mm/page_alloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
In the current code, batch is a local variable, and it cannot be
concurrently modified. It's unnecessary to use READ_ONCE here,
so remove it.
Fixes: 51a755c56dc0 ("mm: tune PCP high automatically")
Signed-off-by: Songtang Liu <liusongtang@bytedance.com>
---
mm/page_alloc.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index e3ea5bf5c459..6edc6e57d4f8 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2661,7 +2661,7 @@ static void free_frozen_page_commit(struct zone *zone,
free_high = (pcp->free_count >= batch &&
(pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) &&
(!(pcp->flags & PCPF_FREE_HIGH_BATCH) ||
- pcp->count >= READ_ONCE(batch)));
+ pcp->count >= batch));
pcp->flags |= PCPF_PREV_FREE_HIGH_ORDER;
} else if (pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) {
pcp->flags &= ~PCPF_PREV_FREE_HIGH_ORDER;
--
2.20.1
On Wed, Apr 02, 2025 at 03:41:16AM -0400, Songtang Liu wrote:
> In the current code, batch is a local variable, and it cannot be
> concurrently modified. It's unnecessary to use READ_ONCE here,
> so remove it.
>
> Fixes: 51a755c56dc0 ("mm: tune PCP high automatically")
> Signed-off-by: Songtang Liu <liusongtang@bytedance.com>
Reviewed-by: Vishal Moola (Oracle) <vishal.moola@gmail.com>
On Wed, Apr 02, 2025 at 03:41:16AM -0400, Songtang Liu wrote:
> In the current code, batch is a local variable, and it cannot be
> concurrently modified. It's unnecessary to use READ_ONCE here,
> so remove it.
>
> Fixes: 51a755c56dc0 ("mm: tune PCP high automatically")
> Signed-off-by: Songtang Liu <liusongtang@bytedance.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
--
Oscar Salvador
SUSE Labs
> In the current code, batch is a local variable, and it cannot be
> concurrently modified. It's unnecessary to use READ_ONCE here,
> so remove it.
>
> Fixes: 51a755c56dc0 ("mm: tune PCP high automatically")
> Signed-off-by: Songtang Liu <liusongtang@bytedance.com>
Reviewed-by: Pankaj Gupta <pankaj.gupta@amd.com>
> ---
> mm/page_alloc.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index e3ea5bf5c459..6edc6e57d4f8 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -2661,7 +2661,7 @@ static void free_frozen_page_commit(struct zone *zone,
> free_high = (pcp->free_count >= batch &&
> (pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) &&
> (!(pcp->flags & PCPF_FREE_HIGH_BATCH) ||
> - pcp->count >= READ_ONCE(batch)));
> + pcp->count >= batch));
> pcp->flags |= PCPF_PREV_FREE_HIGH_ORDER;
> } else if (pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) {
> pcp->flags &= ~PCPF_PREV_FREE_HIGH_ORDER;
On 4/2/25 13:11, Songtang Liu wrote:
> In the current code, batch is a local variable, and it cannot be
> concurrently modified. It's unnecessary to use READ_ONCE here,
> so remove it.
>
> Fixes: 51a755c56dc0 ("mm: tune PCP high automatically")
Although READ_ONCE() should not have been used here with the above
commit - does this cause any problem though ? So in other words is
the Fixes: tag necessary here ?
> Signed-off-by: Songtang Liu <liusongtang@bytedance.com>
> ---
> mm/page_alloc.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index e3ea5bf5c459..6edc6e57d4f8 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -2661,7 +2661,7 @@ static void free_frozen_page_commit(struct zone *zone,
> free_high = (pcp->free_count >= batch &&
> (pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) &&
> (!(pcp->flags & PCPF_FREE_HIGH_BATCH) ||
> - pcp->count >= READ_ONCE(batch)));
> + pcp->count >= batch));
> pcp->flags |= PCPF_PREV_FREE_HIGH_ORDER;
> } else if (pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) {
> pcp->flags &= ~PCPF_PREV_FREE_HIGH_ORDER;
Otherwise LGTM
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Songtang Liu <liusongtang@bytedance.com> writes:
> In the current code, batch is a local variable, and it cannot be
> concurrently modified. It's unnecessary to use READ_ONCE here,
> so remove it.
>
> Fixes: 51a755c56dc0 ("mm: tune PCP high automatically")
> Signed-off-by: Songtang Liu <liusongtang@bytedance.com>
> ---
> mm/page_alloc.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index e3ea5bf5c459..6edc6e57d4f8 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -2661,7 +2661,7 @@ static void free_frozen_page_commit(struct zone *zone,
> free_high = (pcp->free_count >= batch &&
> (pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) &&
> (!(pcp->flags & PCPF_FREE_HIGH_BATCH) ||
> - pcp->count >= READ_ONCE(batch)));
> + pcp->count >= batch));
Nice catch!
Reviewed-by: Huang Ying <ying.huang@linux.alibaba.com>
> pcp->flags |= PCPF_PREV_FREE_HIGH_ORDER;
> } else if (pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) {
> pcp->flags &= ~PCPF_PREV_FREE_HIGH_ORDER;
---
Best Regards,
Huang, Ying
On 4/2/25 3:41 PM, Songtang Liu wrote:
> In the current code, batch is a local variable, and it cannot be
> concurrently modified. It's unnecessary to use READ_ONCE here,
> so remove it.
>
> Fixes: 51a755c56dc0 ("mm: tune PCP high automatically")
> Signed-off-by: Songtang Liu <liusongtang@bytedance.com>
> ---
> mm/page_alloc.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index e3ea5bf5c459..6edc6e57d4f8 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -2661,7 +2661,7 @@ static void free_frozen_page_commit(struct zone *zone,
> free_high = (pcp->free_count >= batch &&
> (pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) &&
> (!(pcp->flags & PCPF_FREE_HIGH_BATCH) ||
> - pcp->count >= READ_ONCE(batch)));
> + pcp->count >= batch));
> pcp->flags |= PCPF_PREV_FREE_HIGH_ORDER;
> } else if (pcp->flags & PCPF_PREV_FREE_HIGH_ORDER) {
> pcp->flags &= ~PCPF_PREV_FREE_HIGH_ORDER;
Reviewed-by: Qi Zheng <zhengqi.arch@bytedance.com>
Thanks!
© 2016 - 2026 Red Hat, Inc.