2020-12-17 06:31:20

by Alex Shi

Subject: [PATCH 3/3] mm/compaction: remove rcu_read_lock during page compaction

rcu_read_lock was used to guard memcg destory, now TestClearPageLRU
could block this happen, so we don't need it. Remove it to reduce
locking load in debugging mode.

Signed-off-by: Alex Shi <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: [email protected]
Cc: [email protected]
---
mm/compaction.c | 5 +----
1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/mm/compaction.c b/mm/compaction.c
index 8049d3530812..02af220fb992 100644
--- a/mm/compaction.c
+++ b/mm/compaction.c
@@ -995,7 +995,6 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
if (!TestClearPageLRU(page))
goto isolate_fail_put;

- rcu_read_lock();
lruvec = mem_cgroup_page_lruvec(page, pgdat);

/* If we already hold the lock, we can skip some rechecking */
@@ -1005,7 +1004,6 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,

compact_lock_irqsave(&lruvec->lru_lock, &flags, cc);
locked = lruvec;
- rcu_read_unlock();

lruvec_memcg_debug(lruvec, page);

@@ -1026,8 +1024,7 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
SetPageLRU(page);
goto isolate_fail_put;
}
- } else
- rcu_read_unlock();
+ }

/* The whole page is taken off the LRU; skip the tail pages. */
if (PageCompound(page))
--
2.29.GIT

2020-12-22 03:51:23

by Hugh Dickins

Subject: Re: [PATCH 3/3] mm/compaction: remove rcu_read_lock during page compaction

On Thu, 17 Dec 2020, Alex Shi wrote:

> rcu_read_lock was used to guard memcg destory, now TestClearPageLRU
> could block this happen, so we don't need it. Remove it to reduce
> locking load in debugging mode.

"
isolate_migratepages_block() used rcu_read_lock() with the intention
of safeguarding against the mem_cgroup being destroyed concurrently;
but its TestClearPageLRU already protects against that. Delete the
unnecessary rcu_read_lock() and _unlock().
"

>
> Signed-off-by: Alex Shi <[email protected]>
> Cc: Hugh Dickins <[email protected]>

Acked-by: Hugh Dickins <[email protected]>

> Cc: Johannes Weiner <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: [email protected]
> Cc: [email protected]
> ---
> mm/compaction.c | 5 +----
> 1 file changed, 1 insertion(+), 4 deletions(-)
>
> diff --git a/mm/compaction.c b/mm/compaction.c
> index 8049d3530812..02af220fb992 100644
> --- a/mm/compaction.c
> +++ b/mm/compaction.c
> @@ -995,7 +995,6 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
> if (!TestClearPageLRU(page))
> goto isolate_fail_put;
>
> - rcu_read_lock();
> lruvec = mem_cgroup_page_lruvec(page, pgdat);
>
> /* If we already hold the lock, we can skip some rechecking */
> @@ -1005,7 +1004,6 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
>
> compact_lock_irqsave(&lruvec->lru_lock, &flags, cc);
> locked = lruvec;
> - rcu_read_unlock();
>
> lruvec_memcg_debug(lruvec, page);
>
> @@ -1026,8 +1024,7 @@ isolate_migratepages_block(struct compact_control *cc, unsigned long low_pfn,
> SetPageLRU(page);
> goto isolate_fail_put;
> }
> - } else
> - rcu_read_unlock();
> + }
>
> /* The whole page is taken off the LRU; skip the tail pages. */
> if (PageCompound(page))
> --
> 2.29.GIT