On 03/14/2013 08:39 PM, Hillf Danton wrote:
On Sat, Mar 9, 2013 at 10:14 AM, Will Huck <will.huckk@xxxxxxxxx> wrote:
Cc experts. Hugh, Johannes,
On 03/04/2013 08:21 PM, Lenky Gao wrote:
2013/3/4 Zlatko Calusic <zlatko.calusic@xxxxxxxx>:
The drop_caches mechanism doesn't free dirty page cache pages. And your
bash
script is creating a lot of dirty pages. Run it like this and see if it
helps your case:
sync; echo 3 > /proc/sys/vm/drop_caches
Thanks for your advice.
The inactive memory still cannot be reclaimed after i execute the sync
command:
# cat /proc/meminfo | grep Inactive\(file\);
Inactive(file): 882824 kB
# sync;
# echo 3 > /proc/sys/vm/drop_caches
# cat /proc/meminfo | grep Inactive\(file\);
Inactive(file): 777664 kB
I find these page becomes orphaned in this function, but do not understand
why:
/*
* If truncate cannot remove the fs-private metadata from the page, the
page
* becomes orphaned. It will be left on the LRU and may even be mapped
into
* user pagetables if we're racing with filemap_fault().
*
* We need to bale out if page->mapping is no longer equal to the
original
* mapping. This happens a) when the VM reclaimed the page while we
waited on
* its lock, b) when a concurrent invalidate_mapping_pages got there
first and
* c) when tmpfs swizzles a page between a tmpfs inode and swapper_space.
*/
static int
truncate_complete_page(struct address_space *mapping, struct page *page)
{
...
My file system type is ext3, mounted with the opteion data=journal and
it is easy to reproduce.
Perhaps we have to consider page count for orphan page if it
could be reproduced with mainline.
Why? /proc/sys/vm/drop_caches will call invalidate_mapping_pages()
instead of truncate_complete_page().
Hillf
---
--- a/mm/vmscan.c Sun Mar 10 13:36:26 2013
+++ b/mm/vmscan.c Thu Mar 14 20:29:40 2013
@@ -315,14 +315,14 @@ out:
return ret;
}
-static inline int is_page_cache_freeable(struct page *page)
+static inline int is_page_cache_freeable(struct page *page, int has_mapping)
{
/*
* A freeable page cache page is referenced only by the caller
* that isolated the page, the page cache radix tree and
* optional buffer heads at page->private.
*/
- return page_count(page) - page_has_private(page) == 2;
+ return page_count(page) - page_has_private(page) == has_mapping + 1;
}
page count 2 is for page cache and isolator, why you check mapping
separately?
static int may_write_to_queue(struct backing_dev_info *bdi,
@@ -393,7 +393,7 @@ static pageout_t pageout(struct page *pa
* swap_backing_dev_info is bust: it doesn't reflect the
* congestion state of the swapdevs. Easy to fix, if needed.
*/
- if (!is_page_cache_freeable(page))
+ if (!is_page_cache_freeable(page, mapping ? 1 : 0))
return PAGE_KEEP;
if (!mapping) {
/*
--
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>