linux_dsm_epyc7002/arch/frv/mm/pgalloc.c
Hugh Dickins 4c21e2f244 [PATCH] mm: split page table lock
Christoph Lameter demonstrated very poor scalability on the SGI 512-way, with
a many-threaded application which concurrently initializes different parts of
a large anonymous area.

This patch corrects that, by using a separate spinlock per page table page, to
guard the page table entries in that page, instead of using the mm's single
page_table_lock.  (But even then, page_table_lock is still used to guard page
table allocation, and anon_vma allocation.)

In this implementation, the spinlock is tucked inside the struct page of the
page table page: with a BUILD_BUG_ON in case it overflows - which it would in
the case of 32-bit PA-RISC with spinlock debugging enabled.

Splitting the lock is not quite for free: another cacheline access.  Ideally,
I suppose we would use split ptlock only for multi-threaded processes on
multi-cpu machines; but deciding that dynamically would have its own costs.
So for now enable it by config, at some number of cpus - since the Kconfig
language doesn't support inequalities, let preprocessor compare that with
NR_CPUS.  But I don't think it's worth being user-configurable: for good
testing of both split and unsplit configs, split now at 4 cpus, and perhaps
change that to 8 later.

There is a benefit even for singly threaded processes: kswapd can be attacking
one part of the mm while another part is busy faulting.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-29 21:40:42 -07:00

160 lines
4.2 KiB
C

/* pgalloc.c: page directory & page table allocation
*
* Copyright (C) 2004 Red Hat, Inc. All Rights Reserved.
* Written by David Howells (dhowells@redhat.com)
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*/
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/mm.h>
#include <linux/highmem.h>
#include <asm/pgalloc.h>
#include <asm/page.h>
#include <asm/cacheflush.h>
pgd_t swapper_pg_dir[PTRS_PER_PGD] __attribute__((aligned(PAGE_SIZE)));
kmem_cache_t *pgd_cache;
pte_t *pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address)
{
pte_t *pte = (pte_t *)__get_free_page(GFP_KERNEL|__GFP_REPEAT);
if (pte)
clear_page(pte);
return pte;
}
struct page *pte_alloc_one(struct mm_struct *mm, unsigned long address)
{
struct page *page;
#ifdef CONFIG_HIGHPTE
page = alloc_pages(GFP_KERNEL|__GFP_HIGHMEM|__GFP_REPEAT, 0);
#else
page = alloc_pages(GFP_KERNEL|__GFP_REPEAT, 0);
#endif
if (page)
clear_highpage(page);
flush_dcache_page(page);
return page;
}
void __set_pmd(pmd_t *pmdptr, unsigned long pmd)
{
unsigned long *__ste_p = pmdptr->ste;
int loop;
if (!pmd) {
memset(__ste_p, 0, PME_SIZE);
}
else {
BUG_ON(pmd & (0x3f00 | xAMPRx_SS | 0xe));
for (loop = PME_SIZE; loop > 0; loop -= 4) {
*__ste_p++ = pmd;
pmd += __frv_PT_SIZE;
}
}
frv_dcache_writeback((unsigned long) pmdptr, (unsigned long) (pmdptr + 1));
}
/*
* List of all pgd's needed for non-PAE so it can invalidate entries
* in both cached and uncached pgd's; not needed for PAE since the
* kernel pmd is shared. If PAE were not to share the pmd a similar
* tactic would be needed. This is essentially codepath-based locking
* against pageattr.c; it is the unique case in which a valid change
* of kernel pagetables can't be lazily synchronized by vmalloc faults.
* vmalloc faults work because attached pagetables are never freed.
* If the locking proves to be non-performant, a ticketing scheme with
* checks at dup_mmap(), exec(), and other mmlist addition points
* could be used. The locking scheme was chosen on the basis of
* manfred's recommendations and having no core impact whatsoever.
* -- wli
*/
DEFINE_SPINLOCK(pgd_lock);
struct page *pgd_list;
static inline void pgd_list_add(pgd_t *pgd)
{
struct page *page = virt_to_page(pgd);
page->index = (unsigned long) pgd_list;
if (pgd_list)
pgd_list->private = (unsigned long) &page->index;
pgd_list = page;
set_page_private(page, (unsigned long)&pgd_list);
}
static inline void pgd_list_del(pgd_t *pgd)
{
struct page *next, **pprev, *page = virt_to_page(pgd);
next = (struct page *) page->index;
pprev = (struct page **)page_private(page);
*pprev = next;
if (next)
next->private = (unsigned long) pprev;
}
void pgd_ctor(void *pgd, kmem_cache_t *cache, unsigned long unused)
{
unsigned long flags;
if (PTRS_PER_PMD == 1)
spin_lock_irqsave(&pgd_lock, flags);
memcpy((pgd_t *) pgd + USER_PGDS_IN_LAST_PML4,
swapper_pg_dir + USER_PGDS_IN_LAST_PML4,
(PTRS_PER_PGD - USER_PGDS_IN_LAST_PML4) * sizeof(pgd_t));
if (PTRS_PER_PMD > 1)
return;
pgd_list_add(pgd);
spin_unlock_irqrestore(&pgd_lock, flags);
memset(pgd, 0, USER_PGDS_IN_LAST_PML4 * sizeof(pgd_t));
}
/* never called when PTRS_PER_PMD > 1 */
void pgd_dtor(void *pgd, kmem_cache_t *cache, unsigned long unused)
{
unsigned long flags; /* can be called from interrupt context */
spin_lock_irqsave(&pgd_lock, flags);
pgd_list_del(pgd);
spin_unlock_irqrestore(&pgd_lock, flags);
}
pgd_t *pgd_alloc(struct mm_struct *mm)
{
pgd_t *pgd;
pgd = kmem_cache_alloc(pgd_cache, GFP_KERNEL);
if (!pgd)
return pgd;
return pgd;
}
void pgd_free(pgd_t *pgd)
{
/* in the non-PAE case, clear_page_tables() clears user pgd entries */
kmem_cache_free(pgd_cache, pgd);
}
void __init pgtable_cache_init(void)
{
pgd_cache = kmem_cache_create("pgd",
PTRS_PER_PGD * sizeof(pgd_t),
PTRS_PER_PGD * sizeof(pgd_t),
0,
pgd_ctor,
pgd_dtor);
if (!pgd_cache)
panic("pgtable_cache_init(): Cannot create pgd cache");
}