Re: Thread implementations...

Stephen C. Tweedie (sct@redhat.com)
Mon, 24 Aug 1998 11:41:01 +0100


Hi,

On 13 Aug 1998 19:05:31 +0200, Magnus Ahltorp <map@stacken.kth.se> said:

>> A possible solution could be extending of the SIGIO concept in the
>> following way:
>>
>> - SIGIO handler gets an extra parameter containing the FD
>> in being reported.

Already done. :)

> I received this reply from Johnny Eriksson <bygg@swip.net>:

> Implement a new set of signals, one for each possible file.

Yikes. 30,000 signal handlers in place at once?!

In 2.1.117, there is a new fcntl function, F_SETSIG (and a corresponding
F_GETSIG). This lets you specify a signal number to be delivered in
place of SIGIO if async IO arrives. The best part is that if you
specify the number of a POSIX realtime signal, then the signal gets
queued with the fd number as an argument to the signal handler.

Right now you need to be using glibc-2.1 to take full advantage of the
posix realtime stuff, but the kernel mechanism is all there.

--Stephen

-
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.altern.org/andrebalsa/doc/lkml-faq.html