arm64: Correct virt_addr_valid
authorLaura Abbott <lauraa@codeaurora.org>
Wed, 11 Dec 2013 01:23:02 +0000 (01:23 +0000)
committerCatalin Marinas <catalin.marinas@arm.com>
Thu, 19 Dec 2013 17:43:02 +0000 (17:43 +0000)
The definition of virt_addr_valid is that virt_addr_valid should
return true if and only if virt_to_page returns a valid pointer.
The current definition of virt_addr_valid only checks against the
virtual address range. There's no guarantee that just because a
virtual address falls bewteen PAGE_OFFSET and high_memory the
associated physical memory has a valid backing struct page. Follow
the example of other architectures and convert to pfn_valid to
verify that the virtual address is actually valid.

Cc: Will Deacon <will.deacon@arm.com>
Cc: Nicolas Pitre <nico@linaro.org>
Signed-off-by: Laura Abbott <lauraa@codeaurora.org>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
arch/arm64/include/asm/memory.h

index 3776217..9dc5dc3 100644 (file)
@@ -146,8 +146,7 @@ static inline void *phys_to_virt(phys_addr_t x)
 #define ARCH_PFN_OFFSET                PHYS_PFN_OFFSET
 
 #define virt_to_page(kaddr)    pfn_to_page(__pa(kaddr) >> PAGE_SHIFT)
-#define        virt_addr_valid(kaddr)  (((void *)(kaddr) >= (void *)PAGE_OFFSET) && \
-                                ((void *)(kaddr) < (void *)high_memory))
+#define        virt_addr_valid(kaddr)  pfn_valid(__pa(kaddr) >> PAGE_SHIFT)
 
 #endif