spl (short for set priority level, after the PDP-11assembler instruction of the same name[1]) is the name for a collection of Unix kernel routines or macros used to change the interrupt priority level.[2][3] This was historically needed to synchronize critical sections of kernel code that should not be interrupted.[4] Newer Unix variants which support symmetric multiprocessing now mostly use mutexes for this purpose, which is a more general solution, so multiple processors can execute kernel code at the same time.[5][1]
On older PDP-11 versions of Unix, there were eight of these routines, ranging from spl0 to spl7, each corresponding to one PDP-11 interrupt priority level,[3] in addition to splx, which restores a previous priority level (returned by one of the other routines).[2] On BSD Unix and its derivatives, these are called splhigh, splserial, splsched, splclock, splstatclock, splvm, spltty, splsofttty, splnet, splbio, splsoftnet, splsoftclock, spllowersoftclock, spl0, and splx.[2]
As of March 2019[update], the spl family of primitives is still heavily used in OpenBSD[6] and NetBSD,[7] which is evidenced by the plentiful calls to splnet() within the networking code;[6][7] whereas FreeBSD and DragonFly BSD use more modern concepts; for example, in DragonFly, LWKT tokens may be used in place of spl.
