Native POSIX Thread Library
Before the 2.6 version of the Linux kernel, processes were the schedulable entities, and there were no special facilities for threads. However, it did have a system call — clone — which creates a copy of the calling process where the copy shares the address space of the caller. The LinuxThreads project used this system call to provide kernel-level threads (most of the previous thread implementations in Linux worked entirely in userland). Unfortunately, it only partially complied with POSIX, particularly in the areas of signal handling, scheduling, and inter-process synchronization primitives.
To improve upon LinuxThreads, it was clear that some kernel support and a new threading library would be required. Two competing projects were started to address the requirement: NGPT (Next Generation POSIX Threads) worked on by a team which included developers from IBM, and NPTL by developers at Red Hat. The NGPT team collaborated closely with the NPTL team and combined the best features of both implementations into NPTL. The NGPT project was subsequently abandoned in mid-2003 after merging its best features into NPTL.
NPTL was first released in Red Hat Linux 9. Old-style Linux POSIX threading is known for having trouble with threads that refuse to yield to the system occasionally, because it does not take the opportunity to preempt them when it arises, something that Windows was known to do better at the time. Red Hat claimed that NPTL fixed this problem in an article on the Java website about Java on Red Hat Linux 9.
NPTL uses a similar approach to LinuxThreads, in that the primary abstraction known by the kernel is still a process, and new threads are created with the clone() system call (called from the NPTL library). However, NPTL requires specialized kernel support to implement (for example) the contended case of synchronisation primitives which might require threads to sleep and wake again. The primitive used for this is known as a futex.
NPTL is a so-called 1×1 threads library, in that threads created by the user (via the
pthread_create() library function) are in 1-1 correspondence with schedulable entities in the kernel (tasks, in the Linux case). This is the simplest possible threading implementation.
An alternative to NPTL's 1×1 model is the m×n model.