[PATCH] mm/mlock: Fix vma iterator conversion of apply_vma_lock_flags()

[PATCH] mm/mlock: Fix vma iterator conversion of apply_vma_lock_flags()

[Liam R. Howlett]

apply_vma_lock_flags() calls mlock_fixup(), which could merge the VMA
after where the vma iterator is located.  Although this is not an issue,
the next iteration of the loop will check the start of the vma to be
equal to the locally saved 'tmp' variable and cause an incorrect failure
scenario.  Fix the error by setting tmp to the end of the vma iterator
value before restarting the loop.

There is also a potential of the error code being overwritten when the
loop terminates early.  Fix the return issue by directly returning when
an error is encountered since there is nothing to undo after the loop.

Reported-by: Ryan Roberts <ryan.roberts@arm.com>
Link: https://lore.kernel.org/linux-mm/50341ca1-d582-b33a-e3d0-acb08a65166f@arm.com/
Cc: <stable@vger.kernel.org>
Fixes: 37598f5a9d8b ("mlock: convert mlock to vma iterator")
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
---
 mm/mlock.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/mm/mlock.c b/mm/mlock.c
index d7db94519884..0a0c996c5c21 100644
--- a/mm/mlock.c
+++ b/mm/mlock.c
@@ -477,7 +477,6 @@ static int apply_vma_lock_flags(unsigned long start, size_t len,
 {
 	unsigned long nstart, end, tmp;
 	struct vm_area_struct *vma, *prev;
-	int error;
 	VMA_ITERATOR(vmi, current->mm, start);
 
 	VM_BUG_ON(offset_in_page(start));
@@ -498,6 +497,7 @@ static int apply_vma_lock_flags(unsigned long start, size_t len,
 	nstart = start;
 	tmp = vma->vm_start;
 	for_each_vma_range(vmi, vma, end) {
+		int error;
 		vm_flags_t newflags;
 
 		if (vma->vm_start != tmp)
@@ -511,14 +511,15 @@ static int apply_vma_lock_flags(unsigned long start, size_t len,
 			tmp = end;
 		error = mlock_fixup(&vmi, vma, &prev, nstart, tmp, newflags);
 		if (error)
-			break;
+			return error;
+		tmp = vma_iter_end(&vmi);
 		nstart = tmp;
 	}
 
-	if (vma_iter_end(&vmi) < end)
+	if (tmp < end)
 		return -ENOMEM;
 
-	return error;
+	return 0;
 }
 
 /*
-- 
2.39.2

Re: [PATCH] mm/mlock: Fix vma iterator conversion of apply_vma_lock_flags()

[Ryan Roberts]

On 11/07/2023 18:50, Liam R. Howlett wrote:
> apply_vma_lock_flags() calls mlock_fixup(), which could merge the VMA
> after where the vma iterator is located.  Although this is not an issue,
> the next iteration of the loop will check the start of the vma to be
> equal to the locally saved 'tmp' variable and cause an incorrect failure
> scenario.  Fix the error by setting tmp to the end of the vma iterator
> value before restarting the loop.
> 
> There is also a potential of the error code being overwritten when the
> loop terminates early.  Fix the return issue by directly returning when
> an error is encountered since there is nothing to undo after the loop.
> 
> Reported-by: Ryan Roberts <ryan.roberts@arm.com>
> Link: https://lore.kernel.org/linux-mm/50341ca1-d582-b33a-e3d0-acb08a65166f@arm.com/
> Cc: <stable@vger.kernel.org>
> Fixes: 37598f5a9d8b ("mlock: convert mlock to vma iterator")
> Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>

Wow; thanks for the quick fix! I've tested this on arm64/v6.4 with the failing
mm selftests that I reported: they are now passing.

Tested-by: Ryan Roberts <ryan.roberts@arm.com>

> ---
>  mm/mlock.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/mm/mlock.c b/mm/mlock.c
> index d7db94519884..0a0c996c5c21 100644
> --- a/mm/mlock.c
> +++ b/mm/mlock.c
> @@ -477,7 +477,6 @@ static int apply_vma_lock_flags(unsigned long start, size_t len,
>  {
>  	unsigned long nstart, end, tmp;
>  	struct vm_area_struct *vma, *prev;
> -	int error;
>  	VMA_ITERATOR(vmi, current->mm, start);
>  
>  	VM_BUG_ON(offset_in_page(start));
> @@ -498,6 +497,7 @@ static int apply_vma_lock_flags(unsigned long start, size_t len,
>  	nstart = start;
>  	tmp = vma->vm_start;
>  	for_each_vma_range(vmi, vma, end) {
> +		int error;
>  		vm_flags_t newflags;
>  
>  		if (vma->vm_start != tmp)
> @@ -511,14 +511,15 @@ static int apply_vma_lock_flags(unsigned long start, size_t len,
>  			tmp = end;
>  		error = mlock_fixup(&vmi, vma, &prev, nstart, tmp, newflags);
>  		if (error)
> -			break;
> +			return error;
> +		tmp = vma_iter_end(&vmi);
>  		nstart = tmp;
>  	}
>  
> -	if (vma_iter_end(&vmi) < end)
> +	if (tmp < end)
>  		return -ENOMEM;
>  
> -	return error;
> +	return 0;
>  }
>  
>  /*