[PATCH] mm/migrate_device: Return number of migrating pages in args->cpages

[PATCH] mm/migrate_device: Return number of migrating pages in args->cpages

[Alistair Popple]

migrate_vma->cpages originally contained a count of the number of
pages migrating including non-present pages which can be poluated
directly on the target.

Commit 241f68859656 ("mm/migrate_device.c: refactor migrate_vma and
migrate_deivce_coherent_page()") inadvertantly changed this to contain
just the number of pages that were unmapped. Usage of
migrate_vma->cpages isn't documented, but most drivers use it to see
if all the requested addresses can be migrated so restore the original
behaviour.

Fixes: 241f68859656 ("mm/migrate_device.c: refactor migrate_vma and migrate_deivce_coherent_page()")
Signed-off-by: Alistair Popple <apopple@nvidia.com>
Reported-by: Ralph Campbell <rcampbell@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Alex Sierra <alex.sierra@amd.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Cc: Lyude Paul <lyude@redhat.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
---

Hi Andrew, hoping you can merge this small fix which Ralph reported to
me for v6.1-rcX.

---
 mm/migrate_device.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/mm/migrate_device.c b/mm/migrate_device.c
index 6fa682eef7a0..721b2365dbca 100644
--- a/mm/migrate_device.c
+++ b/mm/migrate_device.c
@@ -357,7 +357,8 @@ static bool migrate_vma_check_page(struct page *page, struct page *fault_page)
 }
 
 /*
- * Unmaps pages for migration. Returns number of unmapped pages.
+ * Unmaps pages for migration. Returns number of source pfns marked as
+ * migrating.
  */
 static unsigned long migrate_device_unmap(unsigned long *src_pfns,
 					  unsigned long npages,
@@ -373,8 +374,11 @@ static unsigned long migrate_device_unmap(unsigned long *src_pfns,
 		struct page *page = migrate_pfn_to_page(src_pfns[i]);
 		struct folio *folio;
 
-		if (!page)
+		if (!page) {
+			if (src_pfns[i] & MIGRATE_PFN_MIGRATE)
+				unmapped++;
 			continue;
+		}
 
 		/* ZONE_DEVICE pages are not on LRU */
 		if (!is_zone_device_page(page)) {
-- 
2.35.1

Re: [PATCH] mm/migrate_device: Return number of migrating pages in args->cpages

[Ralph Campbell]

On 11/10/22 16:51, Alistair Popple wrote:
> migrate_vma->cpages originally contained a count of the number of
> pages migrating including non-present pages which can be poluated

"populated"

> directly on the target.
>
> Commit 241f68859656 ("mm/migrate_device.c: refactor migrate_vma and
> migrate_deivce_coherent_page()") inadvertantly changed this to contain
> just the number of pages that were unmapped. Usage of
> migrate_vma->cpages isn't documented, but most drivers use it to see
> if all the requested addresses can be migrated so restore the original
> behaviour.
>
> Fixes: 241f68859656 ("mm/migrate_device.c: refactor migrate_vma and migrate_deivce_coherent_page()")
> Signed-off-by: Alistair Popple <apopple@nvidia.com>
> Reported-by: Ralph Campbell <rcampbell@nvidia.com>

You can add
Reviewed-by: Ralph Campbell <rcampbell@nvidia.com>

Thanks!

Re: [PATCH] mm/migrate_device: Return number of migrating pages in args->cpages

[Alistair Popple]

Hi Andrew,

I realize it's getting late in the cycle but any chance you could take
this one as well for v6.1-rc6? It fixes a minor bug introduced in
v6.1-rc1 which could cause some drivers to think they can't migrate
pages when they can. Thanks.

Ralph Campbell <rcampbell@nvidia.com> writes:

> On 11/10/22 16:51, Alistair Popple wrote:
>> migrate_vma->cpages originally contained a count of the number of
>> pages migrating including non-present pages which can be poluated
>
> "populated"
>
>> directly on the target.
>>
>> Commit 241f68859656 ("mm/migrate_device.c: refactor migrate_vma and
>> migrate_deivce_coherent_page()") inadvertantly changed this to contain
>> just the number of pages that were unmapped. Usage of
>> migrate_vma->cpages isn't documented, but most drivers use it to see
>> if all the requested addresses can be migrated so restore the original
>> behaviour.
>>
>> Fixes: 241f68859656 ("mm/migrate_device.c: refactor migrate_vma and migrate_deivce_coherent_page()")
>> Signed-off-by: Alistair Popple <apopple@nvidia.com>
>> Reported-by: Ralph Campbell <rcampbell@nvidia.com>
>
> You can add
> Reviewed-by: Ralph Campbell <rcampbell@nvidia.com>
>
> Thanks!