Re: patch-2.0.1[67] kills automount daemon (Berkeley amd)

Linus Torvalds (torvalds@cs.helsinki.fi)
Thu, 5 Sep 1996 12:34:27 +0300 (EET DST)


On Thu, 5 Sep 1996, Swen Thuemmler wrote:
>
> it seems to work. I had no more hangs with amd, but I could not
> deterministically force a hang before, therefore "seems". But I've tried
> hard (with run-parts /etc/cron.daily, this used to hang the automounter).

Ok, thanks. I'll assume it fixes the problems (it really should - there
haven't been any other changes that should result in problems like these).

> There is another place in rpcsock.c, where rsock->backlog is tested:
>
> rsock->cong -= RPC_CWNDSCALE;
> if (rsock->cong < rsock->cwnd && rsock->backlog)
> wake_up(&rsock->backlog);
>
> Should this be changed, too? (I'm just applying pattern matching, I don't
> really understand this...)

Good spotting, but for various reasons the above is actually ok (you could
even remove the test for "rsock->backlog" completely, in fact). It could
be changed to use "waitqueue_active()", and in some sense it would be
more correct that way, but I'd actually prefer just removing that test..

Linus