Mailing List Archive

> 0 44_ident.0.8.16.patch
>
> Wups. Can someone improve this?
>
> compiler warning:
>
> gcc -c -O2 -DXBITHACK -m486 http_main.c
> http_main.c: In function `child_main':
> http_main.c:786: warning: passing arg 2 of `getsockname' from incompatible pointer type
>
> Note: from 'man getsockname' on FreeBSD we get:
>
> SYNOPSIS
> int
> getsockname(int s, struct sockaddr *name, int *namelen)
>
> I mean, it seems to work (had to install inetd to test it even),
> but I think we just got lucky.
>
> The warning is harmless, but this patch (to be applied after 44_ident)
> should fix it.

Oh, the line number is bogus, sorry, sorry. I had god knows how many
other patches in line at the time. The point was that the call to
getsockname() had the wrong type defined for the *name parameter; this
weird type arising from the #if defined(NEXT)... lines. But like you
say it seems to work all the same.

I can't test this new fix right now, I'll ask Paul R to lend a hand
this afternoon.

> David.
>
> *** http_main.c.orig2 Mon Nov 6 18:20:18 1995
> --- http_main.c Wed Nov 22 11:15:13 1995
> ***************
> *** 718,729 ****
> void child_main(int child_num_arg)
> {
> int clen;
> - #if defined(NEXT) || defined(LINUX) || defined(SOLARIS2)
> struct sockaddr sa_server;
> struct sockaddr sa_client;
> - #else
> - struct sockaddr_in sa_server,sa_client;
> - #endif
>
> #ifdef ULTRIX_BRAIN_DEATH
> extern char *rfc931();
> --- 718,725 ----
> ***************
> *** 776,782 ****
>
> accept_mutex_on(); /* Lock around "accept", if necessary */
>
> ! while ((csd=accept(sd,(struct sockaddr *)&sa_client,&clen)) == -1)
> if (errno != EINTR)
> log_error("socket error: accept failed", server_conf);
>
> --- 772,778 ----
>
> accept_mutex_on(); /* Lock around "accept", if necessary */
>
> ! while ((csd=accept(sd, &sa_client, &clen)) == -1)
> if (errno != EINTR)
> log_error("socket error: accept failed", server_conf);
>
>