Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 4c21e2f2 authored by Hugh Dickins's avatar Hugh Dickins Committed by Linus Torvalds
Browse files

[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: default avatarHugh Dickins <hugh@veritas.com>
Signed-off-by: default avatarAndrew Morton <akpm@osdl.org>
Signed-off-by: default avatarLinus Torvalds <torvalds@osdl.org>
parent b38c6845

arch/arm/mm/mm-armv.c

+1 −0
Original line number Original line Diff line number Diff line
@@ -229,6 +229,7 @@ void free_pgd_slow(pgd_t *pgd) 
	pte = pmd_page(*pmd);

	pte = pmd_page(*pmd);

	pmd_clear(pmd);

	pmd_clear(pmd);

	dec_page_state(nr_page_table_pages);

	dec_page_state(nr_page_table_pages);

	pte_lock_deinit(pte);

	pte_free(pte);

	pte_free(pte);

	pmd_free(pmd);

	pmd_free(pmd);

free:

free:

arch/frv/mm/pgalloc.c

+2 −2
Original line number Original line Diff line number Diff line
@@ -87,14 +87,14 @@ static inline void pgd_list_add(pgd_t *pgd) 
	if (pgd_list)

	if (pgd_list)

		pgd_list->private = (unsigned long) &page->index;

		pgd_list->private = (unsigned long) &page->index;

	pgd_list = page;

	pgd_list = page;

	page->private = (unsigned long) &pgd_list;

	set_page_private(page, (unsigned long)&pgd_list);

}

}





static inline void pgd_list_del(pgd_t *pgd)

static inline void pgd_list_del(pgd_t *pgd)

{

{

	struct page *next, **pprev, *page = virt_to_page(pgd);

	struct page *next, **pprev, *page = virt_to_page(pgd);

	next = (struct page *) page->index;

	next = (struct page *) page->index;

	pprev = (struct page **) page->private;

	pprev = (struct page **)page_private(page);

	*pprev = next;

	*pprev = next;

	if (next)

	if (next)

		next->private = (unsigned long) pprev;

		next->private = (unsigned long) pprev;

arch/i386/mm/pgtable.c

+4 −4
Original line number Original line Diff line number Diff line
@@ -188,19 +188,19 @@ static inline void pgd_list_add(pgd_t *pgd) 
	struct page *page = virt_to_page(pgd);

	struct page *page = virt_to_page(pgd);

	page->index = (unsigned long)pgd_list;

	page->index = (unsigned long)pgd_list;

	if (pgd_list)

	if (pgd_list)

		pgd_list->private = (unsigned long)&page->index;

		set_page_private(pgd_list, (unsigned long)&page->index);

	pgd_list = page;

	pgd_list = page;

	page->private = (unsigned long)&pgd_list;

	set_page_private(page, (unsigned long)&pgd_list);

}

}





static inline void pgd_list_del(pgd_t *pgd)

static inline void pgd_list_del(pgd_t *pgd)

{

{

	struct page *next, **pprev, *page = virt_to_page(pgd);

	struct page *next, **pprev, *page = virt_to_page(pgd);

	next = (struct page *)page->index;

	next = (struct page *)page->index;

	pprev = (struct page **)page->private;

	pprev = (struct page **)page_private(page);

	*pprev = next;

	*pprev = next;

	if (next)

	if (next)

		next->private = (unsigned long)pprev;

		set_page_private(next, (unsigned long)pprev);

}

}





void pgd_ctor(void *pgd, kmem_cache_t *cache, unsigned long unused)

void pgd_ctor(void *pgd, kmem_cache_t *cache, unsigned long unused)

arch/um/kernel/skas/mmu.c

+1 −0
Original line number Original line Diff line number Diff line
@@ -144,6 +144,7 @@ void destroy_context_skas(struct mm_struct *mm) 




	if(!proc_mm || !ptrace_faultinfo){

	if(!proc_mm || !ptrace_faultinfo){

		free_page(mmu->id.stack);

		free_page(mmu->id.stack);

		pte_lock_deinit(virt_to_page(mmu->last_page_table));

		pte_free_kernel((pte_t *) mmu->last_page_table);

		pte_free_kernel((pte_t *) mmu->last_page_table);

                dec_page_state(nr_page_table_pages);

                dec_page_state(nr_page_table_pages);

#ifdef CONFIG_3_LEVEL_PGTABLES

#ifdef CONFIG_3_LEVEL_PGTABLES

fs/afs/file.c

+2 −2
Original line number Original line Diff line number Diff line
@@ -291,8 +291,8 @@ static int afs_file_releasepage(struct page *page, gfp_t gfp_flags) 
		cachefs_uncache_page(vnode->cache, page);

		cachefs_uncache_page(vnode->cache, page);

#endif

#endif





		pageio = (struct cachefs_page *) page->private;

		pageio = (struct cachefs_page *) page_private(page);

		page->private = 0;

		set_page_private(page, 0);

		ClearPagePrivate(page);

		ClearPagePrivate(page);





		if (pageio)

		if (pageio)