[PATCH] habanalabs: full FW hard reset support

[PATCH] habanalabs: full FW hard reset support

[Oded Gabbay]

From: Ofir Bitton <obitton@habana.ai>

Driver must fetch FW hard reset capability at every FW boot stage:
preboot, CPU boot, CPU application.
If hard reset is triggered, driver will take into consideration
only the last capability received.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/common/firmware_if.c | 51 +++++++++++++++-----
 1 file changed, 38 insertions(+), 13 deletions(-)

diff --git a/drivers/misc/habanalabs/common/firmware_if.c b/drivers/misc/habanalabs/common/firmware_if.c
index c970bfc6db66..7720b48c6239 100644
--- a/drivers/misc/habanalabs/common/firmware_if.c
+++ b/drivers/misc/habanalabs/common/firmware_if.c
@@ -629,32 +629,34 @@ int hl_fw_read_preboot_status(struct hl_device *hdev, u32 cpu_boot_status_reg,
 	/* We read security status multiple times during boot:
 	 * 1. preboot - a. Check whether the security status bits are valid
 	 *              b. Check whether fw security is enabled
-	 *              c. Check whether hard reset is done by fw
-	 * 2. boot cpu - we get boot cpu security status
-	 * 3. FW application - we get FW application security status
+	 *              c. Check whether hard reset is done by preboot
+	 * 2. boot cpu - a. Fetch boot cpu security status
+	 *               b. Check whether hard reset is done by boot cpu
+	 * 3. FW application - a. Fetch fw application security status
+	 *                     b. Check whether hard reset is done by fw app
 	 *
 	 * Preboot:
 	 * Check security status bit (CPU_BOOT_DEV_STS0_ENABLED), if it is set
 	 * check security enabled bit (CPU_BOOT_DEV_STS0_SECURITY_EN)
 	 */
 	if (security_status & CPU_BOOT_DEV_STS0_ENABLED) {
-		hdev->asic_prop.fw_security_status_valid = 1;
+		prop->fw_security_status_valid = 1;
 
 		if (!(security_status & CPU_BOOT_DEV_STS0_SECURITY_EN))
 			prop->fw_security_disabled = true;
 
 		if (security_status & CPU_BOOT_DEV_STS0_FW_HARD_RST_EN)
-			hdev->asic_prop.hard_reset_done_by_fw = true;
+			prop->hard_reset_done_by_fw = true;
 	} else {
-		hdev->asic_prop.fw_security_status_valid = 0;
+		prop->fw_security_status_valid = 0;
 		prop->fw_security_disabled = true;
 	}
 
-	dev_dbg(hdev->dev, "Firmware hard-reset is %s\n",
-		hdev->asic_prop.hard_reset_done_by_fw ? "enabled" : "disabled");
+	dev_dbg(hdev->dev, "Firmware preboot hard-reset is %s\n",
+			prop->hard_reset_done_by_fw ? "enabled" : "disabled");
 
 	dev_info(hdev->dev, "firmware-level security is %s\n",
-		prop->fw_security_disabled ? "disabled" : "enabled");
+			prop->fw_security_disabled ? "disabled" : "enabled");
 
 	return 0;
 }
@@ -664,6 +666,7 @@ int hl_fw_init_cpu(struct hl_device *hdev, u32 cpu_boot_status_reg,
 			u32 cpu_security_boot_status_reg, u32 boot_err0_reg,
 			bool skip_bmc, u32 cpu_timeout, u32 boot_fit_timeout)
 {
+	struct asic_fixed_properties *prop = &hdev->asic_prop;
 	u32 status;
 	int rc;
 
@@ -732,11 +735,22 @@ int hl_fw_init_cpu(struct hl_device *hdev, u32 cpu_boot_status_reg,
 	/* Read U-Boot version now in case we will later fail */
 	hdev->asic_funcs->read_device_fw_version(hdev, FW_COMP_UBOOT);
 
+	/* Clear reset status since we need to read it again from boot CPU */
+	prop->hard_reset_done_by_fw = false;
+
 	/* Read boot_cpu security bits */
-	if (hdev->asic_prop.fw_security_status_valid)
-		hdev->asic_prop.fw_boot_cpu_security_map =
+	if (prop->fw_security_status_valid) {
+		prop->fw_boot_cpu_security_map =
 				RREG32(cpu_security_boot_status_reg);
 
+		if (prop->fw_boot_cpu_security_map &
+				CPU_BOOT_DEV_STS0_FW_HARD_RST_EN)
+			prop->hard_reset_done_by_fw = true;
+	}
+
+	dev_dbg(hdev->dev, "Firmware boot CPU hard-reset is %s\n",
+			prop->hard_reset_done_by_fw ? "enabled" : "disabled");
+
 	if (rc) {
 		detect_cpu_boot_status(hdev, status);
 		rc = -EIO;
@@ -805,11 +819,22 @@ int hl_fw_init_cpu(struct hl_device *hdev, u32 cpu_boot_status_reg,
 		goto out;
 	}
 
+	/* Clear reset status since we need to read again from app */
+	prop->hard_reset_done_by_fw = false;
+
 	/* Read FW application security bits */
-	if (hdev->asic_prop.fw_security_status_valid)
-		hdev->asic_prop.fw_app_security_map =
+	if (prop->fw_security_status_valid) {
+		prop->fw_app_security_map =
 				RREG32(cpu_security_boot_status_reg);
 
+		if (prop->fw_app_security_map &
+				CPU_BOOT_DEV_STS0_FW_HARD_RST_EN)
+			prop->hard_reset_done_by_fw = true;
+	}
+
+	dev_dbg(hdev->dev, "Firmware application CPU hard-reset is %s\n",
+			prop->hard_reset_done_by_fw ? "enabled" : "disabled");
+
 	dev_info(hdev->dev, "Successfully loaded firmware to device\n");
 
 out:
-- 
2.17.1

[PATCH] habanalabs/gaudi: disable CGM at HW initialization

[Oded Gabbay]

In case the clock gating was enabled in preboot we need to disable it
at the H/W initialization stage before touching the MME/TPC registers.
Otherwise, the ASIC can get stuck. If the security is enabled in
the firmware level, the CGM is always disabled and the driver can't
enable it.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/gaudi/gaudi.c              | 14 +++++++++++---
 .../misc/habanalabs/include/common/hl_boot_if.h    |  5 +++++
 2 files changed, 16 insertions(+), 3 deletions(-)

diff --git a/drivers/misc/habanalabs/gaudi/gaudi.c b/drivers/misc/habanalabs/gaudi/gaudi.c
index 65895ba075fe..f316b898e8e0 100644
--- a/drivers/misc/habanalabs/gaudi/gaudi.c
+++ b/drivers/misc/habanalabs/gaudi/gaudi.c
@@ -2403,8 +2403,6 @@ static void gaudi_init_golden_registers(struct hl_device *hdev)
 	gaudi_init_e2e(hdev);
 	gaudi_init_hbm_cred(hdev);
 
-	hdev->asic_funcs->disable_clock_gating(hdev);
-
 	for (tpc_id = 0, tpc_offset = 0;
 				tpc_id < TPC_NUMBER_OF_ENGINES;
 				tpc_id++, tpc_offset += TPC_CFG_OFFSET) {
@@ -3416,6 +3414,9 @@ static void gaudi_set_clock_gating(struct hl_device *hdev)
 	if (hdev->in_debug)
 		return;
 
+	if (!hdev->asic_prop.fw_security_disabled)
+		return;
+
 	for (i = GAUDI_PCI_DMA_1, qman_offset = 0 ; i < GAUDI_HBM_DMA_1 ; i++) {
 		enable = !!(hdev->clock_gating_mask &
 				(BIT_ULL(gaudi_dma_assignment[i])));
@@ -3467,7 +3468,7 @@ static void gaudi_disable_clock_gating(struct hl_device *hdev)
 	u32 qman_offset;
 	int i;
 
-	if (!(gaudi->hw_cap_initialized & HW_CAP_CLK_GATE))
+	if (!hdev->asic_prop.fw_security_disabled)
 		return;
 
 	for (i = 0, qman_offset = 0 ; i < DMA_NUMBER_OF_CHANNELS ; i++) {
@@ -3801,6 +3802,13 @@ static int gaudi_hw_init(struct hl_device *hdev)
 		return rc;
 	}
 
+	/* In case the clock gating was enabled in preboot we need to disable
+	 * it here before touching the MME/TPC registers.
+	 * There is no need to take clk gating mutex because when this function
+	 * runs, no other relevant code can run
+	 */
+	hdev->asic_funcs->disable_clock_gating(hdev);
+
 	/* SRAM scrambler must be initialized after CPU is running from HBM */
 	gaudi_init_scrambler_sram(hdev);
 
diff --git a/drivers/misc/habanalabs/include/common/hl_boot_if.h b/drivers/misc/habanalabs/include/common/hl_boot_if.h
index 755c4800f002..7cb5f2d3e565 100644
--- a/drivers/misc/habanalabs/include/common/hl_boot_if.h
+++ b/drivers/misc/habanalabs/include/common/hl_boot_if.h
@@ -150,6 +150,10 @@
  * CPU_BOOT_DEV_STS0_PLL_INFO_EN	FW retrieval of PLL info is enabled.
  *					Initialized in: linux
  *
+ * CPU_BOOT_DEV_STS0_CLK_GATE_EN	Clock Gating enabled.
+ *					FW initialized Clock Gating.
+ *					Initialized in: preboot
+ *
  * CPU_BOOT_DEV_STS0_ENABLED		Device status register enabled.
  *					This is a main indication that the
  *					running FW populates the device status
@@ -171,6 +175,7 @@
 #define CPU_BOOT_DEV_STS0_DRAM_SCR_EN			(1 << 9)
 #define CPU_BOOT_DEV_STS0_FW_HARD_RST_EN		(1 << 10)
 #define CPU_BOOT_DEV_STS0_PLL_INFO_EN			(1 << 11)
+#define CPU_BOOT_DEV_STS0_CLK_GATE_EN			(1 << 13)
 #define CPU_BOOT_DEV_STS0_ENABLED			(1 << 31)
 
 enum cpu_boot_status {
-- 
2.17.1

[PATCH] habanalabs/gaudi: enhance reset message

[Oded Gabbay]

Print the initiator who performs the hard-reset for easier debugging.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/gaudi/gaudi.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/drivers/misc/habanalabs/gaudi/gaudi.c b/drivers/misc/habanalabs/gaudi/gaudi.c
index f316b898e8e0..b0528883a995 100644
--- a/drivers/misc/habanalabs/gaudi/gaudi.c
+++ b/drivers/misc/habanalabs/gaudi/gaudi.c
@@ -3936,11 +3936,15 @@ static void gaudi_hw_fini(struct hl_device *hdev, bool hard_reset)
 
 		WREG32(mmPSOC_GLOBAL_CONF_SW_ALL_RST,
 			1 << PSOC_GLOBAL_CONF_SW_ALL_RST_IND_SHIFT);
-	}
 
-	dev_info(hdev->dev,
-		"Issued HARD reset command, going to wait %dms\n",
-		reset_timeout_ms);
+		dev_info(hdev->dev,
+			"Issued HARD reset command, going to wait %dms\n",
+			reset_timeout_ms);
+	} else {
+		dev_info(hdev->dev,
+			"Firmware performs HARD reset, going to wait %dms\n",
+			reset_timeout_ms);
+	}
 
 	/*
 	 * After hard reset, we can't poll the BTM_FSM register because the PSOC
-- 
2.17.1

[PATCH] habanalabs: update comment in hl_boot_if.h

[Oded Gabbay]

Hard-reset flag is updated in many stages of the boot sequence of the
firmware.

Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/include/common/hl_boot_if.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/misc/habanalabs/include/common/hl_boot_if.h b/drivers/misc/habanalabs/include/common/hl_boot_if.h
index 7cb5f2d3e565..b637dfd69f6e 100644
--- a/drivers/misc/habanalabs/include/common/hl_boot_if.h
+++ b/drivers/misc/habanalabs/include/common/hl_boot_if.h
@@ -145,7 +145,7 @@
  *					implemented. This means that FW will
  *					perform hard reset procedure on
  *					receiving the halt-machine event.
- *					Initialized in: linux
+ *					Initialized in: preboot, u-boot, linux
  *
  * CPU_BOOT_DEV_STS0_PLL_INFO_EN	FW retrieval of PLL info is enabled.
  *					Initialized in: linux
-- 
2.17.1