Implement per-CPU caching for the page table page (vp) pool and the PTE

descriptor (pted) pool in the arm64 pmap implementation.  This
significantly reduces the side-effects of lock contention on the kernel
map lock that is (incorrectly) translated into excessive page daemon
wakeups.  This is not a perfect solution but it does lead to significant
speedups on machines with many CPU cores.

This requires adding a new pmap_init_percpu() function that gets called
at the point where kernel is ready to set up the per-CPU pool caches.
Dummy implementations of this function are added for all non-arm64
architectures.  Some other architectures can probably benefit from
providing an actual implementation that sets up per-CPU caches for
pmap pools as well.

ok phessler@, claudio@, miod@, patrick@

1 files changed, 2 insertions, 1 deletions

diff --git a/sys/arch/powerpc64/include/pmap.h b/sys/arch/powerpc64/include/pmap.h
index c7f4c9e257b..6489f4af094 100644
--- a/sys/arch/powerpc64/include/pmap.h
+++ b/sys/arch/powerpc64/include/pmap.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: pmap.h,v 1.18 2021/10/12 18:06:15 kettenis Exp $	*/
+/*	$OpenBSD: pmap.h,v 1.19 2023/12/11 22:12:53 kettenis Exp $	*/
 
 /*
  * Copyright (c) 2020 Mark Kettenis <kettenis@openbsd.org>
@@ -64,6 +64,7 @@ extern struct pmap kernel_pmap_store;
 #define pmap_resident_count(pm) ((pm)->pm_stats.resident_count)
 #define pmap_wired_count(pm)	((pm)->pm_stats.wired_count)
 
+#define pmap_init_percpu()		do { /* nothing */ } while (0)
 #define pmap_unuse_final(p)
 #define pmap_remove_holes(vm)
 #define pmap_update(pm)