Le 29/12/2012 02:46, Philip Guenther a écrit : > On Thu, Dec 27, 2012 at 5:04 AM, Maxime Villard <rusty...@gmx.fr> wrote: >> Well, as no one seems to give a fuck on tech@, I put a more >> glamourous title here. > > The fd/FILE part of your diff changes the behavior of pfctl to be > incorrect when there are no states. > > > Philip Guenther > >
Hum, Before: - we open the file, go in the loop to do ioctl stuff, and if it fails BAM we lose the fd when returning at l.1893 or l.1889. If it worked, we write data to the file and close it. Now: - we go in the loop, we do ioctl stuff, and then if nothing failed we open the file and write data to it, and close it. If something failed in the loop, we return without leaking f. The fd is not used in the loop, I just moved it down. So I don't see what behaviour it changes.
