Re: [PATCH][RFC] Pollable /proc/<pid>/ - avoid SIGCHLD/poll() races

[Paul Menage <pmenage@ensim.com>]

>I don't think it's contrived --- writing not a byte, but the pid and
>return status of the dead child to a pipe is an old but useful trick.
>It gives a natural serialisation of child deaths, and also eliminates
>the common race where a child dies before its parent has recorded its
>pid in a data structure. See it as a safe way of converting an
>asynchronous signal to a queued event.

Except that this enhancement is not completely safe, as if you get more
than 1024 children reaped (assuming you send two bytes of pid and two
bytes of status) between checks of the pipe, you'll lose notifications.
Admittedly this should be a problem in the sshd case, but it's not a
perfect solution in general. 

At least if you're only using the pipe to stop select() from blocking,
you don't care about overflowing the pipe as there's no important
information in there anyway.

Paul