iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

accel/habanalabs: update device boot error check

Browse Source 
Use a predefined mask which set the device critical boot errors.
Driver will fail and stop its loading, only upon detecting at least
one of those errors defined in this mask.

Signed-off-by: Farah Kassabri <fkassabri@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
This commit is contained in:
Farah Kassabri 2023-10-18 16:22:13 +03:00 committed by Oded Gabbay
parent f64fa33260
commit fbc2a09e09
1 changed files with 32 additions and 83 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

115
drivers/accel/habanalabs/common/firmware_if.c
View File

@ -646,39 +646,27 @@ int hl_fw_send_heartbeat(struct hl_device *hdev)
return rc; return rc;
} }
static bool fw_report_boot_dev0(struct hl_device *hdev, u32 err_val, static bool fw_report_boot_dev0(struct hl_device *hdev, u32 err_val, u32 sts_val)
u32 sts_val)
{ {
bool err_exists = false; bool err_exists = false;
if (!(err_val & CPU_BOOT_ERR0_ENABLED)) if (!(err_val & CPU_BOOT_ERR0_ENABLED))
return false; return false;
if (err_val & CPU_BOOT_ERR0_DRAM_INIT_FAIL) { if (err_val & CPU_BOOT_ERR0_DRAM_INIT_FAIL)
dev_err(hdev->dev, dev_err(hdev->dev, "Device boot error - DRAM initialization failed\n");
"Device boot error - DRAM initialization failed\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_FIT_CORRUPTED) { if (err_val & CPU_BOOT_ERR0_FIT_CORRUPTED)
dev_err(hdev->dev, "Device boot error - FIT image corrupted\n"); dev_err(hdev->dev, "Device boot error - FIT image corrupted\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_TS_INIT_FAIL) { if (err_val & CPU_BOOT_ERR0_TS_INIT_FAIL)
dev_err(hdev->dev, dev_err(hdev->dev, "Device boot error - Thermal Sensor initialization failed\n");
"Device boot error - Thermal Sensor initialization failed\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_BMC_WAIT_SKIPPED) { if (err_val & CPU_BOOT_ERR0_BMC_WAIT_SKIPPED) {
if (hdev->bmc_enable) { if (hdev->bmc_enable) {
dev_err(hdev->dev, dev_err(hdev->dev, "Device boot error - Skipped waiting for BMC\n");
"Device boot error - Skipped waiting for BMC\n");
err_exists = true;
} else { } else {
dev_info(hdev->dev, dev_info(hdev->dev, "Device boot message - Skipped waiting for BMC\n");
"Device boot message - Skipped waiting for BMC\n");
/* This is an info so we don't want it to disable the /* This is an info so we don't want it to disable the
* device * device
*/ */
@ -686,48 +674,29 @@ static bool fw_report_boot_dev0(struct hl_device *hdev, u32 err_val,
} }
} }
if (err_val & CPU_BOOT_ERR0_NIC_DATA_NOT_RDY) { if (err_val & CPU_BOOT_ERR0_NIC_DATA_NOT_RDY)
dev_err(hdev->dev, dev_err(hdev->dev, "Device boot error - Serdes data from BMC not available\n");
"Device boot error - Serdes data from BMC not available\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_NIC_FW_FAIL) { if (err_val & CPU_BOOT_ERR0_NIC_FW_FAIL)
dev_err(hdev->dev, dev_err(hdev->dev, "Device boot error - NIC F/W initialization failed\n");
"Device boot error - NIC F/W initialization failed\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_SECURITY_NOT_RDY) { if (err_val & CPU_BOOT_ERR0_SECURITY_NOT_RDY)
dev_err(hdev->dev, dev_err(hdev->dev, "Device boot warning - security not ready\n");
"Device boot warning - security not ready\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_SECURITY_FAIL) { if (err_val & CPU_BOOT_ERR0_SECURITY_FAIL)
dev_err(hdev->dev, "Device boot error - security failure\n"); dev_err(hdev->dev, "Device boot error - security failure\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_EFUSE_FAIL) { if (err_val & CPU_BOOT_ERR0_EFUSE_FAIL)
dev_err(hdev->dev, "Device boot error - eFuse failure\n"); dev_err(hdev->dev, "Device boot error - eFuse failure\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_SEC_IMG_VER_FAIL) { if (err_val & CPU_BOOT_ERR0_SEC_IMG_VER_FAIL)
dev_err(hdev->dev, "Device boot error - Failed to load preboot secondary image\n"); dev_err(hdev->dev, "Device boot error - Failed to load preboot secondary image\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_PLL_FAIL) { if (err_val & CPU_BOOT_ERR0_PLL_FAIL)
dev_err(hdev->dev, "Device boot error - PLL failure\n"); dev_err(hdev->dev, "Device boot error - PLL failure\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_TMP_THRESH_INIT_FAIL) { if (err_val & CPU_BOOT_ERR0_TMP_THRESH_INIT_FAIL)
dev_err(hdev->dev, "Device boot error - Failed to set threshold for temperature sensor\n"); dev_err(hdev->dev, "Device boot error - Failed to set threshold for temperature sensor\n");
err_exists = true;
}
if (err_val & CPU_BOOT_ERR0_DEVICE_UNUSABLE_FAIL) { if (err_val & CPU_BOOT_ERR0_DEVICE_UNUSABLE_FAIL) {
/* Ignore this bit, don't prevent driver loading */ /* Ignore this bit, don't prevent driver loading */
@ -735,52 +704,32 @@ static bool fw_report_boot_dev0(struct hl_device *hdev, u32 err_val,
err_val &= ~CPU_BOOT_ERR0_DEVICE_UNUSABLE_FAIL; err_val &= ~CPU_BOOT_ERR0_DEVICE_UNUSABLE_FAIL;
} }
if (err_val & CPU_BOOT_ERR0_BINNING_FAIL) { if (err_val & CPU_BOOT_ERR0_BINNING_FAIL)
dev_err(hdev->dev, "Device boot error - binning failure\n"); dev_err(hdev->dev, "Device boot error - binning failure\n");
err_exists = true;
}
if (sts_val & CPU_BOOT_DEV_STS0_ENABLED) if (sts_val & CPU_BOOT_DEV_STS0_ENABLED)
dev_dbg(hdev->dev, "Device status0 %#x\n", sts_val); dev_dbg(hdev->dev, "Device status0 %#x\n", sts_val);
if (err_val & CPU_BOOT_ERR0_DRAM_SKIPPED)
dev_err(hdev->dev, "Device boot warning - Skipped DRAM initialization\n");
if (err_val & CPU_BOOT_ERR_ENG_ARC_MEM_SCRUB_FAIL)
dev_err(hdev->dev, "Device boot error - ARC memory scrub failed\n");
/* All warnings should go here in order not to reach the unknown error validation */
if (err_val & CPU_BOOT_ERR0_EEPROM_FAIL) { if (err_val & CPU_BOOT_ERR0_EEPROM_FAIL) {
dev_err(hdev->dev, "Device boot error - EEPROM failure detected\n"); dev_err(hdev->dev, "Device boot error - EEPROM failure detected\n");
err_exists = true; err_exists = true;
} }
/* All warnings should go here in order not to reach the unknown error validation */ if (err_val & CPU_BOOT_ERR0_PRI_IMG_VER_FAIL)
if (err_val & CPU_BOOT_ERR0_DRAM_SKIPPED) { dev_warn(hdev->dev, "Device boot warning - Failed to load preboot primary image\n");
dev_warn(hdev->dev,
"Device boot warning - Skipped DRAM initialization\n");
/* This is a warning so we don't want it to disable the
* device
*/
err_val &= ~CPU_BOOT_ERR0_DRAM_SKIPPED;
}
if (err_val & CPU_BOOT_ERR0_PRI_IMG_VER_FAIL) { if (err_val & CPU_BOOT_ERR0_TPM_FAIL)
dev_warn(hdev->dev, dev_warn(hdev->dev, "Device boot warning - TPM failure\n");
"Device boot warning - Failed to load preboot primary image\n");
/* This is a warning so we don't want it to disable the
* device as we have a secondary preboot image
*/
err_val &= ~CPU_BOOT_ERR0_PRI_IMG_VER_FAIL;
}
if (err_val & CPU_BOOT_ERR0_TPM_FAIL) { if (err_val & CPU_BOOT_ERR_FATAL_MASK)
dev_warn(hdev->dev,
"Device boot warning - TPM failure\n");
/* This is a warning so we don't want it to disable the
* device
*/
err_val &= ~CPU_BOOT_ERR0_TPM_FAIL;
}
if (!err_exists && (err_val & ~CPU_BOOT_ERR0_ENABLED)) {
dev_err(hdev->dev,
"Device boot error - unknown ERR0 error 0x%08x\n", err_val);
err_exists = true; err_exists = true;
}
/* return error only if it's in the predefined mask */ /* return error only if it's in the predefined mask */
if (err_exists && ((err_val & ~CPU_BOOT_ERR0_ENABLED) & if (err_exists && ((err_val & ~CPU_BOOT_ERR0_ENABLED) &