Re: [PATCH] mm: Don't use radix tree writeback tags for pages in swap cache

[Rik van Riel]

On Thu, 2016-08-25 at 12:27 -0700, Huang, Ying wrote:
> File pages use a set of radix tags (DIRTY, TOWRITE, WRITEBACK, etc.)
> to
> accelerate finding the pages with a specific tag in the radix tree
> during inode writeback.ÂÂBut for anonymous pages in the swap cache,
> there is no inode writeback.ÂÂSo there is no need to find the
> pages with some writeback tags in the radix tree.ÂÂIt is not
> necessary
> to touch radix tree writeback tags for pages in the swap cache.
> 
> With this patch, the swap out bandwidth improved 22.3% (from ~1.2GB/s
> to
> ~ 1.48GBps) in the vm-scalability swap-w-seq test case with 8
> processes.
> The test is done on a Xeon E5 v3 system.ÂÂThe swap device used is a
> RAM
> simulated PMEM (persistent memory) device.ÂÂThe improvement comes
> from
> the reduced contention on the swap cache radix tree lock.ÂÂTo test
> sequential swapping out, the test case uses 8 processes, which
> sequentially allocate and write to the anonymous pages until RAM and
> part of the swap device is used up.
> 
> Details of comparison is as follow,
> 
> baseÂÂÂÂÂÂÂÂÂÂÂÂÂbase+patch
> ---------------- --------------------------
> ÂÂÂÂÂÂÂÂÂ%stddevÂÂÂÂÂ%changeÂÂÂÂÂÂÂÂÂ%stddev
> ÂÂÂÂÂÂÂÂÂÂÂÂÂ\ÂÂÂÂÂÂÂÂÂÂ|ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ\
> ÂÂÂ1207402 ÂÂÂ7%ÂÂÂÂÂ+22.3%ÂÂÂÂ1476578 ÂÂÂ6%ÂÂvmstat.swap.so
> ÂÂÂ2506952 ÂÂÂ2%ÂÂÂÂÂ+28.1%ÂÂÂÂ3212076 ÂÂÂ7%ÂÂvm-
> scalability.throughput
> ÂÂÂÂÂ10.86 Â 12%ÂÂÂÂÂ-23.4%ÂÂÂÂÂÂÂ8.31 Â 16%ÂÂperf-profile.cycles-
> pp._raw_spin_lock_irq.__add_to_swap_cache.add_to_swap_cache.add_to_sw
> ap.shrink_page_list
> ÂÂÂÂÂ10.82 Â 13%ÂÂÂÂÂ-33.1%ÂÂÂÂÂÂÂ7.24 Â 14%ÂÂperf-profile.cycles-
> pp._raw_spin_lock_irqsave.__remove_mapping.shrink_page_list.shrink_in
> active_list.shrink_zone_memcg
> ÂÂÂÂÂ10.36 Â 11%ÂÂÂÂ-100.0%ÂÂÂÂÂÂÂ0.00 Â -1%ÂÂperf-profile.cycles-
> pp._raw_spin_lock_irqsave.__test_set_page_writeback.bdev_write_page._
> _swap_writepage.swap_writepage
> ÂÂÂÂÂ10.52 Â 12%ÂÂÂÂ-100.0%ÂÂÂÂÂÂÂ0.00 Â -1%ÂÂperf-profile.cycles-
> pp._raw_spin_lock_irqsave.test_clear_page_writeback.end_page_writebac
> k.page_endio.pmem_rw_page
> 
> Cc: Hugh Dickins <hughd@xxxxxxxxxx>
> Cc: Shaohua Li <shli@xxxxxxxxxx>
> Cc: Minchan Kim <minchan@xxxxxxxxxx>
> Cc: Rik van Riel <riel@xxxxxxxxxx>
> Cc: Mel Gorman <mgorman@xxxxxxxxxxxxxxxxxxx>
> Cc: Tejun Heo <tj@xxxxxxxxxx>
> Cc: Wu Fengguang <fengguang.wu@xxxxxxxxx>
> Cc: Dave Hansen <dave.hansen@xxxxxxxxx>
> Signed-off-by: "Huang, Ying" <ying.huang@xxxxxxxxx>
> ---
> Âmm/page-writeback.c | 6 ++++--
> Â1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/page-writeback.c b/mm/page-writeback.c
> index 82e7252..599d2f9 100644
> --- a/mm/page-writeback.c
> +++ b/mm/page-writeback.c
> @@ -2728,7 +2728,8 @@ int test_clear_page_writeback(struct page
> *page)
> Â	int ret;
> Â
> Â	lock_page_memcg(page);
> -	if (mapping) {
> +	/* Pages in swap cache don't use writeback tags */
> +	if (mapping && !PageSwapCache(page)) {

I wonder if that should be a mapping_uses_tags(mapping)
macro or similar, and a per-mapping flag?

I suspect there will be another case coming up soon
where we have a page cache radix tree, but no need
for dirty/writeback/... tags.

That use case would be DAX filesystems, where we do
use a struct page, but that struct page points at
persistent storage, and the tags are not necessary.