habanalabs: load CPU device boot loader from host

Load CPU device boot loader during driver boot time in order to avoid flash
write for every boot loader update.

To preserve backward-compatibility, skip the device boot load if the device
doesn't request it.
Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>

drivers/misc/habanalabs/firmware_if.c

@@ -328,8 +328,9 @@ static void fw_read_errors(struct hl_device *hdev, u32 boot_err0_reg)
 }
 
 int hl_fw_init_cpu(struct hl_device *hdev, u32 cpu_boot_status_reg,
-			u32 msg_to_cpu_reg, u32 boot_err0_reg, bool skip_bmc,
-			u32 cpu_timeout)
+			u32 msg_to_cpu_reg, u32 cpu_msg_status_reg,
+			u32 boot_err0_reg, bool skip_bmc,
+			u32 cpu_timeout, u32 boot_fit_timeout)
 {
 	u32 status;
 	int rc;
@@ -337,6 +338,48 @@ int hl_fw_init_cpu(struct hl_device *hdev, u32 cpu_boot_status_reg,
 	dev_info(hdev->dev, "Going to wait for device boot (up to %lds)\n",
 		cpu_timeout / USEC_PER_SEC);
 
+	/* Wait for boot FIT request */
+	rc = hl_poll_timeout(
+		hdev,
+		cpu_boot_status_reg,
+		status,
+		status == CPU_BOOT_STATUS_WAITING_FOR_BOOT_FIT,
+		10000,
+		boot_fit_timeout);
+
+	if (rc) {
+		dev_dbg(hdev->dev,
+			"No boot fit request received, resuming boot\n");
+	} else {
+		rc = hdev->asic_funcs->load_boot_fit_to_device(hdev);
+		if (rc)
+			goto out;
+
+		/* Clear device CPU message status */
+		WREG32(cpu_msg_status_reg, CPU_MSG_CLR);
+
+		/* Signal device CPU that boot loader is ready */
+		WREG32(msg_to_cpu_reg, KMD_MSG_FIT_RDY);
+
+		/* Poll for CPU device ack */
+		rc = hl_poll_timeout(
+			hdev,
+			cpu_msg_status_reg,
+			status,
+			status == CPU_MSG_OK,
+			10000,
+			boot_fit_timeout);
+
+		if (rc) {
+			dev_err(hdev->dev,
+				"Timeout waiting for boot fit load ack\n");
+			goto out;
+		}
+
+		/* Clear message */
+		WREG32(msg_to_cpu_reg, KMD_MSG_NA);
+	}
+
 	/* Make sure CPU boot-loader is running */
 	rc = hl_poll_timeout(
 		hdev,
@@ -396,7 +439,8 @@ int hl_fw_init_cpu(struct hl_device *hdev, u32 cpu_boot_status_reg,
 			break;
 		default:
 			dev_err(hdev->dev,
-				"Device boot error - Invalid status code\n");
+				"Device boot error - Invalid status code %d\n",
+				status);
 			break;
 		}
 
@@ -450,6 +494,9 @@ int hl_fw_init_cpu(struct hl_device *hdev, u32 cpu_boot_status_reg,
 		10000,
 		cpu_timeout);
 
+	/* Clear message */
+	WREG32(msg_to_cpu_reg, KMD_MSG_NA);
+
 	if (rc) {
 		if (status == CPU_BOOT_STATUS_FIT_CORRUPTED)
 			dev_err(hdev->dev,
@@ -458,7 +505,6 @@ int hl_fw_init_cpu(struct hl_device *hdev, u32 cpu_boot_status_reg,
 			dev_err(hdev->dev,
 				"Device failed to load, %d\n", status);
 
-		WREG32(msg_to_cpu_reg, KMD_MSG_NA);
 		rc = -EIO;
 		goto out;
 	}

drivers/misc/habanalabs/goya/goya.c

@@ -72,7 +72,7 @@
  *
  */
 
-#define GOYA_UBOOT_FW_FILE	"habanalabs/goya/goya-u-boot.bin"
+#define GOYA_BOOT_FIT_FILE	"habanalabs/goya/goya-boot-fit.itb"
 #define GOYA_LINUX_FW_FILE	"habanalabs/goya/goya-fit.itb"
 
 #define GOYA_MMU_REGS_NUM		63
@@ -87,6 +87,7 @@
 #define GOYA_TEST_QUEUE_WAIT_USEC	100000		/* 100ms */
 #define GOYA_PLDM_MMU_TIMEOUT_USEC	(MMU_CONFIG_TIMEOUT_USEC * 100)
 #define GOYA_PLDM_QMAN0_TIMEOUT_USEC	(HL_DEVICE_TIMEOUT_USEC * 30)
+#define GOYA_BOOT_FIT_REQ_TIMEOUT_USEC	1000000		/* 1s */
 
 #define GOYA_QMAN0_FENCE_VAL		0xD169B243
 
@@ -2210,80 +2211,37 @@ static void goya_halt_engines(struct hl_device *hdev, bool hard_reset)
 }
 
 /*
- * goya_push_uboot_to_device() - Push u-boot FW code to device.
+ * goya_load_firmware_to_device() - Load LINUX FW code to device.
  * @hdev: Pointer to hl_device structure.
  *
- * Copy u-boot fw code from firmware file to SRAM BAR.
+ * Copy LINUX fw code from firmware file to HBM BAR.
  *
  * Return: 0 on success, non-zero for failure.
  */
-static int goya_push_uboot_to_device(struct hl_device *hdev)
+static int goya_load_firmware_to_device(struct hl_device *hdev)
 {
 	void __iomem *dst;
 
-	dst = hdev->pcie_bar[SRAM_CFG_BAR_ID] + UBOOT_FW_OFFSET;
+	dst = hdev->pcie_bar[DDR_BAR_ID] + LINUX_FW_OFFSET;
 
-	return hl_fw_load_fw_to_device(hdev, GOYA_UBOOT_FW_FILE, dst);
+	return hl_fw_load_fw_to_device(hdev, GOYA_LINUX_FW_FILE, dst);
 }
 
 /*
- * goya_load_firmware_to_device() - Load LINUX FW code to device.
+ * goya_load_boot_fit_to_device() - Load boot fit to device.
  * @hdev: Pointer to hl_device structure.
  *
- * Copy LINUX fw code from firmware file to HBM BAR.
+ * Copy boot fit file to SRAM BAR.
  *
  * Return: 0 on success, non-zero for failure.
  */
-static int goya_load_firmware_to_device(struct hl_device *hdev)
+static int goya_load_boot_fit_to_device(struct hl_device *hdev)
 {
 	void __iomem *dst;
 
-	dst = hdev->pcie_bar[DDR_BAR_ID] + LINUX_FW_OFFSET;
-
-	return hl_fw_load_fw_to_device(hdev, GOYA_LINUX_FW_FILE, dst);
-}
-
-static int goya_pldm_init_cpu(struct hl_device *hdev)
-{
-	u32 unit_rst_val;
-	int rc;
-
-	/* Must initialize SRAM scrambler before pushing u-boot to SRAM */
-	goya_init_golden_registers(hdev);
-
-	/* Put ARM cores into reset */
-	WREG32(mmCPU_CA53_CFG_ARM_RST_CONTROL, CPU_RESET_ASSERT);
-	RREG32(mmCPU_CA53_CFG_ARM_RST_CONTROL);
+	dst = hdev->pcie_bar[SRAM_CFG_BAR_ID] + BOOT_FIT_SRAM_OFFSET;
 
-	/* Reset the CA53 MACRO */
-	unit_rst_val = RREG32(mmPSOC_GLOBAL_CONF_UNIT_RST_N);
-	WREG32(mmPSOC_GLOBAL_CONF_UNIT_RST_N, CA53_RESET);
-	RREG32(mmPSOC_GLOBAL_CONF_UNIT_RST_N);
-	WREG32(mmPSOC_GLOBAL_CONF_UNIT_RST_N, unit_rst_val);
-	RREG32(mmPSOC_GLOBAL_CONF_UNIT_RST_N);
-
-	rc = goya_push_uboot_to_device(hdev);
-	if (rc)
-		return rc;
-
-	rc = goya_load_firmware_to_device(hdev);
-	if (rc)
-		return rc;
-
-	WREG32(mmPSOC_GLOBAL_CONF_UBOOT_MAGIC, KMD_MSG_FIT_RDY);
-	WREG32(mmPSOC_GLOBAL_CONF_WARM_REBOOT, CPU_BOOT_STATUS_NA);
-
-	WREG32(mmCPU_CA53_CFG_RST_ADDR_LSB_0,
-		lower_32_bits(SRAM_BASE_ADDR + UBOOT_FW_OFFSET));
-	WREG32(mmCPU_CA53_CFG_RST_ADDR_MSB_0,
-		upper_32_bits(SRAM_BASE_ADDR + UBOOT_FW_OFFSET));
-
-	/* Release ARM core 0 from reset */
-	WREG32(mmCPU_CA53_CFG_ARM_RST_CONTROL,
-					CPU_RESET_CORE0_DEASSERT);
-	RREG32(mmCPU_CA53_CFG_ARM_RST_CONTROL);
-
-	return 0;
+	return hl_fw_load_fw_to_device(hdev, GOYA_BOOT_FIT_FILE, dst);
 }
 
 /*
@@ -2325,7 +2283,7 @@ static void goya_read_device_fw_version(struct hl_device *hdev,
 	}
 }
 
-static int goya_init_cpu(struct hl_device *hdev, u32 cpu_timeout)
+static int goya_init_cpu(struct hl_device *hdev)
 {
 	struct goya_device *goya = hdev->asic_specific;
 	int rc;
@@ -2346,22 +2304,15 @@ static int goya_init_cpu(struct hl_device *hdev, u32 cpu_timeout)
 		return -EIO;
 	}
 
-	if (hdev->pldm) {
-		rc = goya_pldm_init_cpu(hdev);
-		if (rc)
-			return rc;
-
-		goto out;
-	}
-
 	rc = hl_fw_init_cpu(hdev, mmPSOC_GLOBAL_CONF_CPU_BOOT_STATUS,
-			mmPSOC_GLOBAL_CONF_UBOOT_MAGIC, mmCPU_BOOT_ERR0,
-			false, cpu_timeout);
+			mmPSOC_GLOBAL_CONF_UBOOT_MAGIC,
+			mmCPU_CMD_STATUS_TO_HOST, mmCPU_BOOT_ERR0,
+			false, GOYA_CPU_TIMEOUT_USEC,
+			GOYA_BOOT_FIT_REQ_TIMEOUT_USEC);
 
 	if (rc)
 		return rc;
 
-out:
 	goya->hw_cap_initialized |= HW_CAP_CPU;
 
 	return 0;
@@ -2476,7 +2427,7 @@ static int goya_hw_init(struct hl_device *hdev)
 	 */
 	WREG32(mmHW_STATE, HL_DEVICE_HW_STATE_DIRTY);
 
-	rc = goya_init_cpu(hdev, GOYA_CPU_TIMEOUT_USEC);
+	rc = goya_init_cpu(hdev);
 	if (rc) {
 		dev_err(hdev->dev, "failed to initialize CPU\n");
 		return rc;
@@ -5270,6 +5221,7 @@ static const struct hl_asic_funcs goya_funcs = {
 	.get_queue_id_for_cq = goya_get_queue_id_for_cq,
 	.read_device_fw_version = goya_read_device_fw_version,
 	.load_firmware_to_device = goya_load_firmware_to_device,
+	.load_boot_fit_to_device = goya_load_boot_fit_to_device,
 	.set_dma_mask_from_fw = goya_set_dma_mask_from_fw,
 	.get_device_time = goya_get_device_time
 };

drivers/misc/habanalabs/habanalabs.h

@@ -552,6 +552,7 @@ enum hl_pll_frequency {
  * @read_device_fw_version: read the device's firmware versions that are
  *                          contained in registers
  * @load_firmware_to_device: load the firmware to the device's memory
+ * @load_boot_fit_to_device: load boot fit to device's memory
  * @set_dma_mask_from_fw: set the DMA mask in the driver according to the
  *                        firmware configuration
  * @get_device_time: Get the device time.
@@ -646,6 +647,7 @@ struct hl_asic_funcs {
 	void (*read_device_fw_version)(struct hl_device *hdev,
 					enum hl_fw_component fwc);
 	int (*load_firmware_to_device)(struct hl_device *hdev);
+	int (*load_boot_fit_to_device)(struct hl_device *hdev);
 	void (*set_dma_mask_from_fw)(struct hl_device *hdev);
 	u64 (*get_device_time)(struct hl_device *hdev);
 };
@@ -1644,8 +1646,9 @@ int hl_fw_send_heartbeat(struct hl_device *hdev);
 int hl_fw_armcp_info_get(struct hl_device *hdev);
 int hl_fw_get_eeprom_data(struct hl_device *hdev, void *data, size_t max_size);
 int hl_fw_init_cpu(struct hl_device *hdev, u32 cpu_boot_status_reg,
-			u32 msg_to_cpu_reg, u32 boot_err0_reg, bool skip_bmc,
-			u32 cpu_timeout);
+			u32 msg_to_cpu_reg, u32 cpu_msg_status_reg,
+			u32 boot_err0_reg, bool skip_bmc,
+			u32 cpu_timeout, u32 boot_fit_timeout);
 
 int hl_pci_bars_map(struct hl_device *hdev, const char * const name[3],
 			bool is_wc[3]);

drivers/misc/habanalabs/include/hl_boot_if.h

@@ -11,6 +11,8 @@
 #define LKD_HARD_RESET_MAGIC		0xED7BD694
 #define HL_POWER9_HOST_MAGIC		0x1DA30009
 
+#define BOOT_FIT_SRAM_OFFSET		0x200000
+
 /*
  * CPU error bits in BOOT_ERROR registers
  *
@@ -77,6 +79,7 @@ enum cpu_boot_status {
 	CPU_BOOT_STATUS_BMC_WAITING_SKIPPED, /* deprecated - will be removed */
 	/* Last boot loader progress status, ready to receive commands */
 	CPU_BOOT_STATUS_READY_TO_BOOT = 15,
+	CPU_BOOT_STATUS_WAITING_FOR_BOOT_FIT = 16,
 };
 
 enum kmd_msg {