diff mbox

ARM: mm: fix the boundary checking on bitmaps

Message ID 1404959313-15198-1-git-send-email-haojian.zhuang@linaro.org (mailing list archive)
State New, archived
Headers show

Commit Message

Haojian Zhuang July 10, 2014, 2:28 a.m. UTC
The issue of boundary checking on bitmaps is introduced by this commit
in below.

commit 4d852ef8c2544ce21ae41414099a7504c61164a0
Author: Andreas Herrmann <andreas.herrmann@calxeda.com>
Date:   Tue Feb 25 13:09:53 2014 +0100

    arm: dma-mapping: Add support to extend DMA IOMMU mappings

Multiple bitmaps were introduced as extension. If it needs to extend
a bitmap, it still check whether the allocation exceeding the total
size, not current bitmap size. So change the condition from
mapping->bits to PAGE_SIZE.

Signed-off-by: Haojian Zhuang <haojian.zhuang@linaro.org>
---
 arch/arm/mm/dma-mapping.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

Comments

Will Deacon July 10, 2014, 11:12 a.m. UTC | #1
On Thu, Jul 10, 2014 at 03:28:33AM +0100, Haojian Zhuang wrote:
> The issue of boundary checking on bitmaps is introduced by this commit
> in below.
> 
> commit 4d852ef8c2544ce21ae41414099a7504c61164a0
> Author: Andreas Herrmann <andreas.herrmann@calxeda.com>
> Date:   Tue Feb 25 13:09:53 2014 +0100
> 
>     arm: dma-mapping: Add support to extend DMA IOMMU mappings
> 
> Multiple bitmaps were introduced as extension. If it needs to extend
> a bitmap, it still check whether the allocation exceeding the total
> size, not current bitmap size. So change the condition from
> mapping->bits to PAGE_SIZE.
> 
> Signed-off-by: Haojian Zhuang <haojian.zhuang@linaro.org>
> ---
>  arch/arm/mm/dma-mapping.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c
> index 4c88935..d7da5c3 100644
> --- a/arch/arm/mm/dma-mapping.c
> +++ b/arch/arm/mm/dma-mapping.c
> @@ -1089,9 +1089,9 @@ static inline dma_addr_t __alloc_iova(struct dma_iommu_mapping *mapping,
>  	spin_lock_irqsave(&mapping->lock, flags);
>  	for (i = 0; i < mapping->nr_bitmaps; i++) {
>  		start = bitmap_find_next_zero_area(mapping->bitmaps[i],
> -				mapping->bits, 0, count, align);
> +				PAGE_SIZE, 0, count, align);

I'm probably just missing something, but what guarantees that the bitmap is
PAGE_SIZE? It looks to me like we could easily go off the end of the bitmap
if arm_iommu_create_mapping is called with a small size parameter.

Will
Marek Szyprowski July 14, 2014, 1:02 p.m. UTC | #2
Hello,

On 2014-07-10 04:28, Haojian Zhuang wrote:
> The issue of boundary checking on bitmaps is introduced by this commit
> in below.
>
> commit 4d852ef8c2544ce21ae41414099a7504c61164a0
> Author: Andreas Herrmann <andreas.herrmann@calxeda.com>
> Date:   Tue Feb 25 13:09:53 2014 +0100
>
>      arm: dma-mapping: Add support to extend DMA IOMMU mappings
>
> Multiple bitmaps were introduced as extension. If it needs to extend
> a bitmap, it still check whether the allocation exceeding the total
> size, not current bitmap size. So change the condition from
> mapping->bits to PAGE_SIZE.
>
> Signed-off-by: Haojian Zhuang <haojian.zhuang@linaro.org>

NACK, this patch is not correct. mapping->bits contains the correct size of
a single bitmap, see the beginning of arm_iommu_create_mapping() function.

> ---
>   arch/arm/mm/dma-mapping.c | 8 ++++----
>   1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c
> index 4c88935..d7da5c3 100644
> --- a/arch/arm/mm/dma-mapping.c
> +++ b/arch/arm/mm/dma-mapping.c
> @@ -1089,9 +1089,9 @@ static inline dma_addr_t __alloc_iova(struct dma_iommu_mapping *mapping,
>   	spin_lock_irqsave(&mapping->lock, flags);
>   	for (i = 0; i < mapping->nr_bitmaps; i++) {
>   		start = bitmap_find_next_zero_area(mapping->bitmaps[i],
> -				mapping->bits, 0, count, align);
> +				PAGE_SIZE, 0, count, align);
>   
> -		if (start > mapping->bits)
> +		if (start > PAGE_SIZE)
>   			continue;
>   
>   		bitmap_set(mapping->bitmaps[i], start, count);
> @@ -1110,9 +1110,9 @@ static inline dma_addr_t __alloc_iova(struct dma_iommu_mapping *mapping,
>   		}
>   
>   		start = bitmap_find_next_zero_area(mapping->bitmaps[i],
> -				mapping->bits, 0, count, align);
> +				PAGE_SIZE, 0, count, align);
>   
> -		if (start > mapping->bits) {
> +		if (start > PAGE_SIZE) {
>   			spin_unlock_irqrestore(&mapping->lock, flags);
>   			return DMA_ERROR_CODE;
>   		}

Best regards
diff mbox

Patch

diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c
index 4c88935..d7da5c3 100644
--- a/arch/arm/mm/dma-mapping.c
+++ b/arch/arm/mm/dma-mapping.c
@@ -1089,9 +1089,9 @@  static inline dma_addr_t __alloc_iova(struct dma_iommu_mapping *mapping,
 	spin_lock_irqsave(&mapping->lock, flags);
 	for (i = 0; i < mapping->nr_bitmaps; i++) {
 		start = bitmap_find_next_zero_area(mapping->bitmaps[i],
-				mapping->bits, 0, count, align);
+				PAGE_SIZE, 0, count, align);
 
-		if (start > mapping->bits)
+		if (start > PAGE_SIZE)
 			continue;
 
 		bitmap_set(mapping->bitmaps[i], start, count);
@@ -1110,9 +1110,9 @@  static inline dma_addr_t __alloc_iova(struct dma_iommu_mapping *mapping,
 		}
 
 		start = bitmap_find_next_zero_area(mapping->bitmaps[i],
-				mapping->bits, 0, count, align);
+				PAGE_SIZE, 0, count, align);
 
-		if (start > mapping->bits) {
+		if (start > PAGE_SIZE) {
 			spin_unlock_irqrestore(&mapping->lock, flags);
 			return DMA_ERROR_CODE;
 		}