ARM: DMA-mapping: mark all !DMA_TO_DEVICE pages in unmapping as clean
PATCH 1/1 ARM: DMA-mapping: mark all !DMA_TO_DEVICE pages in unmapping as clean
It is common for one sg to include many pages, so mark all these pages as clean to avoid unnecessary flushing on them in set_pte_at() or update_mmu_cache().
The patch might improve loading performance of applciation code a bit.
On the below test code to read file(~1GByte size) from usb mass storage disk to buffer created with mmap(PROT_READ | PROT_EXEC) on Pandaboard, average ~1% improvement can be observed with the patch on 10 times test.
unsigned int sum = 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 *mbuffer;
int fd;
int i;
unsigned long page_size, size;
struct stat stat;
struct timeval t1, t2;
page_size = getpagesize();
fd = open(argv[1], O_RDONLY);
assert(fd >= 0);
fstat(fd, &stat);
size = stat.st_size;
printf("%s: file %s, file size %lu, page size %lu\n", argv[0],
read_filename, size, page_size);
gettimeofday(&t1, NULL);
mbuffer = mmap(NULL, size, PROT_READ | PROT_EXEC, MAP_SHARED, fd, 0);
for (i = 0 ; i < size ; i += page_size)
sum += mbuffer[i];
munmap(mbuffer, page_size);
gettimeofday(&t2, NULL);
printf("\tread mmaped time: %luus\n", tv_diff(&t1, &t2));
close(fd);
}---
v1:
- fix one mistake on computing pfn, pointed out by Nicolas
arch/arm/mm/dma-mapping.c | 20 +++++++++++++++++--- 1 file changed, 17 insertions(+), 3 deletions(-)
arch/arm/mm/dma-mapping.c +17 -3
@@ -880,10 +880,24 @@ static void __dma_page_dev_to_cpu(struct page *page, unsigned long off,dma_cache_maint_page (page, off, size, dir, dmac_unmap_area);/* -* Mark the D-cache clean for this page to avoid extra flushing. +* Mark the D-cache clean for these pages to avoid extra flushing. */ -if (dir !=DMA_TO_DEVICE && off ==0 && size >=PAGE_SIZE ) -set_bit (PG_dcache_clean, &page->flags); +if (dir !=DMA_TO_DEVICE && size >=PAGE_SIZE ) { +unsigned long pfn; +size_t left = size; + + pfn =page_to_pfn (page) + off /PAGE_SIZE ; + off %=PAGE_SIZE ; +if (off) { + pfn++; + left -=PAGE_SIZE - off; + } +while (left >=PAGE_SIZE ) { + page =pfn_to_page (pfn++); +set_bit (PG_dcache_clean, &page->flags); + left -=PAGE_SIZE ; + } + } }/**
Signature · 1 line
1.7.9.5
PATCH 1/1 ARM: DMA-mapping: mark all !DMA_TO_DEVICE pages in unmapping as clean
Hello,
Quoted 48 lines · Ming Lei
It is common for one sg to include many pages, so mark all these pages as clean to avoid unnecessary flushing on them in
set_pte_at()orupdate_mmu_cache().The patch might improve loading performance of applciation code a bit.
On the below test code to read file(~1GByte size) from usb mass storage disk to buffer created with
mmap(PROT_READ | PROT_EXEC)on Pandaboard, average ~1% improvement can be observed with the patch on 10 times test.unsigned int sum = 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 *mbuffer; int fd; int i; unsigned long page_size, size; struct stat stat; struct timeval t1, t2; page_size = getpagesize(); fd = open(argv[1], O_RDONLY); assert(fd >= 0); fstat(fd, &stat); size = stat.st_size; printf("%s: file %s, file size %lu, page size %lu\n", argv[0], read_filename, size, page_size); gettimeofday(&t1, NULL); mbuffer = mmap(NULL, size, PROT_READ | PROT_EXEC, MAP_SHARED, fd, 0); for (i = 0 ; i < size ; i += page_size) sum += mbuffer[i]; munmap(mbuffer, page_size); gettimeofday(&t2, NULL); printf("\tread mmaped time: %luus\n", tv_diff(&t1, &t2)); close(fd); }
Quoted 5 lines
Acked-by Nicolas Pitre <[email protected]>Cc Catalin Marinas <[email protected]>Cc Marek Szyprowski <[email protected]>Cc Russell King <[email protected]>Signed-off-by Ming Lei <[email protected]>
Thanks for proving this patch. I'm really sorry for a late reply. I've applied it to my dma-mapping tree.
Quoted 29 lines
---
v1:
- fix one mistake on computing pfn, pointed out by Nicolasarch/arm/mm/dma-mapping.c | 20 +++++++++++++++++--- 1 file changed, 17 insertions(+), 3 deletions(-)arch/arm/mm/dma-mapping.c +10
-3@@ -880,10 +880,24 @@ static void __dma_page_dev_to_cpu(struct page *page, unsigned long off,dma_cache_maint_page (page, off, size, dir, dmac_unmap_area);/* -* Mark the D-cache clean for this page to avoid extra flushing. +* Mark the D-cache clean for these pages to avoid extra flushing. */ -if (dir !=DMA_TO_DEVICE && off ==0 && size >=PAGE_SIZE ) -set_bit (PG_dcache_clean, &page->flags); +if (dir !=DMA_TO_DEVICE && size >=PAGE_SIZE ) { +unsigned long pfn; +size_t left = size; + + pfn =page_to_pfn (page) + off /PAGE_SIZE ; + off %=PAGE_SIZE ; +if (off) { + pfn++; + left -=PAGE_SIZE - off;
arch/arm/mm/dma-mapping.c static void __dma_page_dev_to_cpu(struct page *page, unsigned long off, · line 903
Quoted 10 lines
arch/arm/mm/dma-mapping.c +7
-0+ } +while (left >=PAGE_SIZE ) { + page =pfn_to_page (pfn++); +set_bit (PG_dcache_clean, &page->flags); + left -=PAGE_SIZE ; + } + } }/**
Best regards
Signature · 2 lines
Marek Szyprowski Samsung R&D Institute Poland
PATCH 1/1 ARM: DMA-mapping: mark all !DMA_TO_DEVICE pages in unmapping as clean
Quoted 3 lines · Marek Szyprowski
Thanks for proving this patch. I'm really sorry for a late reply. I've applied it to my dma-mapping tree.
That is fine, and the patch has been applied in rmk's patch system.
Thanks,
--
Ming Lei