Skip to contentSkip to conversation

mm/mmap/vma_merge: actually set next to NULL if not applicable

PATCH In review Arnd Bergmann 2 participants 6 messages Mar 23, 2023

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/3 mm/mmap/vma_merge: actually set next to NULL if not applicable

From: Arnd Bergmann <[email protected]>

As clang builds point out, the variable 'next' is now uninitialized in some conditions as a result of a previous patch that tried to rely on it being NULL here:

mm/mmap.c:939:11: error: variable 'next' is used uninitialized whenever 'if' condition is false [-Werror,-Wsometimes-uninitialized]
else if (!curr)
^~~~~
mm/mmap.c:952:15: note: uninitialized use occurs here

        merge_next = next && mpol_equal(policy, vma_policy(next)) &&
                     ^~~~
Fixes e887ecae997e ("mm/mmap/vma_merge: set next to NULL if not applicable")
---
 mm/mmap.c | 2 ++
 1 file changed, 2 insertions(+)
mm/mmap.c +2 -0
index 54099a604cf8..c01d43bd694e 100644
@@ -939,6 +939,8 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
 	else if (!curr)
 		/* Is there a VMA next to a hole (case 1 - 3) or prev (4)? */
 		next = vma_lookup(mm, end);
+	else
+		next = NULL;
 
 	/* Can we merge the predecessor? */
 	if (prev && addr == prev->vm_end && mpol_equal(vma_policy(prev), policy)
Signature · 1 line
2.39.2
#1 lore raw

PATCH 2/3 mm: vmalloc: fix vmap_ram_vread_iter() return value

From: Arnd Bergmann <[email protected]>

The error handling in vmap_ram_vread_iter() can return an uninitialized value in some cases:

mm/vmalloc.c:3539:6: error: variable 'remains' is used uninitialized whenever 'if' condition is true [-Werror,-Wsometimes-uninitialized]

        if (!vb)
            ^~~
mm/vmalloc.c:3587:17: note: uninitialized use occurs here
        return count - remains + zero_iter(iter, remains);
                       ^~~~~~~

Move the initialization up a few lines.

Fixes d9cab54f7737 ("mm: vmalloc: convert vread() to vread_iter()")
---
 mm/vmalloc.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
mm/vmalloc.c +1 -2
index ebfa1e9fe6f9..fb216495fe5a 100644
@@ -3521,7 +3521,7 @@ static size_t vmap_ram_vread_iter(struct iov_iter *iter, const char *addr,
 	struct vmap_block *vb;
 	unsigned long offset;
 	unsigned int rs, re;
-	size_t remains, n;
+	size_t remains = count, n;
 
 	/*
 	 * If it's area created by vm_map_ram() interface directly, but
@@ -3545,7 +3545,6 @@ static size_t vmap_ram_vread_iter(struct iov_iter *iter, const char *addr,
 		goto finished_zero;
 	}
 
-	remains = count;
 	for_each_set_bitrange(rs, re, vb->used_map, VMAP_BBMAP_BITS) {
 		size_t copied;
 
Signature · 1 line
2.39.2
#2 lore raw

Re: [PATCH 2/3] mm: vmalloc: fix vmap_ram_vread_iter() return value

Quoted 42 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

The error handling in vmap_ram_vread_iter() can return an uninitialized value in some cases:

mm/vmalloc.c:3539:6: error: variable 'remains' is used uninitialized whenever 'if' condition is true [-Werror,-Wsometimes-uninitialized]

        if (!vb)
            ^~~
mm/vmalloc.c:3587:17: note: uninitialized use occurs here
        return count - remains + zero_iter(iter, remains);
                       ^~~~~~~

Move the initialization up a few lines.

Fixes d9cab54f7737 ("mm: vmalloc: convert vread() to vread_iter()")
---
 mm/vmalloc.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
mm/vmalloc.c +1 -3
index ebfa1e9fe6f9..fb216495fe5a 100644
@@ -3521,7 +3521,7 @@ static size_t vmap_ram_vread_iter(struct iov_iter *iter, const char *addr,
 	struct vmap_block *vb;
 	unsigned long offset;
 	unsigned int rs, re;
-	size_t remains, n;
+	size_t remains = count, n;
 
 	/*
 	 * If it's area created by vm_map_ram() interface directly, but
@@ -3545,7 +3545,6 @@ static size_t vmap_ram_vread_iter(struct iov_iter *iter, const char *addr,
 		goto finished_zero;
 	}
 
-	remains = count;
 	for_each_set_bitrange(rs, re, vb->used_map, VMAP_BBMAP_BITS) {
 		size_t copied;
 
--
Quoted 3 lines

2.39.2

This was also already fixed in a more recent version of this patch set, in v5 I believe. Again apologies for this!

Latest version at
https://lore.kernel.org/all/[email protected]

#3 lore raw

PATCH 3/3 mm: vmalloc: mark zero_iter() static

In reply to Arnd Bergmann

From: Arnd Bergmann <[email protected]>

The zero_iter() function was put into the global namespace, apparently by accident:

mm/vmalloc.c:3448:8: error: no previous prototype for function 'zero_iter' [-Werror,-Wmissing-prototypes] size_t zero_iter(struct iov_iter *iter, size_t count)

       ^

mm/vmalloc.c:3448:1: note: declare 'static' if the function is not intended to be used outside of this translation unit size_t zero_iter(struct iov_iter *iter, size_t count)
^

Fixes d9cab54f7737 ("mm: vmalloc: convert vread() to vread_iter()")
---
 mm/vmalloc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/vmalloc.c +1 -1
index fb216495fe5a..55f5d6fc8629 100644
@@ -3445,7 +3445,7 @@ EXPORT_SYMBOL(vmalloc_32_user);
  *
  * Returns the number of zeroed bytes.
  */
-size_t zero_iter(struct iov_iter *iter, size_t count)
+static size_t zero_iter(struct iov_iter *iter, size_t count)
 {
 	size_t remains = count;
 
Signature · 1 line
2.39.2
#4 lore raw

Re: [PATCH 3/3] mm: vmalloc: mark zero_iter() static

Quoted 32 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

The zero_iter() function was put into the global namespace, apparently by accident:

mm/vmalloc.c:3448:8: error: no previous prototype for function 'zero_iter' [-Werror,-Wmissing-prototypes] size_t zero_iter(struct iov_iter *iter, size_t count)

       ^

mm/vmalloc.c:3448:1: note: declare 'static' if the function is not intended to be used outside of this translation unit size_t zero_iter(struct iov_iter *iter, size_t count)
^

Fixes d9cab54f7737 ("mm: vmalloc: convert vread() to vread_iter()")
---
 mm/vmalloc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/vmalloc.c +1 -2
index fb216495fe5a..55f5d6fc8629 100644
@@ -3445,7 +3445,7 @@ EXPORT_SYMBOL(vmalloc_32_user);
  *
  * Returns the number of zeroed bytes.
  */
-size_t zero_iter(struct iov_iter *iter, size_t count)
+static size_t zero_iter(struct iov_iter *iter, size_t count)
 {
 	size_t remains = count;
 
--
Quoted 3 lines

2.39.2

And finally, this was also fixed in the latest version (v8) see
https://lore.kernel.org/all/[email protected]

Again do forgive the noise! :)

#5 lore raw
In reply to Arnd Bergmann
Quoted 33 lines · Arnd Bergmann

From: Arnd Bergmann <[email protected]>

As clang builds point out, the variable 'next' is now uninitialized in some conditions as a result of a previous patch that tried to rely on it being NULL here:

mm/mmap.c:939:11: error: variable 'next' is used uninitialized whenever 'if' condition is false [-Werror,-Wsometimes-uninitialized]
else if (!curr)
^~~~~
mm/mmap.c:952:15: note: uninitialized use occurs here

        merge_next = next && mpol_equal(policy, vma_policy(next)) &&
                     ^~~~
Fixes e887ecae997e ("mm/mmap/vma_merge: set next to NULL if not applicable")
---
 mm/mmap.c | 2 ++
 1 file changed, 2 insertions(+)
mm/mmap.c +2 -1
index 54099a604cf8..c01d43bd694e 100644
@@ -939,6 +939,8 @@ struct vm_area_struct *vma_merge(struct vma_iterator *vmi, struct mm_struct *mm,
 	else if (!curr)
 		/* Is there a VMA next to a hole (case 1 - 3) or prev (4)? */
 		next = vma_lookup(mm, end);
+	else
+		next = NULL;
 
 	/* Can we merge the predecessor? */
 	if (prev && addr == prev->vm_end && mpol_equal(vma_policy(prev), policy)
--
Quoted 2 lines

2.39.2

This was already fixed in a more recent series of this patch set (at v3). Sorry for this reaching -next!

See https://lore.kernel.org/all/[email protected]/