[AMD Official Use Only - AMD Internal Distribution Only]
Hi, Tao
At start, there is a null check for each call of mp1->sys_func and
sys_func->mp1_send_eeprom_msg. Thomas suggested we remove the check at these
places, instead only check it in hw_init func, to simplify the code.
-----Original Message-----
From: Zhou1, Tao <[email protected]>
Sent: Tuesday, January 27, 2026 11:54 AM
To: Xie, Patrick <[email protected]>; [email protected]
Cc: Chai, Thomas <[email protected]>
Subject: RE: [PATCH 05/14] drm/amd/ras: add wrapper funcs for pmfw eeprom
[AMD Official Use Only - AMD Internal Distribution Only]
> -----Original Message-----
> From: Xie, Patrick <[email protected]>
> Sent: Monday, January 26, 2026 11:55 AM
> To: [email protected]
> Cc: Zhou1, Tao <[email protected]>; Chai, Thomas
> <[email protected]>; Xie, Patrick <[email protected]>
> Subject: [PATCH 05/14] drm/amd/ras: add wrapper funcs for pmfw eeprom
>
> add wrapper funcs for pmfw eeprom interface to make them easier to be
> called
>
> Signed-off-by: Gangliang Xie <[email protected]>
> ---
> .../gpu/drm/amd/ras/rascore/ras_eeprom_fw.c | 125 ++++++++++++++++++
> .../gpu/drm/amd/ras/rascore/ras_eeprom_fw.h | 16 +++
> 2 files changed, 141 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_eeprom_fw.c
> b/drivers/gpu/drm/amd/ras/rascore/ras_eeprom_fw.c
> index 4a65351569e8..5231dfe8c518 100644
> --- a/drivers/gpu/drm/amd/ras/rascore/ras_eeprom_fw.c
> +++ b/drivers/gpu/drm/amd/ras/rascore/ras_eeprom_fw.c
> @@ -36,3 +36,128 @@ void ras_fw_init_feature_flags(struct
> ras_core_context
> *ras_core)
> if (!sys_func->mp1_get_ras_enabled_mask(ras_core, &flags))
> ras_core->ras_fw_features = flags; }
> +
> +bool ras_fw_eeprom_supported(struct ras_core_context *ras_core) {
> + return !!(ras_core->ras_fw_features &
> +RAS_CORE_FW_FEATURE_BIT__RAS_EEPROM);
> +}
> +
> +int ras_fw_get_table_version(struct ras_core_context *ras_core,
> + uint32_t *table_version) {
> + struct ras_mp1 *mp1 = &ras_core->ras_mp1;
> + const struct ras_mp1_sys_func *sys_func = mp1->sys_func;
> +
> + return sys_func->mp1_send_eeprom_msg(ras_core,
> + RAS_SMU_GetRASTableVersion, 0,
> + table_version); }
[Tao] do we need null check for mp1->sys_func and sys_func->mp1_send_eeprom_msg?
> +
> +int ras_fw_get_badpage_count(struct ras_core_context *ras_core,
> + uint32_t *count, uint32_t timeout) {
> + struct ras_mp1 *mp1 = &ras_core->ras_mp1;
> + const struct ras_mp1_sys_func *sys_func = mp1->sys_func;
> + uint64_t end, now;
> + int ret = 0;
> +
> + now = (uint64_t)ktime_to_ms(ktime_get());
> + end = now + timeout;
> +
> + do {
> + ret = sys_func->mp1_send_eeprom_msg(ras_core,
> + RAS_SMU_GetBadPageCount, 0, count);
> + /* eeprom is not ready */
> + if (ret != -EBUSY)
> + return ret;
> +
> + mdelay(10);
> + now = (uint64_t)ktime_to_ms(ktime_get());
> + } while (now < end);
> +
> + RAS_DEV_ERR(ras_core->dev,
> + "smu get bad page count timeout!\n");
> + return ret;
> +}
> +
> +int ras_fw_get_badpage_mca_addr(struct ras_core_context *ras_core,
> + uint16_t index, uint64_t *mca_addr) {
> + struct ras_mp1 *mp1 = &ras_core->ras_mp1;
> + const struct ras_mp1_sys_func *sys_func = mp1->sys_func;
> + uint32_t temp_arg, temp_addr_lo, temp_addr_high;
> + int ret;
> +
> + temp_arg = index | (1 << 16);
> + ret = sys_func->mp1_send_eeprom_msg(ras_core,
> + RAS_SMU_GetBadPageMcaAddr, temp_arg,
> &temp_addr_lo);
> + if (ret)
> + return ret;
> +
> + temp_arg = index | (2 << 16);
> + ret = sys_func->mp1_send_eeprom_msg(ras_core,
> + RAS_SMU_GetBadPageMcaAddr, temp_arg,
> &temp_addr_high);
> +
> + if (!ret)
> + *mca_addr = (uint64_t)temp_addr_high << 32 |
> + temp_addr_lo;
> +
> + return ret;
> +}
> +
> +int ras_fw_set_timestamp(struct ras_core_context *ras_core,
> + uint64_t timestamp) {
> + struct ras_mp1 *mp1 = &ras_core->ras_mp1;
> + const struct ras_mp1_sys_func *sys_func = mp1->sys_func;
> +
> + return sys_func->mp1_send_eeprom_msg(ras_core,
> + RAS_SMU_SetTimestamp, (uint32_t)timestamp, 0); }
> +
> +int ras_fw_get_timestamp(struct ras_core_context *ras_core,
> + uint16_t index, uint64_t *timestamp) {
> + struct ras_mp1 *mp1 = &ras_core->ras_mp1;
> + const struct ras_mp1_sys_func *sys_func = mp1->sys_func;
> + uint32_t temp = 0;
> + int ret;
> +
> + ret = sys_func->mp1_send_eeprom_msg(ras_core,
> + RAS_SMU_GetTimestamp, 0, &temp);
> + if (!ret)
> + *timestamp = temp;
> +
> + return ret;
> +}
> +
> +int ras_fw_get_badpage_ipid(struct ras_core_context *ras_core,
> + uint16_t index, uint64_t *ipid) {
> + struct ras_mp1 *mp1 = &ras_core->ras_mp1;
> + const struct ras_mp1_sys_func *sys_func = mp1->sys_func;
> + uint32_t temp_arg, temp_ipid_lo, temp_ipid_high;
> + int ret;
> +
> + temp_arg = index | (1 << 16);
> + ret = sys_func->mp1_send_eeprom_msg(ras_core,
> + RAS_SMU_GetBadPageIpid, temp_arg, &temp_ipid_lo);
> + if (ret)
> + return ret;
> +
> + temp_arg = index | (2 << 16);
> + ret = sys_func->mp1_send_eeprom_msg(ras_core,
> + RAS_SMU_GetBadPageIpid, temp_arg, &temp_ipid_high);
> + if (!ret)
> + *ipid = (uint64_t)temp_ipid_high << 32 | temp_ipid_lo;
> +
> + return ret;
> +}
> +
> +int ras_fw_erase_ras_table(struct ras_core_context *ras_core,
> + uint32_t *result) {
> + struct ras_mp1 *mp1 = &ras_core->ras_mp1;
> + const struct ras_mp1_sys_func *sys_func = mp1->sys_func;
> +
> + return sys_func->mp1_send_eeprom_msg(ras_core,
> + RAS_SMU_EraseRasTable, 0, result); }
> diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_eeprom_fw.h
> b/drivers/gpu/drm/amd/ras/rascore/ras_eeprom_fw.h
> index 58472e459470..ad98077d532e 100644
> --- a/drivers/gpu/drm/amd/ras/rascore/ras_eeprom_fw.h
> +++ b/drivers/gpu/drm/amd/ras/rascore/ras_eeprom_fw.h
> @@ -24,6 +24,22 @@
> #ifndef __RAS_EEPROM_FW_H__
> #define __RAS_EEPROM_FW_H__
>
> +
> void ras_fw_init_feature_flags(struct ras_core_context *ras_core);
> +bool ras_fw_eeprom_supported(struct ras_core_context *ras_core); int
> +ras_fw_get_table_version(struct ras_core_context *ras_core,
> + uint32_t *table_version); int
> +ras_fw_get_badpage_count(struct ras_core_context *ras_core,
> + uint32_t *count, uint32_t timeout);
> +int ras_fw_get_badpage_mca_addr(struct ras_core_context *ras_core,
> + uint16_t index, uint64_t
> +*mca_addr); int ras_fw_set_timestamp(struct ras_core_context *ras_core,
> + uint64_t timestamp); int
> +ras_fw_get_timestamp(struct ras_core_context *ras_core,
> + uint16_t index, uint64_t *timestamp);
> +int ras_fw_get_badpage_ipid(struct ras_core_context *ras_core,
> + uint16_t index, uint64_t *ipid); int
> +ras_fw_erase_ras_table(struct ras_core_context *ras_core,
> + uint32_t *result);
>
> #endif
> --
> 2.34.1
