Re: patch for 2.1.123 RPC

H.J. Lu (hjl@lucon.org)
Thu, 1 Oct 1998 08:39:53 -0700 (PDT)

>
> This is a multi-part message in MIME format.
> --------------67BC36E57CB67B1F6F789788
> Content-Type: text/plain; charset=us-ascii
> Content-Transfer-Encoding: 7bit
>
> G. Allen Morris III wrote:
>
> > I tried running `dd' the other day and it did hang. `ps' said that
> > it was hung in `end'. The input Q was close ot 64k and the output
> > Q was 8k less.
> >
> > Now that I have a few seconds I have tried it again, but can't get it
> > to fail. Has anyone else tried this.
>
> I've put together a patch for the RPC layer that appears to fix the problem with
> knfsd getting stuck.
>
> The problem was that knfsd was waiting in the svc_recv routine, and was not being
> awakened when the client-side RPC retransmitted the RPC requests. This seems to
> have been caused by a race condition in testing for whether a socket had been
> assigned after waking up. The patch protects the test with a bh-exclusion, and
> this made the problem go away.
>

I applied your sunrpc patch to the client side running linux 2.1.123.
My knfsd server is running 2.1.122. Now on my server, I am getting
massive

Socket destroy delayed (r=64288 w=0)
Socket destroy delayed (r=64400 w=0)
Socket destroy delayed (r=64288 w=0)
Socket destroy delayed (r=64400 w=0)
Socket destroy delayed (r=64288 w=0)
Socket destroy delayed (r=64400 w=0)

As the result, my knfsd server stops responding. I have to restart my
knfsd server.

H.J.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/