Re: [PATCH] tty: add TIOCVHANGUP: time for revoke() in f_ops ?

[Lennart Poettering <mzxreary@0pointer.de>]

On Wed, 23.02.11 00:09, Alan Cox (alan@lxorguk.ukuu.org.uk) wrote:

> It's basically 3 things
> - Lennarts bits for vhangup on an fd

Uh? Me? I didn't write this patch.

(Though I do like to see patch merged and I would use it, and I have
trouble following your logic.)

vhangup() is different from revoke(). vhangup() does weird SIGHUP
handling and stuff, which I think goes way beyond what revoke() would
eventually do. And that different behaviour becomes visible in various
smaller places. e.g. vhangup() results in POLLHUP on the fd, although I
assume that revoke() would more likely result in POLLERR. And there's
more... Let's not pretend this is really the same thing, because it
isn't.

> Its not a quick patch - that's why its not happened yet, vhangup(fd)
> quickfix Lennart style is unfortunately a useless bodge job which like
> most bodge jobs is simply going to spring leaks and need fixing again.

Thanks. If you are trying to insult me, doesn't really work, because I
didn't do this "bodge job". I'll take it as a compliment though that
you say there's a "Lennart style".

Lennart, style icon

-- 
Lennart Poettering - Red Hat, Inc.