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>
        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;                                             \
 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;                                             \