powerpc/asm: Mark cr0 as clobbered in mftb()
authorOliver O'Halloran <oohall@gmail.com>
Thu, 6 Jul 2017 08:46:43 +0000 (18:46 +1000)
committerMichael Ellerman <mpe@ellerman.id.au>
Tue, 11 Jul 2017 02:53:53 +0000 (12:53 +1000)
The workaround for the CELL timebase bug does not correctly mark cr0 as
being clobbered. This means GCC doesn't know that the asm block changes cr0 and
might leave the result of an unrelated comparison in cr0 across the block, which
we then trash, leading to basically random behaviour.

Fixes: 859deea949c3 ("[POWERPC] Cell timebase bug workaround")
Cc: stable@vger.kernel.org # v2.6.19+
Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
[mpe: Tweak change log and flag for stable]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
arch/powerpc/include/asm/reg.h

index 7e50e47375d63f8cdd11ba710cf18f5f0e442deb..a3b6575c7842ed42261efbde20fd1b55e269e031 100644 (file)
@@ -1303,7 +1303,7 @@ static inline void msr_check_and_clear(unsigned long bits)
                                "       .llong 0\n"                     \
                                ".previous"                             \
                        : "=r" (rval) \
-                       : "i" (CPU_FTR_CELL_TB_BUG), "i" (SPRN_TBRL)); \
+                       : "i" (CPU_FTR_CELL_TB_BUG), "i" (SPRN_TBRL) : "cr0"); \
                        rval;})
 #else
 #define mftb()         ({unsigned long rval;   \