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, 4 insertions, 2 deletions

diff --git a/sys/arch/arm64/include/pmap.h b/sys/arch/arm64/include/pmap.h
index 8d5bbf2eaa9..129fa4079b0 100644
--- a/sys/arch/arm64/include/pmap.h
+++ b/sys/arch/arm64/include/pmap.h
@@ -1,4 +1,4 @@
-/* $OpenBSD: pmap.h,v 1.24 2023/06/10 19:30:48 kettenis Exp $ */
+/* $OpenBSD: pmap.h,v 1.25 2023/12/11 22:12:53 kettenis Exp $ */
 /*
  * Copyright (c) 2008,2009,2014 Dale Rahn <drahn@dalerahn.com>
  *
@@ -101,6 +101,9 @@ extern struct pmap kernel_pmap_;
 
 vaddr_t pmap_bootstrap(long kvo, paddr_t lpt1,  long kernelstart,
     long kernelend, long ram_start, long ram_end);
+void pmap_postinit(void);
+void pmap_init_percpu(void);
+
 void pmap_kenter_cache(vaddr_t va, paddr_t pa, vm_prot_t prot, int cacheable);
 void pmap_page_ro(pmap_t pm, vaddr_t va, vm_prot_t prot);
 void pmap_page_rw(pmap_t pm, vaddr_t va);
@@ -118,7 +121,6 @@ struct pv_entry;
 /* investigate */
 #define pmap_unuse_final(p)		do { /* nothing */ } while (0)
 int	pmap_fault_fixup(pmap_t, vaddr_t, vm_prot_t);
-void pmap_postinit(void);
 
 #define __HAVE_PMAP_MPSAFE_ENTER_COW