powerpc: Fix unpaired __trace_hcall_entry and __trace_hcall_exit

 Unpaired calling of __trace_hcall_entry and __trace_hcall_exit could
 cause incorrect preempt count. And it might happen as the global
 variable hcall_tracepoint_refcount is checked separately before calling
 them.

 Instead, store the value that was used on entry in the stack frame
 and retreive it from there after the call
Reported-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Tested-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

arch/powerpc/platforms/pseries/hvCall.S

@@ -36,6 +36,7 @@ BEGIN_FTR_SECTION;						\
 	b	1f;						\
 END_FTR_SECTION(0, 1);						\
 	ld      r12,hcall_tracepoint_refcount@toc(r2);		\
+	std	r12,32(r1);					\
 	cmpdi	r12,0;						\
 	beq+	1f;						\
 	mflr	r0;						\
@@ -74,7 +75,7 @@ END_FTR_SECTION(0, 1);						\
 BEGIN_FTR_SECTION;						\
 	b	1f;						\
 END_FTR_SECTION(0, 1);						\
-	ld      r12,hcall_tracepoint_refcount@toc(r2);		\
+	ld      r12,32(r1);					\
 	cmpdi	r12,0;						\
 	beq+	1f;						\
 	mflr	r0;						\