netdev - Re: [net-next 1/2] qdisc: Allow qdiscs to provide backpressure up the stack.

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

Message-ID: <1282924797.5208.31.camel@edumazet-laptop>
Date:	Fri, 27 Aug 2010 17:59:57 +0200
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Ben Greear <greearb@...delatech.com>
Cc:	David Miller <davem@...emloft.net>, netdev@...r.kernel.org
Subject: Re: [net-next 1/2] qdisc: Allow qdiscs to provide backpressure up
 the stack.

Le vendredi 27 août 2010 à 08:26 -0700, Ben Greear a écrit :

> For the trivial case, I can just kfree_skb when BUSY is returned, for the
> same overall behaviour as today.  For something like UDP, I might be able
> to poke the SKB back into the queue instead of freeing it.

Its not trivial at all. Actually I bet this is not possible.

There is no queue for UDP (apart CORKing of course). There is at most
one frame per socket. Once this frame is ready (uncork), we send it
immediately via ip_push_pending_frames(), a destructive process, no way
it can be rollbacked in case of failure at the end of the chain.

throttling is only done because some frames are queued in qdisc, keeping
memory accounting refs on socket.

Between UDP and qdisc, they are many layers (netfilter, ip
fragmentation, ...), so requeuing skb is not possible at all.

Netfilter wont rollback all its work.

backpressure should be evaluated by another way.

--
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