arm64: check for upper PAGE_SHIFT bits in pfn_valid()

pfn_valid() returns a false positive when the lower (64 - PAGE_SHIFT)
bits match a valid pfn but some of the upper bits are set.  This caused
a kernel panic in kpageflags_read() when a userspace utility parsed
/proc/*/pagemap, neglected to discard the upper flag bits, and tried to
lseek()+read() from the corresponding offset in /proc/kpageflags.

A valid pfn will never have the upper PAGE_SHIFT bits set, so simply
check for this before passing the pfn to memblock_is_memory().

Change-Id: Ief5d8cd4dd93cbecd545a634a8d5885865cb5970
Signed-off-by: Greg Hackmann <ghackmann@google.com>

diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
index 17bf39ac83ba073109118817c3ab72346ae3824b..325f89ff897ea8f24a32d361939742294a4aaa19 100644 (file)
--- a/arch/arm64/mm/init.c
+++ b/arch/arm64/mm/init.c
@@ -118,9 +118,11 @@ static void __init zone_sizes_init(unsigned long min, unsigned long max)
 }
 
 #ifdef CONFIG_HAVE_ARCH_PFN_VALID
+#define PFN_MASK ((1UL << (64 - PAGE_SHIFT)) - 1)
+
 int pfn_valid(unsigned long pfn)
 {
-       return memblock_is_memory(pfn << PAGE_SHIFT);
+       return (pfn & PFN_MASK) == pfn && memblock_is_memory(pfn << PAGE_SHIFT);
 }
 EXPORT_SYMBOL(pfn_valid);
 #endif