From: Shakeel Butt <shakeel.butt@linux.dev>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>,
Michal Hocko <mhocko@kernel.org>,
Roman Gushchin <roman.gushchin@linux.dev>,
Muchun Song <muchun.song@linux.dev>,
Vlastimil Babka <vbabka@suse.cz>,
Alexei Starovoitov <ast@kernel.org>,
Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
Harry Yoo <harry.yoo@oracle.com>,
Yosry Ahmed <yosry.ahmed@linux.dev>,
bpf@vger.kernel.org, linux-mm@kvack.org, cgroups@vger.kernel.org,
linux-kernel@vger.kernel.org,
Meta kernel team <kernel-team@meta.com>
Subject: [PATCH 7/7] memcg: objcg stock trylock without irq disabling
Date: Tue, 13 May 2025 22:08:13 -0700 [thread overview]
Message-ID: <20250514050813.2526843-8-shakeel.butt@linux.dev> (raw)
In-Reply-To: <20250514050813.2526843-1-shakeel.butt@linux.dev>
There is no need to disable irqs to use objcg per-cpu stock, so let's
just not do that but consume_obj_stock() and refill_obj_stock() will
need to use trylock instead to avoid deadlock against irq. One
consequence of this change is that the charge request from irq context
may take slowpath more often but it should be rare.
Signed-off-by: Shakeel Butt <shakeel.butt@linux.dev>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
---
mm/memcontrol.c | 25 +++++++++++++++----------
1 file changed, 15 insertions(+), 10 deletions(-)
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 78a41378b8f3..73b19137901a 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1907,18 +1907,17 @@ static void drain_local_memcg_stock(struct work_struct *dummy)
static void drain_local_obj_stock(struct work_struct *dummy)
{
struct obj_stock_pcp *stock;
- unsigned long flags;
if (WARN_ONCE(!in_task(), "drain in non-task context"))
return;
- local_lock_irqsave(&obj_stock.lock, flags);
+ local_lock(&obj_stock.lock);
stock = this_cpu_ptr(&obj_stock);
drain_obj_stock(stock);
clear_bit(FLUSHING_CACHED_CHARGE, &stock->flags);
- local_unlock_irqrestore(&obj_stock.lock, flags);
+ local_unlock(&obj_stock.lock);
}
static void refill_stock(struct mem_cgroup *memcg, unsigned int nr_pages)
@@ -2901,10 +2900,10 @@ static bool consume_obj_stock(struct obj_cgroup *objcg, unsigned int nr_bytes,
struct pglist_data *pgdat, enum node_stat_item idx)
{
struct obj_stock_pcp *stock;
- unsigned long flags;
bool ret = false;
- local_lock_irqsave(&obj_stock.lock, flags);
+ if (!local_trylock(&obj_stock.lock))
+ return ret;
stock = this_cpu_ptr(&obj_stock);
if (objcg == READ_ONCE(stock->cached_objcg) && stock->nr_bytes >= nr_bytes) {
@@ -2915,7 +2914,7 @@ static bool consume_obj_stock(struct obj_cgroup *objcg, unsigned int nr_bytes,
__account_obj_stock(objcg, stock, nr_bytes, pgdat, idx);
}
- local_unlock_irqrestore(&obj_stock.lock, flags);
+ local_unlock(&obj_stock.lock);
return ret;
}
@@ -3004,10 +3003,16 @@ static void refill_obj_stock(struct obj_cgroup *objcg, unsigned int nr_bytes,
enum node_stat_item idx)
{
struct obj_stock_pcp *stock;
- unsigned long flags;
unsigned int nr_pages = 0;
- local_lock_irqsave(&obj_stock.lock, flags);
+ if (!local_trylock(&obj_stock.lock)) {
+ if (pgdat)
+ mod_objcg_mlstate(objcg, pgdat, idx, nr_bytes);
+ nr_pages = nr_bytes >> PAGE_SHIFT;
+ nr_bytes = nr_bytes & (PAGE_SIZE - 1);
+ atomic_add(nr_bytes, &objcg->nr_charged_bytes);
+ goto out;
+ }
stock = this_cpu_ptr(&obj_stock);
if (READ_ONCE(stock->cached_objcg) != objcg) { /* reset if necessary */
@@ -3029,8 +3034,8 @@ static void refill_obj_stock(struct obj_cgroup *objcg, unsigned int nr_bytes,
stock->nr_bytes &= (PAGE_SIZE - 1);
}
- local_unlock_irqrestore(&obj_stock.lock, flags);
-
+ local_unlock(&obj_stock.lock);
+out:
if (nr_pages)
obj_cgroup_uncharge_pages(objcg, nr_pages);
}
--
2.47.1
prev parent reply other threads:[~2025-05-14 5:09 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-05-14 5:08 [PATCH 0/7] memcg: make memcg stats irq safe Shakeel Butt
2025-05-14 5:08 ` [PATCH 1/7] memcg: memcg_rstat_updated re-entrant safe against irqs Shakeel Butt
2025-05-14 9:50 ` Vlastimil Babka
2025-05-14 16:39 ` Shakeel Butt
2025-05-14 5:08 ` [PATCH 2/7] memcg: move preempt disable to callers of memcg_rstat_updated Shakeel Butt
2025-05-14 5:08 ` [PATCH 3/7] memcg: make mod_memcg_state re-entrant safe against irqs Shakeel Butt
2025-05-14 5:08 ` [PATCH 4/7] memcg: make count_memcg_events " Shakeel Butt
2025-05-14 5:08 ` [PATCH 5/7] memcg: make __mod_memcg_lruvec_state " Shakeel Butt
2025-05-14 5:08 ` [PATCH 6/7] memcg: no stock lock for cpu hot-unplug Shakeel Butt
2025-05-14 5:08 ` Shakeel Butt [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20250514050813.2526843-8-shakeel.butt@linux.dev \
--to=shakeel.butt@linux.dev \
--cc=akpm@linux-foundation.org \
--cc=ast@kernel.org \
--cc=bigeasy@linutronix.de \
--cc=bpf@vger.kernel.org \
--cc=cgroups@vger.kernel.org \
--cc=hannes@cmpxchg.org \
--cc=harry.yoo@oracle.com \
--cc=kernel-team@meta.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mhocko@kernel.org \
--cc=muchun.song@linux.dev \
--cc=roman.gushchin@linux.dev \
--cc=vbabka@suse.cz \
--cc=yosry.ahmed@linux.dev \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox