[PATCHv2] arm64: Correctly bounds check virt_addr_valid

[PATCHv2] arm64: Correctly bounds check virt_addr_valid

[Laura Abbott]

virt_addr_valid is supposed to return true if and only if virt_to_page
returns a valid page structure. The current macro does math on whatever
address is given and passes that to pfn_valid to verify. vmalloc and
module addresses can happen to generate a pfn that 'happens' to be
valid. Fix this by only performing the pfn_valid check on addresses that
have the potential to be valid.

Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
---
v2: Properly parenthesize macro arguments. Re-factor to common macro.

Also in case it wasn't clear, there's no need to try and squeeze this
into 4.8. Hardened usercopy should have all the checks, this is just for
full correctness.
---
 arch/arm64/include/asm/memory.h | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h
index 31b7322..ba62df8 100644
--- a/arch/arm64/include/asm/memory.h
+++ b/arch/arm64/include/asm/memory.h
@@ -214,7 +214,7 @@ static inline void *phys_to_virt(phys_addr_t x)
 
 #ifndef CONFIG_SPARSEMEM_VMEMMAP
 #define virt_to_page(kaddr)	pfn_to_page(__pa(kaddr) >> PAGE_SHIFT)
-#define virt_addr_valid(kaddr)	pfn_valid(__pa(kaddr) >> PAGE_SHIFT)
+#define _virt_addr_valid(kaddr)	pfn_valid(__pa(kaddr) >> PAGE_SHIFT)
 #else
 #define __virt_to_pgoff(kaddr)	(((u64)(kaddr) & ~PAGE_OFFSET) / PAGE_SIZE * sizeof(struct page))
 #define __page_to_voff(kaddr)	(((u64)(page) & ~VMEMMAP_START) * PAGE_SIZE / sizeof(struct page))
@@ -222,11 +222,15 @@ static inline void *phys_to_virt(phys_addr_t x)
 #define page_to_virt(page)	((void *)((__page_to_voff(page)) | PAGE_OFFSET))
 #define virt_to_page(vaddr)	((struct page *)((__virt_to_pgoff(vaddr)) | VMEMMAP_START))
 
-#define virt_addr_valid(kaddr)	pfn_valid((((u64)(kaddr) & ~PAGE_OFFSET) \
+#define _virt_addr_valid(kaddr)	pfn_valid((((u64)(kaddr) & ~PAGE_OFFSET) \
 					   + PHYS_OFFSET) >> PAGE_SHIFT)
 #endif
 #endif
 
+#define _virt_addr_is_linear(kaddr)	(((u64)(kaddr)) >= PAGE_OFFSET)
+#define virt_addr_valid(kaddr)		(_virt_addr_is_linear(kaddr) && \
+					 _virt_addr_valid(kaddr))
+
 #include <asm-generic/memory_model.h>
 
 #endif
-- 
2.7.4

[PATCHv2] arm64: Correctly bounds check virt_addr_valid

[Kees Cook]

On Wed, Sep 21, 2016 at 3:25 PM, Laura Abbott <labbott@redhat.com> wrote:
>
> virt_addr_valid is supposed to return true if and only if virt_to_page
> returns a valid page structure. The current macro does math on whatever
> address is given and passes that to pfn_valid to verify. vmalloc and
> module addresses can happen to generate a pfn that 'happens' to be
> valid. Fix this by only performing the pfn_valid check on addresses that
> have the potential to be valid.
>
> Acked-by: Mark Rutland <mark.rutland@arm.com>
> Signed-off-by: Laura Abbott <labbott@redhat.com>
> ---
> v2: Properly parenthesize macro arguments. Re-factor to common macro.
>
> Also in case it wasn't clear, there's no need to try and squeeze this
> into 4.8. Hardened usercopy should have all the checks, this is just for
> full correctness.

After this lands for 4.9, I should likely drop the checks that are in
hardened usercopy? That'll speed things up ever so slightly, and will
let us catch other architectures that have a weird
virt_addr_valid()...

-Kees

-- 
Kees Cook
Nexus Security

[PATCHv2] arm64: Correctly bounds check virt_addr_valid

[Will Deacon]

On Wed, Sep 21, 2016 at 03:25:04PM -0700, Laura Abbott wrote:
> 
> virt_addr_valid is supposed to return true if and only if virt_to_page
> returns a valid page structure. The current macro does math on whatever
> address is given and passes that to pfn_valid to verify. vmalloc and
> module addresses can happen to generate a pfn that 'happens' to be
> valid. Fix this by only performing the pfn_valid check on addresses that
> have the potential to be valid.
> 
> Acked-by: Mark Rutland <mark.rutland@arm.com>
> Signed-off-by: Laura Abbott <labbott@redhat.com>
> ---
> v2: Properly parenthesize macro arguments. Re-factor to common macro.
> 
> Also in case it wasn't clear, there's no need to try and squeeze this
> into 4.8. Hardened usercopy should have all the checks, this is just for
> full correctness.

Thanks, I'll push this onto for-next/core later today.

Will