netdev - all this business about ->sk

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-next>] [day] [month] [year] [list]

Date:	Tue, 17 Jun 2008 02:54:46 -0700 (PDT)
From:	David Miller <davem@...emloft.net>
To:	netdev@...r.kernel.org
Subject: all this business about ->sk_sleep

So you're probably wondering why I tidied up all of those code paths
setting sk_sleep inappropriately or not using sock_graft().

If you grep the net-next-2.6 tree now there are only 6 assignments to
sk->sk_sleep:

include/net/sock.h:	sk->sk_sleep  = NULL;
include/net/sock.h:	sk->sk_sleep = &parent->wait;
net/core/sock.c:		newsk->sk_sleep	 = NULL;
net/core/sock.c:		sk->sk_sleep	=	&sock->wait;
net/core/sock.c:		sk->sk_sleep	=	NULL;
net/unix/af_unix.c:	newsk->sk_sleep		= &newu->peer_wait;

The only thing that uses something other than &sk->sk_socket->wait
for sk->sk_sleep is AF_UNIX.  So that's a useful discovery and
we may be able to get rid of sk->sk_sleep somehow.  I'll try to
launch a full investiation into how and why AF_UNIX uses a special
wait queue.

But the original impetus for this consolidation is that I might
move struct socket freeing over to RCU in order to facilitate the
removal of sk->sk_callback_lock.  And if so it's tidier if there
are only a handful of rcu_assign_pointer() calls to add.
--
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