linux-kernel - IOV iterator bug in net tree (was Re: Deadlock when running DevStack on latest pull of net tree)

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:	Fri, 27 May 2016 21:32:39 -0700
From:	Alexander Duyck <alexander.duyck@...il.com>
To:	Netdev <netdev@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	David Miller <davem@...emloft.net>
Subject: IOV iterator bug in net tree (was Re: Deadlock when running DevStack
 on latest pull of net tree)

On Fri, May 27, 2016 at 11:21 AM, Alexander Duyck
<alexander.duyck@...il.com> wrote:
> I started out this morning by trying to run DevStack on the latest
> "net' kernel and it looks like I am hanging on some sort of locking
> problem with RabbitMQ.  Specifically I am seeing one CPU jump to 100%
> with perf showing that I am spinning on a lock.
>
> I'm working to bisect it now, but just thought I would put it out
> there if anybody had already root caused this issue.  Below is a few
> traces to the spin lock call on the spinning CPU:

I misread the perf trace.  I wasn't spinning on a lock I was spinning
on the IOV iterator.  It just happened to push the lock to the top of
the perf trace.  I bisected it down to the recent change in
iterate_and_advance which was already fixed in Linus's tree in commit
19f18459330f "do 'fold checks into iterate_and_advance()" right'.

- Alex