Re: fs/dcache.c - BUG: soft lockup - CPU#5 stuck for 22s! [systemd-udevd:1667]

[Al Viro <viro@ZenIV.linux.org.uk>]

On Thu, May 29, 2014 at 04:52:33AM +0100, Al Viro wrote:
> On Thu, May 29, 2014 at 04:11:49AM +0100, Al Viro wrote:
> > On Wed, May 28, 2014 at 07:39:54PM +0100, Al Viro wrote:
> > 
> > > OK, the warnings about averting your eyes very much apply; the thing below
> > > definitely needs more massage before it becomes acceptable (and no, it's
> > > not a single commit; I'm not that insane), but it changes behaviour in the
> > > way described above.  Could you check if the livelock persists with it?
> > > No trace-generating code in there, so the logs should be compact enough...
> > 
> > Here's an updated patch, hopefully slightly less vomit-inducing.  Should
> > give the same behaviour as the previous one...  Again, it's a cumulative
> > diff - I'm still massaging the splitup here.
> 
> BTW, it still leaves the "proceed to parent" case in shrink_dentry_list();
> in theory, it's also vulnerable to the same livelock.  Can be dealt pretty
> much the same way; I'd rather leave that one for right after -final, though,
> if the already posted variant turns out to be sufficient...

... which is (presumably) dealt with the incremental I'd just sent to Linus;
seeing what kind of dumb mistakes I'm making, I'd better call it quits for
tonight - it's 1:30am here and I didn't have anywhere near enough sleep
yesterday.  I'd appeciate if you could test the patch immediately
upthread (from Message-ID: <20140529031149.GE18016@ZenIV.linux.org.uk>)
and see if it helps.  There's an incremental on top of it (from
Message-ID: <20140529052621.GH18016@ZenIV.linux.org.uk>) that might or
might not be a good idea.

I'm crawling to bed right now; back in ~7 hours...