Skip to contentSkip to conversation

fs/proc/task_mmu: Fix loss of young/dirty bits during pagemap scan

PATCH Applied Ryan Roberts 4 participants 5 messages Apr 29, 2024

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fs/proc/task_mmu: Fix loss of young/dirty bits during pagemap scan

make_uffd_wp_pte() was previously doing:

  pte = ptep_get(ptep);
  ptep_modify_prot_start(ptep);
  pte = pte_mkuffd_wp(pte);
  ptep_modify_prot_commit(ptep, pte);

But if another thread accessed or dirtied the pte between the first 2 calls, this could lead to loss of that information. Since ptep_modify_prot_start() gets and clears atomically, the following is the correct pattern and prevents any possible race. Any access after the first call would see an invalid pte and cause a fault:

  pte = ptep_modify_prot_start(ptep);
  pte = pte_mkuffd_wp(pte);
  ptep_modify_prot_commit(ptep, pte);
Fixes 52526ca7fdb9 ("fs/proc/task_mmu: implement IOCTL to get and optionally clear info about PTEs")
---
 fs/proc/task_mmu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/proc/task_mmu.c +1 -1
index 23fbab954c20..af4bc1da0c01 100644
@@ -1825,7 +1825,7 @@ static void make_uffd_wp_pte(struct vm_area_struct *vma,
 		pte_t old_pte;
 
 		old_pte = ptep_modify_prot_start(vma, addr, pte);
-		ptent = pte_mkuffd_wp(ptent);
+		ptent = pte_mkuffd_wp(old_pte);
 		ptep_modify_prot_commit(vma, addr, pte, old_pte, ptent);
 	} else if (is_swap_pte(ptent)) {
 		ptent = pte_swp_mkuffd_wp(ptent);

--
2.25.1

Acked-by on v1 1/1
#1 lore raw
Quoted 28 lines · Ryan Roberts

make_uffd_wp_pte() was previously doing:

   pte = ptep_get(ptep);
   ptep_modify_prot_start(ptep);
   pte = pte_mkuffd_wp(pte);
   ptep_modify_prot_commit(ptep, pte);

But if another thread accessed or dirtied the pte between the first 2 calls, this could lead to loss of that information. Since ptep_modify_prot_start() gets and clears atomically, the following is the correct pattern and prevents any possible race. Any access after the first call would see an invalid pte and cause a fault:

   pte = ptep_modify_prot_start(ptep);
   pte = pte_mkuffd_wp(pte);
   ptep_modify_prot_commit(ptep, pte);
Fixes 52526ca7fdb9 ("fs/proc/task_mmu: implement IOCTL to get and optionally clear info about PTEs")
---
  fs/proc/task_mmu.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
fs/proc/task_mmu.c +0 -0
index 23fbab954c20..af4bc1da0c01 100644

fs/proc/task_mmu.c static void make_uffd_wp_pte(struct vm_area_struct *vma, · line 1829

Quoted 6 lines
fs/proc/task_mmu.c +1 -1
@@ -1825,7 +1825,7 @@ static void make_uffd_wp_pte(struct vm_area_struct *vma,
  		pte_t old_pte;
 
  		old_pte = ptep_modify_prot_start(vma, addr, pte);
-		ptent = pte_mkuffd_wp(ptent);
+		ptent = pte_mkuffd_wp(old_pte);
  		ptep_modify_prot_commit(vma, addr, pte, old_pte, ptent);
Signature · 3 lines
Cheers,

David / dhildenb
Reviewed-by Tested-by on v1 1/1
#2 lore raw
In reply to Ryan Roberts

Thanks!

Quoted 13 lines · Ryan Roberts

make_uffd_wp_pte() was previously doing:

  pte = ptep_get(ptep);
  ptep_modify_prot_start(ptep);
  pte = pte_mkuffd_wp(pte);
  ptep_modify_prot_commit(ptep, pte);

But if another thread accessed or dirtied the pte between the first 2 calls, this could lead to loss of that information. Since ptep_modify_prot_start() gets and clears atomically, the following is the correct pattern and prevents any possible race. Any access after the first call would see an invalid pte and cause a fault:

Quoted 6 lines
  pte = ptep_modify_prot_start(ptep);
  pte = pte_mkuffd_wp(pte);
  ptep_modify_prot_commit(ptep, pte);
Fixes 52526ca7fdb9 ("fs/proc/task_mmu: implement IOCTL to get and optionally clear info about PTEs")
Quoted 20 lines
---
 fs/proc/task_mmu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/proc/task_mmu.c +1 -2
index 23fbab954c20..af4bc1da0c01 100644
@@ -1825,7 +1825,7 @@ static void make_uffd_wp_pte(struct vm_area_struct *vma,
 		pte_t old_pte;
 
 		old_pte = ptep_modify_prot_start(vma, addr, pte);
-		ptent = pte_mkuffd_wp(ptent);
+		ptent = pte_mkuffd_wp(old_pte);
 		ptep_modify_prot_commit(vma, addr, pte, old_pte, ptent);
 	} else if (is_swap_pte(ptent)) {
 		ptent = pte_swp_mkuffd_wp(ptent);
--

2.25.1

Signature · 2 lines
BR,
Muhammad Usama Anjum
#3 lore raw
In reply to Ryan Roberts
Quoted 13 lines · Ryan Roberts

make_uffd_wp_pte() was previously doing:

  pte = ptep_get(ptep);
  ptep_modify_prot_start(ptep);
  pte = pte_mkuffd_wp(pte);
  ptep_modify_prot_commit(ptep, pte);

But if another thread accessed or dirtied the pte between the first 2 calls, this could lead to loss of that information. Since ptep_modify_prot_start() gets and clears atomically, the following is the correct pattern and prevents any possible race. Any access after the first call would see an invalid pte and cause a fault:

Quoted 6 lines
  pte = ptep_modify_prot_start(ptep);
  pte = pte_mkuffd_wp(pte);
  ptep_modify_prot_commit(ptep, pte);
Fixes 52526ca7fdb9 ("fs/proc/task_mmu: implement IOCTL to get and optionally clear info about PTEs")

I guess this should have cc'ed stable but I forgot to add it. Are you able to fix this up when you take it, Andrew, or do I need to repost?

Quoted 20 lines
---
 fs/proc/task_mmu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/proc/task_mmu.c +1 -2
index 23fbab954c20..af4bc1da0c01 100644
@@ -1825,7 +1825,7 @@ static void make_uffd_wp_pte(struct vm_area_struct *vma,
 		pte_t old_pte;
 
 		old_pte = ptep_modify_prot_start(vma, addr, pte);
-		ptent = pte_mkuffd_wp(ptent);
+		ptent = pte_mkuffd_wp(old_pte);
 		ptep_modify_prot_commit(vma, addr, pte, old_pte, ptent);
 	} else if (is_swap_pte(ptent)) {
 		ptent = pte_swp_mkuffd_wp(ptent);
--

2.25.1

maintainer
#4 lore raw
Quoted 5 lines · On Mon, 29 Apr 2024 16:55:07 +0100 Ryan Roberts <[email protected]> wrote:
Quoted 2 lines
Fixes 52526ca7fdb9 ("fs/proc/task_mmu: implement IOCTL to get and optionally clear info about PTEs")

I guess this should have cc'ed stable but I forgot to add it. Are you able to fix this up when you take it, Andrew, or do I need to repost?

I have made that change.