Re: [PATCH] io_uring/net: punt IORING_OP_BIND async if it needs file create

[Jens Axboe <axboe@kernel.dk>]

On 5/15/26 11:00 AM, Gabriel Krisman Bertazi wrote:
>> diff --git a/io_uring/net.c b/io_uring/net.c
>> index 30cd22c0b934..1f5fd8c37a5a 100644
>> --- a/io_uring/net.c
>> +++ b/io_uring/net.c
>> @@ -4,6 +4,7 @@
>>  #include <linux/file.h>
>>  #include <linux/slab.h>
>>  #include <linux/net.h>
>> +#include <linux/un.h>
>>  #include <linux/compat.h>
>>  #include <net/compat.h>
>>  #include <linux/io_uring.h>
>> @@ -1799,11 +1800,29 @@ int io_connect(struct io_kiocb *req, unsigned int issue_flags)
>>  	return IOU_COMPLETE;
>>  }
>>  
>> +/*
>> + * Check if bind request would potentiall end up with filename_create(),
> 
> potentially

thanks, fixed.

>> + * which in turn end up in mnt_want_write() which will grab the fs
>> + * percpu start write sem. This can trigger a lockdep warning.
>> + */
>> +static int io_bind_file_create(const struct io_async_msghdr *io, int addr_len)
>> +{
>> +	const struct sockaddr_un *sun;
>> +
>> +	if (io->addr.ss_family != AF_UNIX)
>> +		return 0;
>> +	if (addr_len <= offsetof(struct sockaddr_un, sun_path))
>> +		return 0;
>> +	sun = (const struct sockaddr_un *) &io->addr;
>> +	return sun->sun_path[0] != '\0';
>> +}
>> +
>>  int io_bind_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
>>  {
>>  	struct io_bind *bind = io_kiocb_to_cmd(req, struct io_bind);
>>  	struct sockaddr __user *uaddr;
>>  	struct io_async_msghdr *io;
>> +	int ret;
>>  
>>  	if (sqe->len || sqe->buf_index || sqe->rw_flags || sqe->splice_fd_in)
>>  		return -EINVAL;
>> @@ -1814,7 +1833,12 @@ int io_bind_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
>>  	io = io_msg_alloc_async(req);
>>  	if (unlikely(!io))
>>  		return -ENOMEM;
>> -	return move_addr_to_kernel(uaddr, bind->addr_len, &io->addr);
>> +	ret = move_addr_to_kernel(uaddr, bind->addr_len, &io->addr);
>> +	if (unlikely(ret))
>> +		return ret;
>> +	if (io_bind_file_create(io, bind->addr_len))
>> +		req->flags |= REQ_F_FORCE_ASYNC;
>> +	return 0;
>>  }
> 
> Patch looks ok. I will add a TODO to myself to look into the bind call
> getting a flag and returning -EWOULDBLOCK, so we can eventually drop the
> workaround.

That would indeed be better, but probably better darn far down, and it'd
need refactoring to pass down 'flags' or similar as msghdr isn't a thing
there... So not sure it's going to be super viable.

> Reviewed-by: Gabriel Krisman Bertazi <krisman@suse.de>

Thanks, added!

-- 
Jens Axboe