mm/mprotect.c | 23 ++++++++++------------- 1 file changed, 10 insertions(+), 13 deletions(-)
In commit_anon_folio_batch(), we iterate over all pages pointed to by the
PTE batch. Therefore we need to know the first page of the batch;
currently we derive that via folio_page(folio, 0), but, that takes us
to the first (head) page of the folio instead - our PTE batch may lie
in the middle of the folio, leading to incorrectness.
Bite the bullet and throw away the micro-optimization of reusing the
folio in favour of code simplicity. Derive the page and the folio in
change_pte_range, and pass the page too to commit_anon_folio_batch to
fix the aforementioned issue.
Reported-by: syzbot+57bcc752f0df8bb1365c@syzkaller.appspotmail.com
Fixes: cac1db8c3aad ("mm: optimize mprotect() by PTE batching")
Signed-off-by: Dev Jain <dev.jain@arm.com>
---
mm/mprotect.c | 23 ++++++++++-------------
1 file changed, 10 insertions(+), 13 deletions(-)
diff --git a/mm/mprotect.c b/mm/mprotect.c
index 78bded7acf79..113b48985834 100644
--- a/mm/mprotect.c
+++ b/mm/mprotect.c
@@ -120,9 +120,8 @@ static int mprotect_folio_pte_batch(struct folio *folio, pte_t *ptep,
static bool prot_numa_skip(struct vm_area_struct *vma, unsigned long addr,
pte_t oldpte, pte_t *pte, int target_node,
- struct folio **foliop)
+ struct folio *folio)
{
- struct folio *folio = NULL;
bool ret = true;
bool toptier;
int nid;
@@ -131,7 +130,6 @@ static bool prot_numa_skip(struct vm_area_struct *vma, unsigned long addr,
if (pte_protnone(oldpte))
goto skip;
- folio = vm_normal_folio(vma, addr, oldpte);
if (!folio)
goto skip;
@@ -173,7 +171,6 @@ static bool prot_numa_skip(struct vm_area_struct *vma, unsigned long addr,
folio_xchg_access_time(folio, jiffies_to_msecs(jiffies));
skip:
- *foliop = folio;
return ret;
}
@@ -231,10 +228,9 @@ static int page_anon_exclusive_sub_batch(int start_idx, int max_len,
* retrieve sub-batches.
*/
static void commit_anon_folio_batch(struct vm_area_struct *vma,
- struct folio *folio, unsigned long addr, pte_t *ptep,
+ struct folio *folio, struct page *first_page, unsigned long addr, pte_t *ptep,
pte_t oldpte, pte_t ptent, int nr_ptes, struct mmu_gather *tlb)
{
- struct page *first_page = folio_page(folio, 0);
bool expected_anon_exclusive;
int sub_batch_idx = 0;
int len;
@@ -251,7 +247,7 @@ static void commit_anon_folio_batch(struct vm_area_struct *vma,
}
static void set_write_prot_commit_flush_ptes(struct vm_area_struct *vma,
- struct folio *folio, unsigned long addr, pte_t *ptep,
+ struct folio *folio, struct page *page, unsigned long addr, pte_t *ptep,
pte_t oldpte, pte_t ptent, int nr_ptes, struct mmu_gather *tlb)
{
bool set_write;
@@ -270,7 +266,7 @@ static void set_write_prot_commit_flush_ptes(struct vm_area_struct *vma,
/* idx = */ 0, set_write, tlb);
return;
}
- commit_anon_folio_batch(vma, folio, addr, ptep, oldpte, ptent, nr_ptes, tlb);
+ commit_anon_folio_batch(vma, folio, page, addr, ptep, oldpte, ptent, nr_ptes, tlb);
}
static long change_pte_range(struct mmu_gather *tlb,
@@ -305,15 +301,19 @@ static long change_pte_range(struct mmu_gather *tlb,
const fpb_t flags = FPB_RESPECT_SOFT_DIRTY | FPB_RESPECT_WRITE;
int max_nr_ptes = (end - addr) >> PAGE_SHIFT;
struct folio *folio = NULL;
+ struct page *page;
pte_t ptent;
+ page = vm_normal_page(vma, addr, oldpte);
+ if (page)
+ folio = page_folio(page);
/*
* Avoid trapping faults against the zero or KSM
* pages. See similar comment in change_huge_pmd.
*/
if (prot_numa) {
int ret = prot_numa_skip(vma, addr, oldpte, pte,
- target_node, &folio);
+ target_node, folio);
if (ret) {
/* determine batch to skip */
@@ -323,9 +323,6 @@ static long change_pte_range(struct mmu_gather *tlb,
}
}
- if (!folio)
- folio = vm_normal_folio(vma, addr, oldpte);
-
nr_ptes = mprotect_folio_pte_batch(folio, pte, oldpte, max_nr_ptes, flags);
oldpte = modify_prot_start_ptes(vma, addr, pte, nr_ptes);
@@ -351,7 +348,7 @@ static long change_pte_range(struct mmu_gather *tlb,
*/
if ((cp_flags & MM_CP_TRY_CHANGE_WRITABLE) &&
!pte_write(ptent))
- set_write_prot_commit_flush_ptes(vma, folio,
+ set_write_prot_commit_flush_ptes(vma, folio, page,
addr, pte, oldpte, ptent, nr_ptes, tlb);
else
prot_commit_flush_ptes(vma, addr, pte, oldpte, ptent,
--
2.30.2On 06.08.25 16:56, Dev Jain wrote:
> In commit_anon_folio_batch(), we iterate over all pages pointed to by the
> PTE batch. Therefore we need to know the first page of the batch;
> currently we derive that via folio_page(folio, 0), but, that takes us
> to the first (head) page of the folio instead - our PTE batch may lie
> in the middle of the folio, leading to incorrectness.
>
> Bite the bullet and throw away the micro-optimization of reusing the
> folio in favour of code simplicity.
Huh? We are still reusing the folio. There is a single
vm_normal_page()+page_folio() lookup in that code.
Maybe what you meant is "Simplify the code by moving the folio lookup
out of prot_numa_skip() such that we only have a single page+folio lookup."
> Derive the page and the folio in
> change_pte_range, and pass the page too to commit_anon_folio_batch to
> fix the aforementioned issue.
>
> Reported-by: syzbot+57bcc752f0df8bb1365c@syzkaller.appspotmail.com
> Fixes: cac1db8c3aad ("mm: optimize mprotect() by PTE batching")
Fell free to add a
Debugged-by: David Hildenbrand <david@redhat.com>
> Signed-off-by: Dev Jain <dev.jain@arm.com>
Acked-by: David Hildenbrand <david@redhat.com>
--
Cheers,
David / dhildenb
On Wed, Aug 06, 2025 at 08:26:11PM +0530, Dev Jain wrote:
> In commit_anon_folio_batch(), we iterate over all pages pointed to by the
> PTE batch. Therefore we need to know the first page of the batch;
> currently we derive that via folio_page(folio, 0), but, that takes us
> to the first (head) page of the folio instead - our PTE batch may lie
> in the middle of the folio, leading to incorrectness.
>
> Bite the bullet and throw away the micro-optimization of reusing the
> folio in favour of code simplicity. Derive the page and the folio in
> change_pte_range, and pass the page too to commit_anon_folio_batch to
> fix the aforementioned issue.
>
> Reported-by: syzbot+57bcc752f0df8bb1365c@syzkaller.appspotmail.com
> Fixes: cac1db8c3aad ("mm: optimize mprotect() by PTE batching")
> Signed-off-by: Dev Jain <dev.jain@arm.com>
This looks reasonable, fixes the problem and compiles/works on my machine so:
Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
This badly needs refactoring as 13 parameters being passed to a function is
ridiculous, but we can do that later.
Let's get this in as a hotfix asap.
> ---
> mm/mprotect.c | 23 ++++++++++-------------
> 1 file changed, 10 insertions(+), 13 deletions(-)
>
> diff --git a/mm/mprotect.c b/mm/mprotect.c
> index 78bded7acf79..113b48985834 100644
> --- a/mm/mprotect.c
> +++ b/mm/mprotect.c
> @@ -120,9 +120,8 @@ static int mprotect_folio_pte_batch(struct folio *folio, pte_t *ptep,
>
> static bool prot_numa_skip(struct vm_area_struct *vma, unsigned long addr,
> pte_t oldpte, pte_t *pte, int target_node,
> - struct folio **foliop)
> + struct folio *folio)
> {
> - struct folio *folio = NULL;
> bool ret = true;
> bool toptier;
> int nid;
> @@ -131,7 +130,6 @@ static bool prot_numa_skip(struct vm_area_struct *vma, unsigned long addr,
> if (pte_protnone(oldpte))
> goto skip;
>
> - folio = vm_normal_folio(vma, addr, oldpte);
> if (!folio)
> goto skip;
>
> @@ -173,7 +171,6 @@ static bool prot_numa_skip(struct vm_area_struct *vma, unsigned long addr,
> folio_xchg_access_time(folio, jiffies_to_msecs(jiffies));
>
> skip:
> - *foliop = folio;
> return ret;
> }
>
> @@ -231,10 +228,9 @@ static int page_anon_exclusive_sub_batch(int start_idx, int max_len,
> * retrieve sub-batches.
> */
> static void commit_anon_folio_batch(struct vm_area_struct *vma,
> - struct folio *folio, unsigned long addr, pte_t *ptep,
> + struct folio *folio, struct page *first_page, unsigned long addr, pte_t *ptep,
> pte_t oldpte, pte_t ptent, int nr_ptes, struct mmu_gather *tlb)
> {
> - struct page *first_page = folio_page(folio, 0);
> bool expected_anon_exclusive;
> int sub_batch_idx = 0;
> int len;
> @@ -251,7 +247,7 @@ static void commit_anon_folio_batch(struct vm_area_struct *vma,
> }
>
> static void set_write_prot_commit_flush_ptes(struct vm_area_struct *vma,
> - struct folio *folio, unsigned long addr, pte_t *ptep,
> + struct folio *folio, struct page *page, unsigned long addr, pte_t *ptep,
> pte_t oldpte, pte_t ptent, int nr_ptes, struct mmu_gather *tlb)
> {
> bool set_write;
> @@ -270,7 +266,7 @@ static void set_write_prot_commit_flush_ptes(struct vm_area_struct *vma,
> /* idx = */ 0, set_write, tlb);
> return;
> }
> - commit_anon_folio_batch(vma, folio, addr, ptep, oldpte, ptent, nr_ptes, tlb);
> + commit_anon_folio_batch(vma, folio, page, addr, ptep, oldpte, ptent, nr_ptes, tlb);
> }
>
> static long change_pte_range(struct mmu_gather *tlb,
> @@ -305,15 +301,19 @@ static long change_pte_range(struct mmu_gather *tlb,
> const fpb_t flags = FPB_RESPECT_SOFT_DIRTY | FPB_RESPECT_WRITE;
> int max_nr_ptes = (end - addr) >> PAGE_SHIFT;
> struct folio *folio = NULL;
> + struct page *page;
> pte_t ptent;
>
> + page = vm_normal_page(vma, addr, oldpte);
> + if (page)
> + folio = page_folio(page);
> /*
> * Avoid trapping faults against the zero or KSM
> * pages. See similar comment in change_huge_pmd.
> */
> if (prot_numa) {
> int ret = prot_numa_skip(vma, addr, oldpte, pte,
> - target_node, &folio);
> + target_node, folio);
> if (ret) {
>
> /* determine batch to skip */
> @@ -323,9 +323,6 @@ static long change_pte_range(struct mmu_gather *tlb,
> }
> }
>
> - if (!folio)
> - folio = vm_normal_folio(vma, addr, oldpte);
> -
> nr_ptes = mprotect_folio_pte_batch(folio, pte, oldpte, max_nr_ptes, flags);
>
> oldpte = modify_prot_start_ptes(vma, addr, pte, nr_ptes);
> @@ -351,7 +348,7 @@ static long change_pte_range(struct mmu_gather *tlb,
> */
> if ((cp_flags & MM_CP_TRY_CHANGE_WRITABLE) &&
> !pte_write(ptent))
> - set_write_prot_commit_flush_ptes(vma, folio,
> + set_write_prot_commit_flush_ptes(vma, folio, page,
> addr, pte, oldpte, ptent, nr_ptes, tlb);
> else
> prot_commit_flush_ptes(vma, addr, pte, oldpte, ptent,
> --
> 2.30.2
>
On 06/08/25 10:10 pm, Lorenzo Stoakes wrote:
> On Wed, Aug 06, 2025 at 08:26:11PM +0530, Dev Jain wrote:
>> In commit_anon_folio_batch(), we iterate over all pages pointed to by the
>> PTE batch. Therefore we need to know the first page of the batch;
>> currently we derive that via folio_page(folio, 0), but, that takes us
>> to the first (head) page of the folio instead - our PTE batch may lie
>> in the middle of the folio, leading to incorrectness.
>>
>> Bite the bullet and throw away the micro-optimization of reusing the
>> folio in favour of code simplicity. Derive the page and the folio in
>> change_pte_range, and pass the page too to commit_anon_folio_batch to
>> fix the aforementioned issue.
>>
>> Reported-by: syzbot+57bcc752f0df8bb1365c@syzkaller.appspotmail.com
>> Fixes: cac1db8c3aad ("mm: optimize mprotect() by PTE batching")
>> Signed-off-by: Dev Jain <dev.jain@arm.com>
> This looks reasonable, fixes the problem and compiles/works on my machine so:
>
> Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Thanks.
>
> This badly needs refactoring as 13 parameters being passed to a function is
Which function are you talking about?
> ridiculous, but we can do that later.
>
> Let's get this in as a hotfix asap.
>
>> ---
>> mm/mprotect.c | 23 ++++++++++-------------
>> 1 file changed, 10 insertions(+), 13 deletions(-)
>>
>> diff --git a/mm/mprotect.c b/mm/mprotect.c
>> index 78bded7acf79..113b48985834 100644
>> --- a/mm/mprotect.c
>> +++ b/mm/mprotect.c
>> @@ -120,9 +120,8 @@ static int mprotect_folio_pte_batch(struct folio *folio, pte_t *ptep,
>>
>> static bool prot_numa_skip(struct vm_area_struct *vma, unsigned long addr,
>> pte_t oldpte, pte_t *pte, int target_node,
>> - struct folio **foliop)
>> + struct folio *folio)
>> {
>> - struct folio *folio = NULL;
>> bool ret = true;
>> bool toptier;
>> int nid;
>> @@ -131,7 +130,6 @@ static bool prot_numa_skip(struct vm_area_struct *vma, unsigned long addr,
>> if (pte_protnone(oldpte))
>> goto skip;
>>
>> - folio = vm_normal_folio(vma, addr, oldpte);
>> if (!folio)
>> goto skip;
>>
>> @@ -173,7 +171,6 @@ static bool prot_numa_skip(struct vm_area_struct *vma, unsigned long addr,
>> folio_xchg_access_time(folio, jiffies_to_msecs(jiffies));
>>
>> skip:
>> - *foliop = folio;
>> return ret;
>> }
>>
>> @@ -231,10 +228,9 @@ static int page_anon_exclusive_sub_batch(int start_idx, int max_len,
>> * retrieve sub-batches.
>> */
>> static void commit_anon_folio_batch(struct vm_area_struct *vma,
>> - struct folio *folio, unsigned long addr, pte_t *ptep,
>> + struct folio *folio, struct page *first_page, unsigned long addr, pte_t *ptep,
>> pte_t oldpte, pte_t ptent, int nr_ptes, struct mmu_gather *tlb)
>> {
>> - struct page *first_page = folio_page(folio, 0);
>> bool expected_anon_exclusive;
>> int sub_batch_idx = 0;
>> int len;
>> @@ -251,7 +247,7 @@ static void commit_anon_folio_batch(struct vm_area_struct *vma,
>> }
>>
>> static void set_write_prot_commit_flush_ptes(struct vm_area_struct *vma,
>> - struct folio *folio, unsigned long addr, pte_t *ptep,
>> + struct folio *folio, struct page *page, unsigned long addr, pte_t *ptep,
>> pte_t oldpte, pte_t ptent, int nr_ptes, struct mmu_gather *tlb)
>> {
>> bool set_write;
>> @@ -270,7 +266,7 @@ static void set_write_prot_commit_flush_ptes(struct vm_area_struct *vma,
>> /* idx = */ 0, set_write, tlb);
>> return;
>> }
>> - commit_anon_folio_batch(vma, folio, addr, ptep, oldpte, ptent, nr_ptes, tlb);
>> + commit_anon_folio_batch(vma, folio, page, addr, ptep, oldpte, ptent, nr_ptes, tlb);
>> }
>>
>> static long change_pte_range(struct mmu_gather *tlb,
>> @@ -305,15 +301,19 @@ static long change_pte_range(struct mmu_gather *tlb,
>> const fpb_t flags = FPB_RESPECT_SOFT_DIRTY | FPB_RESPECT_WRITE;
>> int max_nr_ptes = (end - addr) >> PAGE_SHIFT;
>> struct folio *folio = NULL;
>> + struct page *page;
>> pte_t ptent;
>>
>> + page = vm_normal_page(vma, addr, oldpte);
>> + if (page)
>> + folio = page_folio(page);
>> /*
>> * Avoid trapping faults against the zero or KSM
>> * pages. See similar comment in change_huge_pmd.
>> */
>> if (prot_numa) {
>> int ret = prot_numa_skip(vma, addr, oldpte, pte,
>> - target_node, &folio);
>> + target_node, folio);
>> if (ret) {
>>
>> /* determine batch to skip */
>> @@ -323,9 +323,6 @@ static long change_pte_range(struct mmu_gather *tlb,
>> }
>> }
>>
>> - if (!folio)
>> - folio = vm_normal_folio(vma, addr, oldpte);
>> -
>> nr_ptes = mprotect_folio_pte_batch(folio, pte, oldpte, max_nr_ptes, flags);
>>
>> oldpte = modify_prot_start_ptes(vma, addr, pte, nr_ptes);
>> @@ -351,7 +348,7 @@ static long change_pte_range(struct mmu_gather *tlb,
>> */
>> if ((cp_flags & MM_CP_TRY_CHANGE_WRITABLE) &&
>> !pte_write(ptent))
>> - set_write_prot_commit_flush_ptes(vma, folio,
>> + set_write_prot_commit_flush_ptes(vma, folio, page,
>> addr, pte, oldpte, ptent, nr_ptes, tlb);
>> else
>> prot_commit_flush_ptes(vma, addr, pte, oldpte, ptent,
>> --
>> 2.30.2
>>
On Wed, Aug 06, 2025 at 10:49:46PM +0530, Dev Jain wrote:
>
> On 06/08/25 10:10 pm, Lorenzo Stoakes wrote:
> > On Wed, Aug 06, 2025 at 08:26:11PM +0530, Dev Jain wrote:
> > > In commit_anon_folio_batch(), we iterate over all pages pointed to by the
> > > PTE batch. Therefore we need to know the first page of the batch;
> > > currently we derive that via folio_page(folio, 0), but, that takes us
> > > to the first (head) page of the folio instead - our PTE batch may lie
> > > in the middle of the folio, leading to incorrectness.
> > >
> > > Bite the bullet and throw away the micro-optimization of reusing the
> > > folio in favour of code simplicity. Derive the page and the folio in
> > > change_pte_range, and pass the page too to commit_anon_folio_batch to
> > > fix the aforementioned issue.
> > >
> > > Reported-by: syzbot+57bcc752f0df8bb1365c@syzkaller.appspotmail.com
> > > Fixes: cac1db8c3aad ("mm: optimize mprotect() by PTE batching")
> > > Signed-off-by: Dev Jain <dev.jain@arm.com>
> > This looks reasonable, fixes the problem and compiles/works on my machine so:
> >
> > Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
>
> Thanks.
>
> >
> > This badly needs refactoring as 13 parameters being passed to a function is
>
> Which function are you talking about?
Misread, 9, but it's still silly. mprotect in general needs rework. I may submit
a patch for this at some point...
Anyway let's just get this hotfix in asap.
>
> > ridiculous, but we can do that later.
> >
> > Let's get this in as a hotfix asap.
© 2016 - 2025 Red Hat, Inc.