habanalabs: send an event notification when CS timeout occurs

The Driver needs to inform the User process whenever one of its
CS is timed out. The Driver shall recognize the CS timeout and shall
send an eventfd notification, towards user space, whenever a timeout
is expired on a CS.

Signed-off-by: Tal Cohen <talcohen@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>

diff --git a/drivers/misc/habanalabs/common/command_submission.c b/drivers/misc/habanalabs/common/command_submission.c
index 47b49cb..cbb7c29 100644 (file)
--- a/drivers/misc/habanalabs/common/command_submission.c
+++ b/drivers/misc/habanalabs/common/command_submission.c
@@ -797,10 +797,11 @@ out:
 static void cs_timedout(struct work_struct *work)
 {
        struct hl_device *hdev;
+       u64 event_mask;
        int rc;
        struct hl_cs *cs = container_of(work, struct hl_cs,
                                                 work_tdr.work);
-       bool skip_reset_on_timeout = cs->skip_reset_on_timeout;
+       bool skip_reset_on_timeout = cs->skip_reset_on_timeout, device_reset = false;
 
        rc = cs_get_unless_zero(cs);
        if (!rc)
@@ -811,9 +812,15 @@ static void cs_timedout(struct work_struct *work)
                return;
        }
 
-       /* Mark the CS is timed out so we won't try to cancel its TDR */
-       if (likely(!skip_reset_on_timeout))
+       if (likely(!skip_reset_on_timeout)) {
+               if (hdev->reset_on_lockup)
+                       device_reset = true;
+               else
+                       hdev->reset_info.needs_reset = true;
+
+               /* Mark the CS is timed out so we won't try to cancel its TDR */
                cs->timedout = true;
+       }
 
        hdev = cs->ctx->hdev;
 
@@ -822,6 +829,11 @@ static void cs_timedout(struct work_struct *work)
        if (rc) {
                hdev->last_error.cs_timeout.timestamp = ktime_get();
                hdev->last_error.cs_timeout.seq = cs->sequence;
+
+               event_mask = device_reset ? (HL_NOTIFIER_EVENT_CS_TIMEOUT |
+                               HL_NOTIFIER_EVENT_DEVICE_RESET) : HL_NOTIFIER_EVENT_CS_TIMEOUT;
+
+               hl_notifier_event_send_all(hdev, event_mask);
        }
 
        switch (cs->type) {
@@ -856,12 +868,8 @@ static void cs_timedout(struct work_struct *work)
 
        cs_put(cs);
 
-       if (likely(!skip_reset_on_timeout)) {
-               if (hdev->reset_on_lockup)
-                       hl_device_reset(hdev, HL_DRV_RESET_TDR);
-               else
-                       hdev->reset_info.needs_reset = true;
-       }
+       if (device_reset)
+               hl_device_reset(hdev, HL_DRV_RESET_TDR);
 }
 
 static int allocate_cs(struct hl_device *hdev, struct hl_ctx *ctx,

diff --git a/include/uapi/misc/habanalabs.h b/include/uapi/misc/habanalabs.h
index 5f9a609..18f86d2 100644 (file)
--- a/include/uapi/misc/habanalabs.h
+++ b/include/uapi/misc/habanalabs.h
@@ -1435,10 +1435,12 @@ struct hl_debug_args {
  * HL_NOTIFIER_EVENT_TPC_ASSERT       - Indicates TPC assert event
  * HL_NOTIFIER_EVENT_UNDEFINED_OPCODE - Indicates undefined operation code
  * HL_NOTIFIER_EVENT_DEVICE_RESET     - Indicates device requires a reset
+ * HL_NOTIFIER_EVENT_CS_TIMEOUT       - Indicates CS timeout error
  */
 #define HL_NOTIFIER_EVENT_TPC_ASSERT           (1ULL << 0)
 #define HL_NOTIFIER_EVENT_UNDEFINED_OPCODE     (1ULL << 1)
 #define HL_NOTIFIER_EVENT_DEVICE_RESET         (1ULL << 2)
+#define HL_NOTIFIER_EVENT_CS_TIMEOUT            (1ULL << 3)
 
 /*
  * Various information operations such as: