[PATCH v2] mm/slub: Fix incorrect interpretation of s->offset

[PATCH v2] mm/slub: Fix incorrect interpretation of s->offset

[Waiman Long]

In a couple of places in the slub memory allocator, the code uses
"s->offset" as a check to see if the free pointer is put right after the
object. That check is no longer true with commit 3202fa62fb43 ("slub:
relocate freelist pointer to middle of object").

As a result, echoing "1" into the validate sysfs file, e.g. of dentry,
may cause a bunch of "Freepointer corrupt" error reports like the
following to appear with the system in panic afterwards.

[   38.579769] =============================================================================
[   38.580845] BUG dentry(666:pmcd.service) (Tainted: G    B): Freepointer corrupt
[   38.581948] -----------------------------------------------------------------------------

To fix it, use the check "s->offset == s->inuse" in the new helper
function freeptr_after_object() instead. Also add another helper function
get_info_end() to return the end of info block (inuse + free pointer
if not overlapping with object).

Fixes: 3202fa62fb43 ("slub: relocate freelist pointer to middle of object")
Signed-off-by: Waiman Long <longman@redhat.com>
---
 mm/slub.c | 37 ++++++++++++++++++++++---------------
 1 file changed, 22 insertions(+), 15 deletions(-)

diff --git a/mm/slub.c b/mm/slub.c
index 0e736d66bb42..68f1b4b1c309 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -551,15 +551,29 @@ static void print_section(char *level, char *text, u8 *addr,
 	metadata_access_disable();
 }
 
+static inline bool freeptr_after_object(struct kmem_cache *s)
+{
+	return s->offset == s->inuse;
+}
+
+/*
+ * Return offset of the end of info block which is inuse + free pointer if
+ * not overlapping with object.
+ */
+static inline unsigned int get_info_end(struct kmem_cache *s)
+{
+	if (freeptr_after_object(s))
+		return s->inuse + sizeof(void *);
+	else
+		return s->inuse;
+}
+
 static struct track *get_track(struct kmem_cache *s, void *object,
 	enum track_item alloc)
 {
 	struct track *p;
 
-	if (s->offset)
-		p = object + s->offset + sizeof(void *);
-	else
-		p = object + s->inuse;
+	p = object + get_info_end(s);
 
 	return p + alloc;
 }
@@ -693,10 +707,7 @@ static void print_trailer(struct kmem_cache *s, struct page *page, u8 *p)
 		print_section(KERN_ERR, "Redzone ", p + s->object_size,
 			s->inuse - s->object_size);
 
-	if (s->offset)
-		off = s->offset + sizeof(void *);
-	else
-		off = s->inuse;
+	off = get_info_end(s);
 
 	if (s->flags & SLAB_STORE_USER)
 		off += 2 * sizeof(struct track);
@@ -790,7 +801,7 @@ static int check_bytes_and_report(struct kmem_cache *s, struct page *page,
  * object address
  * 	Bytes of the object to be managed.
  * 	If the freepointer may overlay the object then the free
- * 	pointer is the first word of the object.
+ *	pointer is at the middle of the object.
  *
  * 	Poisoning uses 0x6b (POISON_FREE) and the last byte is
  * 	0xa5 (POISON_END)
@@ -824,11 +835,7 @@ static int check_bytes_and_report(struct kmem_cache *s, struct page *page,
 
 static int check_pad_bytes(struct kmem_cache *s, struct page *page, u8 *p)
 {
-	unsigned long off = s->inuse;	/* The end of info */
-
-	if (s->offset)
-		/* Freepointer is placed after the object. */
-		off += sizeof(void *);
+	unsigned long off = get_info_end(s);	/* The end of info */
 
 	if (s->flags & SLAB_STORE_USER)
 		/* We also have user information there */
@@ -915,7 +922,7 @@ static int check_object(struct kmem_cache *s, struct page *page,
 		check_pad_bytes(s, page, p);
 	}
 
-	if (!s->offset && val == SLUB_RED_ACTIVE)
+	if (!freeptr_after_object(s) && val == SLUB_RED_ACTIVE)
 		/*
 		 * Object and freepointer overlap. Cannot check
 		 * freepointer while object is allocated.
-- 
2.18.1

Re: [PATCH v2] mm/slub: Fix incorrect interpretation of s->offset

[Matthew Wilcox]

On Mon, Apr 27, 2020 at 10:08:22AM -0400, Waiman Long wrote:
> In a couple of places in the slub memory allocator, the code uses
> "s->offset" as a check to see if the free pointer is put right after the
> object. That check is no longer true with commit 3202fa62fb43 ("slub:
> relocate freelist pointer to middle of object").
> 
> As a result, echoing "1" into the validate sysfs file, e.g. of dentry,
> may cause a bunch of "Freepointer corrupt" error reports like the
> following to appear with the system in panic afterwards.
> 
> [   38.579769] =============================================================================
> [   38.580845] BUG dentry(666:pmcd.service) (Tainted: G    B): Freepointer corrupt
> [   38.581948] -----------------------------------------------------------------------------
> 
> To fix it, use the check "s->offset == s->inuse" in the new helper
> function freeptr_after_object() instead. Also add another helper function
> get_info_end() to return the end of info block (inuse + free pointer
> if not overlapping with object).
> 
> Fixes: 3202fa62fb43 ("slub: relocate freelist pointer to middle of object")
> Signed-off-by: Waiman Long <longman@redhat.com>

Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>

Re: [PATCH v2] mm/slub: Fix incorrect interpretation of s->offset

[Christopher Lameter]

On Mon, 27 Apr 2020, Waiman Long wrote:

>
> To fix it, use the check "s->offset == s->inuse" in the new helper
> function freeptr_after_object() instead. Also add another helper function
> get_info_end() to return the end of info block (inuse + free pointer
> if not overlapping with object).
>
> Fixes: 3202fa62fb43 ("slub: relocate freelist pointer to middle of object")
> Signed-off-by: Waiman Long <longman@redhat.com>
> ---
>  mm/slub.c | 37 ++++++++++++++++++++++---------------
>  1 file changed, 22 insertions(+), 15 deletions(-)
>
> diff --git a/mm/slub.c b/mm/slub.c
> index 0e736d66bb42..68f1b4b1c309 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -551,15 +551,29 @@ static void print_section(char *level, char *text, u8 *addr,
>  	metadata_access_disable();
>  }
>
> +static inline bool freeptr_after_object(struct kmem_cache *s)

bool freeptr_outside_of_object()?

> +{
> +	return s->offset == s->inuse;

s->offset >= s->inuse?

There may be a redzone after the object.


> +static inline unsigned int get_info_end(struct kmem_cache *s)

static inline track_offset()?

Re: [PATCH v2] mm/slub: Fix incorrect interpretation of s->offset

[Waiman Long]

On 4/27/20 12:10 PM, Christopher Lameter wrote:
> On Mon, 27 Apr 2020, Waiman Long wrote:
>
>> To fix it, use the check "s->offset == s->inuse" in the new helper
>> function freeptr_after_object() instead. Also add another helper function
>> get_info_end() to return the end of info block (inuse + free pointer
>> if not overlapping with object).
>>
>> Fixes: 3202fa62fb43 ("slub: relocate freelist pointer to middle of object")
>> Signed-off-by: Waiman Long <longman@redhat.com>
>> ---
>>   mm/slub.c | 37 ++++++++++++++++++++++---------------
>>   1 file changed, 22 insertions(+), 15 deletions(-)
>>
>> diff --git a/mm/slub.c b/mm/slub.c
>> index 0e736d66bb42..68f1b4b1c309 100644
>> --- a/mm/slub.c
>> +++ b/mm/slub.c
>> @@ -551,15 +551,29 @@ static void print_section(char *level, char *text, u8 *addr,
>>   	metadata_access_disable();
>>   }
>>
>> +static inline bool freeptr_after_object(struct kmem_cache *s)
> bool freeptr_outside_of_object()?
>
I can change to that name. It doesn't really matter to me.
>> +{
>> +	return s->offset == s->inuse;
> s->offset >= s->inuse?
>
> There may be a redzone after the object.
>
Technically inuse is object + red zone. According to calculate_sizes():

         s->inuse = size;

         if (((flags & (SLAB_TYPESAFE_BY_RCU | SLAB_POISON)) ||
                 s->ctor)) {
                 /*
                  * Relocate free pointer after the object if it is not
                  * permitted to overwrite the first word of the object on
                  * kmem_cache_free.
                  *
                  * This is the case if we do RCU, have a constructor or
                  * destructor or are poisoning the objects.
                  */
                 s->offset = size;
                 size += sizeof(void *);

So (s->offset == s->inuse) when the free pointer is outside of the object.

>> +static inline unsigned int get_info_end(struct kmem_cache *s)
> static inline track_offset()?
>
The main reason why I don't use that is because there is a track data 
structure in slub. There are functions name get_track() and set_track(). 
I don't want to confuse with them.

Cheers,
Longman

Re: [PATCH v2] mm/slub: Fix incorrect interpretation of s->offset

[Waiman Long]

[-- Attachment #1: Type: text/plain, Size: 512 bytes --]

On 4/27/20 1:42 PM, Waiman Long wrote:
>
>>> +static inline unsigned int get_info_end(struct kmem_cache *s)
>> static inline track_offset()?
>>
> The main reason why I don't use that is because there is a track data 
> structure in slub. There are functions name get_track() and 
> set_track(). I don't want to confuse with them.

Oh, I now realize that Matt meant the offset of the track data 
structure. Depending on the debug option, there may not be a track data 
structure at that offset.

Cheers,
Longman


[-- Attachment #2: Type: text/html, Size: 1083 bytes --]