Re: [RFC] Subtle race in dup2() and permissions on /proc/<pid>/fd/<n>
Edgar Toernig (froese@gmx.de)
Sat, 26 Jun 1999 05:20:39 +0200
Alexander Viro wrote:
>
> OK. I'm taking the 'implementation dependent' as permission to
> make it atomic. It's more sensible behaviour and we can trivially do it
> (moreover, doing it that way makes SMP-safe version much easier). So if
> doing the thing atomic is not prohibited I'll go for it. Clearly program
> can't *rely* on (heavily timing-dependent) race taking place ;-) So that
> will not break anything.
Another thing with CLONE_FILES:
What happens with a blocked I/O-call (read, accept) when another thread
closes the fd?
Ciao, ET.
PS: Once, somebody asked me how to abort an accept. He did it on
winsock by closing the listen socket. It told him that this does not
work in unix. But CLONE_FILES changes a lot of things...
-
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/