Re: [2.6.21 patch] net/sunrpc/svcsock.c: fix a check

[Andrew Morton <akpm@linux-foundation.org>]

On Mon, 19 Mar 2007 10:33:42 +0100 Adrian Bunk <bunk@stusta.de> wrote:

> The return value of kernel_recvmsg() should be assigned to "err", not 
> compared with the random value of a never initialized "err"
> (and the "< 0" check wrongly always returned false since == comparisons 
> never have a result < 0).
> 
> Spotted by the Coverity checker.
> 
> Signed-off-by: Adrian Bunk <bunk@stusta.de>
> 
> ---
> --- linux-2.6.21-rc3-mm2/net/sunrpc/svcsock.c.old	2007-03-19 09:44:40.000000000 +0100
> +++ linux-2.6.21-rc3-mm2/net/sunrpc/svcsock.c	2007-03-19 09:45:18.000000000 +0100
> @@ -779,8 +779,8 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
>  	}
>  
>  	clear_bit(SK_DATA, &svsk->sk_flags);
> -	while ((err == kernel_recvmsg(svsk->sk_sock, &msg, NULL,
> -				      0, 0, MSG_PEEK | MSG_DONTWAIT)) < 0 ||
> +	while ((err = kernel_recvmsg(svsk->sk_sock, &msg, NULL,
> +				     0, 0, MSG_PEEK | MSG_DONTWAIT)) < 0 ||
>  	       (skb = skb_recv_datagram(svsk->sk_sk, 0, 1, &err)) == NULL) {
>  		if (err == -EAGAIN) {
>  			svc_sock_received(svsk);

Cute.  The compiler must have decided to apply the "(a==b) can never be
less than zero" optimisation before performing uninitialised variable
analysis.

Neil, this one needs runtime testing before we can apply it to 2.6.21, I
think.