get_kernel_vsid() has a very stern comment saying that it's only valid
for kernel addresses, but there's nothing in the code to enforce that.
Rather than hoping our callers are well behaved, add a check and return
a VSID of 0 (invalid).
Signed-off-by: Michael Ellerman <m...@ellerman.id.au>
---
arch/powerpc/include/asm/book3s/64/mmu-hash.h | 3 +++
1 file changed, 3 insertions(+)
diff --git a/arch/powerpc/include/asm/book3s/64/mmu-hash.h
b/arch/powerpc/include/asm/book3s/64/mmu-hash.h
index a5ab6f5b8a7f..10a34282829e 100644
--- a/arch/powerpc/include/asm/book3s/64/mmu-hash.h
+++ b/arch/powerpc/include/asm/book3s/64/mmu-hash.h
@@ -682,6 +682,9 @@ static inline unsigned long get_kernel_vsid(unsigned long
ea, int ssize)
{
unsigned long context;
+ if (!is_kernel_addr(ea))
+ return 0;
+
/*
* For kernel space, we use context ids 1-4 to map the address space as
* below:
--
2.7.4
