sh: dynamic ftrace support.

First cut at dynamic ftrace support.

Signed-off-by: Matt Fleming <mjf@gentoo.org>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>

arch/sh/Kconfig

@@ -25,6 +25,8 @@ config SUPERH32
 	select HAVE_KRETPROBES
 	select HAVE_ARCH_TRACEHOOK
 	select HAVE_FUNCTION_TRACER
+	select HAVE_FTRACE_MCOUNT_RECORD
+	select HAVE_DYNAMIC_FTRACE
 
 config SUPERH64
 	def_bool y if CPU_SH5

arch/sh/include/asm/ftrace.h

@@ -1,8 +1,29 @@
 #ifndef __ASM_SH_FTRACE_H
 #define __ASM_SH_FTRACE_H
 
+#ifdef CONFIG_FUNCTION_TRACER
+
+#define MCOUNT_INSN_SIZE	4 /* sizeof mcount call */
+
 #ifndef __ASSEMBLY__
 extern void mcount(void);
+
+#define MCOUNT_ADDR		((long)(mcount))
+
+#ifdef CONFIG_DYNAMIC_FTRACE
+#define CALLER_ADDR		((long)(ftrace_caller))
+#define STUB_ADDR		((long)(ftrace_stub))
+
+#define MCOUNT_INSN_OFFSET	((STUB_ADDR - CALLER_ADDR) >> 1)
 #endif
 
+static inline unsigned long ftrace_call_adjust(unsigned long addr)
+{
+	/* 'addr' is the memory table address. */
+	return addr;
+}
+#endif
+
+#endif /* CONFIG_FUNCTION_TRACER */
+
 #endif /* __ASM_SH_FTRACE_H */

arch/sh/kernel/Makefile_32

@@ -4,6 +4,11 @@
 
 extra-y	:= head_32.o init_task.o vmlinux.lds
 
+ifdef CONFIG_FUNCTION_TRACER
+# Do not profile debug and lowlevel utilities
+CFLAGS_REMOVE_ftrace.o = -pg
+endif
+
 obj-y	:= debugtraps.o io.o io_generic.o irq.o machvec.o process_32.o \
 	   ptrace_32.o setup.o signal_32.o sys_sh.o sys_sh32.o \
 	   syscalls_32.o time_32.o topology.o traps.o traps_32.o
@@ -24,5 +29,6 @@ obj-$(CONFIG_STACKTRACE)	+= stacktrace.o
 obj-$(CONFIG_IO_TRAPPED)	+= io_trapped.o
 obj-$(CONFIG_KPROBES)		+= kprobes.o
 obj-$(CONFIG_GENERIC_GPIO)	+= gpio.o
+obj-$(CONFIG_DYNAMIC_FTRACE)	+= ftrace.o
 
 EXTRA_CFLAGS += -Werror

arch/sh/kernel/entry-common.S

@@ -371,47 +371,3 @@ syscall_exit:
 #endif
 7:	.long	do_syscall_trace_enter
 8:	.long	do_syscall_trace_leave
-
-#ifdef CONFIG_FUNCTION_TRACER
-	.align 2
-	.globl	_mcount
-	.type	_mcount,@function
-	.globl	mcount
-	.type	mcount,@function
-_mcount:
-mcount:
-	mov.l	r4, @-r15
-	mov.l	r5, @-r15
-	mov.l	r6, @-r15
-	mov.l	r7, @-r15
-	sts.l	pr, @-r15
-
-	mov.l	@(20,r15),r4
-	sts	pr, r5
-
-	mov.l	1f, r6
-	mov.l	ftrace_stub, r7	
-	cmp/eq	r6, r7
-	bt	skip_trace
-
-	mov.l	@r6, r6
-	jsr	@r6
-	 nop
-
-skip_trace:
-
-	lds.l	@r15+, pr
-	mov.l	@r15+, r7
-	mov.l	@r15+, r6
-	mov.l	@r15+, r5
-	rts
-	 mov.l	@r15+, r4
-
-	.align 2
-1:	.long	ftrace_trace_function
-
-	.globl	ftrace_stub
-ftrace_stub:
-	rts
-	 nop
-#endif /* CONFIG_FUNCTION_TRACER */

arch/sh/kernel/ftrace.c

@@ -0,0 +1,109 @@
+/*
+ * Copyright (C) 2008 Matt Fleming <mjf@gentoo.org>
+ *
+ * Code for replacing ftrace calls with jumps.
+ *
+ * Copyright (C) 2007-2008 Steven Rostedt <srostedt@redhat.com>
+ *
+ * Thanks goes to Ingo Molnar, for suggesting the idea.
+ * Mathieu Desnoyers, for suggesting postponing the modifications.
+ * Arjan van de Ven, for keeping me straight, and explaining to me
+ * the dangers of modifying code on the run.
+ */
+#include <linux/uaccess.h>
+#include <linux/ftrace.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <asm/ftrace.h>
+#include <asm/cacheflush.h>
+
+static unsigned char ftrace_nop[] = {
+	0x09, 0x00,		/* nop */
+	0x09, 0x00,		/* nop */
+};
+
+static unsigned char ftrace_replaced_code[MCOUNT_INSN_SIZE];
+
+unsigned char *ftrace_nop_replace(void)
+{
+	return ftrace_nop;
+}
+
+static int is_sh_nop(unsigned char *ip)
+{
+	return strncmp(ip, ftrace_nop, sizeof(ftrace_nop));
+}
+
+unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
+{
+	/* Place the address in the memory table. */
+	if (addr == CALLER_ADDR)
+		__raw_writel(addr + MCOUNT_INSN_OFFSET, ftrace_replaced_code);
+	else
+		__raw_writel(addr, ftrace_replaced_code);
+
+	/*
+	 * No locking needed, this must be called via kstop_machine
+	 * which in essence is like running on a uniprocessor machine.
+	 */
+	return ftrace_replaced_code;
+}
+
+int ftrace_modify_code(unsigned long ip, unsigned char *old_code,
+		       unsigned char *new_code)
+{
+	unsigned char replaced[MCOUNT_INSN_SIZE];
+
+	/*
+	 * Note: Due to modules and __init, code can
+	 *  disappear and change, we need to protect against faulting
+	 *  as well as code changing. We do this by using the
+	 *  probe_kernel_* functions.
+	 *
+	 * No real locking needed, this code is run through
+	 * kstop_machine, or before SMP starts.
+	 */
+
+	/*
+	 * If we're trying to nop out a call to a function, we instead
+	 * place a call to the address after the memory table.
+	 */
+	if (is_sh_nop(new_code) == 0)
+		__raw_writel(ip + MCOUNT_INSN_SIZE, (unsigned long)new_code);
+
+	/* read the text we want to modify */
+	if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE))
+		return -EFAULT;
+
+	/* Make sure it is what we expect it to be */
+	if (memcmp(replaced, old_code, MCOUNT_INSN_SIZE) != 0)
+		return -EINVAL;
+
+	/* replace the text with the new text */
+	if (probe_kernel_write((void *)ip, new_code, MCOUNT_INSN_SIZE))
+		return -EPERM;
+
+	flush_icache_range(ip, ip + MCOUNT_INSN_SIZE);
+
+	return 0;
+}
+
+int ftrace_update_ftrace_func(ftrace_func_t func)
+{
+	unsigned long ip = (unsigned long)(&ftrace_call);
+	unsigned char old[MCOUNT_INSN_SIZE], *new;
+
+	memcpy(old, (unsigned char *)(ip + MCOUNT_INSN_OFFSET), MCOUNT_INSN_SIZE);
+	new = ftrace_call_replace(ip, (unsigned long)func);
+
+	return ftrace_modify_code(ip + MCOUNT_INSN_OFFSET, old, new);
+}
+
+int __init ftrace_dyn_arch_init(void *data)
+{
+	/* The return code is retured via data */
+	__raw_writel(0, (unsigned long)data);
+
+	return 0;
+}

arch/sh/lib/Makefile

@@ -11,6 +11,7 @@ memcpy-y			:= memcpy.o
 memcpy-$(CONFIG_CPU_SH4)	:= memcpy-sh4.o
 
 lib-$(CONFIG_MMU)		+= copy_page.o clear_page.o
+lib-$(CONFIG_FUNCTION_TRACER)	+= mcount.o
 lib-y				+= $(memcpy-y)
 
 EXTRA_CFLAGS += -Werror

arch/sh/lib/mcount.S

@@ -0,0 +1,90 @@
+/*
+ * arch/sh/lib/mcount.S
+ *
+ *  Copyright (C) 2008  Paul Mundt
+ *  Copyright (C) 2008  Matt Fleming
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ */
+#include <asm/ftrace.h>
+
+#define MCOUNT_ENTER()		\
+	mov.l	r4, @-r15;	\
+	mov.l	r5, @-r15;	\
+	mov.l	r6, @-r15;	\
+	mov.l	r7, @-r15;	\
+	sts.l	pr, @-r15;	\
+				\
+	mov.l	@(20,r15),r4;	\
+	sts	pr, r5
+
+#define MCOUNT_LEAVE()		\
+	lds.l	@r15+, pr;	\
+	mov.l	@r15+, r7;	\
+	mov.l	@r15+, r6;	\
+	mov.l	@r15+, r5;	\
+	rts;			\
+	 mov.l	@r15+, r4
+
+	.align 2
+	.globl	_mcount
+	.type	_mcount,@function
+	.globl	mcount
+	.type	mcount,@function
+_mcount:
+mcount:
+	MCOUNT_ENTER()
+
+#ifdef CONFIG_DYNAMIC_FTRACE
+	.globl	mcount_call
+mcount_call:
+	mov.l	.Lftrace_stub, r6
+#else
+	mov.l	.Lftrace_trace_function, r6
+	mov.l	ftrace_stub, r7
+	cmp/eq	r6, r7
+	bt	skip_trace
+	mov.l	@r6, r6
+#endif
+
+	jsr	@r6
+	 nop
+
+skip_trace:
+	MCOUNT_LEAVE()
+
+	.align 2
+.Lftrace_trace_function:
+	.long	ftrace_trace_function
+
+#ifdef CONFIG_DYNAMIC_FTRACE
+	.globl ftrace_caller
+ftrace_caller:
+	MCOUNT_ENTER()
+
+	.globl ftrace_call
+ftrace_call:
+	mov.l	.Lftrace_stub, r6
+	jsr	@r6
+	 nop
+
+	MCOUNT_LEAVE()
+#endif /* CONFIG_DYNAMIC_FTRACE */
+
+/*
+ * NOTE: From here on the locations of the .Lftrace_stub label and
+ * ftrace_stub itself are fixed. Adding additional data here will skew
+ * the displacement for the memory table and break the block replacement.
+ * Place new labels either after the ftrace_stub body, or before
+ * ftrace_caller. You have been warned.
+ */
+	.align 2
+.Lftrace_stub:
+	.long	ftrace_stub
+
+	.globl	ftrace_stub
+ftrace_stub:
+	rts
+	 nop

scripts/recordmcount.pl

@@ -167,6 +167,17 @@ if ($arch eq "x86_64") {
     $objcopy .= " -O elf32-i386";
     $cc .= " -m32";
 
+} elsif ($arch eq "sh") {
+    $section_regex = "Disassembly of section\\s+(\\S+):";
+    $function_regex = "^([0-9a-fA-F]+)\\s+<(.*?)>:";
+    $mcount_regex = "^\\s*([0-9a-fA-F]+):.*\\smcount\$";
+    $type = ".long";
+
+    # force flags for this arch
+    $ld .= " -m shlelf_linux";
+    $objcopy .= " -O elf32-sh-linux";
+    $cc .= " -m32";
+
 } else {
     die "Arch $arch is not supported with CONFIG_FTRACE_MCOUNT_RECORD";
 }