Re: [V9fs-developer] Hang triggered by udev coldplug, looks like a race

[Andy Lutomirski]

On Mon, Dec 7, 2015 at 6:33 PM, Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> On Mon, Dec 07, 2015 at 05:59:41PM -0800, Andy Lutomirski wrote:
>
>> If I dump all task states (see attached typescript), I see a bunch of
>> things blocked in 9p rpc.  This makes me think it could be a QEMU bug,
>> not a kernel bug.
>
> Maybe, maybe not - I'd suggest dumping the 9p traffic and checking the
> request and response IDs; at least one bug in that area got fixed
> (see commit a84b69 for gory details).
>
> Really, check if something gets screwed with IDs.  They should be unique
> among outstanding requests and never reused until either a response
> with the matching ID arrives, *or* TFLUSH on that ID is sent and reply
> has arrived.   If those rules get buggered, it's very easy to get hangs
> like that.

Trace attached.  I don't see anything wrong, but I also don't know
what I'm looking for.

--Andy

-- 
Andy Lutomirski
AMA Capital Management, LLC

Attachment: 9ptrace.txt.xz
Description: application/xz