RDMA/mlx5: reduce stack usage in mlx5_ib_ufile_hw_cleanup

This function has an array of eight mlx5_async_cmd structures, which
often fits on the stack, but depending on the configuration can
end up blowing the stack frame warning limit:

drivers/infiniband/hw/mlx5/devx.c:2670:6: error: stack frame size (1392) exceeds limit (1280) in 'mlx5_ib_ufile_hw_cleanup' [-Werror,-Wframe-larger-than]

Change this to a dynamic allocation instead. While a kmalloc()
can theoretically fail, a GFP_KERNEL allocation under a page will
block until memory has been freed up, so in the worst case, this
only adds extra time in an already constrained environment.

Fixes: 7c891a4 ("RDMA/mlx5: Add implementation for ufile_hw_cleanup device operation")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Link: https://patch.msgid.link/20250610092846.2642535-1-arnd@kernel.org
Signed-off-by: Leon Romanovsky <leon@kernel.org>

drivers/infiniband/hw/mlx5/devx.c

@@ -2669,7 +2669,7 @@ static void devx_wait_async_destroy(struct mlx5_async_cmd *cmd)
 
 void mlx5_ib_ufile_hw_cleanup(struct ib_uverbs_file *ufile)
 {
-	struct mlx5_async_cmd async_cmd[MAX_ASYNC_CMDS];
+	struct mlx5_async_cmd *async_cmd;
 	struct ib_ucontext *ucontext = ufile->ucontext;
 	struct ib_device *device = ucontext->device;
 	struct mlx5_ib_dev *dev = to_mdev(device);
@@ -2678,6 +2678,10 @@ void mlx5_ib_ufile_hw_cleanup(struct ib_uverbs_file *ufile)
 	int head = 0;
 	int tail = 0;
 
+	async_cmd = kcalloc(MAX_ASYNC_CMDS, sizeof(*async_cmd), GFP_KERNEL);
+	if (!async_cmd)
+		return;
+
 	list_for_each_entry(uobject, &ufile->uobjects, list) {
 		WARN_ON(uverbs_try_lock_object(uobject, UVERBS_LOOKUP_WRITE));
 
@@ -2713,6 +2717,8 @@ void mlx5_ib_ufile_hw_cleanup(struct ib_uverbs_file *ufile)
 		devx_wait_async_destroy(&async_cmd[head % MAX_ASYNC_CMDS]);
 		head++;
 	}
+
+	kfree(async_cmd);
 }
 
 static ssize_t devx_async_cmd_event_read(struct file *filp, char __user *buf,