Re: [PATCH] thread wakeup fix for 2.4.0-test7

From: kuznet@ms2.inr.ac.ru
Date: Mon Aug 28 2000 - 14:06:31 EST


Hello!

> get -EBADF, in some - data. And that race is going to be a hell to hunt
> down.

Nothing to hunt. If master closes descriptor, he knows what he does.
If timeout is 5 minutes and data arrived exactly after
5 minutes+-0 microseconds, connection may die, may stay alive.
It is normal race condition with principially undefined result.

The only thing, which kernel could provide, is to avoid race
resulting in infinite sleep.

> Yes, programs may need a way to abort blocked IO. But overloading close()
> is a *very* bad idea.

Al, did I ever say that it is good idea? It is *very* dirty. 8)

But, alas, it is not an abstract _idea_, it is plain fucking _life_.

My idea is to change behaviour to _not_ less sane
(more sane to my veiwpoint, but it is difficult to argue about tastes),
and to live in peace with this cursed life after this. 8)

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



This archive was generated by hypermail 2b29 : Thu Aug 31 2000 - 21:00:22 EST