mm/rmap: fix soft-dirty and uffd-wp bit loss when remapping zero-filled mTHP subpage to shared zeropage
Subsystems (3)
PATCH v3 1/1 mm/rmap: fix soft-dirty and uffd-wp bit loss when remapping zero-filled mTHP subpage to shared zeropage
From: Lance Yang <[email protected]>
When splitting an mTHP and replacing a zero-filled subpage with the shared zeropage, try_to_map_unused_to_zeropage() currently drops several important PTE bits.
For userspace tools like CRIU, which rely on the soft-dirty mechanism for incremental snapshots, losing the soft-dirty bit means modified pages are missed, leading to inconsistent memory state after restore.
As pointed out by David, the more critical uffd-wp bit is also dropped. This breaks the userfaultfd write-protection mechanism, causing writes to be silently missed by monitoring applications, which can lead to data corruption.
Preserve both the soft-dirty and uffd-wp bits from the old PTE when creating the new zeropage mapping to ensure they are correctly tracked.
---
v2 -> v3:
- ptep_get() gets called only once per iteration (per Dev)
- https://lore.kernel.org/linux-mm/[email protected]/
v1 -> v2:
- Avoid calling ptep_get() multiple times (per Dev)
- Double-check the uffd-wp bit (per David)
- Collect Acked-by from David - thanks!
- https://lore.kernel.org/linux-mm/[email protected]/
mm/migrate.c | 14 ++++++++++---- 1 file changed, 10 insertions(+), 4 deletions(-)
mm/migrate.c +10 -4
@@ -297,6 +297,7 @@ bool isolate_folio_to_list(struct folio *folio, struct list_head *list)static bool try_to_map_unused_to_zeropage (struct page_vma_mapped_walk *pvmw,struct folio *folio, +pte_t old_pte,unsigned long idx) {struct page *page =folio_page (folio, idx);
@@ -306,7 +307,7 @@ static bool try_to_map_unused_to_zeropage(struct page_vma_mapped_walk *pvmw,return false;VM_BUG_ON_PAGE (!PageAnon (page), page);VM_BUG_ON_PAGE (!PageLocked (page), page); -VM_BUG_ON_PAGE (pte_present (ptep_get (pvmw->pte)), page); +VM_BUG_ON_PAGE (pte_present (old_pte), page);if (folio_test_mlocked (folio) || (pvmw->vma->vm_flags &VM_LOCKED ) ||mm_forbids_zeropage (pvmw->vma->vm_mm))
@@ -322,6 +323,12 @@ static bool try_to_map_unused_to_zeropage(struct page_vma_mapped_walk *pvmw, newpte =pte_mkspecial (pfn_pte (my_zero_pfn (pvmw->address), pvmw->vma->vm_page_prot)); + +if (pte_swp_soft_dirty (old_pte)) + newpte =pte_mksoft_dirty (newpte); +if (pte_swp_uffd_wp (old_pte)) + newpte =pte_mkuffd_wp (newpte); +set_pte_at (pvmw->vma->vm_mm, pvmw->address, pvmw->pte, newpte);dec_mm_counter (pvmw->vma->vm_mm,mm_counter (folio));
@@ -344,7 +351,7 @@ static bool remove_migration_pte(struct folio *folio,while (page_vma_mapped_walk (&pvmw)) {rmap_t rmap_flags =RMAP_NONE ; -pte_t old_pte; +pte_t old_pte =ptep_get (pvmw.pte);pte_t pte;swp_entry_t entry;struct page *new;
@@ -365,12 +372,11 @@ static bool remove_migration_pte(struct folio *folio, }#endif if (rmap_walk_arg->map_unused_to_zeropage && -try_to_map_unused_to_zeropage (&pvmw, folio, idx)) +try_to_map_unused_to_zeropage (&pvmw, folio, old_pte, idx))continue ;folio_get (folio); pte =mk_pte (new,READ_ONCE (vma->vm_page_prot)); - old_pte =ptep_get (pvmw.pte); entry =pte_to_swp_entry (old_pte);if (!is_migration_entry_young (entry))
Signature · 1 line
2.49.0
Quoted 43 lines · Lance Yang
From: Lance Yang <[email protected]>
When splitting an mTHP and replacing a zero-filled subpage with the shared zeropage,
try_to_map_unused_to_zeropage()currently drops several important PTE bits.For userspace tools like CRIU, which rely on the soft-dirty mechanism for incremental snapshots, losing the soft-dirty bit means modified pages are missed, leading to inconsistent memory state after restore.
As pointed out by David, the more critical uffd-wp bit is also dropped. This breaks the userfaultfd write-protection mechanism, causing writes to be silently missed by monitoring applications, which can lead to data corruption.
Preserve both the soft-dirty and uffd-wp bits from the old PTE when creating the new zeropage mapping to ensure they are correctly tracked.
Cc <[email protected]>Fixes b1f202060afe ("mm: remap unused subpages to shared zeropage when splitting isolated thp")Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>---
v2 -> v3:
-ptep_get()gets called only once per iteration (per Dev)
- https://lore.kernel.org/linux-mm/[email protected]/v1 -> v2:
- Avoid callingptep_get()multiple times (per Dev)
- Double-check the uffd-wp bit (per David)
- Collect Acked-by from David - thanks!
- https://lore.kernel.org/linux-mm/[email protected]/mm/migrate.c | 14 ++++++++++---- 1 file changed, 10 insertions(+), 4 deletions(-)mm/migrate.c +0
-0
mm/migrate.c bool isolate_folio_to_list(struct folio *folio, struct list_head *list) · line 301
Quoted 5 lines
mm/migrate.c +1
-0@@ -297,6 +297,7 @@ bool isolate_folio_to_list(struct folio *folio, struct list_head *list)static bool try_to_map_unused_to_zeropage (struct page_vma_mapped_walk *pvmw,struct folio *folio, +pte_t old_pte,unsigned long idx)
Nit:
static bool try_to_map_unused_to_zeropage(struct page_vma_mapped_walk *pvmw,
struct folio *folio, pte_t old_pte, unsigned long idx)
LGTM, Thanks!
Signature · 3 lines
Cheers David / dhildenb
Quoted 52 lines · David Hildenbrand
Quoted 43 lines · Lance Yang
From: Lance Yang <[email protected]>
When splitting an mTHP and replacing a zero-filled subpage with the shared zeropage,
try_to_map_unused_to_zeropage()currently drops several important PTE bits.For userspace tools like CRIU, which rely on the soft-dirty mechanism for incremental snapshots, losing the soft-dirty bit means modified pages are missed, leading to inconsistent memory state after restore.
As pointed out by David, the more critical uffd-wp bit is also dropped. This breaks the userfaultfd write-protection mechanism, causing writes to be silently missed by monitoring applications, which can lead to data corruption.
Preserve both the soft-dirty and uffd-wp bits from the old PTE when creating the new zeropage mapping to ensure they are correctly tracked.
Cc <[email protected]>Fixes b1f202060afe ("mm: remap unused subpages to shared zeropage when splitting isolated thp")Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>---
v2 -> v3:
-ptep_get()gets called only once per iteration (per Dev)
- https://lore.kernel.org/linux-mm/20250930043351.34927-1- [email protected]/v1 -> v2:
- Avoid callingptep_get()multiple times (per Dev)
- Double-check the uffd-wp bit (per David)
- Collect Acked-by from David - thanks!
- https://lore.kernel.org/linux-mm/20250928044855.76359-1- [email protected]/
mm/migrate.c| 14 ++++++++++----1 file changed, 10 insertions(+), 4 deletions(-)mm/migrate.c +0
-0mm/migrate.c bool isolate_folio_to_list(struct folio *folio, · line 301
Quoted 5 lines
mm/migrate.c +1
-0@@ -297,6 +297,7 @@ bool isolate_folio_to_list(struct folio *folio, struct list_head *list) static bool try_to_map_unused_to_zeropage(struct page_vma_mapped_walk *pvmw, struct folio *folio, + pte_t old_pte, unsigned long idx)Nit:
Quoted 2 lines
static bool try_to_map_unused_to_zeropage(struct
page_vma_mapped_walk*pvmw,
struct folio *folio,pte_told_pte, unsigned long idx)
Well, let me clean that up ;p
Quoted 2 lines
LGTM, Thanks!
Cheers!
Quoted 43 lines · Lance Yang
From: Lance Yang <[email protected]>
When splitting an mTHP and replacing a zero-filled subpage with the shared zeropage,
try_to_map_unused_to_zeropage()currently drops several important PTE bits.For userspace tools like CRIU, which rely on the soft-dirty mechanism for incremental snapshots, losing the soft-dirty bit means modified pages are missed, leading to inconsistent memory state after restore.
As pointed out by David, the more critical uffd-wp bit is also dropped. This breaks the userfaultfd write-protection mechanism, causing writes to be silently missed by monitoring applications, which can lead to data corruption.
Preserve both the soft-dirty and uffd-wp bits from the old PTE when creating the new zeropage mapping to ensure they are correctly tracked.
Cc <[email protected]>Fixes b1f202060afe ("mm: remap unused subpages to shared zeropage when splitting isolated thp")Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>---
v2 -> v3:
-ptep_get()gets called only once per iteration (per Dev)
- https://lore.kernel.org/linux-mm/[email protected]/v1 -> v2:
- Avoid callingptep_get()multiple times (per Dev)
- Double-check the uffd-wp bit (per David)
- Collect Acked-by from David - thanks!
- https://lore.kernel.org/linux-mm/[email protected]/mm/migrate.c | 14 ++++++++++---- 1 file changed, 10 insertions(+), 4 deletions(-)mm/migrate.c +0
-0
mm/migrate.c bool isolate_folio_to_list(struct folio *folio, struct list_head *list) · line 301
Quoted 5 lines
mm/migrate.c +1
-0@@ -297,6 +297,7 @@ bool isolate_folio_to_list(struct folio *folio, struct list_head *list)static bool try_to_map_unused_to_zeropage (struct page_vma_mapped_walk *pvmw,struct folio *folio, +pte_t old_pte,unsigned long idx)
Could have just added this in the same line as folio?
Quoted 37 lines
mm/migrate.c +8
-2{struct page *page =folio_page (folio, idx);@@ -306,7 +307,7 @@ static bool try_to_map_unused_to_zeropage(struct page_vma_mapped_walk *pvmw,return false;VM_BUG_ON_PAGE (!PageAnon (page), page);VM_BUG_ON_PAGE (!PageLocked (page), page); -VM_BUG_ON_PAGE (pte_present (ptep_get (pvmw->pte)), page); +VM_BUG_ON_PAGE (pte_present (old_pte), page);if (folio_test_mlocked (folio) || (pvmw->vma->vm_flags &VM_LOCKED ) ||mm_forbids_zeropage (pvmw->vma->vm_mm))@@ -322,6 +323,12 @@ static bool try_to_map_unused_to_zeropage(struct page_vma_mapped_walk *pvmw, newpte =pte_mkspecial (pfn_pte (my_zero_pfn (pvmw->address), pvmw->vma->vm_page_prot)); + +if (pte_swp_soft_dirty (old_pte)) + newpte =pte_mksoft_dirty (newpte); +if (pte_swp_uffd_wp (old_pte)) + newpte =pte_mkuffd_wp (newpte); +set_pte_at (pvmw->vma->vm_mm, pvmw->address, pvmw->pte, newpte);dec_mm_counter (pvmw->vma->vm_mm,mm_counter (folio));@@ -344,7 +351,7 @@ static bool remove_migration_pte(struct folio *folio,while (page_vma_mapped_walk (&pvmw)) {rmap_t rmap_flags =RMAP_NONE ; -pte_t old_pte; +pte_t old_pte =ptep_get (pvmw.pte);pte_t pte;swp_entry_t entry;struct page *new;@@ -365,12 +372,11 @@ static bool remove_migration_pte(struct folio *folio, }#endif if (rmap_walk_arg->map_unused_to_zeropage &&
mm/migrate.c static bool remove_migration_pte(struct folio *folio, · line 382
Quoted 10 lines
mm/migrate.c +1
-2- try_to_map_unused_to_zeropage(&pvmw, folio, idx)) + try_to_map_unused_to_zeropage(&pvmw, folio, old_pte, idx)) continue;folio_get (folio); pte =mk_pte (new,READ_ONCE (vma->vm_page_prot)); - old_pte =ptep_get (pvmw.pte); entry =pte_to_swp_entry (old_pte);if (!is_migration_entry_young (entry))
Looks good, the special bit does not overlay on any arch with the soft-dirty bit. It shouldn't overlay with uffd-wp as well since split_huge_zero_page_pmd does the same bit preservation.
Quoted 40 lines · Dev Jain
Quoted 38 lines · Lance Yang
From: Lance Yang <[email protected]>
When splitting an mTHP and replacing a zero-filled subpage with the shared zeropage,
try_to_map_unused_to_zeropage()currently drops several important PTE bits.For userspace tools like CRIU, which rely on the soft-dirty mechanism for incremental snapshots, losing the soft-dirty bit means modified pages are missed, leading to inconsistent memory state after restore.
As pointed out by David, the more critical uffd-wp bit is also dropped. This breaks the userfaultfd write-protection mechanism, causing writes to be silently missed by monitoring applications, which can lead to data corruption.
Preserve both the soft-dirty and uffd-wp bits from the old PTE when creating the new zeropage mapping to ensure they are correctly tracked.
Cc <[email protected]>Fixes b1f202060afe ("mm: remap unused subpages to shared zeropage when splitting isolated thp")Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>---
v2 -> v3:
-ptep_get()gets called only once per iteration (per Dev)
- https://lore.kernel.org/linux-mm/20250930043351.34927-1- [email protected]/v1 -> v2:
- Avoid callingptep_get()multiple times (per Dev)
- Double-check the uffd-wp bit (per David)
- Collect Acked-by from David - thanks!
- https://lore.kernel.org/linux-mm/20250928044855.76359-1- [email protected]/
mm/migrate.c| 14 ++++++++++----1 file changed, 10 insertions(+), 4 deletions(-)
Quoted 12 lines
mm/migrate.c bool isolate_folio_to_list(struct folio *folio, · line 301
Quoted 10 lines
mm/migrate.c +1
-0@@ -297,6 +297,7 @@ bool isolate_folio_to_list(struct folio *folio, struct list_head *list) static bool try_to_map_unused_to_zeropage(struct page_vma_mapped_walk *pvmw, struct folio *folio, + pte_t old_pte, unsigned long idx)Could have just added this in the same line as folio?
Sure ;p
Quoted 41 lines
Quoted 41 lines
mm/migrate.c +1
-1{ struct page *page = folio_page(folio, idx);@@ -306,7 +307,7 @@ static bool try_to_map_unused_to_zeropage(structpage_vma_mapped_walk *pvmw, return false; VM_BUG_ON_PAGE(!PageAnon(page), page); VM_BUG_ON_PAGE(!PageLocked(page), page); - VM_BUG_ON_PAGE(pte_present(ptep_get(pvmw->pte)), page); + VM_BUG_ON_PAGE(pte_present(old_pte), page);if (
folio_test_mlocked(folio)|| (pvmw->vma->vm_flags&VM_LOCKED) ||mm_forbids_zeropage(pvmw->vma->vm_mm))mm/migrate.c +7
-1@@ -322,6 +323,12 @@ static bool try_to_map_unused_to_zeropage(structpage_vma_mapped_walk *pvmw, newpte = pte_mkspecial(pfn_pte(my_zero_pfn(pvmw->address), pvmw->vma->vm_page_prot)); + + if (pte_swp_soft_dirty(old_pte)) + newpte = pte_mksoft_dirty(newpte); + if (pte_swp_uffd_wp(old_pte)) + newpte = pte_mkuffd_wp(newpte); + set_pte_at(pvmw->vma->vm_mm, pvmw->address, pvmw->pte, newpte); dec_mm_counter(pvmw->vma->vm_mm, mm_counter(folio));@@ -344,7 +351,7 @@ static bool remove_migration_pte(struct folio *folio, while (page_vma_mapped_walk(&pvmw)) { rmap_t rmap_flags = RMAP_NONE; - pte_t old_pte; + pte_t old_pte = ptep_get(pvmw.pte); pte_t pte; swp_entry_t entry; struct page *new;@@ -365,12 +372,11 @@ static bool remove_migration_pte(struct folio *folio, } #endifif (
rmap_walk_arg->map_unused_to_zeropage&&
-try_to_map_unused_to_zeropage(&pvmw, folio, idx)) +try_to_map_unused_to_zeropage(&pvmw, folio, old_pte, idx)) continue;
folio_get(folio);
Quoted 8 lines
Quoted 4 lines
pte =
mk_pte(new, READ_ONCE(vma->vm_page_prot)); -old_pte=ptep_get(pvmw.pte);
entry =pte_to_swp_entry(old_pte);
if (!is_migration_entry_young(entry))Looks good, the special bit does not overlay on any arch with the soft- dirty bit. It shouldn't overlay with uffd-wp as well since
split_huge_zero_page_pmddoes the same bit preservation.
Yeah. Thanks for double-checking the bit overlaps!
Good to know we're on solid ground here ;)
Quoted 2 lines
Reviewed-by Dev Jain <[email protected]>
Cheers!
syzbot ci has tested the following series
syzbot ci has tested the following series
[v3] mm/rmap: fix soft-dirty and uffd-wp bit loss when remapping zero-filled mTHP subpage to shared zeropage
https://lore.kernel.org/all/[email protected]
* [PATCH v3 1/1] mm/rmap: fix soft-dirty and uffd-wp bit loss when remapping zero-filled mTHP subpage to shared zeropage
and found the following issue:
general protection fault in remove_migration_pte
Full report is available here:
https://ci.syzbot.org/series/a2021abd-c238-431c-a92e-cc29beb53cbf
***
general protection fault in remove_migration_pte
tree: torvalds URL: https://kernel.googlesource.com/pub/scm/linux/kernel/git/torvalds/linux base: 1896ce8eb6c61824f6c1125d69d8fda1f44a22f8 arch: amd64 compiler: Debian clang version 20.1.8 (++20250708063551+0c9f909b7976-1~exp1~20250708183702.136), Debian LLD 20.1.8 config: https://ci.syzbot.org/builds/84a2085e-d609-43ea-8b19-f9af8ea3d54a/config C repro: https://ci.syzbot.org/findings/3e211477-5a8d-4d4d-935b-15076499b001/c_repro syz repro: https://ci.syzbot.org/findings/3e211477-5a8d-4d4d-935b-15076499b001/syz_repro
Oops: general protection fault, probably for non-canonical address 0xdffffc0000000000: 0000 [#1] SMP KASAN PTI KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007] CPU: 0 UID: 0 PID: 5985 Comm: syz.0.27 Not tainted syzkaller #0 PREEMPT(full) Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.2-debian-1.16.2-1 04/01/2014 RIP: 0010:ptep_get include/linux/pgtable.h:340 [inline] RIP: 0010:remove_migration_pte+0x37f/0x2340 mm/migrate.c:361 Code: 43 20 48 89 84 24 08 01 00 00 49 8d 47 40 48 89 84 24 00 01 00 00 4c 89 64 24 50 4c 8b b4 24 70 01 00 00 4c 89 f0 48 c1 e8 03 <42> 80 3c 28 00 74 08 4c 89 f7 e8 22 3e ff ff 49 8b 06 48 89 44 24 RSP: 0018:ffffc90002c2f3c0 EFLAGS: 00010246 RAX: 0000000000000000 RBX: ffff888027799080 RCX: 1ffffd40008d1006 RDX: 0000000000000000 RSI: 00000000000387ff RDI: 0000000000038600 RBP: ffffc90002c2f5d0 R08: 0000000000000003 R09: 0000000000000004 R10: dffffc0000000000 R11: fffff52000585e30 R12: ffffea0004688008 R13: dffffc0000000000 R14: 0000000000000000 R15: ffffea0004688000 FS: 0000555589124500(0000) GS:ffff8880b8d7e000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000200000000300 CR3: 0000000026118000 CR4: 00000000000006f0 Call Trace: <TASK> rmap_walk_anon+0x553/0x730 mm/rmap.c:2842 remove_migration_ptes mm/migrate.c:478 [inline] migrate_folio_move mm/migrate.c:1394 [inline] migrate_folios_move mm/migrate.c:1725 [inline] migrate_pages_batch+0x200a/0x35c0 mm/migrate.c:1972 migrate_pages_sync mm/migrate.c:2002 [inline] migrate_pages+0x1bcc/0x2930 mm/migrate.c:2111 migrate_to_node mm/mempolicy.c:1244 [inline] do_migrate_pages+0x5ee/0x800 mm/mempolicy.c:1343 kernel_migrate_pages mm/mempolicy.c:1858 [inline] __do_sys_migrate_pages mm/mempolicy.c:1876 [inline] __se_sys_migrate_pages+0x544/0x650 mm/mempolicy.c:1872 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0xfa/0x3b0 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7f922b98ec29 Code: ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 a8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007ffccaf966f8 EFLAGS: 00000246 ORIG_RAX: 0000000000000100 RAX: ffffffffffffffda RBX: 00007f922bbd5fa0 RCX: 00007f922b98ec29 RDX: 0000200000000300 RSI: 0000000000000003 RDI: 0000000000000000 RBP: 00007f922ba11e41 R08: 0000000000000000 R09: 0000000000000000 R10: 0000200000000040 R11: 0000000000000246 R12: 0000000000000000 R13: 00007f922bbd5fa0 R14: 00007f922bbd5fa0 R15: 0000000000000004 </TASK> Modules linked in: ---[ end trace 0000000000000000 ]--- RIP: 0010:ptep_get include/linux/pgtable.h:340 [inline] RIP: 0010:remove_migration_pte+0x37f/0x2340 mm/migrate.c:361 Code: 43 20 48 89 84 24 08 01 00 00 49 8d 47 40 48 89 84 24 00 01 00 00 4c 89 64 24 50 4c 8b b4 24 70 01 00 00 4c 89 f0 48 c1 e8 03 <42> 80 3c 28 00 74 08 4c 89 f7 e8 22 3e ff ff 49 8b 06 48 89 44 24 RSP: 0018:ffffc90002c2f3c0 EFLAGS: 00010246 RAX: 0000000000000000 RBX: ffff888027799080 RCX: 1ffffd40008d1006 RDX: 0000000000000000 RSI: 00000000000387ff RDI: 0000000000038600 RBP: ffffc90002c2f5d0 R08: 0000000000000003 R09: 0000000000000004 R10: dffffc0000000000 R11: fffff52000585e30 R12: ffffea0004688008 R13: dffffc0000000000 R14: 0000000000000000 R15: ffffea0004688000 FS: 0000555589124500(0000) GS:ffff8880b8d7e000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000200000000300 CR3: 0000000026118000 CR4: 00000000000006f0 ---------------- Code disassembly (best guess): 0: 43 20 48 89 rex.XB and %cl,-0x77(%r8) 4: 84 24 08 test %ah,(%rax,%rcx,1) 7: 01 00 add %eax,(%rax) 9: 00 49 8d add %cl,-0x73(%rcx) c: 47 rex.RXB d: 40 rex e: 48 89 84 24 00 01 00 mov %rax,0x100(%rsp) 15: 00 16: 4c 89 64 24 50 mov %r12,0x50(%rsp) 1b: 4c 8b b4 24 70 01 00 mov 0x170(%rsp),%r14 22: 00 23: 4c 89 f0 mov %r14,%rax 26: 48 c1 e8 03 shr $0x3,%rax * 2a: 42 80 3c 28 00 cmpb $0x0,(%rax,%r13,1) <-- trapping instruction 2f: 74 08 je 0x39 31: 4c 89 f7 mov %r14,%rdi 34: e8 22 3e ff ff call 0xffff3e5b 39: 49 8b 06 mov (%r14),%rax 3c: 48 rex.W 3d: 89 .byte 0x89 3e: 44 rex.R 3f: 24 .byte 0x24
***
If these findings have caused you to resend the series or submit a separate fix, please add the following tag to your commit message:
---
This report is generated by a bot. It may contain errors. syzbot ci engineers can be reached at [email protected].
Thanks for the report!
Quoted 25 lines · syzbot ci
syzbot ci has tested the following series
[v3] mm/rmap: fix soft-dirty and uffd-wp bit loss when remapping zero-filled mTHP subpage to shared zeropage
https://lore.kernel.org/all/[email protected]
* [PATCH v3 1/1] mm/rmap: fix soft-dirty and uffd-wp bit loss when remapping zero-filled mTHP subpage to shared zeropageand found the following issue:
general protection fault in remove_migration_pteFull report is available here:
https://ci.syzbot.org/series/a2021abd-c238-431c-a92e-cc29beb53cbf***
general protection fault in remove_migration_ptetree: torvalds URL: https://kernel.googlesource.com/pub/scm/linux/kernel/git/torvalds/linux base: 1896ce8eb6c61824f6c1125d69d8fda1f44a22f8 arch: amd64 compiler: Debian clang version 20.1.8 (++20250708063551+0c9f909b7976-1~exp1~20250708183702.136), Debian LLD 20.1.8 config: https://ci.syzbot.org/builds/84a2085e-d609-43ea-8b19-f9af8ea3d54a/config C repro: https://ci.syzbot.org/findings/3e211477-5a8d-4d4d-935b-15076499b001/c_repro syz repro: https://ci.syzbot.org/findings/3e211477-5a8d-4d4d-935b-15076499b001/syz_repro
Quoted 2 lines
Oops: general protection fault, probably for non-canonical address 0xdffffc0000000000: 0000 [#1] SMP KASAN PTI KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007]
This is a known issue that I introduced in the v3 patch. I spotted this exact NULL pointer dereference bug[1] myself and have already sent out a v5 version[2] with the fix.
The root cause is that ptep_get() is called before the !pwmw.pte check, which handles PMD-mapped THP migration entries.
[1] https://lore.kernel.org/linux-mm/[email protected]
[2] https://lore.kernel.org/linux-mm/[email protected]
Thanks,
Lance
Quoted 89 lines
CPU: 0 UID: 0 PID: 5985 Comm: syz.0.27 Not tainted syzkaller #0 PREEMPT(full) Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.2-debian-1.16.2-1 04/01/2014 RIP: 0010:ptep_get include/linux/pgtable.h:340 [inline] RIP: 0010:remove_migration_pte+0x37f/0x2340 mm/migrate.c:361 Code: 43 20 48 89 84 24 08 01 00 00 49 8d 47 40 48 89 84 24 00 01 00 00 4c 89 64 24 50 4c 8b b4 24 70 01 00 00 4c 89 f0 48 c1 e8 03 <42> 80 3c 28 00 74 08 4c 89 f7 e8 22 3e ff ff 49 8b 06 48 89 44 24 RSP: 0018:ffffc90002c2f3c0 EFLAGS: 00010246 RAX: 0000000000000000 RBX: ffff888027799080 RCX: 1ffffd40008d1006 RDX: 0000000000000000 RSI: 00000000000387ff RDI: 0000000000038600 RBP: ffffc90002c2f5d0 R08: 0000000000000003 R09: 0000000000000004 R10: dffffc0000000000 R11: fffff52000585e30 R12: ffffea0004688008 R13: dffffc0000000000 R14: 0000000000000000 R15: ffffea0004688000 FS: 0000555589124500(0000) GS:ffff8880b8d7e000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000200000000300 CR3: 0000000026118000 CR4: 00000000000006f0 Call Trace: <TASK> rmap_walk_anon+0x553/0x730 mm/rmap.c:2842 remove_migration_ptes mm/migrate.c:478 [inline] migrate_folio_move mm/migrate.c:1394 [inline] migrate_folios_move mm/migrate.c:1725 [inline] migrate_pages_batch+0x200a/0x35c0 mm/migrate.c:1972 migrate_pages_sync mm/migrate.c:2002 [inline] migrate_pages+0x1bcc/0x2930 mm/migrate.c:2111 migrate_to_node mm/mempolicy.c:1244 [inline] do_migrate_pages+0x5ee/0x800 mm/mempolicy.c:1343 kernel_migrate_pages mm/mempolicy.c:1858 [inline] __do_sys_migrate_pages mm/mempolicy.c:1876 [inline] __se_sys_migrate_pages+0x544/0x650 mm/mempolicy.c:1872 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0xfa/0x3b0 arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7f922b98ec29 Code: ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 a8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007ffccaf966f8 EFLAGS: 00000246 ORIG_RAX: 0000000000000100 RAX: ffffffffffffffda RBX: 00007f922bbd5fa0 RCX: 00007f922b98ec29 RDX: 0000200000000300 RSI: 0000000000000003 RDI: 0000000000000000 RBP: 00007f922ba11e41 R08: 0000000000000000 R09: 0000000000000000 R10: 0000200000000040 R11: 0000000000000246 R12: 0000000000000000 R13: 00007f922bbd5fa0 R14: 00007f922bbd5fa0 R15: 0000000000000004 </TASK> Modules linked in: ---[ end trace 0000000000000000 ]--- RIP: 0010:ptep_get include/linux/pgtable.h:340 [inline] RIP: 0010:remove_migration_pte+0x37f/0x2340 mm/migrate.c:361 Code: 43 20 48 89 84 24 08 01 00 00 49 8d 47 40 48 89 84 24 00 01 00 00 4c 89 64 24 50 4c 8b b4 24 70 01 00 00 4c 89 f0 48 c1 e8 03 <42> 80 3c 28 00 74 08 4c 89 f7 e8 22 3e ff ff 49 8b 06 48 89 44 24 RSP: 0018:ffffc90002c2f3c0 EFLAGS: 00010246 RAX: 0000000000000000 RBX: ffff888027799080 RCX: 1ffffd40008d1006 RDX: 0000000000000000 RSI: 00000000000387ff RDI: 0000000000038600 RBP: ffffc90002c2f5d0 R08: 0000000000000003 R09: 0000000000000004 R10: dffffc0000000000 R11: fffff52000585e30 R12: ffffea0004688008 R13: dffffc0000000000 R14: 0000000000000000 R15: ffffea0004688000 FS: 0000555589124500(0000) GS:ffff8880b8d7e000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000200000000300 CR3: 0000000026118000 CR4: 00000000000006f0 ---------------- Code disassembly (best guess): 0: 43 20 48 89 rex.XB and %cl,-0x77(%r8) 4: 84 24 08 test %ah,(%rax,%rcx,1) 7: 01 00 add %eax,(%rax) 9: 00 49 8d add %cl,-0x73(%rcx) c: 47 rex.RXB d: 40 rex e: 48 89 84 24 00 01 00 mov %rax,0x100(%rsp) 15: 00 16: 4c 89 64 24 50 mov %r12,0x50(%rsp) 1b: 4c 8b b4 24 70 01 00 mov 0x170(%rsp),%r14 22: 00 23: 4c 89 f0 mov %r14,%rax 26: 48 c1 e8 03 shr $0x3,%rax * 2a: 42 80 3c 28 00 cmpb $0x0,(%rax,%r13,1) <-- trapping instruction 2f: 74 08 je 0x39 31: 4c 89 f7 mov %r14,%rdi 34: e8 22 3e ff ff call 0xffff3e5b 39: 49 8b 06 mov (%r14),%rax 3c: 48 rex.W 3d: 89 .byte 0x89 3e: 44 rex.R 3f: 24 .byte 0x24***
If these findings have caused you to resend the series or submit a separate fix, please add the following tag to your commit message:
Tested-by [email protected]---
This report is generated by a bot. It may contain errors. syzbot ci engineers can be reached at [email protected].