[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200527191335.mt5ywrgvpppiqj7v@wittgenstein>
Date: Wed, 27 May 2020 21:13:35 +0200
From: Christian Brauner <christian.brauner@...ntu.com>
To: Sargun Dhillon <sargun@...gun.me>
Cc: Kees Cook <keescook@...omium.org>,
Chris Palmer <palmer@...gle.com>, Jann Horn <jannh@...gle.com>,
Robert Sesek <rsesek@...gle.com>,
Jeffrey Vander Stoep <jeffv@...gle.com>,
Linux Containers <containers@...ts.linux-foundation.org>,
linux-kernel@...r.kernel.org, Matt Denton <mpdenton@...gle.com>,
Andy Lutomirski <luto@...nel.org>
Subject: Re: [PATCH 1/2] seccomp: notify user trap about unused filter
On Wed, May 27, 2020 at 05:37:07PM +0000, Sargun Dhillon wrote:
> On Wed, May 27, 2020 at 01:19:01PM +0200, Christian Brauner wrote:
> > +void seccomp_filter_notify(const struct task_struct *tsk)
> > +{
> > + struct seccomp_filter *orig = tsk->seccomp.filter;
> > +
> > + while (orig && refcount_dec_and_test(&orig->live)) {
> > + if (waitqueue_active(&orig->wqh))
> > + wake_up_poll(&orig->wqh, EPOLLHUP);
> > + orig = orig->prev;
> > + }
> > +}
> > +
> Any reason not to write this as:
> for (orig = tsk->seccomp.filter; refcount_dec_and_test(&orig->live); orig = orig->prev)?
Mainly to follow coding style if you look at:
static void __put_seccomp_filter(struct seccomp_filter *orig)
{
/* Clean up single-reference branches iteratively. */
while (orig && refcount_dec_and_test(&orig->usage)) {
struct seccomp_filter *freeme = orig;
orig = orig->prev;
seccomp_filter_free(freeme);
}
}
seemed easier to just have a visual correspondence between those two
codepaths.
>
> Also, for those of us who are plumbing in the likes of Go code into the
> listener, where we don't have direct access to the epoll interface (at
> least not out of the box), what do you think about exposing this on the RECV
I think requiring users to import
golang.org/x/sys/unix
is reasonable. You'll need to special case this to linux builds anyway
even if you have a client or some such that build on on-unixes. And even
if you don't want to import there's always the possibility to use cgo. :)
> ioctl? Or, do you think we should lump that into the "v2" receive API?
I'm confused how you want to plumb this into the ioctl. That seems
unpleasant and against usual poll/wait semantics. I'm now also wondering
how you're using this whole interface without poll. The idea is to wait
until you're notified you can receive.
>
> Either way, this seems useful, as right now, we're intertwining process
> tree lifetime with manager lifetime. This seems cleaner.
Cool.
Christian
Powered by blists - more mailing lists