[PATCH] Revert "net: do not allow changing SO_REUSEADDR/SO_REUSEPORT on bound sockets"

Revert the patch mentioned in the subject because it breaks at least
the Avahi mDNS daemon. That patch namely causes the Ubuntu 18.04 Avahi
daemon to fail to start:

Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: Successfully called chroot().
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: Successfully dropped remaining capabilities.
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: No service file found in /etc/avahi/services.
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: SO_REUSEADDR failed: Structure needs cleaning
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: SO_REUSEADDR failed: Structure needs cleaning
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: Failed to create server: No suitable network protocol available
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: avahi-daemon 0.7 exiting.
Jun 12 09:49:24 ubuntu-vm systemd[1]: avahi-daemon.service: Main process exited, code=exited, status=255/n/a
Jun 12 09:49:24 ubuntu-vm systemd[1]: avahi-daemon.service: Failed with result 'exit-code'.
Jun 12 09:49:24 ubuntu-vm systemd[1]: Failed to start Avahi mDNS/DNS-SD Stack.

Fixes: f396922d862a ("net: do not allow changing SO_REUSEADDR/SO_REUSEPORT on bound sockets")
Cc: Maciej Żenczykowski <redacted>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Bart Van Assche <redacted>
---
 net/core/sock.c | 15 +--------------
 1 file changed, 1 insertion(+), 14 deletions(-)

diff --git a/net/core/sock.c b/net/core/sock.c
index f333d75ef1a9..bcc41829a16d 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c

@@ -728,22 +728,9 @@ int sock_setsockopt(struct socket *sock, int level, int optname,
 			sock_valbool_flag(sk, SOCK_DBG, valbool);
 		break;
 	case SO_REUSEADDR:
-		val = (valbool ? SK_CAN_REUSE : SK_NO_REUSE);
-		if ((sk->sk_family == PF_INET || sk->sk_family == PF_INET6) &&
-		    inet_sk(sk)->inet_num &&
-		    (sk->sk_reuse != val)) {
-			ret = (sk->sk_state == TCP_ESTABLISHED) ? -EISCONN : -EUCLEAN;
-			break;
-		}
-		sk->sk_reuse = val;
+		sk->sk_reuse = (valbool ? SK_CAN_REUSE : SK_NO_REUSE);
 		break;
 	case SO_REUSEPORT:
-		if ((sk->sk_family == PF_INET || sk->sk_family == PF_INET6) &&
-		    inet_sk(sk)->inet_num &&
-		    (sk->sk_reuseport != valbool)) {
-			ret = (sk->sk_state == TCP_ESTABLISHED) ? -EISCONN : -EUCLEAN;
-			break;
-		}
 		sk->sk_reuseport = valbool;
 		break;
 	case SO_TYPE:

-- 
2.17.0

On 06/12/2018 10:05 AM, Bart Van Assche wrote:

Revert the patch mentioned in the subject because it breaks at least
the Avahi mDNS daemon. That patch namely causes the Ubuntu 18.04 Avahi
daemon to fail to start:

Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: Successfully called chroot().
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: Successfully dropped remaining capabilities.
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: No service file found in /etc/avahi/services.
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: SO_REUSEADDR failed: Structure needs cleaning
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: SO_REUSEADDR failed: Structure needs cleaning
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: Failed to create server: No suitable network protocol available
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: avahi-daemon 0.7 exiting.
Jun 12 09:49:24 ubuntu-vm systemd[1]: avahi-daemon.service: Main process exited, code=exited, status=255/n/a
Jun 12 09:49:24 ubuntu-vm systemd[1]: avahi-daemon.service: Failed with result 'exit-code'.
Jun 12 09:49:24 ubuntu-vm systemd[1]: Failed to start Avahi mDNS/DNS-SD Stack.

Fixes: f396922d862a ("net: do not allow changing SO_REUSEADDR/SO_REUSEPORT on bound sockets")
Cc: Maciej Żenczykowski <redacted>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Bart Van Assche <redacted>
---
 net/core/sock.c | 15 +--------------
 1 file changed, 1 insertion(+), 14 deletions(-)

Yes, this change probably broke a lot of applications, unfortunately.

Acked-by: Eric Dumazet <edumazet@google.com>

From: Bart Van Assche <redacted>
Date: Tue, 12 Jun 2018 10:05:55 -0700

Revert the patch mentioned in the subject because it breaks at least
the Avahi mDNS daemon. That patch namely causes the Ubuntu 18.04 Avahi
daemon to fail to start:

Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: Successfully called chroot().
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: Successfully dropped remaining capabilities.
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: No service file found in /etc/avahi/services.
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: SO_REUSEADDR failed: Structure needs cleaning
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: SO_REUSEADDR failed: Structure needs cleaning
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: Failed to create server: No suitable network protocol available
Jun 12 09:49:24 ubuntu-vm avahi-daemon[529]: avahi-daemon 0.7 exiting.
Jun 12 09:49:24 ubuntu-vm systemd[1]: avahi-daemon.service: Main process exited, code=exited, status=255/n/a
Jun 12 09:49:24 ubuntu-vm systemd[1]: avahi-daemon.service: Failed with result 'exit-code'.
Jun 12 09:49:24 ubuntu-vm systemd[1]: Failed to start Avahi mDNS/DNS-SD Stack.

Fixes: f396922d862a ("net: do not allow changing SO_REUSEADDR/SO_REUSEPORT on bound sockets")
Signed-off-by: Bart Van Assche <redacted>

Applied, thanks.

I held off on submitting the reverted patch to -stable, and have now
thus removed it from my -stable queue.

Any ideas about how to fix the core issue of tb->fast* being
effectively invalid?

ie. currently any reader of tb->fastreuse(port) which isn't simply
testing for it being >= 0 is basically a bug (-1 is the empty tb case,
so that AFAICT keeps on working).

For example sk_reuseport_match(tb, sk) can both fail to match when it
should, and can match when it shouldn't...

(at a quick glance, all the readers, and thus bugs are constrained to
the inet_csk_get_port() function)

Do we just delete that entire 'tb->fast*' optimization?  It would
certainly make the code much simpler...

Do we put special case per-family/protocol code (ie. presumably
another indirect call) to fix up tb->fast in the
setsockopt(SOREUSEADDR/PORT) codepath?

Something else?

(btw. I'm not certain if both 0->1 and 1->0 transitions on a bound
socket are equally buggy, I think one is more dangerous then the
other)