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: <c89256b8-bec3-4bc8-8399-cb44794f70f1@linux.dev>
Date: Tue, 7 Oct 2025 16:32:07 +0800
From: Lance Yang <lance.yang@...ux.dev>
To: Dev Jain <dev.jain@....com>
Cc: ziy@...dia.com, baolin.wang@...ux.alibaba.com, Liam.Howlett@...cle.com,
 npache@...hat.com, ryan.roberts@....com, baohua@...nel.org,
 ioworker0@...il.com, richard.weiyang@...il.com,
 linux-kernel@...r.kernel.org, linux-mm@...ck.org, akpm@...ux-foundation.org,
 lorenzo.stoakes@...cle.com, david@...hat.com
Subject: Re: [PATCH mm-new v2 3/3] mm/khugepaged: merge PTE scanning logic
 into a new helper



On 2025/10/7 14:28, Dev Jain wrote:
> 
> On 06/10/25 8:13 pm, Lance Yang wrote:
>> +static inline int thp_collapse_check_pte(pte_t pte, struct 
>> vm_area_struct *vma,
>> +        unsigned long addr, struct collapse_control *cc,
>> +        struct folio **foliop, int *none_or_zero, int *unmapped,
>> +        int *shared, int *scan_result)
> 
> Nit: Will prefer the cc parameter to go at the last.

Yep, got it.

> 
>> +{
>> +    struct folio *folio = NULL;
>> +    struct page *page = NULL;
>> +
>> +    if (pte_none(pte) || is_zero_pfn(pte_pfn(pte))) {
>> +        (*none_or_zero)++;
>> +        if (!userfaultfd_armed(vma) &&
>> +            (!cc->is_khugepaged ||
>> +             *none_or_zero <= khugepaged_max_ptes_none)) {
>> +            return PTE_CHECK_CONTINUE;
>> +        } else {
>> +            *scan_result = SCAN_EXCEED_NONE_PTE;
>> +            count_vm_event(THP_SCAN_EXCEED_NONE_PTE);
>> +            return PTE_CHECK_FAIL;
>> +        }
>> +    } else if (!pte_present(pte)) {
>> +        if (!unmapped) {
>> +            *scan_result = SCAN_PTE_NON_PRESENT;
>> +            return PTE_CHECK_FAIL;
>> +        }
>> +
>> +        if (non_swap_entry(pte_to_swp_entry(pte))) {
>> +            *scan_result = SCAN_PTE_NON_PRESENT;
>> +            return PTE_CHECK_FAIL;
>> +        }
>> +
>> +        (*unmapped)++;
>> +        if (!cc->is_khugepaged ||
>> +            *unmapped <= khugepaged_max_ptes_swap) {
>> +            /*
>> +             * Always be strict with uffd-wp enabled swap
>> +             * entries. Please see comment below for
>> +             * pte_uffd_wp().
>> +             */
>> +            if (pte_swp_uffd_wp(pte)) {
>> +                *scan_result = SCAN_PTE_UFFD_WP;
>> +                return PTE_CHECK_FAIL;
>> +            }
>> +            return PTE_CHECK_CONTINUE;
>> +        } else {
>> +            *scan_result = SCAN_EXCEED_SWAP_PTE;
>> +            count_vm_event(THP_SCAN_EXCEED_SWAP_PTE);
>> +            return PTE_CHECK_FAIL;
>> +        }
>> +    } else if (pte_uffd_wp(pte)) {
>> +        /*
>> +         * Don't collapse the page if any of the small PTEs are
>> +         * armed with uffd write protection. Here we can also mark
>> +         * the new huge pmd as write protected if any of the small
>> +         * ones is marked but that could bring unknown userfault
>> +         * messages that falls outside of the registered range.
>> +         * So, just be simple.
>> +         */
>> +        *scan_result = SCAN_PTE_UFFD_WP;
>> +        return PTE_CHECK_FAIL;
>> +    }
>> +
>> +    page = vm_normal_page(vma, addr, pte);
> 
> You should use vm_normal_folio here and drop struct page altogether - 
> this was also
> noted during the review of the mTHP collapse patchset.

Right, I missed that vm_normal_folio() was the way to go here :)

Thanks for the pointer!
Lance

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ