netdev - [BUG] af_packet: tpacket_destruct_skb throws BUG under high tx load

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>] [day] [month] [year] [list]

Date:	Mon, 6 Aug 2012 13:21:01 +0200
From:	Daniel Borkmann <danborkmann@...earbox.net>
To:	Johann Baudy <johann.baudy@...-log.net>
Cc:	netdev@...r.kernel.org, Eric Dumazet <eric.dumazet@...il.com>,
	"David S. Miller" <davem@...emloft.net>
Subject: [BUG] af_packet: tpacket_destruct_skb throws BUG under high tx load

Under high tx load from mmap()ed PF_PACKET sockets the statement in
tpacket_destruct_skb():

  BUG_ON(__packet_get_status(po, ph) != TP_STATUS_SENDING);

is being thrown, where __packet_get_status(po, ph) has the value
TP_STATUS_SEND_REQUEST instead of TP_STATUS_SENDING.

The send(2) system call was invoked with the flag MSG_DONTWAIT. The
BUG occurred even with having dev_queue_xmit() returned successfully
in tpacket_snd().

By simply removing this BUG statement, the system seemed stable, but
this might not be the proper solution. It's reproducible and I tested
it on a multi-core system  under the latest net/net-next kernel on x86
and x86_64.

On my system, this BUG was triggered in the following cases:

* Multiple processes allocate and use a TX_RING on one device

* A single process allocates a TX_RING on one device and
/proc/sys/net/core/wmem_default, /proc/sys/net/core/wmem_max were
increased

Cheers,

Daniel
--
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