Re: [PATCH 1/2] get the interface name only when needed
On Tue, Apr 10, 2012 at 12:37:58PM +0200, Nikola Pajkovsky wrote:
Vitezslav Samel <vitezslav(a)samel.cz> writes:
Horrible, horrible c89 is! Please use only c99/gnu99 standard.
I will subordinate to this ;-)))
> +int iface_get_ifname(int ifindex, char *ifname)
> +{
> + int fd;
> + int ir;
> + struct ifreq ifr = {
> + .ifr_ifindex = ifindex
> + };
> + int saved_errno;
> + int cl;
> +
> + fd = socket(PF_INET, SOCK_DGRAM, IPPROTO_UDP);
> + if (fd == -1)
> + return fd;
> +
> + ir = ioctl(fd, SIOCGIFNAME, &ifr);
> +
> + /* need to preserve errno across call to close() */
> + saved_errno = errno;
> +
> + cl = close(fd);
> +
> + /* bug out if ioctl() failed */
> + if (ir != 0) {
> + errno = saved_errno;
> + return ir;
> + }
> +
> + /* bug out if close() failed */
> + if (cl != 0)
> + return cl;
> +
> + strncpy(ifname, ifr.ifr_name, IFNAMSIZ);
> + return ir;
> +}
> +
We dont have to check close() here. According man:
"Not checking the return value of *close()* is a common but nevertheless
serious programming error. It is quite possible that errors on a
previous *write(2)* operation are first reported at the final *close()*.
Not checking the return value when closing the file may lead to silent
loss of data. This can especially be observed with NFS and with disk
quota."
Yes, I read it.
Thus, when you write, check *close()*, when you read, you don't
have
to. Nevertheless we are reading, so don't checking.
OK, won't check.
If we won't check errno and print some common message, than we
don't
have to do the machinery about saving and restoring errno.
No, I will leave this my way (with errno saving) to prevent funny
surprises when we'll learn how to print error string.
Vita