Skip to contentSkip to conversation

iommu: amd: Fix IO_PAGE_FAULT due to __unmap_single() size overflow

PATCH In review Suravee Suthikulpanit 3 participants 5 messages Jun 24, 2020 – Jul 30, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 iommu: amd: Fix IO_PAGE_FAULT due to __unmap_single() size overflow

Currently, an integer is used to specify the size in unmap_sg(). With 2GB worth of pages (512k 4k pages), it requires 31 bits (i.e. (1 << 19) << 12), which overflows the integer, and ends up unmapping more pages than intended. Subsequently, this results in IO_PAGE_FAULT.

Uses size_t instead of int to pass parameter to __unmap_single().

Please note that this patch is only for the stable-kernels tree because the commit be62dbf554c5 ("iommu/amd: Convert AMD iommu driver to the dma-iommu api"), which removes the function unmap_sg() was introduced in v5.5. This patch is not applicable in subsequent kernel versions.

---

 drivers/iommu/amd_iommu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iommu/amd_iommu.c +1 -1
index 32de8e7bb8b4..7adc021932b8 100644
@@ -2670,7 +2670,7 @@ static void unmap_sg(struct device *dev, struct scatterlist *sglist,
 	struct protection_domain *domain;
 	struct dma_ops_domain *dma_dom;
 	unsigned long startaddr;
-	int npages;
+	size_t npages;
 
 	domain = get_domain(dev);
 	if (IS_ERR(domain))
Signature · 1 line
2.17.1
maintainer
#1 lore raw
Quoted 6 lines · Suravee Suthikulpanit

Currently, an integer is used to specify the size in unmap_sg(). With 2GB worth of pages (512k 4k pages), it requires 31 bits (i.e. (1 << 19) << 12), which overflows the integer, and ends up unmapping more pages than intended. Subsequently, this results in IO_PAGE_FAULT.

Quoted 7 lines

Uses size_t instead of int to pass parameter to __unmap_single().

Please note that this patch is only for the stable-kernels tree because the commit be62dbf554c5 ("iommu/amd: Convert AMD iommu driver to the dma-iommu api"), which removes the function unmap_sg() was introduced in v5.5. This patch is not applicable in subsequent kernel versions.

So what exact stable kernel version(s) do you want to see this patch applied to?

thanks,

greg k-h

maintainer
#2 lore raw

Hi Greg,

Quoted 2 lines · Greg KH

So what exact stable kernel version(s) do you want to see this patch applied to?

It is needed in kernels <= v5.4. Linux v5.5 has replaced the code with the bug.

Joerg

maintainer
#3 lore raw
Quoted 8 lines · Joerg Roedel

Hi Greg,

Quoted 2 lines · Greg KH

So what exact stable kernel version(s) do you want to see this patch applied to?

It is needed in kernels <= v5.4. Linux v5.5 has replaced the code with the bug.

This doesn't apply to any older kernels that I can see :(
_______________________________________________
iommu mailing list [email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu

maintainer Acked-by on v2 1/1
#4 lore raw
In reply to Suravee Suthikulpanit
Quoted 14 lines · Suravee Suthikulpanit

Currently, an integer is used to specify the size in unmap_sg(). With 2GB worth of pages (512k 4k pages), it requires 31 bits (i.e. (1 << 19) << 12), which overflows the integer, and ends up unmapping more pages than intended. Subsequently, this results in IO_PAGE_FAULT.

Uses size_t instead of int to pass parameter to __unmap_single().

Please note that this patch is only for the stable-kernels tree because the commit be62dbf554c5 ("iommu/amd: Convert AMD iommu driver to the dma-iommu api"), which removes the function unmap_sg() was introduced in v5.5. This patch is not applicable in subsequent kernel versions.

Quoted 4 lines