ftrace: recordmcount fixes

Ingo,

Please pull the latest tip/perf/recordmcount tree, which can be found at:

git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
tip/perf/recordmcount

Steven Rostedt (2):
ftrace: Do not process kernel/trace/ftrace.o with C recordmcount program
ftrace: Use objtree for C version of recordmcount

----

 scripts/Makefile.build |    2 +-
 scripts/recordmcount.c |   22 ++++++++++++++++++++--
 2 files changed, 21 insertions(+), 3 deletions(-)

From: Steven Rostedt <[email protected]>

The file kernel/trace/ftrace.c references the mcount() call to convert the mcount() callers to nops. But because it references mcount(), the mcount() address is placed in the relocation table.

The C version of recordmcount reads the relocation table of all object files, and it will add all references to mcount to the __mcount_loc table that is used to find the places that call mcount() and change the call to a nop. When recordmcount finds the mcount reference in kernel/trace/ftrace.o, it saves that location even though the code is not a call, but references mcount as data.

On boot up, when all calls are converted to nops, the code has a safety check to determine what op code it is actually replacing before it replaces it. If that op code at the address does not match, then a warning is printed and the function tracer is disabled.

The reference to mcount in ftrace.c, causes this warning to trigger, since the reference is not a call to mcount(). The ftrace.c file is not compiled with the -pg flag, so no calls to mcount() should be expected.

This patch simply makes recordmcount.c skip the kernel/trace/ftrace.c file. This was the same solution used by the perl version of recordmcount.

---
 scripts/recordmcount.c |   22 ++++++++++++++++++++--
 1 files changed, 20 insertions(+), 2 deletions(-)

index 7f7f718..26e1271 100644
@@ -313,12 +313,30 @@ do_file(char const *const fname)
 int
 main(int argc, char const *argv[])
 {
+	const char ftrace[] = "kernel/trace/ftrace.o";
+	int ftrace_size = sizeof(ftrace) - 1;
 	int n_error = 0;  /* gcc-4.3.0 false positive complaint */
-	if (argc <= 1)
+
+	if (argc <= 1) {
 		fprintf(stderr, "usage: recordmcount file.o...\n");
-	else  /* Process each file in turn, allowing deep failure. */
+		return 0;
+	}
+
+	/* Process each file in turn, allowing deep failure. */
 	for (--argc, ++argv; 0 < argc; --argc, ++argv) {
 		int const sjval = setjmp(jmpenv);
+		int len;
+
+		/*
+		 * The file kernel/trace/ftrace.o references the mcount
+		 * function but does not call it. Since ftrace.o should
+		 * not be traced anyway, we just skip it.
+		 */
+		len = strlen(argv[0]);
+		if (len >= ftrace_size &&
+		    strcmp(argv[0] + (len - ftrace_size), ftrace) == 0)
+			continue;
+
 		switch (sjval) {
 		default: {
 			fprintf(stderr, "internal error: %s\n", argv[0]);

1.7.1

From: Steven Rostedt <[email protected]>

The C version of recordmcount is compiled to a binary, which will end up located in the objtree. If the kernel is built with O=path, the srctree will not include the binary recordmcount caller.

---
 scripts/Makefile.build |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

index 4d03a7e..4db60b2 100644
@@ -210,7 +210,7 @@ endif
 
 ifdef CONFIG_FTRACE_MCOUNT_RECORD
 ifdef BUILD_C_RECORDMCOUNT
-cmd_record_mcount = $(srctree)/scripts/recordmcount "$(@)";
+cmd_record_mcount = $(objtree)/scripts/recordmcount "$(@)";
 else
 cmd_record_mcount = set -e ; perl $(srctree)/scripts/recordmcount.pl "$(ARCH)" \
 	"$(if $(CONFIG_CPU_BIG_ENDIAN),big,little)" \

1.7.1

Pulled, thanks Steve!

Ingo