netdev - Re: [RFC][PATCH] Remove refcounting from unix domain sockets

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CALCETrUsEpaLjRzgnwQkz_cXM9gjnvuBz5mzrpk=WLR8VNrTmw@mail.gmail.com>
Date:	Sat, 16 May 2015 16:34:30 -0700
From:	Andy Lutomirski <luto@...capital.net>
To:	"Eric W. Biederman" <ebiederm@...ssion.com>
Cc:	Network Development <netdev@...r.kernel.org>,
	Eric Dumazet <eric.dumazet@...il.com>
Subject: Re: [RFC][PATCH] Remove refcounting from unix domain sockets

On Sat, May 16, 2015 at 9:48 AM, Eric W. Biederman
<ebiederm@...ssion.com> wrote:
>
> While thinking about things it occurred to me that by fixing what is
> arguably in a bug in the unix domain socket implementation I can
> remove the performance penality for credential passing.
>
> That bug is the fact that the cmsg returned on a unix domain socket
> is different depending on who calls recvmsg.  Passing file descriptor
> should not changes how it operates (oops).

I'm having some trouble following what's going on with icred and
ocred.  That being said, I think that you're changing the semantics of
socketpair() followed by unshare(), and I wouldn't be that surprised
if something relies on the current behavior.

>
> There is a second bug this also fixes, pids are now translated as they
> traverse netlink sockets.

Can you describe the bug more clearly?

--Andy
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html