Re: Async IO

Ulrich Drepper (
30 Jan 1997 21:39:55 GMT

In article <5cpam4$> (H. Peter Anvin) writes:

> We probably should create a couple of more signals... SIGUNUSED (31)
> is still available, SIGLOST (29) is commented out, signal number 0 is
> unused, and SIGPWR (30) basically only makes sense to init(8). I
> would suggest creating a SIGTHREAD reserved for Linuxthreads; that way
> it wouldn't have to eat up a user signal...

LinuxThreads requires two signals. And if somebody (for Linux-2.0
compatibility) wants to write the POSIX.1b signal emulation using
message queues we need another symbol.

Sooo, we need two people/groups. One writing the emulation of
POSIX.1b signals using 1 signal and the other group who writes
POSIX.1b signals for the kernel. I already have quite some code for
the AIO implementation in user level so that it should be possible to
use it as soon as the signals are available.

-- Uli
---------------.  ,-.   Rubensstrasse 5
Ulrich Drepper  \    ,-------------------'   \  76149 Karlsruhe/Germany
Cygnus Solutions `--'  `------------------------