Re: [PATCH rdma-next 1/1] RDMA/core: Fix WARNING in gid_table_release_one

[Jason Gunthorpe <jgg@ziepe.ca>]

On Mon, Nov 03, 2025 at 06:08:45PM -0800, Zhu Yanjun wrote:
> @@ -800,13 +800,24 @@ static void release_gid_table(struct ib_device *device,
>  		return;
>  
>  	for (i = 0; i < table->sz; i++) {
> +		int cnt = 200;
> +
>  		if (is_gid_entry_free(table->data_vec[i]))
>  			continue;
>  
> -		WARN_ONCE(true,
> -			  "GID entry ref leak for dev %s index %d ref=%u\n",
> +		WARN_ONCE(table->data_vec[i]->state != GID_TABLE_ENTRY_PENDING_DEL,
> +			  "GID entry ref leak for dev %s index %d ref=%u, state: %d\n",
>  			  dev_name(&device->dev), i,
> -			  kref_read(&table->data_vec[i]->kref));
> +			  kref_read(&table->data_vec[i]->kref), table->data_vec[i]->state);
> +
> +		while ((kref_read(&table->data_vec[i]->kref) > 0) && (cnt > 0)) {
> +			cnt--;
> +			msleep(10);
> +		}

Definately don't want to see this looping.

If it is waiting for the work queue then maybe this should flush the
work queue.

Something like this?

--- a/drivers/infiniband/core/cache.c
+++ b/drivers/infiniband/core/cache.c
@@ -799,7 +799,19 @@ static void release_gid_table(struct ib_device *device,
        if (!table)
                return;
 
+       mutex_lock(&table->lock);
        for (i = 0; i < table->sz; i++) {
+               if (is_gid_entry_free(table->data_vec[i]))
+                       continue;
+
+               /*
+                * The entry may be sitting in the WQ waiting for
+                * free_gid_work(), flush it to try to clean it.
+                */
+               mutex_unlock(&table->lock);
+               flush_workqueue(ib_wq);
+               mutex_lock(&table->lock);
+
                if (is_gid_entry_free(table->data_vec[i]))
                        continue;
 
@@ -808,6 +820,7 @@ static void release_gid_table(struct ib_device *device,
                          dev_name(&device->dev), i,
                          kref_read(&table->data_vec[i]->kref));
        }
+       mutex_unlock(&table->lock);
 
        mutex_destroy(&table->lock);
        kfree(table->data_vec);