[PATCH v2 0/2] mm: fix arithmetic for bdi min_ratio and

[PATCH v2 0/2] mm: fix arithmetic for bdi min_ratio and

[Jingbo Xu]

changes since v1:

- split the previous v1 patch into two separate patches with
  corresponding "Fixes" tag (Andrew Morton)
- patch 2: remove "UL" suffix for the "100" constant

v1: https://lore.kernel.org/all/20231218031640.77983-1-jefflexu@linux.alibaba.com/

Jingbo Xu (2):
  mm: fix arithmetic for bdi min_ratio
  mm: fix arithmetic for max_prop_frac when setting max_ratio

 mm/page-writeback.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

-- 
2.19.1.6.gb485710b

[PATCH v2 1/2] mm: fix arithmetic for bdi min_ratio

[Jingbo Xu]

Since now bdi->min_ratio is part per million, fix the wrong arithmetic.
Otherwise it will fail with -EINVAL when setting a reasonable min_ratio,
as it tries to set min_ratio to (min_ratio * BDI_RATIO_SCALE) in
percentage unit, which exceeds 100% anyway.

    # cat /sys/class/bdi/253\:0/min_ratio
    0
    # cat /sys/class/bdi/253\:0/max_ratio
    100
    # echo 1 > /sys/class/bdi/253\:0/min_ratio
    -bash: echo: write error: Invalid argument

Fixes: 8021fb3232f2 ("mm: split off __bdi_set_min_ratio() function")
Reported-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Signed-off-by: Jingbo Xu <jefflexu@linux.alibaba.com>
---
 mm/page-writeback.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/mm/page-writeback.c b/mm/page-writeback.c
index ee2fd6a6af40..2140382dd768 100644
--- a/mm/page-writeback.c
+++ b/mm/page-writeback.c
@@ -692,7 +692,6 @@ static int __bdi_set_min_ratio(struct backing_dev_info *bdi, unsigned int min_ra
 
 	if (min_ratio > 100 * BDI_RATIO_SCALE)
 		return -EINVAL;
-	min_ratio *= BDI_RATIO_SCALE;
 
 	spin_lock_bh(&bdi_lock);
 	if (min_ratio > bdi->max_ratio) {
-- 
2.19.1.6.gb485710b

[PATCH v2 2/2] mm: fix arithmetic for max_prop_frac when setting max_ratio

[Jingbo Xu]

Since now bdi->max_ratio is part per million, fix the wrong arithmetic
for max_prop_frac when setting max_ratio.  Otherwise the miscalculated
max_prop_frac will affect the incrementing of writeout completion count
when max_ratio is not 100%.

Fixes: efc3e6ad53ea ("mm: split off __bdi_set_max_ratio() function")
Signed-off-by: Jingbo Xu <jefflexu@linux.alibaba.com>
---
 mm/page-writeback.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/mm/page-writeback.c b/mm/page-writeback.c
index 2140382dd768..dda59b368c01 100644
--- a/mm/page-writeback.c
+++ b/mm/page-writeback.c
@@ -728,7 +728,8 @@ static int __bdi_set_max_ratio(struct backing_dev_info *bdi, unsigned int max_ra
 		ret = -EINVAL;
 	} else {
 		bdi->max_ratio = max_ratio;
-		bdi->max_prop_frac = (FPROP_FRAC_BASE * max_ratio) / 100;
+		bdi->max_prop_frac = div64_u64(FPROP_FRAC_BASE * max_ratio,
+					       100 * BDI_RATIO_SCALE);
 	}
 	spin_unlock_bh(&bdi_lock);
 
-- 
2.19.1.6.gb485710b

Re: [PATCH v2 2/2] mm: fix arithmetic for max_prop_frac when setting max_ratio

[Matthew Wilcox]

On Tue, Dec 19, 2023 at 10:42:46AM +0800, Jingbo Xu wrote:
>  	} else {
>  		bdi->max_ratio = max_ratio;
> -		bdi->max_prop_frac = (FPROP_FRAC_BASE * max_ratio) / 100;
> +		bdi->max_prop_frac = div64_u64(FPROP_FRAC_BASE * max_ratio,
> +					       100 * BDI_RATIO_SCALE);
>  	}

Why use div64_u64 here?

FPROP_FRAC_BASE is an unsigned long.  max_ratio is an unsigned int, so
the numerator is an unsigned long.  BDI_RATIO_SCALE is 10,000, so the
numerator is an unsigned int.  There's no 64-bit arithmetic needed here.

Re: [PATCH v2 2/2] mm: fix arithmetic for max_prop_frac when setting max_ratio

[Jingbo Xu]

On 12/19/23 12:06 PM, Matthew Wilcox wrote:
> On Tue, Dec 19, 2023 at 10:42:46AM +0800, Jingbo Xu wrote:
>>  	} else {
>>  		bdi->max_ratio = max_ratio;
>> -		bdi->max_prop_frac = (FPROP_FRAC_BASE * max_ratio) / 100;
>> +		bdi->max_prop_frac = div64_u64(FPROP_FRAC_BASE * max_ratio,
>> +					       100 * BDI_RATIO_SCALE);
>>  	}
> 
> Why use div64_u64 here?
> 
> FPROP_FRAC_BASE is an unsigned long.  max_ratio is an unsigned int, so
> the numerator is an unsigned long.  BDI_RATIO_SCALE is 10,000, so the
> numerator is an unsigned int.  There's no 64-bit arithmetic needed here.

Yes, div64_u64() is actually not needed here. So it seems

bdi->max_prop_frac = FPROP_FRAC_BASE * max_ratio / 100 / BDI_RATIO_SCALE;

is adequate?

-- 
Thanks,
Jingbo

Re: [PATCH v2 2/2] mm: fix arithmetic for max_prop_frac when setting max_ratio

[Matthew Wilcox]

On Tue, Dec 19, 2023 at 01:58:21PM +0800, Jingbo Xu wrote:
> On 12/19/23 12:06 PM, Matthew Wilcox wrote:
> > On Tue, Dec 19, 2023 at 10:42:46AM +0800, Jingbo Xu wrote:
> >>  	} else {
> >>  		bdi->max_ratio = max_ratio;
> >> -		bdi->max_prop_frac = (FPROP_FRAC_BASE * max_ratio) / 100;
> >> +		bdi->max_prop_frac = div64_u64(FPROP_FRAC_BASE * max_ratio,
> >> +					       100 * BDI_RATIO_SCALE);
> >>  	}
> > 
> > Why use div64_u64 here?
> > 
> > FPROP_FRAC_BASE is an unsigned long.  max_ratio is an unsigned int, so
> > the numerator is an unsigned long.  BDI_RATIO_SCALE is 10,000, so the
> > numerator is an unsigned int.  There's no 64-bit arithmetic needed here.
> 
> Yes, div64_u64() is actually not needed here. So it seems
> 
> bdi->max_prop_frac = FPROP_FRAC_BASE * max_ratio / 100 / BDI_RATIO_SCALE;
> 
> is adequate?

I'd rather spell that as:

		bdi->max_prop_frac = (FPROP_FRAC_BASE * max_ratio) /
					(100 * BDI_RATIO_SCALE);

It's closer to how you'd write it out mathematically and so it reads
more easily.  At least for me.

Re: [PATCH v2 2/2] mm: fix arithmetic for max_prop_frac when setting max_ratio

[Jingbo Xu]

On 12/19/23 9:01 PM, Matthew Wilcox wrote:
> On Tue, Dec 19, 2023 at 01:58:21PM +0800, Jingbo Xu wrote:
>> On 12/19/23 12:06 PM, Matthew Wilcox wrote:
>>> On Tue, Dec 19, 2023 at 10:42:46AM +0800, Jingbo Xu wrote:
>>>>  	} else {
>>>>  		bdi->max_ratio = max_ratio;
>>>> -		bdi->max_prop_frac = (FPROP_FRAC_BASE * max_ratio) / 100;
>>>> +		bdi->max_prop_frac = div64_u64(FPROP_FRAC_BASE * max_ratio,
>>>> +					       100 * BDI_RATIO_SCALE);
>>>>  	}
>>>
>>> Why use div64_u64 here?
>>>
>>> FPROP_FRAC_BASE is an unsigned long.  max_ratio is an unsigned int, so
>>> the numerator is an unsigned long.  BDI_RATIO_SCALE is 10,000, so the
>>> numerator is an unsigned int.  There's no 64-bit arithmetic needed here.
>>
>> Yes, div64_u64() is actually not needed here. So it seems
>>
>> bdi->max_prop_frac = FPROP_FRAC_BASE * max_ratio / 100 / BDI_RATIO_SCALE;
>>
>> is adequate?
> 
> I'd rather spell that as:
> 
> 		bdi->max_prop_frac = (FPROP_FRAC_BASE * max_ratio) /
> 					(100 * BDI_RATIO_SCALE);
> 
> It's closer to how you'd write it out mathematically and so it reads
> more easily.  At least for me.

Thanks, I would send v3 soon.

-- 
Thanks,
Jingbo