[PATCH v1 0/1] mm: Fix OOM killer and proc stats inaccuracy on large many-core systems

[PATCH v1 0/1] mm: Fix OOM killer and proc stats inaccuracy on large many-core systems

[Mathieu Desnoyers]

Hi Andrew,

This patch modifies the OOM killer and all proc RSS stats to use the
precise for-each-possible-cpu sum to fix the inaccuracy issues. This
approach was suggested by Michal Hocko as a straightforward fix for the
inaccuracy issue by using more precise (but slower) RSS stats sum.

With this, the hierarchical per-cpu counters become a simple
optimization rather than a bug fix. I will post a new version
of the HPCC soon which will be based on this patch.

Feedback is welcome!

Thanks,

Mathieu

Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Dennis Zhou <dennis@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Martin Liu <liumartin@google.com>
Cc: David Rientjes <rientjes@google.com>
Cc: christian.koenig@amd.com
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: SeongJae Park <sj@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: "Liam R . Howlett" <liam.howlett@oracle.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-mm@kvack.org
Cc: stable@vger.kernel.org
Cc: linux-trace-kernel@vger.kernel.org
Cc: Yu Zhao <yuzhao@google.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Mateusz Guzik <mjguzik@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Aboorva Devarajan <aboorvad@linux.ibm.com>

Mathieu Desnoyers (1):
  mm: Fix OOM killer and proc stats inaccuracy on large many-core
    systems

 fs/proc/task_mmu.c | 14 +++++++-------
 include/linux/mm.h |  5 -----
 2 files changed, 7 insertions(+), 12 deletions(-)

-- 
2.39.5

[PATCH v1 1/1] mm: Fix OOM killer and proc stats inaccuracy on large many-core systems

[Mathieu Desnoyers]

Use the precise, albeit slower, precise RSS counter sums for the OOM
killer task selection and proc statistics. The approximated value is
too imprecise on large many-core systems.

The following rss tracking issues were noted by Sweet Tea Dorminy [1],
which lead to picking wrong tasks as OOM kill target:

  Recently, several internal services had an RSS usage regression as part of a
  kernel upgrade. Previously, they were on a pre-6.2 kernel and were able to
  read RSS statistics in a backup watchdog process to monitor and decide if
  they'd overrun their memory budget. Now, however, a representative service
  with five threads, expected to use about a hundred MB of memory, on a 250-cpu
  machine had memory usage tens of megabytes different from the expected amount
  -- this constituted a significant percentage of inaccuracy, causing the
  watchdog to act.

  This was a result of commit f1a7941243c1 ("mm: convert mm's rss stats
  into percpu_counter") [1].  Previously, the memory error was bounded by
  64*nr_threads pages, a very livable megabyte. Now, however, as a result of
  scheduler decisions moving the threads around the CPUs, the memory error could
  be as large as a gigabyte.

  This is a really tremendous inaccuracy for any few-threaded program on a
  large machine and impedes monitoring significantly. These stat counters are
  also used to make OOM killing decisions, so this additional inaccuracy could
  make a big difference in OOM situations -- either resulting in the wrong
  process being killed, or in less memory being returned from an OOM-kill than
  expected.

Here is a (possibly incomplete) list of the prior approaches that were
used or proposed, along with their downside:

1) Per-thread rss tracking: large error on many-thread processes.

2) Per-CPU counters: up to 12% slower for short-lived processes and 9%
   increased system time in make test workloads [1]. Moreover, the
   inaccuracy increases with O(n^2) with the number of CPUs.

3) Per-NUMA-node counters: requires atomics on fast-path (overhead),
   error is high with systems that have lots of NUMA nodes (32 times
   the number of NUMA nodes).

The simple fix proposed here is to do the precise per-cpu counters sum
every time a counter value needs to be read. This applies to the OOM
killer task selection, to the /proc statistics, and to the oom mark_victim
trace event.

Note that commit 82241a83cd15 ("mm: fix the inaccurate memory statistics
issue for users") introduced get_mm_counter_sum() for precise proc
memory status queries for _some_ proc files. This change renames
get_mm_counter_sum() to get_mm_counter(), thus moving the rest of the
proc files to the precise sum.

This change effectively increases the latency introduced when the OOM
killer executes in favor of doing a more precise OOM target task
selection. Effectively, the OOM killer iterates on all tasks, for all
relevant page types, for which the precise sum iterates on all possible
CPUs.

As a reference, here is the execution time of the OOM killer
before/after the change:

AMD EPYC 9654 96-Core (2 sockets)
Within a KVM, configured with 256 logical cpus.

                                  |  before  |  after   |
----------------------------------|----------|----------|
nr_processes=40                   |  0.3 ms  |   0.5 ms |
nr_processes=10000                |  3.0 ms  |  80.0 ms |

Suggested-by: Michal Hocko <mhocko@suse.com>
Fixes: f1a7941243c1 ("mm: convert mm's rss stats into percpu_counter")
Link: https://lore.kernel.org/lkml/20250331223516.7810-2-sweettea-kernel@dorminy.me/ # [1]
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Dennis Zhou <dennis@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Martin Liu <liumartin@google.com>
Cc: David Rientjes <rientjes@google.com>
Cc: christian.koenig@amd.com
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: SeongJae Park <sj@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: "Liam R . Howlett" <liam.howlett@oracle.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-mm@kvack.org
Cc: stable@vger.kernel.org
Cc: linux-trace-kernel@vger.kernel.org
Cc: Yu Zhao <yuzhao@google.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Mateusz Guzik <mjguzik@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Aboorva Devarajan <aboorvad@linux.ibm.com>
---
 fs/proc/task_mmu.c | 14 +++++++-------
 include/linux/mm.h |  5 -----
 2 files changed, 7 insertions(+), 12 deletions(-)

diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 81dfc26bfae8..8ca4fbf53fc5 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -39,9 +39,9 @@ void task_mem(struct seq_file *m, struct mm_struct *mm)
 	unsigned long text, lib, swap, anon, file, shmem;
 	unsigned long hiwater_vm, total_vm, hiwater_rss, total_rss;
 
-	anon = get_mm_counter_sum(mm, MM_ANONPAGES);
-	file = get_mm_counter_sum(mm, MM_FILEPAGES);
-	shmem = get_mm_counter_sum(mm, MM_SHMEMPAGES);
+	anon = get_mm_counter(mm, MM_ANONPAGES);
+	file = get_mm_counter(mm, MM_FILEPAGES);
+	shmem = get_mm_counter(mm, MM_SHMEMPAGES);
 
 	/*
 	 * Note: to minimize their overhead, mm maintains hiwater_vm and
@@ -62,7 +62,7 @@ void task_mem(struct seq_file *m, struct mm_struct *mm)
 	text = min(text, mm->exec_vm << PAGE_SHIFT);
 	lib = (mm->exec_vm << PAGE_SHIFT) - text;
 
-	swap = get_mm_counter_sum(mm, MM_SWAPENTS);
+	swap = get_mm_counter(mm, MM_SWAPENTS);
 	SEQ_PUT_DEC("VmPeak:\t", hiwater_vm);
 	SEQ_PUT_DEC(" kB\nVmSize:\t", total_vm);
 	SEQ_PUT_DEC(" kB\nVmLck:\t", mm->locked_vm);
@@ -95,12 +95,12 @@ unsigned long task_statm(struct mm_struct *mm,
 			 unsigned long *shared, unsigned long *text,
 			 unsigned long *data, unsigned long *resident)
 {
-	*shared = get_mm_counter_sum(mm, MM_FILEPAGES) +
-			get_mm_counter_sum(mm, MM_SHMEMPAGES);
+	*shared = get_mm_counter(mm, MM_FILEPAGES) +
+			get_mm_counter(mm, MM_SHMEMPAGES);
 	*text = (PAGE_ALIGN(mm->end_code) - (mm->start_code & PAGE_MASK))
 								>> PAGE_SHIFT;
 	*data = mm->data_vm + mm->stack_vm;
-	*resident = *shared + get_mm_counter_sum(mm, MM_ANONPAGES);
+	*resident = *shared + get_mm_counter(mm, MM_ANONPAGES);
 	return mm->total_vm;
 }
 
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 6f959d8ca4b4..d096bb3593ba 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2847,11 +2847,6 @@ static inline bool get_user_page_fast_only(unsigned long addr,
  * per-process(per-mm_struct) statistics.
  */
 static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
-{
-	return percpu_counter_read_positive(&mm->rss_stat[member]);
-}
-
-static inline unsigned long get_mm_counter_sum(struct mm_struct *mm, int member)
 {
 	return percpu_counter_sum_positive(&mm->rss_stat[member]);
 }
-- 
2.39.5

Re: [PATCH v1 1/1] mm: Fix OOM killer and proc stats inaccuracy on large many-core systems

[Andrew Morton]

On Tue, 13 Jan 2026 14:47:34 -0500 Mathieu Desnoyers <mathieu.desnoyers@efficios.com> wrote:

> Use the precise, albeit slower, precise RSS counter sums for the OOM
> killer task selection and proc statistics. The approximated value is
> too imprecise on large many-core systems.

Thanks.

Problem: if I also queue your "mm: Reduce latency of OOM killer task
selection" series then this single patch won't get tested, because the
larger series erases this patch, yes?

Obvious solution: aim this patch at next-merge-window and let's look at
the larger series for the next -rc cycle.  Thoughts?

> The following rss tracking issues were noted by Sweet Tea Dorminy [1],
> which lead to picking wrong tasks as OOM kill target:
> 
>   Recently, several internal services had an RSS usage regression as part of a
>   kernel upgrade. Previously, they were on a pre-6.2 kernel and were able to
>   read RSS statistics in a backup watchdog process to monitor and decide if
>   they'd overrun their memory budget. Now, however, a representative service
>   with five threads, expected to use about a hundred MB of memory, on a 250-cpu
>   machine had memory usage tens of megabytes different from the expected amount
>   -- this constituted a significant percentage of inaccuracy, causing the
>   watchdog to act.
> 
>   This was a result of commit f1a7941243c1 ("mm: convert mm's rss stats
>   into percpu_counter") [1].  Previously, the memory error was bounded by
>   64*nr_threads pages, a very livable megabyte. Now, however, as a result of
>   scheduler decisions moving the threads around the CPUs, the memory error could
>   be as large as a gigabyte.
> 
>   This is a really tremendous inaccuracy for any few-threaded program on a
>   large machine and impedes monitoring significantly. These stat counters are
>   also used to make OOM killing decisions, so this additional inaccuracy could
>   make a big difference in OOM situations -- either resulting in the wrong
>   process being killed, or in less memory being returned from an OOM-kill than
>   expected.
> 
> Here is a (possibly incomplete) list of the prior approaches that were
> used or proposed, along with their downside:
> 
> 1) Per-thread rss tracking: large error on many-thread processes.
> 
> 2) Per-CPU counters: up to 12% slower for short-lived processes and 9%
>    increased system time in make test workloads [1]. Moreover, the
>    inaccuracy increases with O(n^2) with the number of CPUs.
> 
> 3) Per-NUMA-node counters: requires atomics on fast-path (overhead),
>    error is high with systems that have lots of NUMA nodes (32 times
>    the number of NUMA nodes).
> 
> The simple fix proposed here is to do the precise per-cpu counters sum
> every time a counter value needs to be read. This applies to the OOM
> killer task selection, to the /proc statistics, and to the oom mark_victim
> trace event.
> 
> Note that commit 82241a83cd15 ("mm: fix the inaccurate memory statistics
> issue for users") introduced get_mm_counter_sum() for precise proc
> memory status queries for _some_ proc files. This change renames
> get_mm_counter_sum() to get_mm_counter(), thus moving the rest of the
> proc files to the precise sum.

Please confirm - switching /proc functions from get_mm_counter_sum() to
get_mm_counter_sum() doesn't actually change anything, right?  It would
be concerning to add possible overhead to things like task_statm().

> This change effectively increases the latency introduced when the OOM
> killer executes in favor of doing a more precise OOM target task
> selection. Effectively, the OOM killer iterates on all tasks, for all
> relevant page types, for which the precise sum iterates on all possible
> CPUs.
> 
> As a reference, here is the execution time of the OOM killer
> before/after the change:
> 
> AMD EPYC 9654 96-Core (2 sockets)
> Within a KVM, configured with 256 logical cpus.
> 
>                                   |  before  |  after   |
> ----------------------------------|----------|----------|
> nr_processes=40                   |  0.3 ms  |   0.5 ms |
> nr_processes=10000                |  3.0 ms  |  80.0 ms |

That seems acceptable.

Re: [PATCH v1 1/1] mm: Fix OOM killer and proc stats inaccuracy on large many-core systems

[Mathieu Desnoyers]

On 2026-01-13 16:46, Andrew Morton wrote:
> On Tue, 13 Jan 2026 14:47:34 -0500 Mathieu Desnoyers <mathieu.desnoyers@efficios.com> wrote:
> 
>> Use the precise, albeit slower, precise RSS counter sums for the OOM
>> killer task selection and proc statistics. The approximated value is
>> too imprecise on large many-core systems.
> 
> Thanks.
> 
> Problem: if I also queue your "mm: Reduce latency of OOM killer task
> selection" series then this single patch won't get tested, because the
> larger series erases this patch, yes?

That's a good point.

> 
> Obvious solution: aim this patch at next-merge-window and let's look at
> the larger series for the next -rc cycle.  Thoughts?

Yes, that works for me. Does it mean I should re-submit the hpcc
series after the next merge window closes, or do you keep a queue of
stuff waiting for the next -rc cycle somewhere ?

> 
>> The following rss tracking issues were noted by Sweet Tea Dorminy [1],
>> which lead to picking wrong tasks as OOM kill target:
>>
>>    Recently, several internal services had an RSS usage regression as part of a
>>    kernel upgrade. Previously, they were on a pre-6.2 kernel and were able to
>>    read RSS statistics in a backup watchdog process to monitor and decide if
>>    they'd overrun their memory budget. Now, however, a representative service
>>    with five threads, expected to use about a hundred MB of memory, on a 250-cpu
>>    machine had memory usage tens of megabytes different from the expected amount
>>    -- this constituted a significant percentage of inaccuracy, causing the
>>    watchdog to act.
>>
>>    This was a result of commit f1a7941243c1 ("mm: convert mm's rss stats
>>    into percpu_counter") [1].  Previously, the memory error was bounded by
>>    64*nr_threads pages, a very livable megabyte. Now, however, as a result of
>>    scheduler decisions moving the threads around the CPUs, the memory error could
>>    be as large as a gigabyte.
>>
>>    This is a really tremendous inaccuracy for any few-threaded program on a
>>    large machine and impedes monitoring significantly. These stat counters are
>>    also used to make OOM killing decisions, so this additional inaccuracy could
>>    make a big difference in OOM situations -- either resulting in the wrong
>>    process being killed, or in less memory being returned from an OOM-kill than
>>    expected.
>>
>> Here is a (possibly incomplete) list of the prior approaches that were
>> used or proposed, along with their downside:
>>
>> 1) Per-thread rss tracking: large error on many-thread processes.
>>
>> 2) Per-CPU counters: up to 12% slower for short-lived processes and 9%
>>     increased system time in make test workloads [1]. Moreover, the
>>     inaccuracy increases with O(n^2) with the number of CPUs.
>>
>> 3) Per-NUMA-node counters: requires atomics on fast-path (overhead),
>>     error is high with systems that have lots of NUMA nodes (32 times
>>     the number of NUMA nodes).
>>
>> The simple fix proposed here is to do the precise per-cpu counters sum
>> every time a counter value needs to be read. This applies to the OOM
>> killer task selection, to the /proc statistics, and to the oom mark_victim
>> trace event.
>>
>> Note that commit 82241a83cd15 ("mm: fix the inaccurate memory statistics
>> issue for users") introduced get_mm_counter_sum() for precise proc
>> memory status queries for _some_ proc files. This change renames
>> get_mm_counter_sum() to get_mm_counter(), thus moving the rest of the
>> proc files to the precise sum.
> 
> Please confirm - switching /proc functions from get_mm_counter_sum() to
> get_mm_counter_sum() doesn't actually change anything, right?  It would
> be concerning to add possible overhead to things like task_statm().

The approach proposed by this patch is to switch all proc ABIs which
query RSS to the precise sum to eliminate any discrepancy caused by too
imprecise approximate sums. It's a big hammer, and it can slow down
those proc interfaces, including task_statm(). Is it an issue ?

The hpcc series introduces an approximation which provides accuracy
limits on the approximation that make the result is still somewhat
meaninful on large many core systems.

The overall approach here would be to move back those proc interfaces
which care about low overhead to the hpcc approximate sum when it lands
upstream. But in order to learn that, we need to know which proc
interface files are performance-sensitive. How can we get that data ?

Thanks,

Mathieu

-- 
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com