linux-kernel - Re: [PATCH net] rds: Fix endian annotations across various assignments

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <ac09739a-eec5-4025-989e-a6a202fcfd63@lunn.ch>
Date: Sun, 10 Aug 2025 19:32:03 +0200
From: Andrew Lunn <andrew@...n.ch>
To: Ujwal Kundur <ujwal.kundur@...il.com>
Cc: allison.henderson@...cle.com, davem@...emloft.net, edumazet@...gle.com,
	kuba@...nel.org, pabeni@...hat.com, horms@...nel.org,
	netdev@...r.kernel.org, linux-rdma@...r.kernel.org,
	rds-devel@....oracle.com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH net] rds: Fix endian annotations across various
 assignments

On Sun, Aug 10, 2025 at 10:41:55PM +0530, Ujwal Kundur wrote:
> Sparse reports the following warnings:
> 
> net/rds/af_rds.c:245:22: warning: invalid assignment: |=
> net/rds/af_rds.c:245:22: left side has type restricted __poll_t
> net/rds/af_rds.c:245:22: right side has type int
> 
> __poll_t is typedef'ed to __bitwise while POLLERR is defined as 0x0008,
> force conversion.

Please could you split this up, one patch per type of problem.

> diff --git a/net/rds/af_rds.c b/net/rds/af_rds.c
> index 086a13170e09..9cd5905d916a 100644
> --- a/net/rds/af_rds.c
> +++ b/net/rds/af_rds.c
> @@ -242,7 +242,7 @@ static __poll_t rds_poll(struct file *file, struct socket *sock,
>  	if (rs->rs_snd_bytes < rds_sk_sndbuf(rs))
>  		mask |= (EPOLLOUT | EPOLLWRNORM);
>  	if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
> -		mask |= POLLERR;
> +		mask |= (__force __poll_t)POLLERR;

I don't like __force, it suggests something is wrong with the
design. If it is needed, it should be hidden away.

However:

~/linux/net$ grep -r POLLERR
caif/caif_socket.c:	wake_up_interruptible_poll(sk_sleep(sk), EPOLLERR|EPOLLHUP);
caif/caif_socket.c:		mask |= EPOLLERR;
rds/af_rds.c:		mask |= POLLERR;
bluetooth/af_bluetooth.c:		mask |= EPOLLERR |
sctp/socket.c:		mask |= EPOLLERR |
vmw_vsock/af_vsock.c:		mask |= EPOLLERR;
vmw_vsock/af_vsock.c:				mask |= EPOLLERR;
vmw_vsock/af_vsock.c:					mask |= EPOLLERR;
9p/trans_fd.c:		return EPOLLERR;
9p/trans_fd.c:	if (n & (EPOLLERR | EPOLLHUP | EPOLLNVAL)) {
mptcp/protocol.c:		mask |= EPOLLERR;
core/datagram.c:	if (key && !(key_to_poll(key) & (EPOLLIN | EPOLLERR)))
core/datagram.c:		mask |= EPOLLERR |
core/sock.c:		wake_up_interruptible_poll(&wq->wait, EPOLLERR);
nfc/llcp_sock.c:		mask |= EPOLLERR |
smc/af_smc.c:		else if (flags & EPOLLERR)
smc/af_smc.c:			mask |= EPOLLERR;
phonet/socket.c:		return EPOLLERR;
iucv/af_iucv.c:		mask |= EPOLLERR |
unix/af_unix.c:		mask |= EPOLLERR;
unix/af_unix.c:		mask |= EPOLLERR |
ipv4/tcp.c:		mask |= EPOLLERR;
sunrpc/rpc_pipe.c:		mask |= EPOLLERR | EPOLLHUP;
atm/common.c:		mask = EPOLLERR;

So why is af_rds.c special? Or do all these files also give the same
warning?

Also:

https://elixir.bootlin.com/linux/v6.16/source/include/uapi/linux/eventpoll.h#L34

#define EPOLLERR	(__force __poll_t)0x00000008

So your patch does nothing.

    Andrew

---
pw-bot: cr