s390/ftrace: optimize function graph caller code

When the function graph tracer is disabled we can skip three additional
instructions. So let's just do this.

So if function tracing is enabled but function graph tracing is
runtime disabled, we get away with a single unconditional branch.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

arch/s390/include/asm/ftrace.h

@@ -4,6 +4,7 @@
 #ifndef __ASSEMBLY__
 
 extern void _mcount(void);
+extern char ftrace_graph_caller_end;
 
 struct dyn_arch_ftrace { };
 

arch/s390/kernel/ftrace.c

@@ -170,6 +170,29 @@ unsigned long __kprobes prepare_ftrace_return(unsigned long parent,
  * directly after the instructions. To enable the call we calculate
  * the original offset to prepare_ftrace_return and put it back.
  */
+
+#ifdef CONFIG_64BIT
+
+int ftrace_enable_ftrace_graph_caller(void)
+{
+	static unsigned short offset = 0x0002;
+
+	return probe_kernel_write((void *) ftrace_graph_caller + 2,
+				  &offset, sizeof(offset));
+}
+
+int ftrace_disable_ftrace_graph_caller(void)
+{
+	unsigned short offset;
+
+	offset = ((void *) &ftrace_graph_caller_end -
+		  (void *) ftrace_graph_caller) / 2;
+	return probe_kernel_write((void *) ftrace_graph_caller + 2,
+				  &offset, sizeof(offset));
+}
+
+#else /* CONFIG_64BIT */
+
 int ftrace_enable_ftrace_graph_caller(void)
 {
 	unsigned short offset;
@@ -188,5 +211,6 @@ int ftrace_disable_ftrace_graph_caller(void)
 				  &offset, sizeof(offset));
 }
 
+#endif /* CONFIG_64BIT */
 #endif /* CONFIG_DYNAMIC_FTRACE */
 #endif /* CONFIG_FUNCTION_GRAPH_TRACER */

arch/s390/kernel/mcount64.S

@@ -32,14 +32,17 @@ ENTRY(ftrace_caller)
 	lg	%r14,0(%r14)
 	basr	%r14,%r14
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
+# The j instruction gets runtime patched to a nop instruction.
+# See ftrace_enable_ftrace_graph_caller. The patched instruction is:
+#	j	.+4
+ENTRY(ftrace_graph_caller)
+	j	ftrace_graph_caller_end
 	lg	%r2,168(%r15)
 	lg	%r3,272(%r15)
-ENTRY(ftrace_graph_caller)
-# The bras instruction gets runtime patched to call prepare_ftrace_return.
-# See ftrace_enable_ftrace_graph_caller. The patched instruction is:
-#	bras	%r14,prepare_ftrace_return
-	bras	%r14,0f
-0:	stg	%r2,168(%r15)
+	brasl	%r14,prepare_ftrace_return
+	stg	%r2,168(%r15)
+ftrace_graph_caller_end:
+	.globl	ftrace_graph_caller_end
 #endif
 	aghi	%r15,160
 	lmg	%r2,%r5,32(%r15)