[PATCH v2] memcg: Convert memcg->socket_pressure to u64.

[PATCH v2] memcg: Convert memcg->socket_pressure to u64.

[Kuniyuki Iwashima]

memcg->socket_pressure is initialised with jiffies when the memcg
is created.

Once vmpressure detects that the cgroup is under memory pressure,
the field is updated with jiffies + HZ to signal the fact to the
socket layer and suppress memory allocation for one second.

Otherwise, the field is not updated.

mem_cgroup_under_socket_pressure() uses time_before() to check if
jiffies is less than memcg->socket_pressure, and this has a bug on
32-bit kernel.

  if (time_before(jiffies, memcg->socket_pressure))
          return true;

As time_before() casts the final result to long, the acceptable delta
between two timestamps is 2 ^ (BITS_PER_LONG - 1).

On 32-bit kernel with CONFIG_HZ=1000, this is about 24 days.

  >>> (2 ** 31) / 1000 / 60 / 60 / 24
  24.855134814814818

Once 24 days have passed since the last update of socket_pressure,
mem_cgroup_under_socket_pressure() starts to lie until the next
24 days pass.

We don't need to worry about this on 64-bit machines unless they
serve for 300 million years.

  >>> (2 ** 63) / 1000 / 60 / 60 / 24 / 365
  292471208.6775361

Let's convert memcg->socket_pressure to u64.

Fixes: 8e8ae645249b8 ("mm: memcontrol: hook up vmpressure to socket pressure")
Reported-by: Neal Cardwell <ncardwell@google.com>
Suggested-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
---
v2: No delayed work, simply make socket_pressure u64

v1: https://lore.kernel.org/all/20250716042925.106239-1-kuniyu@google.com/T/#u
---
 include/linux/memcontrol.h | 44 +++++++++++++++++++++++++++++++++++---
 mm/memcontrol.c            |  5 ++++-
 mm/vmpressure.c            |  2 +-
 3 files changed, 46 insertions(+), 5 deletions(-)

diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index 87b6688f124a7..785173aa0739c 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -251,8 +251,10 @@ struct mem_cgroup {
 	 * that this indicator should NOT be used in legacy cgroup mode
 	 * where socket memory is accounted/charged separately.
 	 */
-	unsigned long		socket_pressure;
-
+	u64			socket_pressure;
+#if BITS_PER_LONG < 64
+	seqlock_t		socket_pressure_seqlock;
+#endif
 	int kmemcg_id;
 	/*
 	 * memcg->objcg is wiped out as a part of the objcg repaprenting
@@ -1602,6 +1604,42 @@ extern struct static_key_false memcg_sockets_enabled_key;
 #define mem_cgroup_sockets_enabled static_branch_unlikely(&memcg_sockets_enabled_key)
 void mem_cgroup_sk_alloc(struct sock *sk);
 void mem_cgroup_sk_free(struct sock *sk);
+
+#if BITS_PER_LONG < 64
+static inline void mem_cgroup_set_socket_pressure(struct mem_cgroup *memcg)
+{
+	u64 val = get_jiffies_64() + HZ;
+	unsigned long flags;
+
+	write_seqlock_irqsave(&memcg->socket_pressure_seqlock, flags);
+	memcg->socket_pressure = val;
+	write_sequnlock_irqrestore(&memcg->socket_pressure_seqlock, flags);
+}
+
+static inline u64 mem_cgroup_get_socket_pressure(struct mem_cgroup *memcg)
+{
+	unsigned int seq;
+	u64 val;
+
+	do {
+		seq = read_seqbegin(&memcg->socket_pressure_seqlock);
+		val = memcg->socket_pressure;
+	} while (read_seqretry(&memcg->socket_pressure_seqlock, seq));
+
+	return val;
+}
+#else
+static inline void mem_cgroup_set_socket_pressure(struct mem_cgroup *memcg)
+{
+	WRITE_ONCE(memcg->socket_pressure, jiffies + HZ);
+}
+
+static inline u64 mem_cgroup_get_socket_pressure(struct mem_cgroup *memcg)
+{
+	return READ_ONCE(memcg->socket_pressure);
+}
+#endif
+
 static inline bool mem_cgroup_under_socket_pressure(struct mem_cgroup *memcg)
 {
 #ifdef CONFIG_MEMCG_V1
@@ -1609,7 +1647,7 @@ static inline bool mem_cgroup_under_socket_pressure(struct mem_cgroup *memcg)
 		return !!memcg->tcpmem_pressure;
 #endif /* CONFIG_MEMCG_V1 */
 	do {
-		if (time_before(jiffies, READ_ONCE(memcg->socket_pressure)))
+		if (time_before64(get_jiffies_64(), mem_cgroup_get_socket_pressure(memcg)))
 			return true;
 	} while ((memcg = parent_mem_cgroup(memcg)));
 	return false;
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 902da8a9c643a..2900594e21766 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -3757,7 +3757,10 @@ static struct mem_cgroup *mem_cgroup_alloc(struct mem_cgroup *parent)
 	INIT_LIST_HEAD(&memcg->memory_peaks);
 	INIT_LIST_HEAD(&memcg->swap_peaks);
 	spin_lock_init(&memcg->peaks_lock);
-	memcg->socket_pressure = jiffies;
+	memcg->socket_pressure = get_jiffies_64();
+#if BITS_PER_LONG < 64
+	seqlock_init(&memcg->socket_pressure_seqlock);
+#endif
 	memcg1_memcg_init(memcg);
 	memcg->kmemcg_id = -1;
 	INIT_LIST_HEAD(&memcg->objcg_list);
diff --git a/mm/vmpressure.c b/mm/vmpressure.c
index bd5183dfd8791..c197ed47bcc45 100644
--- a/mm/vmpressure.c
+++ b/mm/vmpressure.c
@@ -316,7 +316,7 @@ void vmpressure(gfp_t gfp, struct mem_cgroup *memcg, bool tree,
 			 * asserted for a second in which subsequent
 			 * pressure events can occur.
 			 */
-			WRITE_ONCE(memcg->socket_pressure, jiffies + HZ);
+			mem_cgroup_set_socket_pressure(memcg);
 		}
 	}
 }
-- 
2.50.0.727.gbf7dc18ff4-goog

Re: [PATCH v2] memcg: Convert memcg->socket_pressure to u64.

[Shakeel Butt]

On Thu, Jul 17, 2025 at 07:46:43PM +0000, Kuniyuki Iwashima wrote:
> memcg->socket_pressure is initialised with jiffies when the memcg
> is created.
> 
> Once vmpressure detects that the cgroup is under memory pressure,
> the field is updated with jiffies + HZ to signal the fact to the
> socket layer and suppress memory allocation for one second.
> 
> Otherwise, the field is not updated.
> 
> mem_cgroup_under_socket_pressure() uses time_before() to check if
> jiffies is less than memcg->socket_pressure, and this has a bug on
> 32-bit kernel.
> 
>   if (time_before(jiffies, memcg->socket_pressure))
>           return true;
> 
> As time_before() casts the final result to long, the acceptable delta
> between two timestamps is 2 ^ (BITS_PER_LONG - 1).
> 
> On 32-bit kernel with CONFIG_HZ=1000, this is about 24 days.
> 
>   >>> (2 ** 31) / 1000 / 60 / 60 / 24
>   24.855134814814818
> 
> Once 24 days have passed since the last update of socket_pressure,
> mem_cgroup_under_socket_pressure() starts to lie until the next
> 24 days pass.
> 
> We don't need to worry about this on 64-bit machines unless they
> serve for 300 million years.
> 
>   >>> (2 ** 63) / 1000 / 60 / 60 / 24 / 365
>   292471208.6775361
> 
> Let's convert memcg->socket_pressure to u64.
> 
> Fixes: 8e8ae645249b8 ("mm: memcontrol: hook up vmpressure to socket pressure")
> Reported-by: Neal Cardwell <ncardwell@google.com>
> Suggested-by: Andrew Morton <akpm@linux-foundation.org>
> Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>

Acked-by: Shakeel Butt <shakeel.butt@linux.dev>

Re: [PATCH v2] memcg: Convert memcg->socket_pressure to u64.

[Johannes Weiner]

On Thu, Jul 17, 2025 at 07:46:43PM +0000, Kuniyuki Iwashima wrote:
> memcg->socket_pressure is initialised with jiffies when the memcg
> is created.
> 
> Once vmpressure detects that the cgroup is under memory pressure,
> the field is updated with jiffies + HZ to signal the fact to the
> socket layer and suppress memory allocation for one second.
> 
> Otherwise, the field is not updated.
> 
> mem_cgroup_under_socket_pressure() uses time_before() to check if
> jiffies is less than memcg->socket_pressure, and this has a bug on
> 32-bit kernel.
> 
>   if (time_before(jiffies, memcg->socket_pressure))
>           return true;
> 
> As time_before() casts the final result to long, the acceptable delta
> between two timestamps is 2 ^ (BITS_PER_LONG - 1).
> 
> On 32-bit kernel with CONFIG_HZ=1000, this is about 24 days.
> 
>   >>> (2 ** 31) / 1000 / 60 / 60 / 24
>   24.855134814814818
> 
> Once 24 days have passed since the last update of socket_pressure,
> mem_cgroup_under_socket_pressure() starts to lie until the next
> 24 days pass.
> 
> We don't need to worry about this on 64-bit machines unless they
> serve for 300 million years.
> 
>   >>> (2 ** 63) / 1000 / 60 / 60 / 24 / 365
>   292471208.6775361
> 
> Let's convert memcg->socket_pressure to u64.
> 
> Fixes: 8e8ae645249b8 ("mm: memcontrol: hook up vmpressure to socket pressure")
> Reported-by: Neal Cardwell <ncardwell@google.com>
> Suggested-by: Andrew Morton <akpm@linux-foundation.org>
> Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
> ---
> v2: No delayed work, simply make socket_pressure u64
> 
> v1: https://lore.kernel.org/all/20250716042925.106239-1-kuniyu@google.com/T/#u
> ---

Acked-by: Johannes Weiner <hannes@cmpxchg.org>