netdev - Re: [PATCH 2/2] IP_GRE: Linearize skb before csum.

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: <20130122.232235.643904201225800004.davem@davemloft.net>
Date:	Tue, 22 Jan 2013 23:22:35 -0500 (EST)
From:	David Miller <davem@...emloft.net>
To:	eric.dumazet@...il.com
Cc:	jesse@...ira.com, pshelar@...ira.com, netdev@...r.kernel.org
Subject: Re: [PATCH 2/2] IP_GRE: Linearize skb before csum.

From: David Miller <davem@...emloft.net>
Date: Tue, 22 Jan 2013 20:46:29 -0500 (EST)

> We can't emit bogus checksums on the wire when this is completely
> under our control.

BTW consider a legitimate use case where an application has a shared
memory page between threads that keeps track of statistic counters,
and it uses sendfile() over an fd backing that mmap()'d area to
send the statistics out remotely over TCP.

That's completely legitimate.

As is an application constantly overwriting a page in the filesystem
page cache, and another app doing sendfile() over TCP from that
area.

In such a situation everyone of your magic retransmits will have a
bad checksum too.

You can't say that this behavior is anything other than broken, and
I don't want to be responsible for a system that's doing broken
things like that.

If your GSO over GRE tunnels stuff fundamentally depends upon this
magic retransmit fantasy working, then yes we have to disable it
completely.
--
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