Merge branch 'mlx5-misc-fixes-2025-12-09'

Tariq Toukan says:

====================
mlx5 misc fixes 2025-12-09

This patchset provides misc bug fixes from the team to the mlx5 core and
Eth drivers.
====================

Link: https://patch.msgid.link/1765284977-1363052-1-git-send-email-tariqt@nvidia.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

Author: Paolo Abeni

drivers/net/ethernet/mellanox/mlx5/core/devlink.c

@@ -197,6 +197,11 @@ static int mlx5_devlink_reload_down(struct devlink *devlink, bool netns_change,
 	struct pci_dev *pdev = dev->pdev;
 	int ret = 0;
 
+	if (mlx5_fw_reset_in_progress(dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Can't reload during firmware reset");
+		return -EBUSY;
+	}
+
 	if (mlx5_dev_is_lightweight(dev)) {
 		if (action != DEVLINK_RELOAD_ACTION_DRIVER_REINIT)
 			return -EOPNOTSUPP;

drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c

@@ -33,6 +33,7 @@
 #include "lib/eq.h"
 #include "fw_tracer.h"
 #include "fw_tracer_tracepoint.h"
+#include <linux/ctype.h>
 
 static int mlx5_query_mtrc_caps(struct mlx5_fw_tracer *tracer)
 {
@@ -358,6 +359,47 @@ static const char *VAL_PARM		= "%llx";
 static const char *REPLACE_64_VAL_PARM	= "%x%x";
 static const char *PARAM_CHAR		= "%";
 
+static bool mlx5_is_valid_spec(const char *str)
+{
+	/* Parse format specifiers to find the actual type.
+	 * Structure: %[flags][width][.precision][length]type
+	 * Skip flags, width, precision & length.
+	 */
+	while (isdigit(*str) || *str == '#' || *str == '.' || *str == 'l')
+		str++;
+
+	/* Check if it's a valid integer/hex specifier or %%:
+	 * Valid formats: %x, %d, %i, %u, etc.
+	 */
+	if (*str != 'x' && *str != 'X' && *str != 'd' && *str != 'i' &&
+	    *str != 'u' && *str != 'c' && *str != '%')
+		return false;
+
+	return true;
+}
+
+static bool mlx5_tracer_validate_params(const char *str)
+{
+	const char *substr = str;
+
+	if (!str)
+		return false;
+
+	substr = strstr(substr, PARAM_CHAR);
+	while (substr) {
+		if (!mlx5_is_valid_spec(substr + 1))
+			return false;
+
+		if (*(substr + 1) == '%')
+			substr = strstr(substr + 2, PARAM_CHAR);
+		else
+			substr = strstr(substr + 1, PARAM_CHAR);
+
+	}
+
+	return true;
+}
+
 static int mlx5_tracer_message_hash(u32 message_id)
 {
 	return jhash_1word(message_id, 0) & (MESSAGE_HASH_SIZE - 1);
@@ -419,6 +461,10 @@ static int mlx5_tracer_get_num_of_params(char *str)
 	char *substr, *pstr = str;
 	int num_of_params = 0;
 
+	/* Validate that all parameters are valid before processing */
+	if (!mlx5_tracer_validate_params(str))
+		return -EINVAL;
+
 	/* replace %llx with %x%x */
 	substr = strstr(pstr, VAL_PARM);
 	while (substr) {
@@ -427,11 +473,15 @@ static int mlx5_tracer_get_num_of_params(char *str)
 		substr = strstr(pstr, VAL_PARM);
 	}
 
-	/* count all the % characters */
+	/* count all the % characters, but skip %% (escaped percent) */
 	substr = strstr(str, PARAM_CHAR);
 	while (substr) {
-		num_of_params += 1;
-		str = substr + 1;
+		if (*(substr + 1) != '%') {
+			num_of_params += 1;
+			str = substr + 1;
+		} else {
+			str = substr + 2;
+		}
 		substr = strstr(str, PARAM_CHAR);
 	}
 
@@ -570,14 +620,17 @@ void mlx5_tracer_print_trace(struct tracer_string_format *str_frmt,
 {
 	char	tmp[512];
 
-	snprintf(tmp, sizeof(tmp), str_frmt->string,
-		 str_frmt->params[0],
-		 str_frmt->params[1],
-		 str_frmt->params[2],
-		 str_frmt->params[3],
-		 str_frmt->params[4],
-		 str_frmt->params[5],
-		 str_frmt->params[6]);
+	if (str_frmt->invalid_string)
+		snprintf(tmp, sizeof(tmp), "BAD_FORMAT: %s", str_frmt->string);
+	else
+		snprintf(tmp, sizeof(tmp), str_frmt->string,
+			 str_frmt->params[0],
+			 str_frmt->params[1],
+			 str_frmt->params[2],
+			 str_frmt->params[3],
+			 str_frmt->params[4],
+			 str_frmt->params[5],
+			 str_frmt->params[6]);
 
 	trace_mlx5_fw(dev->tracer, trace_timestamp, str_frmt->lost,
 		      str_frmt->event_id, tmp);
@@ -609,6 +662,13 @@ static int mlx5_tracer_handle_raw_string(struct mlx5_fw_tracer *tracer,
 	return 0;
 }
 
+static void mlx5_tracer_handle_bad_format_string(struct mlx5_fw_tracer *tracer,
+						 struct tracer_string_format *cur_string)
+{
+	cur_string->invalid_string = true;
+	list_add_tail(&cur_string->list, &tracer->ready_strings_list);
+}
+
 static int mlx5_tracer_handle_string_trace(struct mlx5_fw_tracer *tracer,
 					   struct tracer_event *tracer_event)
 {
@@ -619,12 +679,18 @@ static int mlx5_tracer_handle_string_trace(struct mlx5_fw_tracer *tracer,
 		if (!cur_string)
 			return mlx5_tracer_handle_raw_string(tracer, tracer_event);
 
-		cur_string->num_of_params = mlx5_tracer_get_num_of_params(cur_string->string);
-		cur_string->last_param_num = 0;
 		cur_string->event_id = tracer_event->event_id;
 		cur_string->tmsn = tracer_event->string_event.tmsn;
 		cur_string->timestamp = tracer_event->string_event.timestamp;
 		cur_string->lost = tracer_event->lost_event;
+		cur_string->last_param_num = 0;
+		cur_string->num_of_params = mlx5_tracer_get_num_of_params(cur_string->string);
+		if (cur_string->num_of_params < 0) {
+			pr_debug("%s Invalid format string parameters\n",
+				 __func__);
+			mlx5_tracer_handle_bad_format_string(tracer, cur_string);
+			return 0;
+		}
 		if (cur_string->num_of_params == 0) /* trace with no params */
 			list_add_tail(&cur_string->list, &tracer->ready_strings_list);
 	} else {
@@ -634,6 +700,11 @@ static int mlx5_tracer_handle_string_trace(struct mlx5_fw_tracer *tracer,
 				 __func__, tracer_event->string_event.tmsn);
 			return mlx5_tracer_handle_raw_string(tracer, tracer_event);
 		}
+		if (cur_string->num_of_params < 0) {
+			pr_debug("%s string parameter of invalid string, dumping\n",
+				 __func__);
+			return 0;
+		}
 		cur_string->last_param_num += 1;
 		if (cur_string->last_param_num > TRACER_MAX_PARAMS) {
 			pr_debug("%s Number of params exceeds the max (%d)\n",

drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.h

@@ -125,6 +125,7 @@ struct tracer_string_format {
 	struct list_head list;
 	u32 timestamp;
 	bool lost;
+	bool invalid_string;
 };
 
 enum mlx5_fw_tracer_ownership_state {

drivers/net/ethernet/mellanox/mlx5/core/en.h

@@ -69,7 +69,7 @@ struct page_pool;
 #define MLX5E_METADATA_ETHER_TYPE (0x8CE4)
 #define MLX5E_METADATA_ETHER_LEN 8
 
-#define MLX5E_ETH_HARD_MTU (ETH_HLEN + PSP_ENCAP_HLEN + PSP_TRL_SIZE + VLAN_HLEN + ETH_FCS_LEN)
+#define MLX5E_ETH_HARD_MTU (ETH_HLEN + VLAN_HLEN + ETH_FCS_LEN)
 
 #define MLX5E_HW2SW_MTU(params, hwmtu) ((hwmtu) - ((params)->hard_mtu))
 #define MLX5E_SW2HW_MTU(params, swmtu) ((swmtu) + ((params)->hard_mtu))

drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c

@@ -342,9 +342,8 @@ static void mlx5e_ipsec_init_macs(struct mlx5e_ipsec_sa_entry *sa_entry,
 		rt_dst_entry = &rt->dst;
 		break;
 	case AF_INET6:
-		rt_dst_entry = ipv6_stub->ipv6_dst_lookup_flow(
-			dev_net(netdev), NULL, &fl6, NULL);
-		if (IS_ERR(rt_dst_entry))
+		if (!IS_ENABLED(CONFIG_IPV6) ||
+		    ip6_dst_lookup(dev_net(netdev), NULL, &rt_dst_entry, &fl6))
 			goto neigh;
 		break;
 	default:
@@ -359,6 +358,9 @@ static void mlx5e_ipsec_init_macs(struct mlx5e_ipsec_sa_entry *sa_entry,
 
 	neigh_ha_snapshot(addr, n, netdev);
 	ether_addr_copy(dst, addr);
+	if (attrs->dir == XFRM_DEV_OFFLOAD_OUT &&
+	    is_zero_ether_addr(addr))
+		neigh_event_send(n, NULL);
 	dst_release(rt_dst_entry);
 	neigh_release(n);
 	return;

drivers/net/ethernet/mellanox/mlx5/core/en_tx.c

@@ -939,7 +939,11 @@ void mlx5e_free_txqsq_descs(struct mlx5e_txqsq *sq)
 	sq->dma_fifo_cc = dma_fifo_cc;
 	sq->cc = sqcc;
 
-	netdev_tx_completed_queue(sq->txq, npkts, nbytes);
+	/* Do not update BQL for TXQs that got replaced by new active ones, as
+	 * netdev_tx_reset_queue() is called for them in mlx5e_activate_txqsq().
+	 */
+	if (sq == sq->priv->txq2sq[sq->txq_ix])
+		netdev_tx_completed_queue(sq->txq, npkts, nbytes);
 }
 
 #ifdef CONFIG_MLX5_CORE_IPOIB

drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c

@@ -52,6 +52,7 @@
 #include "devlink.h"
 #include "lag/lag.h"
 #include "en/tc/post_meter.h"
+#include "fw_reset.h"
 
 /* There are two match-all miss flows, one for unicast dst mac and
  * one for multicast.
@@ -3991,6 +3992,11 @@ int mlx5_devlink_eswitch_mode_set(struct devlink *devlink, u16 mode,
 	if (IS_ERR(esw))
 		return PTR_ERR(esw);
 
+	if (mlx5_fw_reset_in_progress(esw->dev)) {
+		NL_SET_ERR_MSG_MOD(extack, "Can't change eswitch mode during firmware reset");
+		return -EBUSY;
+	}
+
 	if (esw_mode_from_devlink(mode, &mlx5_mode))
 		return -EINVAL;
 

drivers/net/ethernet/mellanox/mlx5/core/fw_reset.c

@@ -15,6 +15,7 @@ enum {
 	MLX5_FW_RESET_FLAGS_DROP_NEW_REQUESTS,
 	MLX5_FW_RESET_FLAGS_RELOAD_REQUIRED,
 	MLX5_FW_RESET_FLAGS_UNLOAD_EVENT,
+	MLX5_FW_RESET_FLAGS_RESET_IN_PROGRESS,
 };
 
 struct mlx5_fw_reset {
@@ -128,6 +129,16 @@ int mlx5_fw_reset_query(struct mlx5_core_dev *dev, u8 *reset_level, u8 *reset_ty
 	return mlx5_reg_mfrl_query(dev, reset_level, reset_type, NULL, NULL);
 }
 
+bool mlx5_fw_reset_in_progress(struct mlx5_core_dev *dev)
+{
+	struct mlx5_fw_reset *fw_reset = dev->priv.fw_reset;
+
+	if (!fw_reset)
+		return false;
+
+	return test_bit(MLX5_FW_RESET_FLAGS_RESET_IN_PROGRESS, &fw_reset->reset_flags);
+}
+
 static int mlx5_fw_reset_get_reset_method(struct mlx5_core_dev *dev,
 					  u8 *reset_method)
 {
@@ -243,6 +254,8 @@ static void mlx5_fw_reset_complete_reload(struct mlx5_core_dev *dev)
 							BIT(DEVLINK_RELOAD_ACTION_FW_ACTIVATE));
 		devl_unlock(devlink);
 	}
+
+	clear_bit(MLX5_FW_RESET_FLAGS_RESET_IN_PROGRESS, &fw_reset->reset_flags);
 }
 
 static void mlx5_stop_sync_reset_poll(struct mlx5_core_dev *dev)
@@ -462,27 +475,48 @@ static void mlx5_sync_reset_request_event(struct work_struct *work)
 	struct mlx5_fw_reset *fw_reset = container_of(work, struct mlx5_fw_reset,
 						      reset_request_work);
 	struct mlx5_core_dev *dev = fw_reset->dev;
+	bool nack_request = false;
+	struct devlink *devlink;
 	int err;
 
 	err = mlx5_fw_reset_get_reset_method(dev, &fw_reset->reset_method);
-	if (err)
+	if (err) {
+		nack_request = true;
 		mlx5_core_warn(dev, "Failed reading MFRL, err %d\n", err);
+	} else if (!mlx5_is_reset_now_capable(dev, fw_reset->reset_method) ||
+		   test_bit(MLX5_FW_RESET_FLAGS_NACK_RESET_REQUEST,
+			    &fw_reset->reset_flags)) {
+		nack_request = true;
+	}
 
-	if (err || test_bit(MLX5_FW_RESET_FLAGS_NACK_RESET_REQUEST, &fw_reset->reset_flags) ||
-	    !mlx5_is_reset_now_capable(dev, fw_reset->reset_method)) {
+	devlink = priv_to_devlink(dev);
+	/* For external resets, try to acquire devl_lock. Skip if devlink reset is
+	 * pending (lock already held)
+	 */
+	if (nack_request ||
+	    (!test_bit(MLX5_FW_RESET_FLAGS_PENDING_COMP,
+		       &fw_reset->reset_flags) &&
+	     !devl_trylock(devlink))) {
 		err = mlx5_fw_reset_set_reset_sync_nack(dev);
 		mlx5_core_warn(dev, "PCI Sync FW Update Reset Nack %s",
 			       err ? "Failed" : "Sent");
 		return;
 	}
+
 	if (mlx5_sync_reset_set_reset_requested(dev))
-		return;
+		goto unlock;
+
+	set_bit(MLX5_FW_RESET_FLAGS_RESET_IN_PROGRESS, &fw_reset->reset_flags);
 
 	err = mlx5_fw_reset_set_reset_sync_ack(dev);
 	if (err)
 		mlx5_core_warn(dev, "PCI Sync FW Update Reset Ack Failed. Error code: %d\n", err);
 	else
 		mlx5_core_warn(dev, "PCI Sync FW Update Reset Ack. Device reset is expected.\n");
+
+unlock:
+	if (!test_bit(MLX5_FW_RESET_FLAGS_PENDING_COMP, &fw_reset->reset_flags))
+		devl_unlock(devlink);
 }
 
 static int mlx5_pci_link_toggle(struct mlx5_core_dev *dev, u16 dev_id)
@@ -722,6 +756,8 @@ static void mlx5_sync_reset_abort_event(struct work_struct *work)
 
 	if (mlx5_sync_reset_clear_reset_requested(dev, true))
 		return;
+
+	clear_bit(MLX5_FW_RESET_FLAGS_RESET_IN_PROGRESS, &fw_reset->reset_flags);
 	mlx5_core_warn(dev, "PCI Sync FW Update Reset Aborted.\n");
 }
 
@@ -758,6 +794,7 @@ static void mlx5_sync_reset_timeout_work(struct work_struct *work)
 
 	if (mlx5_sync_reset_clear_reset_requested(dev, true))
 		return;
+	clear_bit(MLX5_FW_RESET_FLAGS_RESET_IN_PROGRESS, &fw_reset->reset_flags);
 	mlx5_core_warn(dev, "PCI Sync FW Update Reset Timeout.\n");
 }
 
@@ -844,7 +881,8 @@ void mlx5_drain_fw_reset(struct mlx5_core_dev *dev)
 	cancel_work_sync(&fw_reset->reset_reload_work);
 	cancel_work_sync(&fw_reset->reset_now_work);
 	cancel_work_sync(&fw_reset->reset_abort_work);
-	cancel_delayed_work(&fw_reset->reset_timeout_work);
+	if (test_bit(MLX5_FW_RESET_FLAGS_RESET_REQUESTED, &fw_reset->reset_flags))
+		mlx5_sync_reset_clear_reset_requested(dev, true);
 }
 
 static const struct devlink_param mlx5_fw_reset_devlink_params[] = {

drivers/net/ethernet/mellanox/mlx5/core/fw_reset.h

@@ -10,6 +10,7 @@ int mlx5_fw_reset_query(struct mlx5_core_dev *dev, u8 *reset_level, u8 *reset_ty
 int mlx5_fw_reset_set_reset_sync(struct mlx5_core_dev *dev, u8 reset_type_sel,
 				 struct netlink_ext_ack *extack);
 int mlx5_fw_reset_set_live_patch(struct mlx5_core_dev *dev);
+bool mlx5_fw_reset_in_progress(struct mlx5_core_dev *dev);
 
 int mlx5_fw_reset_wait_reset_done(struct mlx5_core_dev *dev);
 void mlx5_sync_reset_unload_flow(struct mlx5_core_dev *dev, bool locked);

drivers/net/ethernet/mellanox/mlx5/core/main.c

@@ -2232,6 +2232,7 @@ static void shutdown(struct pci_dev *pdev)
 
 	mlx5_core_info(dev, "Shutdown was called\n");
 	set_bit(MLX5_BREAK_FW_WAIT, &dev->intf_state);
+	mlx5_drain_fw_reset(dev);
 	mlx5_drain_health_wq(dev);
 	err = mlx5_try_fast_unload(dev);
 	if (err)