mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Quanmin Yan <yanquanmin1@huawei.com>
To: SeongJae Park <sj@kernel.org>
Cc: <akpm@linux-foundation.org>, <damon@lists.linux.dev>,
	<linux-kernel@vger.kernel.org>, <linux-mm@kvack.org>,
	<wangkefeng.wang@huawei.com>, <zuoze1@huawei.com>
Subject: Re: [PATCH] mm/damon/stat: set last_refresh_jiffies to jiffies at startup
Date: Wed, 29 Oct 2025 10:02:59 +0800	[thread overview]
Message-ID: <75364faa-4cce-4e0b-ae43-dc725e6918da@huawei.com> (raw)
In-Reply-To: <20251029013038.66625-1-sj@kernel.org>


在 2025/10/29 9:30, SeongJae Park 写道:
> On Tue, 28 Oct 2025 07:32:49 -0700 SeongJae Park <sj@kernel.org> wrote:
>
>> On Tue, 28 Oct 2025 07:19:14 -0700 SeongJae Park <sj@kernel.org> wrote:
>>
>>> On Tue, 28 Oct 2025 14:19:27 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>>>
>>>> In DAMON_STAT's damon_stat_damon_call_fn(), time_before_eq() is used to
>>>> avoid unnecessarily frequent stat update.
>>>>
>>>> On 32-bit systems, the kernel initializes jiffies to "-5 minutes" to make
>>>> jiffies wrap bugs appear earlier. However, this causes time_before_eq()
>>>> in DAMON_STAT to unexpectedly return true during the first 5 minutes
>>>> after boot on 32-bit systems (see [1] for more explanation, which fixes
>>>> another jiffies-related issue in DAMON). As a result, DAMON_STAT does not
>>>> update any monitoring results during that period, which can be more
>>>> confusing when DAMON_STAT_ENABLED_DEFAULT is enabled.
>>>>
>>>> Fix it by setting last_refresh_jiffies to jiffies at startup.
>>> Nice catch, thank you for this patch!
>>>
>>>> [1] https://lkml.kernel.org/r/20250822025057.1740854-1-ekffu200098@gmail.com
>>>>
>>>> Fixes: fabdd1e911da ("mm/damon/stat: calculate and expose estimated memory bandwidth")
>>>> Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
>>>> ---
>>>>   mm/damon/stat.c | 3 +++
>>>>   1 file changed, 3 insertions(+)
>>>>
>>>> diff --git a/mm/damon/stat.c b/mm/damon/stat.c
>>>> index 6c4503d2aee3..6dc3e18de910 100644
>>>> --- a/mm/damon/stat.c
>>>> +++ b/mm/damon/stat.c
>>>> @@ -132,6 +132,9 @@ static int damon_stat_damon_call_fn(void *data)
>>>>   	struct damon_ctx *c = data;
>>>>   	static unsigned long last_refresh_jiffies;
>>>>   
>>>> +	if (unlikely(!last_refresh_jiffies))
>>>> +		last_refresh_jiffies = jiffies;
>>>> +
>>> How about doing the initialization together with the declaration?  E.g.,
>>>
>>>   static int damon_stat_damon_call_fn(void *data)
>>>   {
>>>          struct damon_ctx *c = data;
>>> -       static unsigned long last_refresh_jiffies;
>>> +       static unsigned long last_refresh_jiffies = jiffies;
> Please ignore the above suggestion.  It will even not build, like below...
>
> .../mm/damon/stat.c: In function ‘damon_stat_damon_call_fn’:
> .../mm/damon/stat.c:133:53: error: initializer element is not constant
>    133 |         static unsigned long last_refresh_jiffies = jiffies;
>        |                                                     ^~~~~~~
>
>> Actually, a similar issue can happen again if DAMON_STAT is stopped and
>> restarted by user.  That is, if user stops DAMON_STAT just after
>> last_refresh_jiffies is updated, and restart it after 5 seconds or more, the
>> time_before_eq() on damon_call_fn() will return true, so stat updates will
>> happen earlier than expected.  Shouldn't be a real problem, but better to avoid
>> if possible.
>>
>> How about making last_refresh_jiffies a global variable and initialize it on
>> damon_stat_start()?  To avoid unnecessary name conflicts, the variable name
>> would also better to be changed, e.g., damon_stat_last_refresh_jiffies.
> But, please consider the above one.
>
> And I just realized a similar issue exist for next_update_jiffies in
> mm/damon/sysfs.c file.  Please feel free to send a patch for that if you
> willing to.
>
OK, I’ll review all these issues and send a new patch set once everything
is ready.🙂

Thanks,
Quanmin Yan

[...]


  reply	other threads:[~2025-10-29  2:03 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-10-28  6:19 Quanmin Yan
2025-10-28 14:19 ` SeongJae Park
2025-10-28 14:32   ` SeongJae Park
2025-10-29  1:30     ` SeongJae Park
2025-10-29  2:02       ` Quanmin Yan [this message]
2025-10-29  1:32   ` Quanmin Yan

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=75364faa-4cce-4e0b-ae43-dc725e6918da@huawei.com \
    --to=yanquanmin1@huawei.com \
    --cc=akpm@linux-foundation.org \
    --cc=damon@lists.linux.dev \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=sj@kernel.org \
    --cc=wangkefeng.wang@huawei.com \
    --cc=zuoze1@huawei.com \
    /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

all inboxes | Powered by JetHome®