The fill_page_cache_func() is invoked and start from zero to
allocate nr_pages of page, if the kfree_rcu_work() executes before
the fill_page_cache_func(), the krcp->nr_bkv_objs is updated before
enter for-loop and equal to nr_pages, since the page is allocated
first, and then check the krcp->nr_bkv_objs in put_cached_bnode(),
this produces a meaningless __get_free_page() call, this commit
therefore make allocate page start from krcp->nr_bkv_objs and
check krcp->nr_bkv_objs before allocate page.
Signed-off-by: Zqiang <[email protected]>
---
kernel/rcu/tree.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 41daae3239b5..dcb86f9f2dd6 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -3223,7 +3223,7 @@ static void fill_page_cache_func(struct work_struct *work)
nr_pages = atomic_read(&krcp->backoff_page_cache_fill) ?
1 : rcu_min_cached_objs;
- for (i = 0; i < nr_pages; i++) {
+ for (i = READ_ONCE(krcp->nr_bkv_objs); i < nr_pages; i++) {
bnode = (struct kvfree_rcu_bulk_data *)
__get_free_page(GFP_KERNEL | __GFP_NORETRY | __GFP_NOMEMALLOC | __GFP_NOWARN);
--
2.32.0
On Wed, Apr 12, 2023 at 10:31:27PM +0800, Zqiang wrote:
> The fill_page_cache_func() is invoked and start from zero to
> allocate nr_pages of page, if the kfree_rcu_work() executes before
> the fill_page_cache_func(), the krcp->nr_bkv_objs is updated before
> enter for-loop and equal to nr_pages, since the page is allocated
> first, and then check the krcp->nr_bkv_objs in put_cached_bnode(),
> this produces a meaningless __get_free_page() call, this commit
> therefore make allocate page start from krcp->nr_bkv_objs and
> check krcp->nr_bkv_objs before allocate page.
>
> Signed-off-by: Zqiang <[email protected]>
> ---
> kernel/rcu/tree.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index 41daae3239b5..dcb86f9f2dd6 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -3223,7 +3223,7 @@ static void fill_page_cache_func(struct work_struct *work)
> nr_pages = atomic_read(&krcp->backoff_page_cache_fill) ?
> 1 : rcu_min_cached_objs;
>
> - for (i = 0; i < nr_pages; i++) {
> + for (i = READ_ONCE(krcp->nr_bkv_objs); i < nr_pages; i++) {
> bnode = (struct kvfree_rcu_bulk_data *)
> __get_free_page(GFP_KERNEL | __GFP_NORETRY | __GFP_NOMEMALLOC | __GFP_NOWARN);
>
> --
> 2.32.0
>
Reviewed-by: Uladzislau Rezki (Sony) <[email protected]>
--
Uladzislau Rezki
On Thu, Apr 13, 2023 at 11:07:09AM +0200, Uladzislau Rezki wrote:
> On Wed, Apr 12, 2023 at 10:31:27PM +0800, Zqiang wrote:
> > The fill_page_cache_func() is invoked and start from zero to
> > allocate nr_pages of page, if the kfree_rcu_work() executes before
> > the fill_page_cache_func(), the krcp->nr_bkv_objs is updated before
> > enter for-loop and equal to nr_pages, since the page is allocated
> > first, and then check the krcp->nr_bkv_objs in put_cached_bnode(),
> > this produces a meaningless __get_free_page() call, this commit
> > therefore make allocate page start from krcp->nr_bkv_objs and
> > check krcp->nr_bkv_objs before allocate page.
> >
> > Signed-off-by: Zqiang <[email protected]>
> > ---
> > kernel/rcu/tree.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> > index 41daae3239b5..dcb86f9f2dd6 100644
> > --- a/kernel/rcu/tree.c
> > +++ b/kernel/rcu/tree.c
> > @@ -3223,7 +3223,7 @@ static void fill_page_cache_func(struct work_struct *work)
> > nr_pages = atomic_read(&krcp->backoff_page_cache_fill) ?
> > 1 : rcu_min_cached_objs;
> >
> > - for (i = 0; i < nr_pages; i++) {
> > + for (i = READ_ONCE(krcp->nr_bkv_objs); i < nr_pages; i++) {
> > bnode = (struct kvfree_rcu_bulk_data *)
> > __get_free_page(GFP_KERNEL | __GFP_NORETRY | __GFP_NOMEMALLOC | __GFP_NOWARN);
> >
> > --
> > 2.32.0
> >
> Reviewed-by: Uladzislau Rezki (Sony) <[email protected]>
Queued and pushed, thank you both!
Thanx, Paul