linux-kernel - Re: [RFC PATCH 1/3] hugetlb: skip to end of PT page mapping when pte not present

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 for Android: free password hash cracker in your pocket

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

Date:   Tue, 31 May 2022 09:56:52 -0700
From:   Mike Kravetz <mike.kravetz@...cle.com>
To:     Baolin Wang <baolin.wang@...ux.alibaba.com>,
        linux-kernel@...r.kernel.org, linux-mm@...ck.org
Cc:     Muchun Song <songmuchun@...edance.com>,
        Michal Hocko <mhocko@...e.com>, Peter Xu <peterx@...hat.com>,
        Naoya Horiguchi <naoya.horiguchi@...ux.dev>,
        James Houghton <jthoughton@...gle.com>,
        Mina Almasry <almasrymina@...gle.com>,
        "Aneesh Kumar K . V" <aneesh.kumar@...ux.vnet.ibm.com>,
        Anshuman Khandual <anshuman.khandual@....com>,
        Paul Walmsley <paul.walmsley@...ive.com>,
        Christian Borntraeger <borntraeger@...ux.ibm.com>,
        Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [RFC PATCH 1/3] hugetlb: skip to end of PT page mapping when pte
 not present

On 5/30/22 03:10, Baolin Wang wrote:
> On 5/28/2022 6:58 AM, Mike Kravetz wrote:
> 
> I tested on my ARM64 machine with implementing arm64 specific hugetlb_mask_last_hp() as below, and it works well.
> 
> Just a few nits inline, otherwise looks good to me.
> Tested-by: Baolin Wang <baolin.wang@...ux.alibaba.com>
> Reviewed-by: Baolin Wang <baolin.wang@...ux.alibaba.com>
> 
> diff --git a/arch/arm64/mm/hugetlbpage.c b/arch/arm64/mm/hugetlbpage.c
> index d93ba128a2b0..e04a097ffcc4 100644
> --- a/arch/arm64/mm/hugetlbpage.c
> +++ b/arch/arm64/mm/hugetlbpage.c
> @@ -376,6 +376,28 @@ pte_t *huge_pte_offset(struct mm_struct *mm,
>         return NULL;
>  }
> 
> +unsigned long hugetlb_mask_last_hp(struct hstate *h)
> +{
> +       unsigned long hp_size = huge_page_size(h);
> +
> +       switch (hp_size) {
> +       case P4D_SIZE:
> +               return PGDIR_SIZE - P4D_SIZE;
> +       case PUD_SIZE:
> +               return P4D_SIZE - PUD_SIZE;
> +       case CONT_PMD_SIZE:
> +               return PUD_SIZE - CONT_PMD_SIZE;
> +       case PMD_SIZE:
> +               return PUD_SIZE - PMD_SIZE;
> +       case CONT_PTE_SIZE:
> +               return PMD_SIZE - CONT_PTE_SIZE;
> +       default:
> +               break;
> +       }
> +
> +       return ~(0UL);
> +}
> 

Thanks!  I was hesitant to put something together for those CONT_* sizes.

>>   +/*
>> + * Return a mask that can be used to update an address to the last huge
>> + * page in a page table page mapping size.  Used to skip non-present
>> + * page table entries when linearly scanning address ranges.  Architectures
>> + * with unique huge page to page table relationships can define their own
>> + * version of this routine.
>> + */
>> +unsigned long hugetlb_mask_last_hp(struct hstate *h)
>> +{
>> +    unsigned long hp_size = huge_page_size(h);
>> +
>> +    if (hp_size == P4D_SIZE)
>> +        return PGDIR_SIZE - P4D_SIZE;
>> +    else if (hp_size == PUD_SIZE)
>> +        return P4D_SIZE - PUD_SIZE;
>> +    else if (hp_size == PMD_SIZE)
>> +        return PUD_SIZE - PMD_SIZE;
> 
> Changing to use 'switch' looks more readable?

Agree.  Or, I might just go with Peter's simplification.
> 
>> +
>> +    return ~(0);
> 
> Better to return '~(0UL)' to keep function type consistent.

Yes, thanks!
-- 
Mike Kravetz