[PATCH] mm/vmscan: Initialize dirty to prevent uninitialized use

[PATCH] mm/vmscan: Initialize dirty to prevent uninitialized use

[Purva Yeshi]

Fix Smatch-detected error:
mm/vmscan.c:3509 walk_pte_range() error: uninitialized symbol 'dirty'.
mm/vmscan.c:3522 walk_pte_range() error: uninitialized symbol 'dirty'.
mm/vmscan.c:3600 walk_pmd_range_locked() error: uninitialized symbol 'dirty'.
mm/vmscan.c:3614 walk_pmd_range_locked() error: uninitialized symbol 'dirty'.
mm/vmscan.c:4220 lru_gen_look_around() error: uninitialized symbol 'dirty'.
mm/vmscan.c:4232 lru_gen_look_around() error: uninitialized symbol 'dirty'.

Smatch reports 'dirty' as uninitialized, leading to potential
undefined behavior.

Explicitly initialize dirty to 0 in walk_pte_range(),
walk_pmd_range_locked(), and lru_gen_look_around() in mm/vmscan.c
to fix Smatch error.

Signed-off-by: Purva Yeshi <purvayeshi550@gmail.com>
---
 mm/vmscan.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index c767d71c43d7..39c49fcd960c 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -3456,7 +3456,7 @@ static bool walk_pte_range(pmd_t *pmd, unsigned long start, unsigned long end,
 			   struct mm_walk *args)
 {
 	int i;
-	bool dirty;
+	bool dirty = 0;
 	pte_t *pte;
 	spinlock_t *ptl;
 	unsigned long addr;
@@ -3535,7 +3535,7 @@ static void walk_pmd_range_locked(pud_t *pud, unsigned long addr, struct vm_area
 				  struct mm_walk *args, unsigned long *bitmap, unsigned long *first)
 {
 	int i;
-	bool dirty;
+	bool dirty = 0;
 	pmd_t *pmd;
 	spinlock_t *ptl;
 	struct folio *last = NULL;
@@ -4147,7 +4147,7 @@ static void lru_gen_age_node(struct pglist_data *pgdat, struct scan_control *sc)
 bool lru_gen_look_around(struct page_vma_mapped_walk *pvmw)
 {
 	int i;
-	bool dirty;
+	bool dirty = 0;
 	unsigned long start;
 	unsigned long end;
 	struct lru_gen_mm_walk *walk;
-- 
2.34.1

Re: [PATCH] mm/vmscan: Initialize dirty to prevent uninitialized use

[Yu Zhao]

On Tue, Mar 25, 2025 at 1:49 PM Purva Yeshi <purvayeshi550@gmail.com> wrote:
>
> Fix Smatch-detected error:
> mm/vmscan.c:3509 walk_pte_range() error: uninitialized symbol 'dirty'.
> mm/vmscan.c:3522 walk_pte_range() error: uninitialized symbol 'dirty'.
> mm/vmscan.c:3600 walk_pmd_range_locked() error: uninitialized symbol 'dirty'.
> mm/vmscan.c:3614 walk_pmd_range_locked() error: uninitialized symbol 'dirty'.
> mm/vmscan.c:4220 lru_gen_look_around() error: uninitialized symbol 'dirty'.
> mm/vmscan.c:4232 lru_gen_look_around() error: uninitialized symbol 'dirty'.
>
> Smatch reports 'dirty' as uninitialized, leading to potential
> undefined behavior.

Thanks -- this seems like false positives from Smatch, where the
problem should be fixed.

> Explicitly initialize dirty to 0 in walk_pte_range(),
> walk_pmd_range_locked(), and lru_gen_look_around() in mm/vmscan.c
> to fix Smatch error.
>
> Signed-off-by: Purva Yeshi <purvayeshi550@gmail.com>

Re: [PATCH] mm/vmscan: Initialize dirty to prevent uninitialized use

[Purva Yeshi]

On 26/03/25 02:05, Yu Zhao wrote:
> On Tue, Mar 25, 2025 at 1:49 PM Purva Yeshi <purvayeshi550@gmail.com> wrote:
>>
>> Fix Smatch-detected error:
>> mm/vmscan.c:3509 walk_pte_range() error: uninitialized symbol 'dirty'.
>> mm/vmscan.c:3522 walk_pte_range() error: uninitialized symbol 'dirty'.
>> mm/vmscan.c:3600 walk_pmd_range_locked() error: uninitialized symbol 'dirty'.
>> mm/vmscan.c:3614 walk_pmd_range_locked() error: uninitialized symbol 'dirty'.
>> mm/vmscan.c:4220 lru_gen_look_around() error: uninitialized symbol 'dirty'.
>> mm/vmscan.c:4232 lru_gen_look_around() error: uninitialized symbol 'dirty'.
>>
>> Smatch reports 'dirty' as uninitialized, leading to potential
>> undefined behavior.
> 
> Thanks -- this seems like false positives from Smatch, where the
> problem should be fixed.

Should we make modifications to address this Smatch error, or is it safe 
to ignore?

> 
>> Explicitly initialize dirty to 0 in walk_pte_range(),
>> walk_pmd_range_locked(), and lru_gen_look_around() in mm/vmscan.c
>> to fix Smatch error.
>>
>> Signed-off-by: Purva Yeshi <purvayeshi550@gmail.com>