Re: mutex syscall patch, questions about clone and its possible bugs

Theodore Y. Ts'o (tytso@mit.edu)
Mon, 13 May 1996 19:15:36 -0400


Date: Mon, 13 May 1996 08:28:44 -0500
From: Miguel de Icaza <miguel@roxanne.nuclecu.unam.mx>

I remember when at one time Chris Provenzano said something about
people wanting one thousand threads in one application. The kernel
can't cope with that number of threads very nicely, he proposes to use
a number of kernel threads (ie, our clone() created threads) and then
keep a pool of his userlevel threads to do the work.

Chris Provenzano thinks that a valid programming paradigm might
require using a thousand threads in one application. But keep in mind
that he's thread fanatic.

Sure, I could rewrite a program so that there was a separate
thread for each key on the keyboard, to be woken up when the user typed
that particular key. Most people don't do it now because it's horribly
inefficient on most thread implementations, especially one that uses
kernel threads. But just because you *can* do something like that using
Chris's pthreads package doesn't mean that it's a sane thing to do!

- Ted