Re: [PATCH 1/2] mm: remove redundant check in page_vma_mapped_walk

Re: [PATCH 1/2] mm: remove redundant check in page_vma_mapped_walk

[Andrew Morton]

On Wed,  5 Jul 2023 05:39:31 +0800 Kemeng Shi <shikemeng@huaweicloud.com> wrote:

> For PVMW_SYNC case, we always take pte lock when get first pte of
> PTE-mapped THP in map_pte and hold it until:
> 1. scan of pmd range finished or
> 2. scan of user input range finished or
> 3. user stop walk with page_vma_mapped_walk_done.
> In each case. pte lock will not be freed during middle scan of PTE-mapped
> THP.
> 
> ...
>
> --- a/mm/page_vma_mapped.c
> +++ b/mm/page_vma_mapped.c
> @@ -275,10 +275,6 @@ bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
>  				goto restart;
>  			}
>  			pvmw->pte++;
> -			if ((pvmw->flags & PVMW_SYNC) && !pvmw->ptl) {
> -				pvmw->ptl = pte_lockptr(mm, pvmw->pmd);
> -				spin_lock(pvmw->ptl);
> -			}
>  		} while (pte_none(*pvmw->pte));
>  
>  		if (!pvmw->ptl) {

This code has changed significantly since 6.4.  Please develop against
the mm-unstable branch at
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm, thanks.

[PATCH 0/2] Two minor cleanups for page_vma_mapped.c

[Kemeng Shi]

Hi all, This series contains two minor cleanups for page_vma_mapped.c
to remove redundant check and correct stale coment. More details can be
found in respective patches. Thanks!

Kemeng Shi (2):
  mm: remove redundant check in page_vma_mapped_walk
  mm: correct stale comment of function check_pte

 mm/page_vma_mapped.c | 17 ++++++++---------
 1 file changed, 8 insertions(+), 9 deletions(-)

-- 
2.30.0

[PATCH 1/2] mm: remove redundant check in page_vma_mapped_walk

[Kemeng Shi]

For PVMW_SYNC case, we always take pte lock when get first pte of
PTE-mapped THP in map_pte and hold it until:
1. scan of pmd range finished or
2. scan of user input range finished or
3. user stop walk with page_vma_mapped_walk_done.
In each case. pte lock will not be freed during middle scan of PTE-mapped
THP.

Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
---
 mm/page_vma_mapped.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/mm/page_vma_mapped.c b/mm/page_vma_mapped.c
index 4e448cfbc6ef..83858758e239 100644
--- a/mm/page_vma_mapped.c
+++ b/mm/page_vma_mapped.c
@@ -275,10 +275,6 @@ bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
 				goto restart;
 			}
 			pvmw->pte++;
-			if ((pvmw->flags & PVMW_SYNC) && !pvmw->ptl) {
-				pvmw->ptl = pte_lockptr(mm, pvmw->pmd);
-				spin_lock(pvmw->ptl);
-			}
 		} while (pte_none(*pvmw->pte));
 
 		if (!pvmw->ptl) {
-- 
2.30.0

[PATCH 2/2] mm: correct stale comment of function check_pte

[Kemeng Shi]

Commit 2aff7a4755bed ("mm: Convert page_vma_mapped_walk to work on PFNs")
replaced page with pfns in page_vma_mapped_walk structure and updated
"@pvmw->page" to "@pvmw->pfn" in comment of function page_vma_mapped_walk.

This patch update stale "page" to "pfn" in comment of check_pte.

Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
---
 mm/page_vma_mapped.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/mm/page_vma_mapped.c b/mm/page_vma_mapped.c
index 83858758e239..dd0a820a0e6c 100644
--- a/mm/page_vma_mapped.c
+++ b/mm/page_vma_mapped.c
@@ -54,20 +54,23 @@ static bool map_pte(struct page_vma_mapped_walk *pvmw)
 }
 
 /**
- * check_pte - check if @pvmw->page is mapped at the @pvmw->pte
- * @pvmw: page_vma_mapped_walk struct, includes a pair pte and page for checking
+ * check_pte - check if [pvmw->pfn, @pvmw->pfn + @pvmw->nr_pages) is
+ * mapped at the @pvmw->pte
+ * @pvmw: page_vma_mapped_walk struct, includes a pair pte and pfn range
+ * for checking
  *
- * page_vma_mapped_walk() found a place where @pvmw->page is *potentially*
+ * page_vma_mapped_walk() found a place where @pvmw->pfn is *potentially*
  * mapped. check_pte() has to validate this.
  *
  * pvmw->pte may point to empty PTE, swap PTE or PTE pointing to
  * arbitrary page.
  *
  * If PVMW_MIGRATION flag is set, returns true if @pvmw->pte contains migration
- * entry that points to @pvmw->page or any subpage in case of THP.
+ * entry that points to @pvmw->pfn or [pvmw->pfn, @pvmw->pfn + @pvmw->nr_pages)
+ * in case of THP.
  *
  * If PVMW_MIGRATION flag is not set, returns true if pvmw->pte points to
- * pvmw->page or any subpage in case of THP.
+ * pvmw->pfn or [pvmw->pfn, @pvmw->pfn + @pvmw->nr_pages) in case of THP.
  *
  * Otherwise, return false.
  *
-- 
2.30.0

Re: [PATCH 1/2] mm: remove redundant check in page_vma_mapped_walk

[Kemeng Shi]

on 7/5/2023 1:05 AM, Andrew Morton wrote:
> On Wed,  5 Jul 2023 05:39:31 +0800 Kemeng Shi <shikemeng@huaweicloud.com> wrote:
> 
>> For PVMW_SYNC case, we always take pte lock when get first pte of
>> PTE-mapped THP in map_pte and hold it until:
>> 1. scan of pmd range finished or
>> 2. scan of user input range finished or
>> 3. user stop walk with page_vma_mapped_walk_done.
>> In each case. pte lock will not be freed during middle scan of PTE-mapped
>> THP.
>>
>> ...
>>
>> --- a/mm/page_vma_mapped.c
>> +++ b/mm/page_vma_mapped.c
>> @@ -275,10 +275,6 @@ bool page_vma_mapped_walk(struct page_vma_mapped_walk *pvmw)
>>  				goto restart;
>>  			}
>>  			pvmw->pte++;
>> -			if ((pvmw->flags & PVMW_SYNC) && !pvmw->ptl) {
>> -				pvmw->ptl = pte_lockptr(mm, pvmw->pmd);
>> -				spin_lock(pvmw->ptl);
>> -			}
>>  		} while (pte_none(*pvmw->pte));
>>  
>>  		if (!pvmw->ptl) {
> 
> This code has changed significantly since 6.4.  Please develop against
> the mm-unstable branch at
> git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm, thanks.
> 
> 
Thanks for reminding me of this, I will check my changes in updated code.

-- 
Best wishes
Kemeng Shi