diff mbox series

[5/6] habanalabs: use notifications and graceful reset for decoder

Message ID 20230227111306.3985896-5-ogabbay@kernel.org (mailing list archive)
State New, archived
Headers show
Series [1/6] habanalabs: add helper function to get vm hash node | expand

Commit Message

Oded Gabbay Feb. 27, 2023, 11:13 a.m. UTC 
From: Tomer Tayar <ttayar@habana.ai>

Add notifications to user in case of decoder abnormal interrupts, and
use the graceful reset mechanism if reset is required.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/accel/habanalabs/common/decoder.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)

Comments

Stanislaw Gruszka March 1, 2023, 1:34 p.m. UTC | #1 
On Mon, Feb 27, 2023 at 01:13:05PM +0200, Oded Gabbay wrote:
> From: Tomer Tayar <ttayar@habana.ai>
> 
> Add notifications to user in case of decoder abnormal interrupts, and
> use the graceful reset mechanism if reset is required.
> 
> Signed-off-by: Tomer Tayar <ttayar@habana.ai>
> Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
> Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
diff mbox series

Patch

diff --git a/drivers/accel/habanalabs/common/decoder.c b/drivers/accel/habanalabs/common/decoder.c
index 2aab14d74b53..69c78c1784b4 100644
--- a/drivers/accel/habanalabs/common/decoder.c
+++ b/drivers/accel/habanalabs/common/decoder.c
@@ -46,7 +46,7 @@  static void dec_print_abnrm_intr_source(struct hl_device *hdev, u32 irq_status)
 static void dec_error_intr_work(struct hl_device *hdev, u32 base_addr, u32 core_id)
 {
 	bool reset_required = false;
-	u32 irq_status;
+	u32 irq_status, event_mask;
 
 	irq_status = RREG32(base_addr + VCMD_IRQ_STATUS_OFFSET);
 
@@ -54,17 +54,27 @@  static void dec_error_intr_work(struct hl_device *hdev, u32 base_addr, u32 core_
 
 	dec_print_abnrm_intr_source(hdev, irq_status);
 
-	if (irq_status & VCMD_IRQ_STATUS_TIMEOUT_MASK)
-		reset_required = true;
-
 	/* Clear the interrupt */
 	WREG32(base_addr + VCMD_IRQ_STATUS_OFFSET, irq_status);
 
 	/* Flush the interrupt clear */
 	RREG32(base_addr + VCMD_IRQ_STATUS_OFFSET);
 
-	if (reset_required)
-		hl_device_reset(hdev, HL_DRV_RESET_HARD);
+	if (irq_status & VCMD_IRQ_STATUS_TIMEOUT_MASK) {
+		reset_required = true;
+		event_mask = HL_NOTIFIER_EVENT_GENERAL_HW_ERR;
+	} else if (irq_status & VCMD_IRQ_STATUS_CMDERR_MASK) {
+		event_mask = HL_NOTIFIER_EVENT_UNDEFINED_OPCODE;
+	} else {
+		event_mask = HL_NOTIFIER_EVENT_USER_ENGINE_ERR;
+	}
+
+	if (reset_required) {
+		event_mask |= HL_NOTIFIER_EVENT_DEVICE_RESET;
+		hl_device_cond_reset(hdev, 0, event_mask);
+	} else {
+		hl_notifier_event_send_all(hdev, event_mask);
+	}
 }
 
 static void dec_completion_abnrm(struct work_struct *work)