>The Linux® compatibility layer has been substantially improved and now is capab...

kev009 · on April 4, 2016

There are at least three theoretical implementations of userland threads in an Operating System:

* 1:1, where each userland thread maps directly to a kernel thread

* 1:N, where several userland threads map directly to one kernel thread

* M:N, where the kernel maintains a pool of threads and dispatches userland threads to them

History has shown that 1:1 threading has the best trade offs. It's worth more independent research if you are interested in Operating Systems.

akerro · on April 4, 2016

Thanks

anthk · on April 4, 2016

That's reminds me of database record relations.

akerro · on April 4, 2016

Or just re-use of basic structures ;)

xtf · on April 4, 2016

https://en.wikipedia.org/wiki/Thread_%28computing%29#1:1_.28...: 1:1 (kernel-level threading)

Threads created by the user in a 1:1 correspondence with schedulable entities in the kernel[10] are the simplest possible threading implementation. OS/2 and Win32 used this approach from the start, while on Linux the usual C library implements this approach (via the NPTL or older LinuxThreads). This approach is also used by Solaris, NetBSD, FreeBSD, OS X, and iOS.

mkup · on April 4, 2016

It's also worth noting that FreeBSD used to rely on M:N threading model (via libkse) then switched to 1:1 threading model (via libthr).

jdub · on April 4, 2016

Though LinuxThreads was M:N, and Solaris migrated to 1:1 relatively recently too (as did others).

the_why_of_y · on April 4, 2016

No, LinuxThreads was 1:1. Maybe you are thinking of IBM's NGPT proposal?

http://www.onlamp.com/pub/a/onlamp/2002/11/07/linux_threads....

"Relatively recently" would be Solaris 8, in 2000.