| 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
| ||
|
Message-ID: <699b143a-cca4-486c-a4ad-d0be561d4ab2@lucifer.local>
Date: Fri, 17 Oct 2025 16:44:23 +0100
From: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
To: Lance Yang <lance.yang@...ux.dev>
Cc: akpm@...ux-foundation.org, david@...hat.com, ziy@...dia.com,
baolin.wang@...ux.alibaba.com, Liam.Howlett@...cle.com,
npache@...hat.com, ryan.roberts@....com, dev.jain@....com,
baohua@...nel.org, ioworker0@...il.com, linux-kernel@...r.kernel.org,
linux-mm@...ck.org, Wei Yang <richard.weiyang@...il.com>
Subject: Re: [PATCH mm-new v2 1/1] mm/khugepaged: guard is_zero_pfn() calls
with pte_present()
On Fri, Oct 17, 2025 at 05:38:47PM +0800, Lance Yang wrote:
> From: Lance Yang <lance.yang@...ux.dev>
>
> A non-present entry, like a swap PTE, contains completely different data
> (swap type and offset). pte_pfn() doesn't know this, so if we feed it a
> non-present entry, it will spit out a junk PFN.
It feels like this somewhat contradicts points I've made on the original series
re the is_swap_pte() stuff. Sigh.
I guess that's _such a mess_ it's hard to avoid though.
And I guess it's reasonable that !pte_present() means we can't expect a valid
PFN though.
>
> What if that junk PFN happens to match the zeropage's PFN by sheer
> chance? While really unlikely, this would be really bad if it did.
>
> So, let's fix this potential bug by ensuring all calls to is_zero_pfn()
> in khugepaged.c are properly guarded by a pte_present() check.
>
> Suggested-by: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
Not sure I really suggested something that strictly contradicts points I
made... but I guess I did suggest guarding this stuff more carefully.
> Reviewed-by: Dev Jain <dev.jain@....com>
> Reviewed-by: Baolin Wang <baolin.wang@...ux.alibaba.com>
> Reviewed-by: Wei Yang <richard.weiyang@...il.com>
> Signed-off-by: Lance Yang <lance.yang@...ux.dev>
> ---
> Applies against commit 0f22abd9096e in mm-new.
>
> v1 -> v2:
> - Collect Reviewed-by from Dev, Wei and Baolin - thanks!
> - Reduce a level of indentation (per Dev)
> - https://lore.kernel.org/linux-mm/20251016033643.10848-1-lance.yang@linux.dev/
>
> mm/khugepaged.c | 29 ++++++++++++++++-------------
> 1 file changed, 16 insertions(+), 13 deletions(-)
>
> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> index d635d821f611..648d9335de00 100644
> --- a/mm/khugepaged.c
> +++ b/mm/khugepaged.c
> @@ -516,7 +516,7 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte,
> pte_t pteval = ptep_get(_pte);
> unsigned long pfn;
>
> - if (pte_none(pteval))
> + if (!pte_present(pteval))
> continue;
> pfn = pte_pfn(pteval);
> if (is_zero_pfn(pfn))
> @@ -690,17 +690,18 @@ static void __collapse_huge_page_copy_succeeded(pte_t *pte,
> address += nr_ptes * PAGE_SIZE) {
> nr_ptes = 1;
> pteval = ptep_get(_pte);
> - if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) {
> + if (pte_none(pteval) ||
> + (pte_present(pteval) && is_zero_pfn(pte_pfn(pteval)))) {
> add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1);
> - if (is_zero_pfn(pte_pfn(pteval))) {
> - /*
> - * ptl mostly unnecessary.
> - */
> - spin_lock(ptl);
> - ptep_clear(vma->vm_mm, address, _pte);
> - spin_unlock(ptl);
> - ksm_might_unmap_zero_page(vma->vm_mm, pteval);
> - }
> + if (pte_none(pteval))
> + continue;
Yeah I'm not sure I really love this refactoring.
Can be:
if (!is_swap_pte(pteval)) {
add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1);
if (!is_zero_pfn(pte_pfn(pteval)))
continue;
...
}
Doing pte_pfn() on a pte_none() PTE is fine.
Obviously as theree's a lot of hate for is_swap_pte() you could also do:
if (pte_none(pteval) || pte_present(pteval)) {
...
}
Which literally open-codes !is_swap_pte().
At the same time, this makes very clear that PTE none is OK.
> + /*
> + * ptl mostly unnecessary.
> + */
> + spin_lock(ptl);
> + ptep_clear(vma->vm_mm, address, _pte);
> + spin_unlock(ptl);
> + ksm_might_unmap_zero_page(vma->vm_mm, pteval);
> } else {
> struct page *src_page = pte_page(pteval);
>
> @@ -794,7 +795,8 @@ static int __collapse_huge_page_copy(pte_t *pte, struct folio *folio,
> unsigned long src_addr = address + i * PAGE_SIZE;
> struct page *src_page;
>
> - if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) {
> + if (pte_none(pteval) ||
> + (pte_present(pteval) && is_zero_pfn(pte_pfn(pteval)))) {
> clear_user_highpage(page, src_addr);
> continue;
> }
> @@ -1294,7 +1296,8 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,
> goto out_unmap;
> }
> }
> - if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) {
> + if (pte_none(pteval) ||
> + (pte_present(pteval) && is_zero_pfn(pte_pfn(pteval)))) {
> ++none_or_zero;
> if (!userfaultfd_armed(vma) &&
> (!cc->is_khugepaged ||
> --
> 2.49.0
>
I mean all of this seems super gross anyway. We're constantly open-coding the
same check over and over again.
static inline bool pte_is_none_or_zero(pte_t pteval)
{
if (is_swap_pte(pteval))
return false;
return is_zero_pfn(pte_pfn(pteval));
}
Put somewhere in a relevant header file.
Or again, if there's distaste at is_swap_pte(), and here maybe it's more valid
not to use it (given name of function).
static inline bool pte_is_none_or_zero(pte_t pteval)
{
/* Non-present entries do not have a PFN to check. */
if (!pte_present(pteval))
return false;
if (pte_none(pteval))
return true;
return is_zero_pfn(pte_pfn(pteval));
}
I think I'm going to do a series to addres the is_swap_pte() mess actually, as
this whole thing is very frustrating.
Thanks, Lorenzo
Powered by blists - more mailing lists