On Mon, 8 Mar 2021 17:59:47 -0800, Davidlohr Bueso wrote: > A few updates while going through the powerpc port of the qspinlock. > > Patches 1 and 2 are straightforward, while patch 3 can be considered > more of an rfc as I've only tested on a single machine, and there > could be an alternative way if it doesn't end up being nearly a > universal performance win. > > [...]
Applied to powerpc/next. [1/3] powerpc/spinlock: Define smp_mb__after_spinlock only once https://git.kernel.org/powerpc/c/2bf3604c415c9d75311141b8eb6ac8780ef74420 [2/3] powerpc/spinlock: Unserialize spin_is_locked https://git.kernel.org/powerpc/c/66f60522138c2e0d8a3518edd4979df11a2d7525 [3/3] powerpc/qspinlock: Use generic smp_cond_load_relaxed https://git.kernel.org/powerpc/c/deb9b13eb2571fbde164ae012c77985fd14f2f02 cheers