freebsd-skq/sys/powerpc/ps3/mmu_ps3.c
nwhitehorn 8d11dd01a1 New pmap implementation for 64-bit PowerPC processors. The main focus of
this change is to improve concurrency:
- Drop global state stored in the shadow overflow page table (and all other
  global state)
- Remove all global locks
- Use per-PTE lock bits to allow parallel page insertion
- Reconstruct state when requested for evicted PTEs instead of buffering
  it during overflow

This drops total wall time for make buildworld on a 32-thread POWER8 system
by a factor of two and system time by a factor of three, providing performance
20% better than similarly clocked Core i7 Xeons per-core. Performance on
smaller SMP systems, where PMAP lock contention was not as much of an issue,
is nearly unchanged.

Tested on:	POWER8, POWER5+, G5 UP, G5 SMP (64-bit and 32-bit kernels)
Merged from:	user/nwhitehorn/ppc64-pmap-rework
Looked over by:	jhibbits, andreast
MFC after:	3 months
Relnotes:	yes
Sponsored by:	FreeBSD Foundation
2015-02-24 21:37:20 +00:00

284 lines
7.2 KiB
C

/*-
* Copyright (C) 2010 Nathan Whitehorn
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL TOOLS GMBH BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
* OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/msgbuf.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/sysctl.h>
#include <sys/systm.h>
#include <sys/vmmeter.h>
#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/vm_kern.h>
#include <vm/vm_page.h>
#include <vm/vm_map.h>
#include <vm/vm_object.h>
#include <vm/vm_extern.h>
#include <vm/vm_pageout.h>
#include <vm/uma.h>
#include <powerpc/aim/mmu_oea64.h>
#include "mmu_if.h"
#include "moea64_if.h"
#include "ps3-hvcall.h"
#define VSID_HASH_MASK 0x0000007fffffffffUL
#define PTESYNC() __asm __volatile("ptesync")
extern int ps3fb_remap(void);
static uint64_t mps3_vas_id;
/*
* Kernel MMU interface
*/
static void mps3_bootstrap(mmu_t mmup, vm_offset_t kernelstart,
vm_offset_t kernelend);
static void mps3_cpu_bootstrap(mmu_t mmup, int ap);
static int64_t mps3_pte_synch(mmu_t, struct pvo_entry *);
static int64_t mps3_pte_clear(mmu_t, struct pvo_entry *, uint64_t ptebit);
static int64_t mps3_pte_unset(mmu_t, struct pvo_entry *);
static int mps3_pte_insert(mmu_t, struct pvo_entry *);
static mmu_method_t mps3_methods[] = {
MMUMETHOD(mmu_bootstrap, mps3_bootstrap),
MMUMETHOD(mmu_cpu_bootstrap, mps3_cpu_bootstrap),
MMUMETHOD(moea64_pte_synch, mps3_pte_synch),
MMUMETHOD(moea64_pte_clear, mps3_pte_clear),
MMUMETHOD(moea64_pte_unset, mps3_pte_unset),
MMUMETHOD(moea64_pte_insert, mps3_pte_insert),
{ 0, 0 }
};
MMU_DEF_INHERIT(ps3_mmu, "mmu_ps3", mps3_methods, 0, oea64_mmu);
static struct mtx mps3_table_lock;
static void
mps3_bootstrap(mmu_t mmup, vm_offset_t kernelstart, vm_offset_t kernelend)
{
uint64_t final_pteg_count;
mtx_init(&mps3_table_lock, "page table", NULL, MTX_DEF);
moea64_early_bootstrap(mmup, kernelstart, kernelend);
lv1_construct_virtual_address_space(
20 /* log_2(moea64_pteg_count) */, 2 /* n page sizes */,
(24UL << 56) | (16UL << 48) /* page sizes 16 MB + 64 KB */,
&mps3_vas_id, &final_pteg_count
);
moea64_pteg_count = final_pteg_count / sizeof(struct lpteg);
moea64_mid_bootstrap(mmup, kernelstart, kernelend);
moea64_late_bootstrap(mmup, kernelstart, kernelend);
}
static void
mps3_cpu_bootstrap(mmu_t mmup, int ap)
{
struct slb *slb = PCPU_GET(slb);
register_t seg0;
int i;
mtmsr(mfmsr() & ~PSL_DR & ~PSL_IR);
/*
* Destroy the loader's address space if we are coming up for
* the first time, and redo the FB mapping so we can continue
* having a console.
*/
if (!ap)
lv1_destruct_virtual_address_space(0);
lv1_select_virtual_address_space(mps3_vas_id);
if (!ap)
ps3fb_remap();
/*
* Install kernel SLB entries
*/
__asm __volatile ("slbia");
__asm __volatile ("slbmfee %0,%1; slbie %0;" : "=r"(seg0) : "r"(0));
for (i = 0; i < 64; i++) {
if (!(slb[i].slbe & SLBE_VALID))
continue;
__asm __volatile ("slbmte %0, %1" ::
"r"(slb[i].slbv), "r"(slb[i].slbe));
}
}
static int64_t
mps3_pte_synch_locked(struct pvo_entry *pvo)
{
uint64_t halfbucket[4], rcbits;
PTESYNC();
lv1_read_htab_entries(mps3_vas_id, pvo->pvo_pte.slot & ~0x3UL,
&halfbucket[0], &halfbucket[1], &halfbucket[2], &halfbucket[3],
&rcbits);
/* Check if present in page table */
if ((halfbucket[pvo->pvo_pte.slot & 0x3] & LPTE_AVPN_MASK) !=
((pvo->pvo_vpn >> (ADDR_API_SHFT64 - ADDR_PIDX_SHFT)) &
LPTE_AVPN_MASK))
return (-1);
if (!(halfbucket[pvo->pvo_pte.slot & 0x3] & LPTE_VALID))
return (-1);
/*
* rcbits contains the low 12 bits of each PTE's 2nd part,
* spaced at 16-bit intervals
*/
return ((rcbits >> ((3 - (pvo->pvo_pte.slot & 0x3))*16)) &
(LPTE_CHG | LPTE_REF));
}
static int64_t
mps3_pte_synch(mmu_t mmu, struct pvo_entry *pvo)
{
int64_t retval;
mtx_lock(&mps3_table_lock);
retval = mps3_pte_synch_locked(pvo);
mtx_unlock(&mps3_table_lock);
return (retval);
}
static int64_t
mps3_pte_clear(mmu_t mmu, struct pvo_entry *pvo, uint64_t ptebit)
{
int64_t refchg;
struct lpte pte;
mtx_lock(&mps3_table_lock);
refchg = mps3_pte_synch_locked(pvo);
if (refchg < 0) {
mtx_unlock(&mps3_table_lock);
return (refchg);
}
moea64_pte_from_pvo(pvo, &pte);
pte.pte_lo |= refchg;
pte.pte_lo &= ~ptebit;
/* XXX: race on RC bits between write and sync. Anything to do? */
lv1_write_htab_entry(mps3_vas_id, pvo->pvo_pte.slot, pte.pte_hi,
pte.pte_lo);
mtx_unlock(&mps3_table_lock);
return (refchg);
}
static int64_t
mps3_pte_unset(mmu_t mmu, struct pvo_entry *pvo)
{
int64_t refchg;
mtx_lock(&mps3_table_lock);
refchg = mps3_pte_synch_locked(pvo);
if (refchg < 0) {
moea64_pte_overflow--;
mtx_unlock(&mps3_table_lock);
return (-1);
}
/* XXX: race on RC bits between unset and sync. Anything to do? */
lv1_write_htab_entry(mps3_vas_id, pvo->pvo_pte.slot, 0, 0);
mtx_unlock(&mps3_table_lock);
moea64_pte_valid--;
return (refchg & (LPTE_REF | LPTE_CHG));
}
static int
mps3_pte_insert(mmu_t mmu, struct pvo_entry *pvo)
{
int result;
struct lpte pte, evicted;
uint64_t index;
if (pvo->pvo_vaddr & PVO_HID) {
/* Hypercall needs primary PTEG */
pvo->pvo_vaddr &= ~PVO_HID;
pvo->pvo_pte.slot ^= (moea64_pteg_mask << 3);
}
pvo->pvo_pte.slot &= ~7UL;
moea64_pte_from_pvo(pvo, &pte);
evicted.pte_hi = 0;
PTESYNC();
mtx_lock(&mps3_table_lock);
result = lv1_insert_htab_entry(mps3_vas_id, pvo->pvo_pte.slot,
pte.pte_hi, pte.pte_lo, LPTE_LOCKED | LPTE_WIRED, 0,
&index, &evicted.pte_hi, &evicted.pte_lo);
mtx_unlock(&mps3_table_lock);
if (result != 0) {
/* No freeable slots in either PTEG? We're hosed. */
panic("mps3_pte_insert: overflow (%d)", result);
return (-1);
}
/*
* See where we ended up.
*/
if ((index & ~7UL) != pvo->pvo_pte.slot)
pvo->pvo_vaddr |= PVO_HID;
pvo->pvo_pte.slot = index;
moea64_pte_valid++;
if (evicted.pte_hi) {
KASSERT((evicted.pte_hi & (LPTE_WIRED | LPTE_LOCKED)) == 0,
("Evicted a wired PTE"));
moea64_pte_valid--;
moea64_pte_overflow++;
}
return (0);
}