wifi: iwlwifi: back off on continuous errors

When errors occur repeatedly, the driver shouldn't go into a
tight loop trying to reset the device. Implement the backoff
I had already defined IWL_TRANS_RESET_DELAY for, but clearly
forgotten the implementation of.

Fixes: 9a2f13c ("wifi: iwlwifi: implement reset escalation")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com>
Link: https://patch.msgid.link/20250420095642.8816e299efa2.I82cde34e2345a2b33b1f03dbb040f5ad3439a5aa@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

Author: jmberg-intel

drivers/net/wireless/intel/iwlwifi/iwl-trans.c

@@ -21,6 +21,7 @@ struct iwl_trans_dev_restart_data {
 	struct list_head list;
 	unsigned int restart_count;
 	time64_t last_error;
+	bool backoff;
 	char name[];
 };
 
@@ -125,13 +126,20 @@ iwl_trans_determine_restart_mode(struct iwl_trans *trans)
 	if (!data)
 		return at_least;
 
-	if (ktime_get_boottime_seconds() - data->last_error >=
+	if (!data->backoff &&
+	    ktime_get_boottime_seconds() - data->last_error >=
 			IWL_TRANS_RESET_OK_TIME)
 		data->restart_count = 0;
 
 	index = data->restart_count;
-	if (index >= ARRAY_SIZE(escalation_list))
+	if (index >= ARRAY_SIZE(escalation_list)) {
 		index = ARRAY_SIZE(escalation_list) - 1;
+		if (!data->backoff) {
+			data->backoff = true;
+			return IWL_RESET_MODE_BACKOFF;
+		}
+		data->backoff = false;
+	}
 
 	return max(at_least, escalation_list[index]);
 }
@@ -140,7 +148,8 @@ iwl_trans_determine_restart_mode(struct iwl_trans *trans)
 
 static void iwl_trans_restart_wk(struct work_struct *wk)
 {
-	struct iwl_trans *trans = container_of(wk, typeof(*trans), restart.wk);
+	struct iwl_trans *trans = container_of(wk, typeof(*trans),
+					       restart.wk.work);
 	struct iwl_trans_reprobe *reprobe;
 	enum iwl_reset_mode mode;
 
@@ -168,6 +177,12 @@ static void iwl_trans_restart_wk(struct work_struct *wk)
 		return;
 
 	mode = iwl_trans_determine_restart_mode(trans);
+	if (mode == IWL_RESET_MODE_BACKOFF) {
+		IWL_ERR(trans, "Too many device errors - delay next reset\n");
+		queue_delayed_work(system_unbound_wq, &trans->restart.wk,
+				   IWL_TRANS_RESET_DELAY);
+		return;
+	}
 
 	iwl_trans_inc_restart_count(trans->dev);
 
@@ -227,7 +242,7 @@ struct iwl_trans *iwl_trans_alloc(unsigned int priv_size,
 	trans->dev = dev;
 	trans->num_rx_queues = 1;
 
-	INIT_WORK(&trans->restart.wk, iwl_trans_restart_wk);
+	INIT_DELAYED_WORK(&trans->restart.wk, iwl_trans_restart_wk);
 
 	return trans;
 }
@@ -271,7 +286,7 @@ int iwl_trans_init(struct iwl_trans *trans)
 
 void iwl_trans_free(struct iwl_trans *trans)
 {
-	cancel_work_sync(&trans->restart.wk);
+	cancel_delayed_work_sync(&trans->restart.wk);
 	kmem_cache_destroy(trans->dev_cmd_pool);
 }
 
@@ -403,7 +418,7 @@ void iwl_trans_op_mode_leave(struct iwl_trans *trans)
 
 	iwl_trans_pcie_op_mode_leave(trans);
 
-	cancel_work_sync(&trans->restart.wk);
+	cancel_delayed_work_sync(&trans->restart.wk);
 
 	trans->op_mode = NULL;
 

drivers/net/wireless/intel/iwlwifi/iwl-trans.h

@@ -961,7 +961,7 @@ struct iwl_trans {
 	struct iwl_dma_ptr invalid_tx_cmd;
 
 	struct {
-		struct work_struct wk;
+		struct delayed_work wk;
 		struct iwl_fw_error_dump_mode mode;
 		bool during_reset;
 	} restart;
@@ -1162,7 +1162,7 @@ static inline void iwl_trans_schedule_reset(struct iwl_trans *trans,
 	 */
 	trans->restart.during_reset = test_bit(STATUS_IN_SW_RESET,
 					       &trans->status);
-	queue_work(system_unbound_wq, &trans->restart.wk);
+	queue_delayed_work(system_unbound_wq, &trans->restart.wk, 0);
 }
 
 static inline void iwl_trans_fw_error(struct iwl_trans *trans,
@@ -1261,6 +1261,9 @@ enum iwl_reset_mode {
 	IWL_RESET_MODE_RESCAN,
 	IWL_RESET_MODE_FUNC_RESET,
 	IWL_RESET_MODE_PROD_RESET,
+
+	/* keep last - special backoff value */
+	IWL_RESET_MODE_BACKOFF,
 };
 
 void iwl_trans_pcie_reset(struct iwl_trans *trans, enum iwl_reset_mode mode);

drivers/net/wireless/intel/iwlwifi/pcie/trans.c

@@ -2351,7 +2351,8 @@ void iwl_trans_pcie_reset(struct iwl_trans *trans, enum iwl_reset_mode mode)
 	struct iwl_trans_pcie_removal *removal;
 	char _msg = 0, *msg = &_msg;
 
-	if (WARN_ON(mode < IWL_RESET_MODE_REMOVE_ONLY))
+	if (WARN_ON(mode < IWL_RESET_MODE_REMOVE_ONLY ||
+		    mode == IWL_RESET_MODE_BACKOFF))
 		return;
 
 	if (test_bit(STATUS_TRANS_DEAD, &trans->status))