Re: [syzbot] [damon?] divide error in damon_set_attrs

[SeongJae Park]

Hi Kefeng and syzbot,

On Fri, 26 May 2023 20:59:12 +0800 Kefeng Wang <wangkefeng.wang@xxxxxxxxxx> wrote:

> 
> 
> On 2023/5/26 19:51, syzbot wrote:
> > Hello,
> > 
> > syzbot found the following issue on:
> > 
> > HEAD commit:    44c026a73be8 Linux 6.4-rc3
> > git tree:       upstream
> > console output: https://syzkaller.appspot.com/x/log.txt?x=13a92b31280000
> > kernel config:  https://syzkaller.appspot.com/x/.config?x=f389ffdf4e9ba3f0
> > dashboard link: https://syzkaller.appspot.com/bug?extid=841a46899768ec7bec67
> > compiler:       gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2
> > userspace arch: i386
> > 
> > Unfortunately, I don't have any reproducer for this issue yet.
> > 
> > Downloadable assets:
> > disk image: https://storage.googleapis.com/syzbot-assets/35f16ee05df7/disk-44c026a7.raw.xz
> > vmlinux: https://storage.googleapis.com/syzbot-assets/10399498a570/vmlinux-44c026a7.xz
> > kernel image: https://storage.googleapis.com/syzbot-assets/5c72201ea4ba/bzImage-44c026a7.xz
> > 
> > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > Reported-by: syzbot+841a46899768ec7bec67@xxxxxxxxxxxxxxxxxxxxxxxxx
> > 
> > divide error: 0000 [#1] PREEMPT SMP KASAN
> > CPU: 1 PID: 13527 Comm: syz-executor.1 Not tainted 6.4.0-rc3-syzkaller #0
> > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 04/28/2023
> > RIP: 0010:damon_nr_accesses_to_accesses_bp mm/damon/core.c:491 [inline]
> > RIP: 0010:damon_nr_accesses_for_new_attrs mm/damon/core.c:497 [inline]
> > RIP: 0010:damon_update_monitoring_result mm/damon/core.c:506 [inline]
> > RIP: 0010:damon_update_monitoring_results mm/damon/core.c:534 [inline]
> > RIP: 0010:damon_set_attrs+0x224/0x460 mm/damon/core.c:555

Thank you for finding and reporting this bug!

The code of the problem is as below:

    /* convert nr_accesses to access ratio in bp (per 10,000) */
    static unsigned int damon_nr_accesses_to_accesses_bp(
                    unsigned int nr_accesses, struct damon_attrs *attrs)
    {
            unsigned int max_nr_accesses =
                    attrs->aggr_interval / attrs->sample_interval;
    
            return nr_accesses * 10000 / max_nr_accesses;
    }

The problem can happen when 'aggr_interval' is smaller than 'sample_interval',
because 'max_nr_accesses' becomes zero in the case, and resulting in divide by
zero.

Same problem is in damon_accesses_bp_to_nr_accesses().

> 
> make aggr_interval great than or equal sample_interval?
> 
> diff --git a/mm/damon/core.c b/mm/damon/core.c
> index d9ef62047bf5..6fe1960f3d6b 100644
> --- a/mm/damon/core.c
> +++ b/mm/damon/core.c
> @@ -525,8 +525,8 @@ static void damon_update_monitoring_results(struct 
> damon_ctx *ctx,
> 
>          /* if any interval is zero, simply forgive conversion */
>          if (!old_attrs->sample_interval || !old_attrs->aggr_interval ||
> -                       !new_attrs->sample_interval ||
> -                       !new_attrs->aggr_interval)
> +           !new_attrs->sample_interval || !new_attrs->aggr_interval ||
> +           new_attrs->aggr_interval < new_attrs->sample_interval)
>                  return;

Nice and effective fix!  Nevertheless, I think aggregation interval smaller
than sample interval is just a wrong input.  How about adding the check in
damon_set_attrs()'s already existing attributes validation, like below?

--- a/mm/damon/core.c
+++ b/mm/damon/core.c
@@ -580,6 +580,8 @@ int damon_set_attrs(struct damon_ctx *ctx, struct damon_attrs *attrs)
                return -EINVAL;
        if (attrs->min_nr_regions > attrs->max_nr_regions)
                return -EINVAL;
+       if (attrs->aggr_interval < attrs->sample_interval)
+               return -EINVAL;

        damon_update_monitoring_results(ctx, attrs);
        ctx->attrs = *attrs;

Thanks,
SJ