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

habanalabs/gaudi2: add device unavailable notification

Browse Source 
Device unavailable notifies the user that there isn't an option to
retrieve debug information from the device.
When a critical device error occurs and the f/w performs the device
reset, a device unavailable notification shall be sent to the user
process.

Signed-off-by: Tal Cohen <talcohen@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
This commit is contained in:
Tal Cohen 2022-09-28 18:33:19 +03:00 committed by Oded Gabbay
parent 16448d6444
commit 5731b6e6f0
1 changed files with 10 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
drivers/misc/habanalabs/gaudi2/gaudi2.c
View File

@ -8576,7 +8576,7 @@ static void gaudi2_handle_eqe(struct hl_device *hdev, struct hl_eq_entry *eq_ent
{ {
u32 ctl, reset_flags = HL_DRV_RESET_HARD | HL_DRV_RESET_DELAY; u32 ctl, reset_flags = HL_DRV_RESET_HARD | HL_DRV_RESET_DELAY;
struct gaudi2_device *gaudi2 = hdev->asic_specific; struct gaudi2_device *gaudi2 = hdev->asic_specific;
bool reset_required = false, skip_reset = false; bool reset_required = false, skip_reset = false, is_critical = false;
int index, sbte_index; int index, sbte_index;
u64 event_mask = 0; u64 event_mask = 0;
u16 event_type; u16 event_type;
@ -8602,6 +8602,7 @@ static void gaudi2_handle_eqe(struct hl_device *hdev, struct hl_eq_entry *eq_ent
reset_flags |= HL_DRV_RESET_FW_FATAL_ERR; reset_flags |= HL_DRV_RESET_FW_FATAL_ERR;
event_mask |= HL_NOTIFIER_EVENT_GENERAL_HW_ERR; event_mask |= HL_NOTIFIER_EVENT_GENERAL_HW_ERR;
reset_required = gaudi2_handle_ecc_event(hdev, event_type, &eq_entry->ecc_data); reset_required = gaudi2_handle_ecc_event(hdev, event_type, &eq_entry->ecc_data);
is_critical = eq_entry->ecc_data.is_critical;
break; break;
case GAUDI2_EVENT_TPC0_QM ... GAUDI2_EVENT_PDMA1_QM: case GAUDI2_EVENT_TPC0_QM ... GAUDI2_EVENT_PDMA1_QM:
@ -8976,9 +8977,16 @@ static void gaudi2_handle_eqe(struct hl_device *hdev, struct hl_eq_entry *eq_ent
return; return;
reset_device: reset_device:
if (hdev->hard_reset_on_fw_events) { if (hdev->asic_prop.fw_security_enabled && is_critical) {
reset_flags = HL_DRV_RESET_HARD | HL_DRV_RESET_BYPASS_REQ_TO_FW;
/* notify on device unavailable while the reset triggered by fw */
event_mask |= (HL_NOTIFIER_EVENT_DEVICE_RESET |
HL_NOTIFIER_EVENT_DEVICE_UNAVAILABLE);
hl_device_reset(hdev, reset_flags); hl_device_reset(hdev, reset_flags);
} else if (hdev->hard_reset_on_fw_events) {
event_mask |= HL_NOTIFIER_EVENT_DEVICE_RESET; event_mask |= HL_NOTIFIER_EVENT_DEVICE_RESET;
hl_device_reset(hdev, reset_flags);
} else { } else {
if (!gaudi2_irq_map_table[event_type].msg) if (!gaudi2_irq_map_table[event_type].msg)
hl_fw_unmask_irq(hdev, event_type); hl_fw_unmask_irq(hdev, event_type);