Re: [Qemu-devel] [PATCH v2] Support for UDP unicast network backend

[Benjamin]

On 11/28/11 20:39, Stefan Hajnoczi wrote:
> On Fri, Nov 25, 2011 at 12:49 PM, Benjamin<address@hidden>  wrote:
> > +    fd = qemu_socket(PF_INET, SOCK_DGRAM, 0);
> > +    if (fd<  0) {
> > +        perror("socket(PF_INET, SOCK_DGRAM)");
> > +        return -1;
> > +    }
> > +    val = 1;
> > +    ret = setsockopt(fd, SOL_SOCKET, SO_REUSEADDR,
> > +                   (const char *)&val, sizeof(val));
> > +    if (ret<  0) {
> > +        perror("setsockopt(SOL_SOCKET, SO_REUSEADDR)");
>
> Please avoid leaking the file descriptor on error:
> closesocket(fd);
>
> Since existing code also does this it may be more appropriate to send
> a follow-up patch that cleans up all of net/socket.c.
>
> Reviewed-by: Stefan Hajnoczi<address@hidden>
>
> Stefan

I can do that. However is it really a leak considering the fact that
the program will call exit just after?
If it's a matter of consistency and coding style I would understand
though.

One more thing, git-format-patch added a "From" field to the header and
caused this glitch in the mail. I thought git-send-mail or the mail
server would handle it well but apparently not:

From 2f5b85fcadcfee3b75a6a21dc86d10b945c99f0f Mon Sep 17 00:00:00 2001
From: Benjamin MARSILI <address@hidden>

git-am didn't complain with the patch that I sent but it may break after
gmail relayed it 
(http://lists.gnu.org/archive/html/qemu-devel/2011-11/msg03152.html).
The second from header is interpreted as text... Should I remove the
first "From" field before sending the patch?

Sorry for the noise on the ML and thanks to all those who helped me get
involved.

Benjamin