linux-kernel - Re: [PATCH] mm: fix arithmetic for bdi min_ratio and max

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <112dd87b-1072-4568-93d5-3b13ed818511@linux.alibaba.com>
Date: Tue, 19 Dec 2023 10:03:40 +0800
From: Jingbo Xu <jefflexu@...ux.alibaba.com>
To: Andrew Morton <akpm@...ux-foundation.org>
Cc: shr@...kernel.io, linux-mm@...ck.org, linux-kernel@...r.kernel.org,
 joseph.qi@...ux.alibaba.com
Subject: Re: [PATCH] mm: fix arithmetic for bdi min_ratio and max_ratio



On 12/19/23 2:21 AM, Andrew Morton wrote:
> On Mon, 18 Dec 2023 11:16:40 +0800 Jingbo Xu <jefflexu@...ux.alibaba.com> wrote:
> 
>> bdi->[min|max]_ratio are both part per million.  Fix the wrong
>> arithmetic when setting bdi's min_ratio and max_ratio.
>>
>> Fixes: efc3e6ad53ea ("mm: split off __bdi_set_max_ratio() function")
>> Fixes: 8021fb3232f2 ("mm: split off __bdi_set_min_ratio() function")
> 
> As we have two Fixes:, it would be better to have two patches please.

Sounds reasonable.  Thanks.

> 
>> --- 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;

This indeed has visible runtime effect as follows:

# 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

Setting min_ratio will fail with -EINVAL, as the above code tries to set
min_ratio with (min_ratio * BDI_RATIO_SCALE)%, i.e. 10000% in the above
example.

>>  
>>  	spin_lock_bh(&bdi_lock);
>>  	if (min_ratio > bdi->max_ratio) {
>> @@ -729,7 +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,
>> +					       100UL * BDI_RATIO_SCALE);
>>  	}
>>  	spin_unlock_bh(&bdi_lock);

This one has no visible runtime effect, but I believe this would affect
the incrementing of writeout completion count when max_ratio is not 100%.


-- 
Thanks,
Jingbo