Skip to contentSkip to conversation

mm: fix up softleaf vs thp device-private in remove_migration_pmd()

PATCH In review Balbir Singh 2 participants 7 messages Nov 14, 2025

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mm: fix up softleaf vs thp device-private in remove_migration_pmd()

A fix-up in the thp device-private patch
mm/rmap: extend rmap and migration support device-private entries introduces conflicts with the softleaf series, this is a fix up patch to resolve the issue

Cc Andrew Morton <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Zi Yan <[email protected]>
Cc Joshua Hahn <[email protected]>
Cc Rakie Kim <[email protected]>
Cc Byungchul Park <[email protected]>
Cc Gregory Price <[email protected]>
Cc Ying Huang <[email protected]>
Cc Alistair Popple <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Baolin Wang <[email protected]>
Cc "Liam R. Howlett" <[email protected]>
Cc Nico Pache <[email protected]>
Cc Ryan Roberts <[email protected]>
Cc Dev Jain <[email protected]>
Cc Barry Song <[email protected]>
Cc Lyude Paul <[email protected]>
Cc Danilo Krummrich <[email protected]>
Cc David Airlie <[email protected]>
Cc Simona Vetter <[email protected]>
Cc Ralph Campbell <[email protected]>
Cc Mika Penttilä <[email protected]>
Cc Matthew Brost <[email protected]>
Cc Francois Dugast <[email protected]>
---
 mm/huge_memory.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
mm/huge_memory.c +1 -3
index 3a8f0c4143c5..4d65c51bc0fe 100644
@@ -4760,7 +4760,7 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
 	folio_get(folio);
 	pmde = folio_mk_pmd(folio, READ_ONCE(vma->vm_page_prot));
 
-	if (is_writable_migration_entry(entry))
+	if (softleaf_is_migration_write(entry))
 		pmde = pmd_mkwrite(pmde, vma);
 
 	if (folio_is_device_private(folio)) {
@@ -4777,8 +4777,6 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
 
 	if (pmd_swp_soft_dirty(*pvmw->pmd))
 		pmde = pmd_mksoft_dirty(pmde);
-	if (softleaf_is_migration_write(entry))
-		pmde = pmd_mkwrite(pmde, vma);
 	if (pmd_swp_uffd_wp(*pvmw->pmd))
 		pmde = pmd_mkuffd_wp(pmde);
 	if (!softleaf_is_migration_young(entry))
Signature · 1 line
2.51.1
#1 lore raw

PATCH 1/1 mm/huge_memory: fix override of entry in remove_migration_pmd

Recent changes exposed a BUG in remove_migration_pmd() where the migration entry was being overridden when the folio is device private.

Use scope local entry for creating the device private pmde. Make the pmde writable if the migration entry is writable by moving the check is_migration_write() prior to creating the device private entry.

Cc Andrew Morton <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Zi Yan <[email protected]>
Cc Joshua Hahn <[email protected]>
Cc Rakie Kim <[email protected]>
Cc Byungchul Park <[email protected]>
Cc Gregory Price <[email protected]>
Cc Ying Huang <[email protected]>
Cc Alistair Popple <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Baolin Wang <[email protected]>
Cc "Liam R. Howlett" <[email protected]>
Cc Nico Pache <[email protected]>
Cc Ryan Roberts <[email protected]>
Cc Dev Jain <[email protected]>
Cc Barry Song <[email protected]>
Cc Lyude Paul <[email protected]>
Cc Danilo Krummrich <[email protected]>
Cc David Airlie <[email protected]>
Cc Simona Vetter <[email protected]>
Cc Ralph Campbell <[email protected]>
Cc Mika Penttilä <[email protected]>
Cc Matthew Brost <[email protected]>
Cc Francois Dugast <[email protected]>
---
 mm/huge_memory.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
mm/huge_memory.c +5 -2
index 9dda8c48daca..df93768a6e15 100644
@@ -4698,7 +4698,12 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
 	folio_get(folio);
 	pmde = folio_mk_pmd(folio, READ_ONCE(vma->vm_page_prot));
 
+	if (is_writable_migration_entry(entry))
+		pmde = pmd_mkwrite(pmde, vma);
+
 	if (folio_is_device_private(folio)) {
+		swp_entry_t entry;
+
 		if (pmd_write(pmde))
 			entry = make_writable_device_private_entry(
 							page_to_pfn(new));
@@ -4710,8 +4715,6 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
 
 	if (pmd_swp_soft_dirty(*pvmw->pmd))
 		pmde = pmd_mksoft_dirty(pmde);
-	if (is_writable_migration_entry(entry))
-		pmde = pmd_mkwrite(pmde, vma);
 	if (pmd_swp_uffd_wp(*pvmw->pmd))
 		pmde = pmd_mkuffd_wp(pmde);
 	if (!is_migration_entry_young(entry))
Signature · 1 line
2.51.1
#2 lore raw

Re: [PATCH] mm/huge_memory: fix override of entry in remove_migration_pmd

Quoted 33 lines · Balbir Singh

Recent changes exposed a BUG in remove_migration_pmd() where the migration entry was being overridden when the folio is device private.

Use scope local entry for creating the device private pmde. Make the pmde writable if the migration entry is writable by moving the check is_migration_write() prior to creating the device private entry.

Cc Andrew Morton <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Zi Yan <[email protected]>
Cc Joshua Hahn <[email protected]>
Cc Rakie Kim <[email protected]>
Cc Byungchul Park <[email protected]>
Cc Gregory Price <[email protected]>
Cc Ying Huang <[email protected]>
Cc Alistair Popple <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Baolin Wang <[email protected]>
Cc "Liam R. Howlett" <[email protected]>
Cc Nico Pache <[email protected]>
Cc Ryan Roberts <[email protected]>
Cc Dev Jain <[email protected]>
Cc Barry Song <[email protected]>
Cc Lyude Paul <[email protected]>
Cc Danilo Krummrich <[email protected]>
Cc David Airlie <[email protected]>
Cc Simona Vetter <[email protected]>
Cc Ralph Campbell <[email protected]>
Cc Mika Penttilä <[email protected]>
Cc Matthew Brost <[email protected]>
Cc Francois Dugast <[email protected]>
Quoted 6 lines
---
 mm/huge_memory.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

Hi, Andrew

This patch fixes a bug in mm/rmap: extend rmap and migration support device-private entries in the thp device-private series.

Thanks,
Balbir

#3 lore raw

Re: [PATCH] mm/huge_memory: fix override of entry in remove_migration_pmd

Quoted 41 lines · Balbir Singh
Quoted 33 lines · Balbir Singh

Recent changes exposed a BUG in remove_migration_pmd() where the migration entry was being overridden when the folio is device private.

Use scope local entry for creating the device private pmde. Make the pmde writable if the migration entry is writable by moving the check is_migration_write() prior to creating the device private entry.

Cc Andrew Morton <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Zi Yan <[email protected]>
Cc Joshua Hahn <[email protected]>
Cc Rakie Kim <[email protected]>
Cc Byungchul Park <[email protected]>
Cc Gregory Price <[email protected]>
Cc Ying Huang <[email protected]>
Cc Alistair Popple <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Baolin Wang <[email protected]>
Cc "Liam R. Howlett" <[email protected]>
Cc Nico Pache <[email protected]>
Cc Ryan Roberts <[email protected]>
Cc Dev Jain <[email protected]>
Cc Barry Song <[email protected]>
Cc Lyude Paul <[email protected]>
Cc Danilo Krummrich <[email protected]>
Cc David Airlie <[email protected]>
Cc Simona Vetter <[email protected]>
Cc Ralph Campbell <[email protected]>
Cc Mika Penttilä <[email protected]>
Cc Matthew Brost <[email protected]>
Cc Francois Dugast <[email protected]>
Quoted 6 lines
---
  mm/huge_memory.c | 7 +++++--
  1 file changed, 5 insertions(+), 2 deletions(-)
Quoted 4 lines

Hi, Andrew

This patch fixes a bug in mm/rmap: extend rmap and migration support device-private entries in the thp device-private series.

When you resend, best to add under the "--" something like

"This fixup should be squashed into the patch "$NAMEOFCOMMIT" in mm/mm-unstable"

It is also helpful to call the patch then

"fixup: $NAMEOFCOMMIT"

So it's immediately clear that this is something to be squashed.

Signature · 3 lines
Cheers

David
#4 lore raw

Re: [PATCH] mm/huge_memory: fix override of entry in remove_migration_pmd

In reply to Balbir Singh
Quoted 57 lines · Balbir Singh

Recent changes exposed a BUG in remove_migration_pmd() where the migration entry was being overridden when the folio is device private.

Use scope local entry for creating the device private pmde. Make the pmde writable if the migration entry is writable by moving the check is_migration_write() prior to creating the device private entry.

Cc Andrew Morton <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Zi Yan <[email protected]>
Cc Joshua Hahn <[email protected]>
Cc Rakie Kim <[email protected]>
Cc Byungchul Park <[email protected]>
Cc Gregory Price <[email protected]>
Cc Ying Huang <[email protected]>
Cc Alistair Popple <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Baolin Wang <[email protected]>
Cc "Liam R. Howlett" <[email protected]>
Cc Nico Pache <[email protected]>
Cc Ryan Roberts <[email protected]>
Cc Dev Jain <[email protected]>
Cc Barry Song <[email protected]>
Cc Lyude Paul <[email protected]>
Cc Danilo Krummrich <[email protected]>
Cc David Airlie <[email protected]>
Cc Simona Vetter <[email protected]>
Cc Ralph Campbell <[email protected]>
Cc Mika Penttilä <[email protected]>
Cc Matthew Brost <[email protected]>
Cc Francois Dugast <[email protected]>
---
  mm/huge_memory.c | 7 +++++--
  1 file changed, 5 insertions(+), 2 deletions(-)
mm/huge_memory.c +5 -0
index 9dda8c48daca..df93768a6e15 100644
@@ -4698,7 +4698,12 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
  	folio_get(folio);
  	pmde = folio_mk_pmd(folio, READ_ONCE(vma->vm_page_prot));
 
+	if (is_writable_migration_entry(entry))
+		pmde = pmd_mkwrite(pmde, vma);
+
  	if (folio_is_device_private(folio)) {
+		swp_entry_t entry;
+
  		if (pmd_write(pmde))
  			entry = make_writable_device_private_entry(
  							page_to_pfn(new));

mm/huge_memory.c void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new) · line 4720

Quoted 8 lines
mm/huge_memory.c +0 -2
@@ -4710,8 +4715,6 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
 
  	if (pmd_swp_soft_dirty(*pvmw->pmd))
  		pmde = pmd_mksoft_dirty(pmde);
-	if (is_writable_migration_entry(entry))
-		pmde = pmd_mkwrite(pmde, vma);
  	if (pmd_swp_uffd_wp(*pvmw->pmd))
  		pmde = pmd_mkuffd_wp(pmde);
  	if (!is_migration_entry_young(entry))

There are more problems here: you cannot call pmd_mksoft_dirty() etc on something that is not a present pmd! We have pmd_swp_mksoft_dirty() and friends for that.

So you'll have to completely split both paths.

Signature · 3 lines
Cheers

David
#5 lore raw

Re: [PATCH] mm/huge_memory: fix override of entry in remove_migration_pmd

Quoted 67 lines · David Hildenbrand (Red Hat)
Quoted 56 lines · Balbir Singh

Recent changes exposed a BUG in remove_migration_pmd() where the migration entry was being overridden when the folio is device private.

Use scope local entry for creating the device private pmde. Make the pmde writable if the migration entry is writable by moving the check is_migration_write() prior to creating the device private entry.

Cc Andrew Morton <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Zi Yan <[email protected]>
Cc Joshua Hahn <[email protected]>
Cc Rakie Kim <[email protected]>
Cc Byungchul Park <[email protected]>
Cc Gregory Price <[email protected]>
Cc Ying Huang <[email protected]>
Cc Alistair Popple <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Baolin Wang <[email protected]>
Cc "Liam R. Howlett" <[email protected]>
Cc Nico Pache <[email protected]>
Cc Ryan Roberts <[email protected]>
Cc Dev Jain <[email protected]>
Cc Barry Song <[email protected]>
Cc Lyude Paul <[email protected]>
Cc Danilo Krummrich <[email protected]>
Cc David Airlie <[email protected]>
Cc Simona Vetter <[email protected]>
Cc Ralph Campbell <[email protected]>
Cc Mika Penttilä <[email protected]>
Cc Matthew Brost <[email protected]>
Cc Francois Dugast <[email protected]>

---
  mm/huge_memory.c | 7 +++++--

  1 file changed, 5 insertions(+), 2 deletions(-)
mm/huge_memory.c +4 -0
index 9dda8c48daca..df93768a6e15 100644
@@ -4698,7 +4698,12 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
       folio_get(folio);
       pmde = folio_mk_pmd(folio, READ_ONCE(vma->vm_page_prot));
   +    if (is_writable_migration_entry(entry))
+        pmde = pmd_mkwrite(pmde, vma);
+
       if (folio_is_device_private(folio)) {
+        swp_entry_t entry;
+
           if (pmd_write(pmde))

              entry = make_writable_device_private_entry( page_to_pfn(new));

mm/huge_memory.c void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new) · line 4719

Quoted 7 lines
mm/huge_memory.c +0 -2
@@ -4710,8 +4715,6 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
         if (pmd_swp_soft_dirty(*pvmw->pmd))
           pmde = pmd_mksoft_dirty(pmde);
-    if (is_writable_migration_entry(entry))
-        pmde = pmd_mkwrite(pmde, vma);
       if (pmd_swp_uffd_wp(*pvmw->pmd))
           pmde = pmd_mkuffd_wp(pmde);
       if (!is_migration_entry_young(entry))

There are more problems here: you cannot call pmd_mksoft_dirty() etc on something that is not a present pmd! We have pmd_swp_mksoft_dirty() and friends for that.

Quoted 2 lines

So you'll have to completely split both paths.

I'll follow the pattern of remove_migration_pte(), where all the device private handling and that of pte_swp_soft_dirty and pte_swp_mkuffd_wp is done at the every end after migration processing.

Thanks for the review! That might also take up any softleaf fix-up's required, I'll check

Balbir

Balbir

#6 lore raw
In reply to Balbir Singh
Quoted 50 lines · Balbir Singh

A fix-up in the thp device-private patch
mm/rmap: extend rmap and migration support device-private entries introduces conflicts with the softleaf series, this is a fix up patch to resolve the issue

Cc Andrew Morton <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Zi Yan <[email protected]>
Cc Joshua Hahn <[email protected]>
Cc Rakie Kim <[email protected]>
Cc Byungchul Park <[email protected]>
Cc Gregory Price <[email protected]>
Cc Ying Huang <[email protected]>
Cc Alistair Popple <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Baolin Wang <[email protected]>
Cc "Liam R. Howlett" <[email protected]>
Cc Nico Pache <[email protected]>
Cc Ryan Roberts <[email protected]>
Cc Dev Jain <[email protected]>
Cc Barry Song <[email protected]>
Cc Lyude Paul <[email protected]>
Cc Danilo Krummrich <[email protected]>
Cc David Airlie <[email protected]>
Cc Simona Vetter <[email protected]>
Cc Ralph Campbell <[email protected]>
Cc Mika Penttilä <[email protected]>
Cc Matthew Brost <[email protected]>
Cc Francois Dugast <[email protected]>
---
 mm/huge_memory.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
mm/huge_memory.c +1 -1
index 3a8f0c4143c5..4d65c51bc0fe 100644
@@ -4760,7 +4760,7 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
 	folio_get(folio);
 	pmde = folio_mk_pmd(folio, READ_ONCE(vma->vm_page_prot));
 
-	if (is_writable_migration_entry(entry))
+	if (softleaf_is_migration_write(entry))
 		pmde = pmd_mkwrite(pmde, vma);
 
 	if (folio_is_device_private(folio)) {

mm/huge_memory.c void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new) · line 4782

Quoted 8 lines
mm/huge_memory.c +0 -2
@@ -4777,8 +4777,6 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
 
 	if (pmd_swp_soft_dirty(*pvmw->pmd))
 		pmde = pmd_mksoft_dirty(pmde);
-	if (softleaf_is_migration_write(entry))
-		pmde = pmd_mkwrite(pmde, vma);
 	if (pmd_swp_uffd_wp(*pvmw->pmd))
 		pmde = pmd_mkuffd_wp(pmde);
 	if (!softleaf_is_migration_young(entry))

Hi, Andrew

This is the fixup for the rebase issue that might occur after applying
the previous patch to fixup mm/rmap: extend rmap and migration support device-private entries. I've not done fixs up rebases as a separate patch, but I resolved the conflict in favour of the existing patch and applied this on top of the series.

Balbir