Re: cgroup and FALLOC_FL_PUNCH_HOLE: WARNING: CPU: 13 PID: 2438 at mm/page_counter.c:57 page_counter_uncharge+0x4b/0x5

[Mike Kravetz]

> On 21.10.20 15:11, David Hildenbrand wrote:
>> On 21.10.20 14:57, Michal Privoznik wrote:
>>> On 10/21/20 5:35 AM, Mike Kravetz wrote:
>>>> On 10/20/20 6:38 AM, David Hildenbrand wrote:
>>>>
>>>> It would be good if Mina (at least) would look these over.  Would also
>>>> be interesting to know if these fixes address the bug seen with the qemu
>>>> use case.
>>>>
>>>> I'm still doing more testing and code inspection to look for other issues.
>>>>
...
...
>>>
>>> I've applied, rebuilt and tested, but unfortunately I still hit the problem:
>>> [ 6472.719047] ------------[ cut here ]------------
>>> [ 6472.719052] WARNING: CPU: 6 PID: 11773 at mm/page_counter.c:57 
...
...
>>
>> Agreed, same over here. :(
>>
> 
> I *think* the following on top makes it fly
> 
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index 67fc6383995b..5cf7f6a6c1a6 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -656,6 +656,9 @@ static long region_del(struct resv_map *resv, long
> f, long t)
> 
>                         del += t - f;
> 
> +                       hugetlb_cgroup_uncharge_file_region(
> +                               resv, rg, t - f);
> +
>                         /* New entry for end of split region */
>                         nrg->from = t;
>                         nrg->to = rg->to;
> @@ -667,9 +670,6 @@ static long region_del(struct resv_map *resv, long
> f, long t)
>                         /* Original entry is trimmed */
>                         rg->to = f;
> 
> -                       hugetlb_cgroup_uncharge_file_region(
> -                               resv, rg, nrg->to - nrg->from);
> -
>                         list_add(&nrg->link, &rg->link);
>                         nrg = NULL;
>                         break;
> 
> 

Thanks, yes that certainly does look like a bug in that code.

Does that resolve the issue with quemu?

I want to do a little more testing/research before sending a patch later
today.
-- 
Mike Kravetz