| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <1476967085-89647-1-git-send-email-zhouxianrong@huawei.com>
Date: Thu, 20 Oct 2016 20:38:05 +0800
From: <zhouxianrong@...wei.com>
To: <linux-mm@...ck.org>
CC: <linux-kernel@...r.kernel.org>, <akpm@...ux-foundation.org>,
<viro@...iv.linux.org.uk>, <mingo@...hat.com>,
<peterz@...radead.org>, <hannes@...xchg.org>,
<mgorman@...hsingularity.net>, <vbabka@...e.cz>, <mhocko@...e.com>,
<vdavydov.dev@...il.com>, <minchan@...nel.org>, <riel@...hat.com>,
<zhouxianrong@...wei.com>, <zhouxiyu@...wei.com>,
<zhangshiming5@...wei.com>, <won.ho.park@...wei.com>,
<tuxiaobing@...wei.com>
Subject: [PATCH] bdi flusher should not be throttled here when it fall into buddy slow path
From: z00281421 <z00281421@...esmail.huawei.com>
The bdi flusher should be throttled only depends on
own bdi and is decoupled with others.
separate PGDAT_WRITEBACK into PGDAT_ANON_WRITEBACK and
PGDAT_FILE_WRITEBACK avoid scanning anon lru and it is ok
then throttled on file WRITEBACK.
i think above may be not right.
Signed-off-by: z00281421 <z00281421@...esmail.huawei.com>
---
fs/fs-writeback.c | 8 ++++++--
include/linux/mmzone.h | 7 +++++--
mm/vmscan.c | 20 ++++++++++++--------
3 files changed, 23 insertions(+), 12 deletions(-)
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 05713a5..ddcc70f 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -1905,10 +1905,13 @@ void wb_workfn(struct work_struct *work)
{
struct bdi_writeback *wb = container_of(to_delayed_work(work),
struct bdi_writeback, dwork);
+ struct backing_dev_info *bdi = container_of(to_delayed_work(work),
+ struct backing_dev_info, wb.dwork);
long pages_written;
set_worker_desc("flush-%s", dev_name(wb->bdi->dev));
- current->flags |= PF_SWAPWRITE;
+ current->flags |= (PF_SWAPWRITE | PF_LESS_THROTTLE);
+ current->bdi = bdi;
if (likely(!current_is_workqueue_rescuer() ||
!test_bit(WB_registered, &wb->state))) {
@@ -1938,7 +1941,8 @@ void wb_workfn(struct work_struct *work)
else if (wb_has_dirty_io(wb) && dirty_writeback_interval)
wb_wakeup_delayed(wb);
- current->flags &= ~PF_SWAPWRITE;
+ current->bdi = NULL;
+ current->flags &= ~(PF_SWAPWRITE | PF_LESS_THROTTLE);
}
/*
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 7f2ae99..fa602e9 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -528,8 +528,11 @@ enum pgdat_flags {
* many dirty file pages at the tail
* of the LRU.
*/
- PGDAT_WRITEBACK, /* reclaim scanning has recently found
- * many pages under writeback
+ PGDAT_ANON_WRITEBACK, /* reclaim scanning has recently found
+ * many anonymous pages under writeback
+ */
+ PGDAT_FILE_WRITEBACK, /* reclaim scanning has recently found
+ * many file pages under writeback
*/
PGDAT_RECLAIM_LOCKED, /* prevents concurrent reclaim */
};
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 0fe8b71..3f08ba3 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -917,6 +917,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
unsigned long nr_reclaimed = 0;
unsigned long nr_writeback = 0;
unsigned long nr_immediate = 0;
+ int file;
cond_resched();
@@ -954,6 +955,8 @@ static unsigned long shrink_page_list(struct list_head *page_list,
may_enter_fs = (sc->gfp_mask & __GFP_FS) ||
(PageSwapCache(page) && (sc->gfp_mask & __GFP_IO));
+ file = page_is_file_cache(page)
+
/*
* The number of dirty pages determines if a zone is marked
* reclaim_congested which affects wait_iff_congested. kswapd
@@ -1016,7 +1019,7 @@ static unsigned long shrink_page_list(struct list_head *page_list,
/* Case 1 above */
if (current_is_kswapd() &&
PageReclaim(page) &&
- test_bit(PGDAT_WRITEBACK, &pgdat->flags)) {
+ test_bit(PGDAT_ANON_WRITEBACK + file, &pgdat->flags)) {
nr_immediate++;
goto keep_locked;
@@ -1643,13 +1646,14 @@ putback_inactive_pages(struct lruvec *lruvec, struct list_head *page_list)
* If a kernel thread (such as nfsd for loop-back mounts) services
* a backing device by writing to the page cache it sets PF_LESS_THROTTLE.
* In that case we should only throttle if the backing device it is
- * writing to is congested. In other cases it is safe to throttle.
+ * writing to is congested. The bdi flusher should be throttled only depends
+ * on own bdi and is decoupled with others. In other cases it is safe to throttle.
*/
-static int current_may_throttle(void)
+static int current_may_throttle(int file)
{
return !(current->flags & PF_LESS_THROTTLE) ||
- current->backing_dev_info == NULL ||
- bdi_write_congested(current->backing_dev_info);
+ (file && (current->backing_dev_info == NULL ||
+ bdi_write_congested(current->backing_dev_info)));
}
static bool inactive_reclaimable_pages(struct lruvec *lruvec,
@@ -1774,7 +1778,7 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
* are encountered in the nr_immediate check below.
*/
if (nr_writeback && nr_writeback == nr_taken)
- set_bit(PGDAT_WRITEBACK, &pgdat->flags);
+ set_bit(PGDAT_ANON_WRITEBACK + file, &pgdat->flags);
/*
* Legacy memcg will stall in page writeback so avoid forcibly
@@ -1803,7 +1807,7 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
* that pages are cycling through the LRU faster than
* they are written so also forcibly stall.
*/
- if (nr_immediate && current_may_throttle())
+ if (nr_immediate && current_may_throttle(file))
congestion_wait(BLK_RW_ASYNC, HZ/10);
}
@@ -1813,7 +1817,7 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
* unqueued dirty pages or cycling through the LRU too quickly.
*/
if (!sc->hibernation_mode && !current_is_kswapd() &&
- current_may_throttle())
+ current_may_throttle(file))
wait_iff_congested(pgdat, BLK_RW_ASYNC, HZ/10);
trace_mm_vmscan_lru_shrink_inactive(pgdat->node_id,
--
1.7.9.5
Powered by blists - more mailing lists