tile: ftrace: fix function_graph tracer issues

- Add support for ARCH_SUPPORTS_FTRACE_OPS
- Replace the instruction in ftrace_call with the bundle {move r10, lr;
jal ftrace_stub}, so that the lr contains the right value after returning
from ftrace_stub.  An alternative fix might be to leave the instruction
in ftrace_call alone when it is being updated with ftrace_stub.

Signed-off-by: Tony Lu <zlu@ezchip.com>
Signed-off-by: Chris Metcalf <cmetcalf@ezchip.com>

arch/tile/include/asm/ftrace.h

@@ -23,6 +23,8 @@
 #ifndef __ASSEMBLY__
 extern void __mcount(void);
 
+#define ARCH_SUPPORTS_FTRACE_OPS 1
+
 #ifdef CONFIG_DYNAMIC_FTRACE
 static inline unsigned long ftrace_call_adjust(unsigned long addr)
 {

arch/tile/kernel/ftrace.c

@@ -74,7 +74,11 @@ static unsigned long ftrace_gen_branch(unsigned long pc, unsigned long addr,
 			create_JumpOff_X1(pcrel_by_instr);
 	}
 
-	if (addr == FTRACE_ADDR) {
+	/*
+	 * Also put { move r10, lr; jal ftrace_stub } in a bundle, which
+	 * is used to replace the instruction in address ftrace_call.
+	 */
+	if (addr == FTRACE_ADDR || addr == (unsigned long)ftrace_stub) {
 		/* opcode: or r10, lr, zero */
 		opcode_x0 =
 			create_Dest_X0(10) |

arch/tile/kernel/mcount_64.S

@@ -81,7 +81,12 @@ STD_ENTRY(ftrace_caller)
 
 	/* arg1: self return address */
 	/* arg2: parent's return address */
-	{ move	r0, lr; move	r1, r10 }
+	/* arg3: ftrace_ops */
+	/* arg4: regs (but make it NULL) */
+	{ move	r0, lr;  moveli        r2, hw2_last(function_trace_op) }
+	{ move	r1, r10; shl16insli    r2, r2, hw1(function_trace_op) }
+	{ movei	r3, 0;   shl16insli    r2, r2, hw0(function_trace_op) }
+	ld	r2,r2
 
 	.global	ftrace_call
 ftrace_call: