SO_DEBUG in proxy_ftp.c

[Marc Slemko <ma...@znep.com>]

In proxy_ftp.c:

    if (setsockopt(dsock, SOL_SOCKET, SO_DEBUG, (const char *)&one,
      sizeof (int)) == -1)
    {
        proxy_log_uerror("setsockopt", NULL,
            "proxy: error setting PASV debug option", r->server);
        pclosef(pool, dsock);
        pclosef(pool, sock);
        return SERVER_ERROR;
    }

Why is SO_DEBUG being set?  PR 166 details someone on a platform that
it apparently does not work on so the proxy is failing.

Re: SO_DEBUG in proxy_ftp.c

[Chuck Murcko <ch...@topsail.org>]

Marc Slemko wrote:
> 
> In proxy_ftp.c:
> 
>     if (setsockopt(dsock, SOL_SOCKET, SO_DEBUG, (const char *)&one,
>       sizeof (int)) == -1)
>     {
>         proxy_log_uerror("setsockopt", NULL,
>             "proxy: error setting PASV debug option", r->server);
>         pclosef(pool, dsock);
>         pclosef(pool, sock);
>         return SERVER_ERROR;
>     }
> 
> Why is SO_DEBUG being set?  PR 166 details someone on a platform that
> it apparently does not work on so the proxy is failing.

Interesting. I don't have this in my proxy bug database here. I assume
this was only reported to the GNATS setup on hyperreal, and not in
email? That would explain why I missed it. Time to start checking the
GNATS deal too, I guess.

Anyway, it was in for me to get additional data via trpt(8) while
debugging. It's a cut-and-paste error, and shouldn't be aborting, just
noting an info log message and moving on. I'll put it on the queue of
proxy stuff (currently about 4-5 items). I used to report this stuff
around beta release time, but no one seemed too interested, so I just
work on it and put in the patches when they're fixed.

I'm curious to see which platform this is failing on.
-- 
chuck
Chuck Murcko
The Topsail Group, West Chester PA USA
chuck@topsail.org