dcsimg
RealTime IT News

Linux 2.6.37 kills the Big Kernel Lock

tux.jpg
From the 'Locks are Bad' files:

Linus Torvalds released the first RC for the 2.6.37 Linux kernel today, including nearly 10,000 comitts.

Yeaah, that's a lot of code. Perhaps the most noteworthy item at this early stage though is about the final stage of a fix that has been in the working for years in the Linux kernel.

"The part that I think deserves some extra mention is that we've
finally largely gotten rid of the BKL (big kernel lock) in all the
core stuff, and you can easily compile a kernel without any BKL support at all," Torvalds wrote.

The problem with the BKL is that it's an older less-elegant approach to locking, than more modern fine-grained spinlocks and other locking mechanisms.

Why this matters from my perspective is that with the removal of BKL, Linux can now become even more real-time in the mainline. as well as potentially improve performance and control through the kernel.

It's also a sign of the maturity of Linux that kernel developers have been able to look back in the past of Linux and correct a long standing issue with new code approaches. Sure there is some bloat in the Linux kernel code-base, but the BKL fix to me looks like proof-positive that you can teach an old dog new tricks (not that Linux is that old either but...).