[RFC] ARM: mm: lazy cache flushing on non-mapped pages

Commit Message

Ming Lei May 22, 2013, 11:55 a.m. UTC

Currently flush_dcache_page() thinks pages as non-mapped if
mapping_mapped(mapping) return false. This approach is very
coase:
	- mmap on part of file may cause all pages backed on
	the file being thought as mmaped

	- file-backed pages aren't mapped into user space actually
	if the memory mmaped on the file isn't accessed

This patch also uses page_mapcount() to decide if the page has been
mapped.

From the attached test code, we can see there is much performance
improvement(>25%) when accessing page caches via read under this
situations, so memcpy benefits a lot from not flushing cache
under this situations.

Seq.   read time without the patch	Seq. read time with the patch

Comments

Catalin Marinas May 22, 2013, 2:42 p.m. UTC | #1

On Wed, May 22, 2013 at 12:55:11PM +0100, Ming Lei wrote:
> Currently flush_dcache_page() thinks pages as non-mapped if
> mapping_mapped(mapping) return false. This approach is very
> coase:
> 	- mmap on part of file may cause all pages backed on
> 	the file being thought as mmaped
> 
> 	- file-backed pages aren't mapped into user space actually
> 	if the memory mmaped on the file isn't accessed

That's the meaning of mapping_mapped(), it doesn't tell whether the page
has a user virtual address.

> This patch also uses page_mapcount() to decide if the page has been
> mapped.

Better use page_mapped() instead.

Ming Lei May 23, 2013, 2:47 a.m. UTC | #2

On Wed, May 22, 2013 at 10:42 PM, Catalin Marinas
<[e-mail]> wrote:
>
> That's the meaning of mapping_mapped(), it doesn't tell whether the page
> has a user virtual address.

Yes, I mean pages only need to flush if they has been mapped, so we can
use page_mapped() to decide that.

>
>> This patch also uses page_mapcount() to decide if the page has been
>> mapped.
>
> Better use page_mapped() instead.

Good idea, will do it v1, :-)

Thanks,
--
Ming Lei

Patch

================================================================
No. 0, time  22615636 us		No. 0, time  22014717 us
No. 1, time  4387851 us 		No. 1, time  3113184 us
No. 2, time  4276535 us 		No. 2, time  3005244 us
No. 3, time  4259821 us 		No. 3, time  3001565 us
No. 4, time  4263811 us 		No. 4, time  3002748 us
No. 5, time  4258486 us 		No. 5, time  3004104 us
No. 6, time  4253009 us 		No. 6, time  3002188 us
No. 7, time  4262809 us 		No. 7, time  2998196 us
No. 8, time  4264525 us 		No. 8, time  3007255 us
No. 9, time  4267795 us 		No. 9, time  3005094 us

1), No.0. is to read the file from storage device, and others are
to read the file from page caches basically.
2), file size is 512M, and is on ext4 over usb mass storage.
3), the test is done on Pandaboard.

unsigned int  sum = 0;
unsigned long sum_val = 0;

static unsigned long tv_diff(struct timeval *tv1, struct timeval *tv2)
{
	return (tv2->tv_sec - tv1->tv_sec) * 1000000 +
		(tv2->tv_usec - tv1->tv_usec);
}

int main(int argc, char *argv[])
{
	char *mbuf, fbuf;
	int fd;
	int i;
	unsigned long page_size, size;
	struct stat stat;
	struct timeval t1, t2;
	unsigned char *rbuf = malloc(32 * page_size);

	if (!rbuf) {
		printf("	%s\n", "malloc failed");
		exit(-1);
	}

	page_size = getpagesize();
	fd = open(argv[1], O_RDWR);
	assert(fd >= 0);

	fstat(fd, &stat);
	size = stat.st_size;
	printf("%s: file %s, size %lu, page size %lu\n",
		argv[0],
		argv[1], size, page_size);

	gettimeofday(&t1, NULL);
	mbuf = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
	if (!mbuf) {
		printf("	%s\n", "mmap failed");
		exit(-1);
	}

	for (i = 0 ; i < size ; i += (page_size * 32)) {
		int rcnt;
		lseek(fd, i, SEEK_SET);
		rcnt = read(fd, rbuf, page_size * 32);
		if (rcnt != page_size * 32) {
			printf("%s: read faild\n", __func__);
			exit(-1);
		}
	}
	free(rbuf);
	munmap(mbuf, size);
	gettimeofday(&t2, NULL);
	printf("\tread mmaped time: %luus\n", tv_diff(&t1, &t2));

	close(fd);
}

Cc: Michel Lespinasse <[e-mail]>
Cc: Andrew Morton <[e-mail]>
Cc: Will Deacon <[e-mail]>
Cc: Nicolas Pitre <[e-mail]>
Cc: Catalin Marinas <[e-mail]>
Cc: Russell King <[e-mail]>
Signed-off-by: Ming Lei <[e-mail]>
---
 arch/arm/mm/flush.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mm/flush.c b/arch/arm/mm/flush.c
index 9251d56..1d0e8d5 100644
--- a/arch/arm/mm/flush.c
+++ b/arch/arm/mm/flush.c
@@ -283,8 +283,8 @@  void flush_dcache_page(struct page *page)
 
 	mapping = page_mapping(page);
 
-	if (!cache_ops_need_broadcast() &&
-	    mapping && !mapping_mapped(mapping))
+	if (!cache_ops_need_broadcast() && mapping &&
+		(!mapping_mapped(mapping) || !page_mapcount(page)))
 		clear_bit(PG_dcache_clean, &page->flags);
 	else {
 		__flush_dcache_page(mapping, page);