error_input, the sub block uses lower 12 bits -- cgit From edfdde9013b7930674c8231720351e0ebd42cc34 Mon Sep 17 00:00:00 2001 From: Tao Zhou Date: Fri, 26 Jan 2024 19:43:37 +0800 Subject: drm/amdgpu: disable RAS feature when fini Send RAS disable feature command in fini. Signed-off-by: Tao Zhou 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 084289db23f9..46f3d1013e8c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3430,7 +3430,7 @@ int amdgpu_ras_fini(struct amdgpu_device *adev) WARN(AMDGPU_RAS_GET_FEATURES(con->features), "Feature mask is not cleared"); if (AMDGPU_RAS_GET_FEATURES(con->features)) - amdgpu_ras_disable_all_features(adev, 1); + amdgpu_ras_disable_all_features(adev, 0); cancel_delayed_work_sync(&con->ras_counte_delay_work); -- cgit From 1b6ef74b2b03b54776778476f8adf87dd4f8beb1 Mon Sep 17 00:00:00 2001 From: Lijo Lazar Date: Thu, 22 Feb 2024 14:16:57 +0530 Subject: drm/amdgpu: Add fatal error detected flag For a RAS error that needs a full reset to recover, set the fatal error status. Clear the status once the device is reset. Signed-off-by: Lijo Lazar Reviewed-by: Asad Kamal Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 1 + drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 32 ++++++++++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h | 6 ++++++ 3 files changed, 39 insertions(+) (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 b0ea4ddc8e72..d88309228a26 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -5321,6 +5321,7 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle, list_for_each_entry(tmp_adev, device_list_handle, reset_list) { if (need_full_reset) { /* post card */ + amdgpu_ras_set_fed(tmp_adev, false); r = amdgpu_device_asic_init(tmp_adev); if (r) { dev_warn(tmp_adev->dev, "asic atom init failed!"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 46f3d1013e8c..2c94de305c69 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2439,6 +2439,18 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) ras->gpu_reset_flags &= ~AMDGPU_RAS_GPU_RESET_MODE1_RESET; set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); + /* For any RAS error that needs a full reset to + * recover, set the fatal error status + */ + if (hive) { + list_for_each_entry(remote_adev, + &hive->device_list, + gmc.xgmi.head) + amdgpu_ras_set_fed(remote_adev, + true); + } else { + amdgpu_ras_set_fed(adev, true); + } psp_fatal_error_recovery_quirk(&adev->psp); } } @@ -3440,6 +3452,26 @@ int amdgpu_ras_fini(struct amdgpu_device *adev) return 0; } +bool amdgpu_ras_get_fed_status(struct amdgpu_device *adev) +{ + struct amdgpu_ras *ras; + + ras = amdgpu_ras_get_context(adev); + if (!ras) + return false; + + return atomic_read(&ras->fed); +} + +void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status) +{ + struct amdgpu_ras *ras; + + ras = amdgpu_ras_get_context(adev); + if (ras) + atomic_set(&ras->fed, !!status); +} + void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev) { if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index d10e5bb0e52f..e0f8ce9d8440 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -477,6 +477,8 @@ struct amdgpu_ras { wait_queue_head_t page_retirement_wq; struct mutex page_retirement_lock; atomic_t page_retirement_req_cnt; + /* Fatal error detected flag */ + atomic_t fed; }; struct ras_fs_data { @@ -873,4 +875,8 @@ void amdgpu_ras_add_mca_err_addr(struct ras_err_info *err_info, void amdgpu_ras_del_mca_err_addr(struct ras_err_info *err_info, struct ras_err_addr *mca_err_addr); + +void amdgpu_ras_set_fed(struct amdgpu_device *adev, bool status); +bool amdgpu_ras_get_fed_status(struct amdgpu_device *adev); + #endif -- cgit From 7ec11c2f65d0dad8d3bd05f1ca32a6ed66baebaa Mon Sep 17 00:00:00 2001 From: "Stanley.Yang" Date: Wed, 21 Feb 2024 17:42:20 +0800 Subject: drm/amdgpu: Fix ineffective ras_mask settings Check amdgpu_ras_mask to fix ineffective ras_mask setting due to special asic without sram ecc enable but with poison supported. Signed-off-by: Stanley.Yang Reviewed-by: Tao Zhou 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 2c94de305c69..8ebab6f22e5a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -3652,6 +3652,7 @@ int amdgpu_ras_is_supported(struct amdgpu_device *adev, block == AMDGPU_RAS_BLOCK__SDMA || block == AMDGPU_RAS_BLOCK__VCN || block == AMDGPU_RAS_BLOCK__JPEG) && + (amdgpu_ras_mask & (1 << block)) && amdgpu_ras_is_poison_mode_supported(adev) && amdgpu_ras_get_ras_block(adev, block, 0)) ret = 1; -- cgit