[dm-devel] [PATCH -next v2 01/28] md: use READ_ONCE/WRITE_ONCE for 'suspend_lo' and 'suspend_hi'

Yu Kuai yukuai1 at huaweicloud.com
Mon Sep 25 01:18:11 UTC 2023


Hi,

在 2023/09/14 10:53, Xiao Ni 写道:
> On Mon, Aug 28, 2023 at 10:04 AM Yu Kuai <yukuai1 at huaweicloud.com> wrote:
>>
>> From: Yu Kuai <yukuai3 at huawei.com>
>>
>> Because reading 'suspend_lo' and 'suspend_hi' from md_handle_request()
>> is not protected, use READ_ONCE/WRITE_ONCE to prevent reading abnormal
>> value.
> 
> Hi Kuai
> 
> If we don't use READ_ONCE/WRITE_ONCE, What's the risk here? Could you
> explain in detail or give an example?

Sorry for the late reply.

That depends on the architecture, a load/store may not be atomice,
for example:

// assume a is 10
t1 write 01
// write half first
a = 11
		t2 read
		//read
		a = 11 -> read abnormal value.
// write other half
a = 01

READ_ONCE/WRITE_ONCE can guarantee that either old value or new value is
read.

Thanks,
Kuai

> 
> Regards
> Xiao
>>
>> Signed-off-by: Yu Kuai <yukuai3 at huawei.com>
>> ---
>>   drivers/md/md.c | 16 +++++++++-------
>>   1 file changed, 9 insertions(+), 7 deletions(-)
>>
>> diff --git a/drivers/md/md.c b/drivers/md/md.c
>> index 46badd13a687..9d8dff9d923c 100644
>> --- a/drivers/md/md.c
>> +++ b/drivers/md/md.c
>> @@ -359,11 +359,11 @@ static bool is_suspended(struct mddev *mddev, struct bio *bio)
>>                  return true;
>>          if (bio_data_dir(bio) != WRITE)
>>                  return false;
>> -       if (mddev->suspend_lo >= mddev->suspend_hi)
>> +       if (READ_ONCE(mddev->suspend_lo) >= READ_ONCE(mddev->suspend_hi))
>>                  return false;
>> -       if (bio->bi_iter.bi_sector >= mddev->suspend_hi)
>> +       if (bio->bi_iter.bi_sector >= READ_ONCE(mddev->suspend_hi))
>>                  return false;
>> -       if (bio_end_sector(bio) < mddev->suspend_lo)
>> +       if (bio_end_sector(bio) < READ_ONCE(mddev->suspend_lo))
>>                  return false;
>>          return true;
>>   }
>> @@ -5171,7 +5171,8 @@ __ATTR(sync_max, S_IRUGO|S_IWUSR, max_sync_show, max_sync_store);
>>   static ssize_t
>>   suspend_lo_show(struct mddev *mddev, char *page)
>>   {
>> -       return sprintf(page, "%llu\n", (unsigned long long)mddev->suspend_lo);
>> +       return sprintf(page, "%llu\n",
>> +                      (unsigned long long)READ_ONCE(mddev->suspend_lo));
>>   }
>>
>>   static ssize_t
>> @@ -5191,7 +5192,7 @@ suspend_lo_store(struct mddev *mddev, const char *buf, size_t len)
>>                  return err;
>>
>>          mddev_suspend(mddev);
>> -       mddev->suspend_lo = new;
>> +       WRITE_ONCE(mddev->suspend_lo, new);
>>          mddev_resume(mddev);
>>
>>          mddev_unlock(mddev);
>> @@ -5203,7 +5204,8 @@ __ATTR(suspend_lo, S_IRUGO|S_IWUSR, suspend_lo_show, suspend_lo_store);
>>   static ssize_t
>>   suspend_hi_show(struct mddev *mddev, char *page)
>>   {
>> -       return sprintf(page, "%llu\n", (unsigned long long)mddev->suspend_hi);
>> +       return sprintf(page, "%llu\n",
>> +                      (unsigned long long)READ_ONCE(mddev->suspend_hi));
>>   }
>>
>>   static ssize_t
>> @@ -5223,7 +5225,7 @@ suspend_hi_store(struct mddev *mddev, const char *buf, size_t len)
>>                  return err;
>>
>>          mddev_suspend(mddev);
>> -       mddev->suspend_hi = new;
>> +       WRITE_ONCE(mddev->suspend_hi, new);
>>          mddev_resume(mddev);
>>
>>          mddev_unlock(mddev);
>> --
>> 2.39.2
>>
> 
> .
> 



More information about the dm-devel mailing list