* [PATCH 0/9] ftrace: updates for tip
@ 2009-01-13 6:11 Steven Rostedt
2009-01-13 6:11 ` [PATCH 1/9] ftrace, ia64: make recordmcount destinct module compile Steven Rostedt
` (9 more replies)
0 siblings, 10 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig
The following patches are in:
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
branch: tip/devel
Frederic Weisbecker (2):
tracing/ftrace: separate events tracing and stats tracing engine
tracing: add a new workqueue tracer
Shaohua Li (6):
ftrace, ia64: make recordmcount destinct module compile
ftrace, ia64: explictly ignore a file in recordmcount.pl
ftrace, ia64: Add macro for ftrace_caller
ftrace, ia64: Add recordmcount for ia64
ftrace, ia64: IA64 static ftrace support
ftrace, ia64: IA64 dynamic ftrace support
Uwe Kleine-K?nig (1):
ftrace, trivial: fix typo "resgister" -> "register"
----
arch/ia64/Kconfig | 3 +
arch/ia64/include/asm/ftrace.h | 28 ++++
arch/ia64/kernel/Makefile | 5 +
arch/ia64/kernel/entry.S | 100 ++++++++++++++
arch/ia64/kernel/ftrace.c | 206 ++++++++++++++++++++++++++++
arch/ia64/kernel/ia64_ksyms.c | 6 +
include/linux/ftrace.h | 4 +
include/trace/workqueue.h | 25 ++++
kernel/trace/Kconfig | 11 ++
kernel/trace/Makefile | 1 +
kernel/trace/ftrace.c | 14 +--
kernel/trace/trace.c | 2 -
kernel/trace/trace.h | 20 ---
kernel/trace/trace_branch.c | 108 +++++++++-------
kernel/trace/trace_stat.c | 191 +++++++++++---------------
kernel/trace/trace_stat.h | 31 +++++
kernel/trace/trace_workqueue.c | 287 ++++++++++++++++++++++++++++++++++++++++
kernel/workqueue.c | 16 ++-
scripts/Makefile.build | 13 +-
scripts/recordmcount.pl | 18 ++-
20 files changed, 888 insertions(+), 201 deletions(-)
--
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH 1/9] ftrace, ia64: make recordmcount destinct module compile
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
@ 2009-01-13 6:11 ` Steven Rostedt
2009-01-13 6:11 ` [PATCH 2/9] ftrace, ia64: explictly ignore a file in recordmcount.pl Steven Rostedt
` (8 subsequent siblings)
9 siblings, 0 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig, Steven Rostedt
[-- Attachment #1: 0001-ftrace-ia64-make-recordmcount-destinct-module-comp.patch --]
[-- Type: text/plain, Size: 2454 bytes --]
From: Shaohua Li <shaohua.li@intel.com>
In IA64, module build and kernel build use different option.
Make recordmcount.pl differentiate the two cases.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
scripts/Makefile.build | 13 +++++++------
scripts/recordmcount.pl | 6 +++---
2 files changed, 10 insertions(+), 9 deletions(-)
diff --git a/scripts/Makefile.build b/scripts/Makefile.build
index 5d90030..b5efa98 100644
--- a/scripts/Makefile.build
+++ b/scripts/Makefile.build
@@ -112,13 +112,13 @@ endif
# ---------------------------------------------------------------------------
# Default is built-in, unless we know otherwise
-modkern_cflags := $(CFLAGS_KERNEL)
+modkern_cflags = $(if $(part-of-module), $(CFLAGS_MODULE), $(CFLAGS_KERNEL))
quiet_modtag := $(empty) $(empty)
-$(real-objs-m) : modkern_cflags := $(CFLAGS_MODULE)
-$(real-objs-m:.o=.i) : modkern_cflags := $(CFLAGS_MODULE)
-$(real-objs-m:.o=.s) : modkern_cflags := $(CFLAGS_MODULE)
-$(real-objs-m:.o=.lst): modkern_cflags := $(CFLAGS_MODULE)
+$(real-objs-m) : part-of-module := y
+$(real-objs-m:.o=.i) : part-of-module := y
+$(real-objs-m:.o=.s) : part-of-module := y
+$(real-objs-m:.o=.lst): part-of-module := y
$(real-objs-m) : quiet_modtag := [M]
$(real-objs-m:.o=.i) : quiet_modtag := [M]
@@ -215,7 +215,8 @@ endif
ifdef CONFIG_FTRACE_MCOUNT_RECORD
cmd_record_mcount = perl $(srctree)/scripts/recordmcount.pl "$(ARCH)" \
"$(if $(CONFIG_64BIT),64,32)" \
- "$(OBJDUMP)" "$(OBJCOPY)" "$(CC)" "$(LD)" "$(NM)" "$(RM)" "$(MV)" "$(@)";
+ "$(OBJDUMP)" "$(OBJCOPY)" "$(CC)" "$(LD)" "$(NM)" "$(RM)" "$(MV)" \
+ "$(if $(part-of-module),1,0)" "$(@)";
endif
define rule_cc_o_c
diff --git a/scripts/recordmcount.pl b/scripts/recordmcount.pl
index fe83141..282485a 100755
--- a/scripts/recordmcount.pl
+++ b/scripts/recordmcount.pl
@@ -100,14 +100,14 @@ $P =~ s@.*/@@g;
my $V = '0.1';
-if ($#ARGV < 6) {
- print "usage: $P arch objdump objcopy cc ld nm rm mv inputfile\n";
+if ($#ARGV < 7) {
+ print "usage: $P arch objdump objcopy cc ld nm rm mv is_module inputfile\n";
print "version: $V\n";
exit(1);
}
my ($arch, $bits, $objdump, $objcopy, $cc,
- $ld, $nm, $rm, $mv, $inputfile) = @ARGV;
+ $ld, $nm, $rm, $mv, $is_module, $inputfile) = @ARGV;
# Acceptable sections to record.
my %text_sections = (
--
1.5.6.5
--
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH 2/9] ftrace, ia64: explictly ignore a file in recordmcount.pl
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
2009-01-13 6:11 ` [PATCH 1/9] ftrace, ia64: make recordmcount destinct module compile Steven Rostedt
@ 2009-01-13 6:11 ` Steven Rostedt
2009-01-13 6:11 ` [PATCH 3/9] ftrace, ia64: Add macro for ftrace_caller Steven Rostedt
` (7 subsequent siblings)
9 siblings, 0 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig, Steven Rostedt
[-- Attachment #1: 0002-ftrace-ia64-explictly-ignore-a-file-in-recordmcoun.patch --]
[-- Type: text/plain, Size: 1917 bytes --]
From: Shaohua Li <shaohua.li@intel.com>
In IA64, a function pointer isn't a 'unsigned long' but a
'struct {unsigned long ip, unsigned long gp}'. MCOUNT_ADDR is determined
at link time not compile time, so explictly ignore kernel/trace/ftrace.o
in recordmcount.pl.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
kernel/trace/ftrace.c | 10 +---------
scripts/recordmcount.pl | 5 +++++
2 files changed, 6 insertions(+), 9 deletions(-)
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 9e54a6c..76bb884 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -263,14 +263,6 @@ static void ftrace_update_pid_func(void)
# error Dynamic ftrace depends on MCOUNT_RECORD
#endif
-/*
- * Since MCOUNT_ADDR may point to mcount itself, we do not want
- * to get it confused by reading a reference in the code as we
- * are parsing on objcopy output of text. Use a variable for
- * it instead.
- */
-static unsigned long mcount_addr = MCOUNT_ADDR;
-
enum {
FTRACE_ENABLE_CALLS = (1 << 0),
FTRACE_DISABLE_CALLS = (1 << 1),
@@ -575,7 +567,7 @@ ftrace_code_disable(struct module *mod, struct dyn_ftrace *rec)
ip = rec->ip;
- ret = ftrace_make_nop(mod, rec, mcount_addr);
+ ret = ftrace_make_nop(mod, rec, MCOUNT_ADDR);
if (ret) {
ftrace_bug(ret, ip);
rec->flags |= FTRACE_FL_FAILED;
diff --git a/scripts/recordmcount.pl b/scripts/recordmcount.pl
index 282485a..070042b 100755
--- a/scripts/recordmcount.pl
+++ b/scripts/recordmcount.pl
@@ -109,6 +109,11 @@ if ($#ARGV < 7) {
my ($arch, $bits, $objdump, $objcopy, $cc,
$ld, $nm, $rm, $mv, $is_module, $inputfile) = @ARGV;
+# This file refers to mcount and shouldn't be ftraced, so lets' ignore it
+if ($inputfile eq "kernel/trace/ftrace.o") {
+ exit(0);
+}
+
# Acceptable sections to record.
my %text_sections = (
".text" => 1,
--
1.5.6.5
--
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH 3/9] ftrace, ia64: Add macro for ftrace_caller
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
2009-01-13 6:11 ` [PATCH 1/9] ftrace, ia64: make recordmcount destinct module compile Steven Rostedt
2009-01-13 6:11 ` [PATCH 2/9] ftrace, ia64: explictly ignore a file in recordmcount.pl Steven Rostedt
@ 2009-01-13 6:11 ` Steven Rostedt
2009-01-13 6:11 ` [PATCH 4/9] ftrace, ia64: Add recordmcount for ia64 Steven Rostedt
` (6 subsequent siblings)
9 siblings, 0 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig, Steven Rostedt
[-- Attachment #1: 0003-ftrace-ia64-Add-macro-for-ftrace_caller.patch --]
[-- Type: text/plain, Size: 1337 bytes --]
From: Shaohua Li <shaohua.li@intel.com>
Define FTRACE_ADDR. In IA64, a function pointer isn't a 'unsigned long' but a
'struct {unsigned long ip, unsigned long gp}'.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
include/linux/ftrace.h | 4 ++++
kernel/trace/ftrace.c | 2 +-
2 files changed, 5 insertions(+), 1 deletions(-)
diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
index 677432b..0547214 100644
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h
@@ -126,6 +126,10 @@ extern int ftrace_update_ftrace_func(ftrace_func_t func);
extern void ftrace_caller(void);
extern void ftrace_call(void);
extern void mcount_call(void);
+
+#ifndef FTRACE_ADDR
+#define FTRACE_ADDR ((unsigned long)ftrace_caller)
+#endif
#ifdef CONFIG_FUNCTION_GRAPH_TRACER
extern void ftrace_graph_caller(void);
extern int ftrace_enable_ftrace_graph_caller(void);
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 76bb884..9f53610 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -455,7 +455,7 @@ __ftrace_replace_code(struct dyn_ftrace *rec, int enable)
unsigned long ip, fl;
unsigned long ftrace_addr;
- ftrace_addr = (unsigned long)ftrace_caller;
+ ftrace_addr = (unsigned long)FTRACE_ADDR;
ip = rec->ip;
--
1.5.6.5
--
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH 4/9] ftrace, ia64: Add recordmcount for ia64
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
` (2 preceding siblings ...)
2009-01-13 6:11 ` [PATCH 3/9] ftrace, ia64: Add macro for ftrace_caller Steven Rostedt
@ 2009-01-13 6:11 ` Steven Rostedt
2009-01-13 6:11 ` [PATCH 5/9] ftrace, ia64: IA64 static ftrace support Steven Rostedt
` (5 subsequent siblings)
9 siblings, 0 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig, Steven Rostedt
[-- Attachment #1: 0004-ftrace-ia64-Add-recordmcount-for-ia64.patch --]
[-- Type: text/plain, Size: 807 bytes --]
From: Shaohua Li <shaohua.li@intel.com>
Add recordmcount for ia64.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
scripts/recordmcount.pl | 7 +++++++
1 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/scripts/recordmcount.pl b/scripts/recordmcount.pl
index 070042b..2ded5c8 100755
--- a/scripts/recordmcount.pl
+++ b/scripts/recordmcount.pl
@@ -206,6 +206,13 @@ if ($arch eq "x86_64") {
$alignment = 2;
$section_type = '%progbits';
+} elsif ($arch eq "ia64") {
+ $mcount_regex = "^\\s*([0-9a-fA-F]+):.*\\s_mcount\$";
+ $type = "data8";
+
+ if ($is_module eq "0") {
+ $cc .= " -mconstant-gp";
+ }
} else {
die "Arch $arch is not supported with CONFIG_FTRACE_MCOUNT_RECORD";
}
--
1.5.6.5
--
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH 5/9] ftrace, ia64: IA64 static ftrace support
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
` (3 preceding siblings ...)
2009-01-13 6:11 ` [PATCH 4/9] ftrace, ia64: Add recordmcount for ia64 Steven Rostedt
@ 2009-01-13 6:11 ` Steven Rostedt
2009-01-13 6:11 ` [PATCH 6/9] ftrace, ia64: IA64 dynamic " Steven Rostedt
` (4 subsequent siblings)
9 siblings, 0 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig, Steven Rostedt
[-- Attachment #1: 0005-ftrace-ia64-IA64-static-ftrace-support.patch --]
[-- Type: text/plain, Size: 3248 bytes --]
From: Shaohua Li <shaohua.li@intel.com>
IA64 ftrace suppport. In IA64, below code will be added in each function
if -pg is enabled.
alloc r40=ar.pfs,12,8,0
mov r43=r0;;
mov r42=b0
mov r41=r1
nop.i 0x0
br.call.sptk.many b0 = _mcount;;
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
arch/ia64/Kconfig | 1 +
arch/ia64/include/asm/ftrace.h | 15 ++++++++++++
arch/ia64/kernel/entry.S | 49 ++++++++++++++++++++++++++++++++++++++++
arch/ia64/kernel/ia64_ksyms.c | 6 +++++
4 files changed, 71 insertions(+), 0 deletions(-)
create mode 100644 arch/ia64/include/asm/ftrace.h
diff --git a/arch/ia64/Kconfig b/arch/ia64/Kconfig
index 3d31636..b992ba4 100644
--- a/arch/ia64/Kconfig
+++ b/arch/ia64/Kconfig
@@ -21,6 +21,7 @@ config IA64
select HAVE_OPROFILE
select HAVE_KPROBES
select HAVE_KRETPROBES
+ select HAVE_FUNCTION_TRACER
select HAVE_DMA_ATTRS
select HAVE_KVM
select HAVE_ARCH_TRACEHOOK
diff --git a/arch/ia64/include/asm/ftrace.h b/arch/ia64/include/asm/ftrace.h
new file mode 100644
index 0000000..48694b3
--- /dev/null
+++ b/arch/ia64/include/asm/ftrace.h
@@ -0,0 +1,15 @@
+#ifndef _ASM_IA64_FTRACE_H
+#define _ASM_IA64_FTRACE_H
+
+#ifdef CONFIG_FUNCTION_TRACER
+#define MCOUNT_INSN_SIZE 32 /* sizeof mcount call */
+
+#ifndef __ASSEMBLY__
+extern void _mcount(unsigned long pfs, unsigned long r1, unsigned long b0, unsigned long r0);
+#define mcount _mcount
+
+#endif
+
+#endif /* CONFIG_FUNCTION_TRACER */
+
+#endif /* _ASM_IA64_FTRACE_H */
diff --git a/arch/ia64/kernel/entry.S b/arch/ia64/kernel/entry.S
index d435f4a..c2f7d79 100644
--- a/arch/ia64/kernel/entry.S
+++ b/arch/ia64/kernel/entry.S
@@ -47,6 +47,7 @@
#include <asm/processor.h>
#include <asm/thread_info.h>
#include <asm/unistd.h>
+#include <asm/ftrace.h>
#include "minstate.h"
@@ -1404,6 +1405,54 @@ GLOBAL_ENTRY(unw_init_running)
br.ret.sptk.many rp
END(unw_init_running)
+#ifdef CONFIG_FUNCTION_TRACER
+GLOBAL_ENTRY(_mcount)
+ movl r2 = ftrace_stub
+ movl r3 = ftrace_trace_function;;
+ ld8 r3 = [r3];;
+ ld8 r3 = [r3];;
+ cmp.eq p7,p0 = r2, r3
+(p7) br.sptk.many ftrace_stub
+ ;;
+
+ alloc loc0 = ar.pfs, 4, 4, 2, 0
+ ;;
+ mov loc1 = b0
+ mov out0 = b0
+ mov loc2 = r8
+ mov loc3 = r15
+ ;;
+ adds out0 = -MCOUNT_INSN_SIZE, out0
+ mov out1 = in2
+ mov b6 = r3
+
+ br.call.sptk.many b0 = b6
+ ;;
+ mov ar.pfs = loc0
+ mov b0 = loc1
+ mov r8 = loc2
+ mov r15 = loc3
+ br ftrace_stub
+ ;;
+END(_mcount)
+
+GLOBAL_ENTRY(ftrace_stub)
+ mov r3 = b0
+ movl r2 = _mcount_ret_helper
+ ;;
+ mov b6 = r2
+ mov b7 = r3
+ br.ret.sptk.many b6
+
+_mcount_ret_helper:
+ mov b0 = r42
+ mov r1 = r41
+ mov ar.pfs = r40
+ br b7
+END(ftrace_stub)
+
+#endif /* CONFIG_FUNCTION_TRACER */
+
.rodata
.align 8
.globl sys_call_table
diff --git a/arch/ia64/kernel/ia64_ksyms.c b/arch/ia64/kernel/ia64_ksyms.c
index 6da1f20..2d31186 100644
--- a/arch/ia64/kernel/ia64_ksyms.c
+++ b/arch/ia64/kernel/ia64_ksyms.c
@@ -112,3 +112,9 @@ EXPORT_SYMBOL_GPL(esi_call_phys);
#endif
extern char ia64_ivt[];
EXPORT_SYMBOL(ia64_ivt);
+
+#include <asm/ftrace.h>
+#ifdef CONFIG_FUNCTION_TRACER
+/* mcount is defined in assembly */
+EXPORT_SYMBOL(_mcount);
+#endif
--
1.5.6.5
--
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH 6/9] ftrace, ia64: IA64 dynamic ftrace support
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
` (4 preceding siblings ...)
2009-01-13 6:11 ` [PATCH 5/9] ftrace, ia64: IA64 static ftrace support Steven Rostedt
@ 2009-01-13 6:11 ` Steven Rostedt
2009-01-13 6:11 ` [PATCH 7/9] tracing/ftrace: separate events tracing and stats tracing engine Steven Rostedt
` (3 subsequent siblings)
9 siblings, 0 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig, Steven Rostedt
[-- Attachment #1: 0006-ftrace-ia64-IA64-dynamic-ftrace-support.patch --]
[-- Type: text/plain, Size: 10536 bytes --]
From: Shaohua Li <shaohua.li@intel.com>
IA64 dynamic ftrace support.
The original _mcount stub for each function is like:
alloc r40=ar.pfs,12,8,0
mov r43=r0;;
mov r42=b0
mov r41=r1
nop.i 0x0
br.call.sptk.many b0 = _mcount;;
The patch convert it to below for nop:
[MII] nop.m 0x0
mov r3=ip
nop.i 0x0
[MLX] nop.m 0x0
nop.x 0x0;;
This isn't completely nop, as there is one instuction 'mov r3=ip', but
it should be light and harmless for code follow it.
And below is for call
[MII] nop.m 0x0
mov r3=ip
nop.i 0x0
[MLX] nop.m 0x0
brl.many .;;
In this way, only one instruction is changed to convert code between nop
and call. This should meet dyn-ftrace's requirement.
But this requires CPU support brl instruction, so dyn-ftrace isn't
supported for old Itanium system. Assume there are quite few such old
system running.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
arch/ia64/Kconfig | 2 +
arch/ia64/include/asm/ftrace.h | 13 +++
arch/ia64/kernel/Makefile | 5 +
arch/ia64/kernel/entry.S | 51 ++++++++++
arch/ia64/kernel/ftrace.c | 206 ++++++++++++++++++++++++++++++++++++++++
5 files changed, 277 insertions(+), 0 deletions(-)
create mode 100644 arch/ia64/kernel/ftrace.c
diff --git a/arch/ia64/Kconfig b/arch/ia64/Kconfig
index b992ba4..e20c1d4 100644
--- a/arch/ia64/Kconfig
+++ b/arch/ia64/Kconfig
@@ -21,6 +21,8 @@ config IA64
select HAVE_OPROFILE
select HAVE_KPROBES
select HAVE_KRETPROBES
+ select HAVE_FTRACE_MCOUNT_RECORD
+ select HAVE_DYNAMIC_FTRACE if (!ITANIUM)
select HAVE_FUNCTION_TRACER
select HAVE_DMA_ATTRS
select HAVE_KVM
diff --git a/arch/ia64/include/asm/ftrace.h b/arch/ia64/include/asm/ftrace.h
index 48694b3..d20db3c 100644
--- a/arch/ia64/include/asm/ftrace.h
+++ b/arch/ia64/include/asm/ftrace.h
@@ -8,6 +8,19 @@
extern void _mcount(unsigned long pfs, unsigned long r1, unsigned long b0, unsigned long r0);
#define mcount _mcount
+#include <asm/kprobes.h>
+/* In IA64, MCOUNT_ADDR is set in link time, so it's not a constant at compile time */
+#define MCOUNT_ADDR (((struct fnptr *)mcount)->ip)
+#define FTRACE_ADDR (((struct fnptr *)ftrace_caller)->ip)
+
+static inline unsigned long ftrace_call_adjust(unsigned long addr)
+{
+ /* second bundle, insn 2 */
+ return addr - 0x12;
+}
+
+struct dyn_arch_ftrace {
+};
#endif
#endif /* CONFIG_FUNCTION_TRACER */
diff --git a/arch/ia64/kernel/Makefile b/arch/ia64/kernel/Makefile
index f2778f2..dc62df0 100644
--- a/arch/ia64/kernel/Makefile
+++ b/arch/ia64/kernel/Makefile
@@ -2,6 +2,10 @@
# Makefile for the linux kernel.
#
+ifdef CONFIG_DYNAMIC_FTRACE
+CFLAGS_REMOVE_ftrace.o = -pg
+endif
+
extra-y := head.o init_task.o vmlinux.lds
obj-y := acpi.o entry.o efi.o efi_stub.o gate-data.o fsys.o ia64_ksyms.o irq.o irq_ia64.o \
@@ -28,6 +32,7 @@ obj-$(CONFIG_IA64_CYCLONE) += cyclone.o
obj-$(CONFIG_CPU_FREQ) += cpufreq/
obj-$(CONFIG_IA64_MCA_RECOVERY) += mca_recovery.o
obj-$(CONFIG_KPROBES) += kprobes.o jprobes.o
+obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o
obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o crash.o
obj-$(CONFIG_CRASH_DUMP) += crash_dump.o
obj-$(CONFIG_IA64_UNCACHED_ALLOCATOR) += uncached.o
diff --git a/arch/ia64/kernel/entry.S b/arch/ia64/kernel/entry.S
index c2f7d79..e0be92a 100644
--- a/arch/ia64/kernel/entry.S
+++ b/arch/ia64/kernel/entry.S
@@ -1406,6 +1406,56 @@ GLOBAL_ENTRY(unw_init_running)
END(unw_init_running)
#ifdef CONFIG_FUNCTION_TRACER
+#ifdef CONFIG_DYNAMIC_FTRACE
+GLOBAL_ENTRY(_mcount)
+ br ftrace_stub
+END(_mcount)
+
+.here:
+ br.ret.sptk.many b0
+
+GLOBAL_ENTRY(ftrace_caller)
+ alloc out0 = ar.pfs, 8, 0, 4, 0
+ mov out3 = r0
+ ;;
+ mov out2 = b0
+ add r3 = 0x20, r3
+ mov out1 = r1;
+ br.call.sptk.many b0 = ftrace_patch_gp
+ //this might be called from module, so we must patch gp
+ftrace_patch_gp:
+ movl gp=__gp
+ mov b0 = r3
+ ;;
+.global ftrace_call;
+ftrace_call:
+{
+ .mlx
+ nop.m 0x0
+ movl r3 = .here;;
+}
+ alloc loc0 = ar.pfs, 4, 4, 2, 0
+ ;;
+ mov loc1 = b0
+ mov out0 = b0
+ mov loc2 = r8
+ mov loc3 = r15
+ ;;
+ adds out0 = -MCOUNT_INSN_SIZE, out0
+ mov out1 = in2
+ mov b6 = r3
+
+ br.call.sptk.many b0 = b6
+ ;;
+ mov ar.pfs = loc0
+ mov b0 = loc1
+ mov r8 = loc2
+ mov r15 = loc3
+ br ftrace_stub
+ ;;
+END(ftrace_caller)
+
+#else
GLOBAL_ENTRY(_mcount)
movl r2 = ftrace_stub
movl r3 = ftrace_trace_function;;
@@ -1435,6 +1485,7 @@ GLOBAL_ENTRY(_mcount)
br ftrace_stub
;;
END(_mcount)
+#endif
GLOBAL_ENTRY(ftrace_stub)
mov r3 = b0
diff --git a/arch/ia64/kernel/ftrace.c b/arch/ia64/kernel/ftrace.c
new file mode 100644
index 0000000..7fc8c96
--- /dev/null
+++ b/arch/ia64/kernel/ftrace.c
@@ -0,0 +1,206 @@
+/*
+ * Dynamic function tracing support.
+ *
+ * Copyright (C) 2008 Shaohua Li <shaohua.li@intel.com>
+ *
+ * For licencing details, see COPYING.
+ *
+ * Defines low-level handling of mcount calls when the kernel
+ * is compiled with the -pg flag. When using dynamic ftrace, the
+ * mcount call-sites get patched lazily with NOP till they are
+ * enabled. All code mutation routines here take effect atomically.
+ */
+
+#include <linux/uaccess.h>
+#include <linux/ftrace.h>
+
+#include <asm/cacheflush.h>
+#include <asm/patch.h>
+
+/* In IA64, each function will be added below two bundles with -pg option */
+static unsigned char __attribute__((aligned(8)))
+ftrace_orig_code[MCOUNT_INSN_SIZE] = {
+ 0x02, 0x40, 0x31, 0x10, 0x80, 0x05, /* alloc r40=ar.pfs,12,8,0 */
+ 0xb0, 0x02, 0x00, 0x00, 0x42, 0x40, /* mov r43=r0;; */
+ 0x05, 0x00, 0xc4, 0x00, /* mov r42=b0 */
+ 0x11, 0x48, 0x01, 0x02, 0x00, 0x21, /* mov r41=r1 */
+ 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, /* nop.i 0x0 */
+ 0x08, 0x00, 0x00, 0x50 /* br.call.sptk.many b0 = _mcount;; */
+};
+
+struct ftrace_orig_insn {
+ u64 dummy1, dummy2, dummy3;
+ u64 dummy4:64-41+13;
+ u64 imm20:20;
+ u64 dummy5:3;
+ u64 sign:1;
+ u64 dummy6:4;
+};
+
+/* mcount stub will be converted below for nop */
+static unsigned char ftrace_nop_code[MCOUNT_INSN_SIZE] = {
+ 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, /* [MII] nop.m 0x0 */
+ 0x30, 0x00, 0x00, 0x60, 0x00, 0x00, /* mov r3=ip */
+ 0x00, 0x00, 0x04, 0x00, /* nop.i 0x0 */
+ 0x05, 0x00, 0x00, 0x00, 0x01, 0x00, /* [MLX] nop.m 0x0 */
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* nop.x 0x0;; */
+ 0x00, 0x00, 0x04, 0x00
+};
+
+static unsigned char *ftrace_nop_replace(void)
+{
+ return ftrace_nop_code;
+}
+
+/*
+ * mcount stub will be converted below for call
+ * Note: Just the last instruction is changed against nop
+ * */
+static unsigned char __attribute__((aligned(8)))
+ftrace_call_code[MCOUNT_INSN_SIZE] = {
+ 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, /* [MII] nop.m 0x0 */
+ 0x30, 0x00, 0x00, 0x60, 0x00, 0x00, /* mov r3=ip */
+ 0x00, 0x00, 0x04, 0x00, /* nop.i 0x0 */
+ 0x05, 0x00, 0x00, 0x00, 0x01, 0x00, /* [MLX] nop.m 0x0 */
+ 0xff, 0xff, 0xff, 0xff, 0x7f, 0x00, /* brl.many .;;*/
+ 0xf8, 0xff, 0xff, 0xc8
+};
+
+struct ftrace_call_insn {
+ u64 dummy1, dummy2;
+ u64 dummy3:48;
+ u64 imm39_l:16;
+ u64 imm39_h:23;
+ u64 dummy4:13;
+ u64 imm20:20;
+ u64 dummy5:3;
+ u64 i:1;
+ u64 dummy6:4;
+};
+
+static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
+{
+ struct ftrace_call_insn *code = (void *)ftrace_call_code;
+ unsigned long offset = addr - (ip + 0x10);
+
+ code->imm39_l = offset >> 24;
+ code->imm39_h = offset >> 40;
+ code->imm20 = offset >> 4;
+ code->i = offset >> 63;
+ return ftrace_call_code;
+}
+
+static int
+ftrace_modify_code(unsigned long ip, unsigned char *old_code,
+ unsigned char *new_code, int do_check)
+{
+ 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 (!do_check)
+ goto skip_check;
+
+ /* 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;
+
+skip_check:
+ /* 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;
+}
+
+static int ftrace_make_nop_check(struct dyn_ftrace *rec, unsigned long addr)
+{
+ unsigned char __attribute__((aligned(8))) replaced[MCOUNT_INSN_SIZE];
+ unsigned long ip = rec->ip;
+
+ if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE))
+ return -EFAULT;
+ if (rec->flags & FTRACE_FL_CONVERTED) {
+ struct ftrace_call_insn *call_insn, *tmp_call;
+
+ call_insn = (void *)ftrace_call_code;
+ tmp_call = (void *)replaced;
+ call_insn->imm39_l = tmp_call->imm39_l;
+ call_insn->imm39_h = tmp_call->imm39_h;
+ call_insn->imm20 = tmp_call->imm20;
+ call_insn->i = tmp_call->i;
+ if (memcmp(replaced, ftrace_call_code, MCOUNT_INSN_SIZE) != 0)
+ return -EINVAL;
+ return 0;
+ } else {
+ struct ftrace_orig_insn *call_insn, *tmp_call;
+
+ call_insn = (void *)ftrace_orig_code;
+ tmp_call = (void *)replaced;
+ call_insn->sign = tmp_call->sign;
+ call_insn->imm20 = tmp_call->imm20;
+ if (memcmp(replaced, ftrace_orig_code, MCOUNT_INSN_SIZE) != 0)
+ return -EINVAL;
+ return 0;
+ }
+}
+
+int ftrace_make_nop(struct module *mod,
+ struct dyn_ftrace *rec, unsigned long addr)
+{
+ int ret;
+ char *new;
+
+ ret = ftrace_make_nop_check(rec, addr);
+ if (ret)
+ return ret;
+ new = ftrace_nop_replace();
+ return ftrace_modify_code(rec->ip, NULL, new, 0);
+}
+
+int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
+{
+ unsigned long ip = rec->ip;
+ unsigned char *old, *new;
+
+ old= ftrace_nop_replace();
+ new = ftrace_call_replace(ip, addr);
+ return ftrace_modify_code(ip, old, new, 1);
+}
+
+/* in IA64, _mcount can't directly call ftrace_stub. Only jump is ok */
+int ftrace_update_ftrace_func(ftrace_func_t func)
+{
+ unsigned long ip;
+ unsigned long addr = ((struct fnptr *)ftrace_call)->ip;
+
+ if (func == ftrace_stub)
+ return 0;
+ ip = ((struct fnptr *)func)->ip;
+
+ ia64_patch_imm64(addr + 2, ip);
+
+ flush_icache_range(addr, addr + 16);
+ return 0;
+}
+
+/* run from kstop_machine */
+int __init ftrace_dyn_arch_init(void *data)
+{
+ *(unsigned long *)data = 0;
+
+ return 0;
+}
--
1.5.6.5
--
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH 7/9] tracing/ftrace: separate events tracing and stats tracing engine
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
` (5 preceding siblings ...)
2009-01-13 6:11 ` [PATCH 6/9] ftrace, ia64: IA64 dynamic " Steven Rostedt
@ 2009-01-13 6:11 ` Steven Rostedt
2009-01-13 6:11 ` [PATCH 8/9] tracing: add a new workqueue tracer Steven Rostedt
` (2 subsequent siblings)
9 siblings, 0 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig, Steven Rostedt
[-- Attachment #1: 0007-tracing-ftrace-separate-events-tracing-and-stats-tr.patch --]
[-- Type: text/plain, Size: 14174 bytes --]
From: Frederic Weisbecker <fweisbec@gmail.com>
Impact: tracing's Api change
Currently, the stat tracing depends on the events tracing.
When you switch to a new tracer, the stats files of the previous tracer
will disappear. But it's more scalable to separate those two engines.
This way, we can keep the stat files of one or several tracers when we
want, without bothering of multiple tracer stat files or tracer switching.
To build/destroys its stats files, a tracer just have to call
register_stat_tracer/unregister_stat_tracer everytimes it wants to.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
kernel/trace/trace.c | 2 -
kernel/trace/trace.h | 20 -----
kernel/trace/trace_branch.c | 108 ++++++++++++++-----------
kernel/trace/trace_stat.c | 191 ++++++++++++++++++-------------------------
kernel/trace/trace_stat.h | 31 +++++++
5 files changed, 172 insertions(+), 180 deletions(-)
create mode 100644 kernel/trace/trace_stat.h
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 0418fc3..40217fb 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -2353,7 +2353,6 @@ static int tracing_set_tracer(char *buf)
if (ret)
goto out;
}
- init_tracer_stat(t);
trace_branch_enable(tr);
out:
@@ -3218,7 +3217,6 @@ __init static int tracer_alloc_buffers(void)
#else
current_trace = &nop_trace;
#endif
- init_tracer_stat(current_trace);
/* All seems OK, enable tracing */
tracing_disabled = 0;
diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index b3f9ad1..79c8721 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -334,24 +334,6 @@ struct tracer_flags {
/* Makes more easy to define a tracer opt */
#define TRACER_OPT(s, b) .name = #s, .bit = b
-/*
- * If you want to provide a stat file (one-shot statistics), fill
- * an iterator with stat_start/stat_next and a stat_show callbacks.
- * The others callbacks are optional.
- */
-struct tracer_stat {
- /* The name of your stat file */
- const char *name;
- /* Iteration over statistic entries */
- void *(*stat_start)(void);
- void *(*stat_next)(void *prev, int idx);
- /* Compare two entries for sorting (optional) for stats */
- int (*stat_cmp)(void *p1, void *p2);
- /* Print a stat entry */
- int (*stat_show)(struct seq_file *s, void *p);
- /* Print the headers of your stat entries */
- int (*stat_headers)(struct seq_file *s);
-};
/*
* A specific tracer, represented by methods that operate on a trace array:
@@ -466,8 +448,6 @@ void tracing_start_sched_switch_record(void);
int register_tracer(struct tracer *type);
void unregister_tracer(struct tracer *type);
-void init_tracer_stat(struct tracer *trace);
-
extern unsigned long nsecs_to_usecs(unsigned long nsecs);
extern unsigned long tracing_max_latency;
diff --git a/kernel/trace/trace_branch.c b/kernel/trace/trace_branch.c
index da5cf3e..ca017e0 100644
--- a/kernel/trace/trace_branch.c
+++ b/kernel/trace/trace_branch.c
@@ -16,12 +16,12 @@
#include <asm/local.h>
#include "trace.h"
+#include "trace_stat.h"
#include "trace_output.h"
-static struct tracer branch_trace;
-
#ifdef CONFIG_BRANCH_TRACER
+static struct tracer branch_trace;
static int branch_tracing_enabled __read_mostly;
static DEFINE_MUTEX(branch_tracing_mutex);
@@ -191,6 +191,30 @@ static struct trace_event trace_branch_event = {
.binary = trace_nop_print,
};
+static struct tracer branch_trace __read_mostly =
+{
+ .name = "branch",
+ .init = branch_trace_init,
+ .reset = branch_trace_reset,
+#ifdef CONFIG_FTRACE_SELFTEST
+ .selftest = trace_selftest_startup_branch,
+#endif /* CONFIG_FTRACE_SELFTEST */
+};
+
+__init static int init_branch_tracer(void)
+{
+ int ret;
+
+ ret = register_ftrace_event(&trace_branch_event);
+ if (!ret) {
+ printk(KERN_WARNING "Warning: could not register "
+ "branch events\n");
+ return 1;
+ }
+ return register_tracer(&branch_trace);
+}
+device_initcall(init_branch_tracer);
+
#else
static inline
void trace_likely_condition(struct ftrace_branch_data *f, int val, int expect)
@@ -305,6 +329,29 @@ static int annotated_branch_stat_cmp(void *p1, void *p2)
return 0;
}
+static struct tracer_stat annotated_branch_stats = {
+ .name = "branch_annotated",
+ .stat_start = annotated_branch_stat_start,
+ .stat_next = annotated_branch_stat_next,
+ .stat_cmp = annotated_branch_stat_cmp,
+ .stat_headers = annotated_branch_stat_headers,
+ .stat_show = branch_stat_show
+};
+
+__init static int init_annotated_branch_stats(void)
+{
+ int ret;
+
+ ret = register_stat_tracer(&annotated_branch_stats);
+ if (!ret) {
+ printk(KERN_WARNING "Warning: could not register "
+ "annotated branches stats\n");
+ return 1;
+ }
+ return 0;
+}
+fs_initcall(init_annotated_branch_stats);
+
#ifdef CONFIG_PROFILE_ALL_BRANCHES
extern unsigned long __start_branch_profile[];
@@ -339,60 +386,25 @@ all_branch_stat_next(void *v, int idx)
return p;
}
-static struct tracer_stat branch_stats[] = {
- {.name = "annotated",
- .stat_start = annotated_branch_stat_start,
- .stat_next = annotated_branch_stat_next,
- .stat_cmp = annotated_branch_stat_cmp,
- .stat_headers = annotated_branch_stat_headers,
- .stat_show = branch_stat_show},
-
- {.name = "all",
+static struct tracer_stat all_branch_stats = {
+ .name = "branch_all",
.stat_start = all_branch_stat_start,
.stat_next = all_branch_stat_next,
.stat_headers = all_branch_stat_headers,
- .stat_show = branch_stat_show},
-
- { }
-};
-#else
-static struct tracer_stat branch_stats[] = {
- {.name = "annotated",
- .stat_start = annotated_branch_stat_start,
- .stat_next = annotated_branch_stat_next,
- .stat_cmp = annotated_branch_stat_cmp,
- .stat_headers = annotated_branch_stat_headers,
- .stat_show = branch_stat_show},
-
- { }
+ .stat_show = branch_stat_show
};
-#endif /* CONFIG_PROFILE_ALL_BRANCHES */
-
-static struct tracer branch_trace __read_mostly =
+__init static int all_annotated_branch_stats(void)
{
- .name = "branch",
-#ifdef CONFIG_BRANCH_TRACER
- .init = branch_trace_init,
- .reset = branch_trace_reset,
-#ifdef CONFIG_FTRACE_SELFTEST
- .selftest = trace_selftest_startup_branch,
-#endif /* CONFIG_FTRACE_SELFTEST */
-#endif
- .stats = branch_stats
-};
-
-__init static int init_branch_trace(void)
-{
-#ifdef CONFIG_BRANCH_TRACER
int ret;
- ret = register_ftrace_event(&trace_branch_event);
+
+ ret = register_stat_tracer(&all_branch_stats);
if (!ret) {
- printk(KERN_WARNING "Warning: could not register branch events\n");
+ printk(KERN_WARNING "Warning: could not register "
+ "all branches stats\n");
return 1;
}
-#endif
-
- return register_tracer(&branch_trace);
+ return 0;
}
-device_initcall(init_branch_trace);
+fs_initcall(all_annotated_branch_stats);
+#endif /* CONFIG_PROFILE_ALL_BRANCHES */
diff --git a/kernel/trace/trace_stat.c b/kernel/trace/trace_stat.c
index 1515f9e..cb29282 100644
--- a/kernel/trace/trace_stat.c
+++ b/kernel/trace/trace_stat.c
@@ -10,28 +10,32 @@
#include <linux/list.h>
-#include <linux/seq_file.h>
#include <linux/debugfs.h>
+#include "trace_stat.h"
#include "trace.h"
/* List of stat entries from a tracer */
struct trace_stat_list {
- struct list_head list;
- void *stat;
+ struct list_head list;
+ void *stat;
};
/* A stat session is the stats output in one file */
struct tracer_stat_session {
- struct tracer_stat *ts;
- struct list_head stat_list;
- struct mutex stat_mutex;
+ struct list_head session_list;
+ struct tracer_stat *ts;
+ struct list_head stat_list;
+ struct mutex stat_mutex;
+ struct dentry *file;
};
/* All of the sessions currently in use. Each stat file embeed one session */
-static struct tracer_stat_session **all_stat_sessions;
-static int nb_sessions;
-static struct dentry *stat_dir, **stat_files;
+static LIST_HEAD(all_stat_sessions);
+static DEFINE_MUTEX(all_stat_sessions_mutex);
+
+/* The root directory for all stat files */
+static struct dentry *stat_dir;
static void reset_stat_session(struct tracer_stat_session *session)
@@ -44,66 +48,77 @@ static void reset_stat_session(struct tracer_stat_session *session)
INIT_LIST_HEAD(&session->stat_list);
}
-/* Called when a tracer is initialized */
-static int init_all_sessions(int nb, struct tracer_stat *ts)
+static void destroy_session(struct tracer_stat_session *session)
{
- int i, j;
- struct tracer_stat_session *session;
+ debugfs_remove(session->file);
+ reset_stat_session(session);
+ mutex_destroy(&session->stat_mutex);
+ kfree(session);
+}
- nb_sessions = 0;
- if (all_stat_sessions) {
- for (i = 0; i < nb_sessions; i++) {
- session = all_stat_sessions[i];
- reset_stat_session(session);
- mutex_destroy(&session->stat_mutex);
- kfree(session);
- }
- }
- all_stat_sessions = kmalloc(sizeof(struct tracer_stat_session *) * nb,
- GFP_KERNEL);
- if (!all_stat_sessions)
- return -ENOMEM;
+static int init_stat_file(struct tracer_stat_session *session);
- for (i = 0; i < nb; i++) {
- session = kmalloc(sizeof(struct tracer_stat_session) * nb,
- GFP_KERNEL);
- if (!session)
- goto free_sessions;
+int register_stat_tracer(struct tracer_stat *trace)
+{
+ struct tracer_stat_session *session, *node, *tmp;
+ int ret;
+
+ if (!trace)
+ return -EINVAL;
+
+ if (!trace->stat_start || !trace->stat_next || !trace->stat_show)
+ return -EINVAL;
- INIT_LIST_HEAD(&session->stat_list);
- mutex_init(&session->stat_mutex);
- session->ts = &ts[i];
- all_stat_sessions[i] = session;
+ /* Already registered? */
+ mutex_lock(&all_stat_sessions_mutex);
+ list_for_each_entry_safe(node, tmp, &all_stat_sessions, session_list) {
+ if (node->ts == trace)
+ return -EINVAL;
}
- nb_sessions = nb;
- return 0;
+ mutex_unlock(&all_stat_sessions_mutex);
+
+ /* Init the session */
+ session = kmalloc(sizeof(struct tracer_stat_session), GFP_KERNEL);
+ if (!session)
+ return -ENOMEM;
-free_sessions:
+ session->ts = trace;
+ INIT_LIST_HEAD(&session->session_list);
+ INIT_LIST_HEAD(&session->stat_list);
+ mutex_init(&session->stat_mutex);
+ session->file = NULL;
- for (j = 0; j < i; j++)
- kfree(all_stat_sessions[i]);
+ ret = init_stat_file(session);
+ if (ret) {
+ destroy_session(session);
+ return ret;
+ }
- kfree(all_stat_sessions);
- all_stat_sessions = NULL;
+ /* Register */
+ mutex_lock(&all_stat_sessions_mutex);
+ list_add_tail(&session->session_list, &all_stat_sessions);
+ mutex_unlock(&all_stat_sessions_mutex);
- return -ENOMEM;
+ return 0;
}
-static int basic_tracer_stat_checks(struct tracer_stat *ts)
+void unregister_stat_tracer(struct tracer_stat *trace)
{
- int i;
+ struct tracer_stat_session *node, *tmp;
- if (!ts)
- return 0;
-
- for (i = 0; ts[i].name; i++) {
- if (!ts[i].stat_start || !ts[i].stat_next || !ts[i].stat_show)
- return -EBUSY;
+ mutex_lock(&all_stat_sessions_mutex);
+ list_for_each_entry_safe(node, tmp, &all_stat_sessions, session_list) {
+ if (node->ts == trace) {
+ list_del(&node->session_list);
+ destroy_session(node);
+ break;
+ }
}
- return i;
+ mutex_unlock(&all_stat_sessions_mutex);
}
+
/*
* For tracers that don't provide a stat_cmp callback.
* This one will force an immediate insertion on tail of
@@ -280,63 +295,7 @@ static const struct file_operations tracing_stat_fops = {
.release = tracing_stat_release
};
-
-static void destroy_trace_stat_files(void)
-{
- int i;
-
- if (stat_files) {
- for (i = 0; i < nb_sessions; i++)
- debugfs_remove(stat_files[i]);
- kfree(stat_files);
- stat_files = NULL;
- }
-}
-
-static void init_trace_stat_files(void)
-{
- int i;
-
- if (!stat_dir || !nb_sessions)
- return;
-
- stat_files = kmalloc(sizeof(struct dentry *) * nb_sessions, GFP_KERNEL);
-
- if (!stat_files) {
- pr_warning("trace stat: not enough memory\n");
- return;
- }
-
- for (i = 0; i < nb_sessions; i++) {
- struct tracer_stat_session *session = all_stat_sessions[i];
- stat_files[i] = debugfs_create_file(session->ts->name, 0644,
- stat_dir,
- session, &tracing_stat_fops);
- if (!stat_files[i])
- pr_warning("cannot create %s entry\n",
- session->ts->name);
- }
-}
-
-void init_tracer_stat(struct tracer *trace)
-{
- int nb = basic_tracer_stat_checks(trace->stats);
-
- destroy_trace_stat_files();
-
- if (nb < 0) {
- pr_warning("stat tracing: missing stat callback on %s\n",
- trace->name);
- return;
- }
- if (!nb)
- return;
-
- init_all_sessions(nb, trace->stats);
- init_trace_stat_files();
-}
-
-static int __init tracing_stat_init(void)
+static int tracing_stat_init(void)
{
struct dentry *d_tracing;
@@ -348,4 +307,16 @@ static int __init tracing_stat_init(void)
"'trace_stat' entry\n");
return 0;
}
-fs_initcall(tracing_stat_init);
+
+static int init_stat_file(struct tracer_stat_session *session)
+{
+ if (!stat_dir && tracing_stat_init())
+ return -ENODEV;
+
+ session->file = debugfs_create_file(session->ts->name, 0644,
+ stat_dir,
+ session, &tracing_stat_fops);
+ if (!session->file)
+ return -ENOMEM;
+ return 0;
+}
diff --git a/kernel/trace/trace_stat.h b/kernel/trace/trace_stat.h
new file mode 100644
index 0000000..202274c
--- /dev/null
+++ b/kernel/trace/trace_stat.h
@@ -0,0 +1,31 @@
+#ifndef __TRACE_STAT_H
+#define __TRACE_STAT_H
+
+#include <linux/seq_file.h>
+
+/*
+ * If you want to provide a stat file (one-shot statistics), fill
+ * an iterator with stat_start/stat_next and a stat_show callbacks.
+ * The others callbacks are optional.
+ */
+struct tracer_stat {
+ /* The name of your stat file */
+ const char *name;
+ /* Iteration over statistic entries */
+ void *(*stat_start)(void);
+ void *(*stat_next)(void *prev, int idx);
+ /* Compare two entries for stats sorting */
+ int (*stat_cmp)(void *p1, void *p2);
+ /* Print a stat entry */
+ int (*stat_show)(struct seq_file *s, void *p);
+ /* Print the headers of your stat entries */
+ int (*stat_headers)(struct seq_file *s);
+};
+
+/*
+ * Destroy or create a stat file
+ */
+extern int register_stat_tracer(struct tracer_stat *trace);
+extern void unregister_stat_tracer(struct tracer_stat *trace);
+
+#endif /* __TRACE_STAT_H */
--
1.5.6.5
--
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH 8/9] tracing: add a new workqueue tracer
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
` (6 preceding siblings ...)
2009-01-13 6:11 ` [PATCH 7/9] tracing/ftrace: separate events tracing and stats tracing engine Steven Rostedt
@ 2009-01-13 6:11 ` Steven Rostedt
2009-01-13 6:11 ` [PATCH 9/9] ftrace, trivial: fix typo "resgister" -> "register" Steven Rostedt
2009-01-13 14:06 ` [PATCH 0/9] ftrace: updates for tip Steven Rostedt
9 siblings, 0 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig, Steven Rostedt
[-- Attachment #1: 0008-tracing-add-a-new-workqueue-tracer.patch --]
[-- Type: text/plain, Size: 14649 bytes --]
From: Frederic Weisbecker <fweisbec@gmail.com>
Impact: new tracer
The workqueue tracer provides some statistical informations
about each cpu workqueue thread such as the number of the
works inserted and executed since their creation. It can help
to evaluate the amount of work each of them have to perform.
For example it can help a developer to decide whether he should
choose a per cpu workqueue instead of a singlethreaded one.
It only traces statistical informations for now but it will probably later
provide event tracing too.
Such a tracer could help too, and be improved, to help rt priority sorted
workqueue development.
To have a snapshot of the workqueues state at any time, just do
cat /debugfs/tracing/trace_stat/workqueues
Ie:
1 125 125 reiserfs/1
1 0 0 scsi_tgtd/1
1 0 0 aio/1
1 0 0 ata/1
1 114 114 kblockd/1
1 0 0 kintegrityd/1
1 2147 2147 events/1
0 0 0 kpsmoused
0 105 105 reiserfs/0
0 0 0 scsi_tgtd/0
0 0 0 aio/0
0 0 0 ata_aux
0 0 0 ata/0
0 0 0 cqueue
0 0 0 kacpi_notify
0 0 0 kacpid
0 149 149 kblockd/0
0 0 0 kintegrityd/0
0 1000 1000 khelper
0 2270 2270 events/0
Changes in V2:
_ Drop the static array based on NR_CPU and dynamically allocate the stat array
with num_possible_cpus() and other cpu mask facilities....
_ Trace workqueue insertion at a bit lower level (insert_work instead of queue_work) to handle
even the workqueue barriers.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
include/trace/workqueue.h | 25 ++++
kernel/trace/Kconfig | 11 ++
kernel/trace/Makefile | 1 +
kernel/trace/trace_workqueue.c | 287 ++++++++++++++++++++++++++++++++++++++++
kernel/workqueue.c | 16 ++-
5 files changed, 339 insertions(+), 1 deletions(-)
create mode 100644 include/trace/workqueue.h
create mode 100644 kernel/trace/trace_workqueue.c
diff --git a/include/trace/workqueue.h b/include/trace/workqueue.h
new file mode 100644
index 0000000..867829d
--- /dev/null
+++ b/include/trace/workqueue.h
@@ -0,0 +1,25 @@
+#ifndef __TRACE_WORKQUEUE_H
+#define __TRACE_WORKQUEUE_H
+
+#include <linux/tracepoint.h>
+#include <linux/workqueue.h>
+#include <linux/sched.h>
+
+DECLARE_TRACE(workqueue_insertion,
+ TPPROTO(struct task_struct *wq_thread, struct work_struct *work),
+ TPARGS(wq_thread, work));
+
+DECLARE_TRACE(workqueue_execution,
+ TPPROTO(struct task_struct *wq_thread, struct work_struct *work),
+ TPARGS(wq_thread, work));
+
+/* Trace the creation of one workqueue thread on a cpu */
+DECLARE_TRACE(workqueue_creation,
+ TPPROTO(struct task_struct *wq_thread, int cpu),
+ TPARGS(wq_thread, cpu));
+
+DECLARE_TRACE(workqueue_destruction,
+ TPPROTO(struct task_struct *wq_thread),
+ TPARGS(wq_thread));
+
+#endif /* __TRACE_WORKQUEUE_H */
diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
index 9442392..dde1d46 100644
--- a/kernel/trace/Kconfig
+++ b/kernel/trace/Kconfig
@@ -284,6 +284,17 @@ config KMEMTRACE
If unsure, say N.
+config WORKQUEUE_TRACER
+ bool "Trace workqueues"
+ select TRACING
+ help
+ The workqueue tracer provides some statistical informations
+ about each cpu workqueue thread such as the number of the
+ works inserted and executed since their creation. It can help
+ to evaluate the amount of work each of them have to perform.
+ For example it can help a developer to decide whether he should
+ choose a per cpu workqueue instead of a singlethreaded one.
+
config DYNAMIC_FTRACE
bool "enable/disable ftrace tracepoints dynamically"
diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile
index 05c9182..f76d48f 100644
--- a/kernel/trace/Makefile
+++ b/kernel/trace/Makefile
@@ -36,5 +36,6 @@ obj-$(CONFIG_TRACE_BRANCH_PROFILING) += trace_branch.o
obj-$(CONFIG_HW_BRANCH_TRACER) += trace_hw_branches.o
obj-$(CONFIG_POWER_TRACER) += trace_power.o
obj-$(CONFIG_KMEMTRACE) += kmemtrace.o
+obj-$(CONFIG_WORKQUEUE_TRACER) += trace_workqueue.o
libftrace-y := ftrace.o
diff --git a/kernel/trace/trace_workqueue.c b/kernel/trace/trace_workqueue.c
new file mode 100644
index 0000000..f8118d3
--- /dev/null
+++ b/kernel/trace/trace_workqueue.c
@@ -0,0 +1,287 @@
+/*
+ * Workqueue statistical tracer.
+ *
+ * Copyright (C) 2008 Frederic Weisbecker <fweisbec@gmail.com>
+ *
+ */
+
+
+#include <trace/workqueue.h>
+#include <linux/list.h>
+#include "trace_stat.h"
+#include "trace.h"
+
+
+/* A cpu workqueue thread */
+struct cpu_workqueue_stats {
+ struct list_head list;
+/* Useful to know if we print the cpu headers */
+ bool first_entry;
+ int cpu;
+ pid_t pid;
+/* Can be inserted from interrupt or user context, need to be atomic */
+ atomic_t inserted;
+/*
+ * Don't need to be atomic, works are serialized in a single workqueue thread
+ * on a single CPU.
+ */
+ unsigned int executed;
+};
+
+/* List of workqueue threads on one cpu */
+struct workqueue_global_stats {
+ struct list_head list;
+ spinlock_t lock;
+};
+
+/* Don't need a global lock because allocated before the workqueues, and
+ * never freed.
+ */
+static struct workqueue_global_stats *all_workqueue_stat;
+
+/* Insertion of a work */
+static void
+probe_workqueue_insertion(struct task_struct *wq_thread,
+ struct work_struct *work)
+{
+ int cpu = cpumask_first(&wq_thread->cpus_allowed);
+ struct cpu_workqueue_stats *node, *next;
+ unsigned long flags;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ list_for_each_entry_safe(node, next, &all_workqueue_stat[cpu].list,
+ list) {
+ if (node->pid == wq_thread->pid) {
+ atomic_inc(&node->inserted);
+ goto found;
+ }
+ }
+ pr_debug("trace_workqueue: entry not found\n");
+found:
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+}
+
+/* Execution of a work */
+static void
+probe_workqueue_execution(struct task_struct *wq_thread,
+ struct work_struct *work)
+{
+ int cpu = cpumask_first(&wq_thread->cpus_allowed);
+ struct cpu_workqueue_stats *node, *next;
+ unsigned long flags;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ list_for_each_entry_safe(node, next, &all_workqueue_stat[cpu].list,
+ list) {
+ if (node->pid == wq_thread->pid) {
+ node->executed++;
+ goto found;
+ }
+ }
+ pr_debug("trace_workqueue: entry not found\n");
+found:
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+}
+
+/* Creation of a cpu workqueue thread */
+static void probe_workqueue_creation(struct task_struct *wq_thread, int cpu)
+{
+ struct cpu_workqueue_stats *cws;
+ unsigned long flags;
+
+ WARN_ON(cpu < 0 || cpu >= num_possible_cpus());
+
+ /* Workqueues are sometimes created in atomic context */
+ cws = kzalloc(sizeof(struct cpu_workqueue_stats), GFP_ATOMIC);
+ if (!cws) {
+ pr_warning("trace_workqueue: not enough memory\n");
+ return;
+ }
+ tracing_record_cmdline(wq_thread);
+
+ INIT_LIST_HEAD(&cws->list);
+ cws->cpu = cpu;
+
+ cws->pid = wq_thread->pid;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ if (list_empty(&all_workqueue_stat[cpu].list))
+ cws->first_entry = true;
+ list_add_tail(&cws->list, &all_workqueue_stat[cpu].list);
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+}
+
+/* Destruction of a cpu workqueue thread */
+static void probe_workqueue_destruction(struct task_struct *wq_thread)
+{
+ /* Workqueue only execute on one cpu */
+ int cpu = cpumask_first(&wq_thread->cpus_allowed);
+ struct cpu_workqueue_stats *node, *next;
+ unsigned long flags;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ list_for_each_entry_safe(node, next, &all_workqueue_stat[cpu].list,
+ list) {
+ if (node->pid == wq_thread->pid) {
+ list_del(&node->list);
+ kfree(node);
+ goto found;
+ }
+ }
+
+ pr_debug("trace_workqueue: don't find workqueue to destroy\n");
+found:
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+
+}
+
+static struct cpu_workqueue_stats *workqueue_stat_start_cpu(int cpu)
+{
+ unsigned long flags;
+ struct cpu_workqueue_stats *ret = NULL;
+
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+
+ if (!list_empty(&all_workqueue_stat[cpu].list))
+ ret = list_entry(all_workqueue_stat[cpu].list.next,
+ struct cpu_workqueue_stats, list);
+
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+
+ return ret;
+}
+
+static void *workqueue_stat_start(void)
+{
+ int cpu;
+ void *ret = NULL;
+
+ for_each_possible_cpu(cpu) {
+ ret = workqueue_stat_start_cpu(cpu);
+ if (ret)
+ return ret;
+ }
+ return NULL;
+}
+
+static void *workqueue_stat_next(void *prev, int idx)
+{
+ struct cpu_workqueue_stats *prev_cws = prev;
+ int cpu = prev_cws->cpu;
+ unsigned long flags;
+ void *ret = NULL;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ if (list_is_last(&prev_cws->list, &all_workqueue_stat[cpu].list)) {
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+ for (++cpu ; cpu < num_possible_cpus(); cpu++) {
+ ret = workqueue_stat_start_cpu(cpu);
+ if (ret)
+ return ret;
+ }
+ return NULL;
+ }
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+
+ return list_entry(prev_cws->list.next, struct cpu_workqueue_stats,
+ list);
+}
+
+static int workqueue_stat_show(struct seq_file *s, void *p)
+{
+ struct cpu_workqueue_stats *cws = p;
+ unsigned long flags;
+ int cpu = cws->cpu;
+
+ seq_printf(s, "%3d %6d %6u %s\n", cws->cpu,
+ atomic_read(&cws->inserted),
+ cws->executed,
+ trace_find_cmdline(cws->pid));
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ if (&cws->list == all_workqueue_stat[cpu].list.next)
+ seq_printf(s, "\n");
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+
+ return 0;
+}
+
+static int workqueue_stat_headers(struct seq_file *s)
+{
+ seq_printf(s, "# CPU INSERTED EXECUTED NAME\n");
+ seq_printf(s, "# | | | |\n\n");
+ return 0;
+}
+
+struct tracer_stat workqueue_stats __read_mostly = {
+ .name = "workqueues",
+ .stat_start = workqueue_stat_start,
+ .stat_next = workqueue_stat_next,
+ .stat_show = workqueue_stat_show,
+ .stat_headers = workqueue_stat_headers
+};
+
+
+int __init stat_workqueue_init(void)
+{
+ if (register_stat_tracer(&workqueue_stats)) {
+ pr_warning("Unable to register workqueue stat tracer\n");
+ return 1;
+ }
+
+ return 0;
+}
+fs_initcall(stat_workqueue_init);
+
+/*
+ * Workqueues are created very early, just after pre-smp initcalls.
+ * So we must register our tracepoints at this stage.
+ */
+int __init trace_workqueue_early_init(void)
+{
+ int ret, cpu;
+
+ ret = register_trace_workqueue_insertion(probe_workqueue_insertion);
+ if (ret)
+ goto out;
+
+ ret = register_trace_workqueue_execution(probe_workqueue_execution);
+ if (ret)
+ goto no_insertion;
+
+ ret = register_trace_workqueue_creation(probe_workqueue_creation);
+ if (ret)
+ goto no_execution;
+
+ ret = register_trace_workqueue_destruction(probe_workqueue_destruction);
+ if (ret)
+ goto no_creation;
+
+ all_workqueue_stat = kmalloc(sizeof(struct workqueue_global_stats)
+ * num_possible_cpus(), GFP_KERNEL);
+
+ if (!all_workqueue_stat) {
+ pr_warning("trace_workqueue: not enough memory\n");
+ goto no_creation;
+ }
+
+ for_each_possible_cpu(cpu) {
+ spin_lock_init(&all_workqueue_stat[cpu].lock);
+ INIT_LIST_HEAD(&all_workqueue_stat[cpu].list);
+ }
+
+ return 0;
+
+no_creation:
+ unregister_trace_workqueue_creation(probe_workqueue_creation);
+no_execution:
+ unregister_trace_workqueue_execution(probe_workqueue_execution);
+no_insertion:
+ unregister_trace_workqueue_insertion(probe_workqueue_insertion);
+out:
+ pr_warning("trace_workqueue: unable to trace workqueues\n");
+
+ return 1;
+}
+early_initcall(trace_workqueue_early_init);
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 2f44583..1fc2bc2 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -33,6 +33,7 @@
#include <linux/kallsyms.h>
#include <linux/debug_locks.h>
#include <linux/lockdep.h>
+#include <trace/workqueue.h>
/*
* The per-CPU workqueue (if single thread, we always use the first
@@ -125,9 +126,13 @@ struct cpu_workqueue_struct *get_wq_data(struct work_struct *work)
return (void *) (atomic_long_read(&work->data) & WORK_STRUCT_WQ_DATA_MASK);
}
+DEFINE_TRACE(workqueue_insertion);
+
static void insert_work(struct cpu_workqueue_struct *cwq,
struct work_struct *work, struct list_head *head)
{
+ trace_workqueue_insertion(cwq->thread, work);
+
set_wq_data(work, cwq);
/*
* Ensure that we get the right work->data if we see the
@@ -259,6 +264,8 @@ int queue_delayed_work_on(int cpu, struct workqueue_struct *wq,
}
EXPORT_SYMBOL_GPL(queue_delayed_work_on);
+DEFINE_TRACE(workqueue_execution);
+
static void run_workqueue(struct cpu_workqueue_struct *cwq)
{
spin_lock_irq(&cwq->lock);
@@ -284,7 +291,7 @@ static void run_workqueue(struct cpu_workqueue_struct *cwq)
*/
struct lockdep_map lockdep_map = work->lockdep_map;
#endif
-
+ trace_workqueue_execution(cwq->thread, work);
cwq->current_work = work;
list_del_init(cwq->worklist.next);
spin_unlock_irq(&cwq->lock);
@@ -765,6 +772,8 @@ init_cpu_workqueue(struct workqueue_struct *wq, int cpu)
return cwq;
}
+DEFINE_TRACE(workqueue_creation);
+
static int create_workqueue_thread(struct cpu_workqueue_struct *cwq, int cpu)
{
struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
@@ -787,6 +796,8 @@ static int create_workqueue_thread(struct cpu_workqueue_struct *cwq, int cpu)
sched_setscheduler_nocheck(p, SCHED_FIFO, ¶m);
cwq->thread = p;
+ trace_workqueue_creation(cwq->thread, cpu);
+
return 0;
}
@@ -868,6 +879,8 @@ struct workqueue_struct *__create_workqueue_key(const char *name,
}
EXPORT_SYMBOL_GPL(__create_workqueue_key);
+DEFINE_TRACE(workqueue_destruction);
+
static void cleanup_workqueue_thread(struct cpu_workqueue_struct *cwq)
{
/*
@@ -891,6 +904,7 @@ static void cleanup_workqueue_thread(struct cpu_workqueue_struct *cwq)
* checks list_empty(), and a "normal" queue_work() can't use
* a dead CPU.
*/
+ trace_workqueue_destruction(cwq->thread);
kthread_stop(cwq->thread);
cwq->thread = NULL;
}
--
1.5.6.5
--
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [PATCH 9/9] ftrace, trivial: fix typo "resgister" -> "register"
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
` (7 preceding siblings ...)
2009-01-13 6:11 ` [PATCH 8/9] tracing: add a new workqueue tracer Steven Rostedt
@ 2009-01-13 6:11 ` Steven Rostedt
2009-01-13 8:03 ` Uwe Kleine-König
2009-01-13 14:06 ` [PATCH 0/9] ftrace: updates for tip Steven Rostedt
9 siblings, 1 reply; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 6:11 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig, Uwe Kleine-K?nig,
trivial, Steven Rostedt
[-- Attachment #1: 0009-ftrace-trivial-fix-typo-resgister-register.patch --]
[-- Type: text/plain, Size: 803 bytes --]
From: Uwe Kleine-K?nig <u.kleine-koenig@pengutronix.de>
Signed-off-by: Uwe Kleine-K?nig <ukleinek@strlen.de>
Cc: trivial@kernel.org
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
kernel/trace/ftrace.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 9f53610..8c1c9c0 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -1894,7 +1894,7 @@ int register_ftrace_function(struct ftrace_ops *ops)
}
/**
- * unregister_ftrace_function - unresgister a function for profiling.
+ * unregister_ftrace_function - unregister a function for profiling.
* @ops - ops structure that holds the function to unregister
*
* Unregister a function that was added to be called by ftrace profiling.
--
1.5.6.5
--
^ permalink raw reply related [flat|nested] 15+ messages in thread
* Re: [PATCH 9/9] ftrace, trivial: fix typo "resgister" -> "register"
2009-01-13 6:11 ` [PATCH 9/9] ftrace, trivial: fix typo "resgister" -> "register" Steven Rostedt
@ 2009-01-13 8:03 ` Uwe Kleine-König
2009-01-13 13:04 ` Steven Rostedt
0 siblings, 1 reply; 15+ messages in thread
From: Uwe Kleine-König @ 2009-01-13 8:03 UTC (permalink / raw)
To: Steven Rostedt
Cc: linux-kernel, Ingo Molnar, Andrew Morton, Frederic Weisbecker,
Shaohua Li, ia64, Tony Luck, Sam Ravnborg, trivial,
Steven Rostedt
On Tue, Jan 13, 2009 at 01:11:19AM -0500, Steven Rostedt wrote:
> From: Uwe Kleine-K?nig <u.kleine-koenig@pengutronix.de>
>
> Signed-off-by: Uwe Kleine-K?nig <ukleinek@strlen.de>
who ever will apply this patch, can you please take care that the '?' is
substituted by 'ö' or if that is too hard, by 'oe'?
Thanks
Uwe
--
Pengutronix e.K. | Uwe Kleine-König |
Industrial Linux Solutions | http://www.pengutronix.de/ |
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 9/9] ftrace, trivial: fix typo "resgister" -> "register"
2009-01-13 8:03 ` Uwe Kleine-König
@ 2009-01-13 13:04 ` Steven Rostedt
0 siblings, 0 replies; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 13:04 UTC (permalink / raw)
To: Uwe Kleine-König
Cc: linux-kernel, Ingo Molnar, Andrew Morton, Frederic Weisbecker,
Shaohua Li, ia64, Tony Luck, Sam Ravnborg, trivial,
Steven Rostedt
On Tue, 13 Jan 2009, Uwe Kleine-K?nig wrote:
> On Tue, Jan 13, 2009 at 01:11:19AM -0500, Steven Rostedt wrote:
> > From: Uwe Kleine-K?nig <u.kleine-koenig@pengutronix.de>
> >
> > Signed-off-by: Uwe Kleine-K?nig <ukleinek@strlen.de>
> who ever will apply this patch, can you please take care that the '?' is
> substituted by '?' or if that is too hard, by 'oe'?
Ouch, that looks bad. I just used git am to apply it and ran it through a
compile and boot up (never know when a comment change will break the build
;-)
I did not look at how git checked it in.
Ingo, if you did not pull this, and since it is the last patch, I'll fix
it in my tree.
-- Steve
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 0/9] ftrace: updates for tip
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
` (8 preceding siblings ...)
2009-01-13 6:11 ` [PATCH 9/9] ftrace, trivial: fix typo "resgister" -> "register" Steven Rostedt
@ 2009-01-13 14:06 ` Steven Rostedt
2009-01-13 16:54 ` Steven Rostedt
9 siblings, 1 reply; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 14:06 UTC (permalink / raw)
To: linux-kernel
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig
On Tue, 13 Jan 2009, Steven Rostedt wrote:
>
> The following patches are in:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
>
> branch: tip/devel
Ingo,
Don't pull any of this yet please. I'm going to rebase some of the patches
to fix the change logs (typos). I'll let you know when that's finished.
Thanks,
-- Steve
>
>
> Frederic Weisbecker (2):
> tracing/ftrace: separate events tracing and stats tracing engine
> tracing: add a new workqueue tracer
>
> Shaohua Li (6):
> ftrace, ia64: make recordmcount destinct module compile
> ftrace, ia64: explictly ignore a file in recordmcount.pl
> ftrace, ia64: Add macro for ftrace_caller
> ftrace, ia64: Add recordmcount for ia64
> ftrace, ia64: IA64 static ftrace support
> ftrace, ia64: IA64 dynamic ftrace support
>
> Uwe Kleine-K?nig (1):
> ftrace, trivial: fix typo "resgister" -> "register"
>
> ----
> arch/ia64/Kconfig | 3 +
> arch/ia64/include/asm/ftrace.h | 28 ++++
> arch/ia64/kernel/Makefile | 5 +
> arch/ia64/kernel/entry.S | 100 ++++++++++++++
> arch/ia64/kernel/ftrace.c | 206 ++++++++++++++++++++++++++++
> arch/ia64/kernel/ia64_ksyms.c | 6 +
> include/linux/ftrace.h | 4 +
> include/trace/workqueue.h | 25 ++++
> kernel/trace/Kconfig | 11 ++
> kernel/trace/Makefile | 1 +
> kernel/trace/ftrace.c | 14 +--
> kernel/trace/trace.c | 2 -
> kernel/trace/trace.h | 20 ---
> kernel/trace/trace_branch.c | 108 +++++++++-------
> kernel/trace/trace_stat.c | 191 +++++++++++---------------
> kernel/trace/trace_stat.h | 31 +++++
> kernel/trace/trace_workqueue.c | 287 ++++++++++++++++++++++++++++++++++++++++
> kernel/workqueue.c | 16 ++-
> scripts/Makefile.build | 13 +-
> scripts/recordmcount.pl | 18 ++-
> 20 files changed, 888 insertions(+), 201 deletions(-)
> --
>
>
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH 0/9] ftrace: updates for tip
2009-01-13 14:06 ` [PATCH 0/9] ftrace: updates for tip Steven Rostedt
@ 2009-01-13 16:54 ` Steven Rostedt
2009-01-14 11:12 ` Ingo Molnar
0 siblings, 1 reply; 15+ messages in thread
From: Steven Rostedt @ 2009-01-13 16:54 UTC (permalink / raw)
To: LKML
Cc: Ingo Molnar, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig
On Tue, 13 Jan 2009, Steven Rostedt wrote:
>
>
> On Tue, 13 Jan 2009, Steven Rostedt wrote:
>
> >
> > The following patches are in:
> >
> > git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
> >
> > branch: tip/devel
>
> Ingo,
>
> Don't pull any of this yet please. I'm going to rebase some of the patches
> to fix the change logs (typos). I'll let you know when that's finished.
>
OK, I rebased in changing the change logs. Here's what I did:
s/destinct/distinct/
s/K?nig/Koenig/
Here's the changes:
The following patches are in:
git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
branch: tip/devel
Frederic Weisbecker (2):
tracing/ftrace: separate events tracing and stats tracing engine
tracing: add a new workqueue tracer
Shaohua Li (6):
ftrace, ia64: make recordmcount distinct module compile
ftrace, ia64: explictly ignore a file in recordmcount.pl
ftrace, ia64: Add macro for ftrace_caller
ftrace, ia64: Add recordmcount for ia64
ftrace, ia64: IA64 static ftrace support
ftrace, ia64: IA64 dynamic ftrace support
Uwe Kleine-Koenig (1):
ftrace, trivial: fix typo "resgister" -> "register"
----
arch/ia64/Kconfig | 3 +
arch/ia64/include/asm/ftrace.h | 28 ++++
arch/ia64/kernel/Makefile | 5 +
arch/ia64/kernel/entry.S | 100 ++++++++++++++
arch/ia64/kernel/ftrace.c | 206 ++++++++++++++++++++++++++++
arch/ia64/kernel/ia64_ksyms.c | 6 +
include/linux/ftrace.h | 4 +
include/trace/workqueue.h | 25 ++++
kernel/trace/Kconfig | 11 ++
kernel/trace/Makefile | 1 +
kernel/trace/ftrace.c | 14 +--
kernel/trace/trace.c | 2 -
kernel/trace/trace.h | 20 ---
kernel/trace/trace_branch.c | 108 +++++++++-------
kernel/trace/trace_stat.c | 191 +++++++++++---------------
kernel/trace/trace_stat.h | 31 +++++
kernel/trace/trace_workqueue.c | 287 ++++++++++++++++++++++++++++++++++++++++
kernel/workqueue.c | 16 ++-
scripts/Makefile.build | 13 +-
scripts/recordmcount.pl | 18 ++-
20 files changed, 888 insertions(+), 201 deletions(-)
diff --git a/arch/ia64/Kconfig b/arch/ia64/Kconfig
index 3d31636..e20c1d4 100644
--- a/arch/ia64/Kconfig
+++ b/arch/ia64/Kconfig
@@ -21,6 +21,9 @@ config IA64
select HAVE_OPROFILE
select HAVE_KPROBES
select HAVE_KRETPROBES
+ select HAVE_FTRACE_MCOUNT_RECORD
+ select HAVE_DYNAMIC_FTRACE if (!ITANIUM)
+ select HAVE_FUNCTION_TRACER
select HAVE_DMA_ATTRS
select HAVE_KVM
select HAVE_ARCH_TRACEHOOK
diff --git a/arch/ia64/include/asm/ftrace.h b/arch/ia64/include/asm/ftrace.h
new file mode 100644
index 0000000..d20db3c
--- /dev/null
+++ b/arch/ia64/include/asm/ftrace.h
@@ -0,0 +1,28 @@
+#ifndef _ASM_IA64_FTRACE_H
+#define _ASM_IA64_FTRACE_H
+
+#ifdef CONFIG_FUNCTION_TRACER
+#define MCOUNT_INSN_SIZE 32 /* sizeof mcount call */
+
+#ifndef __ASSEMBLY__
+extern void _mcount(unsigned long pfs, unsigned long r1, unsigned long b0, unsigned long r0);
+#define mcount _mcount
+
+#include <asm/kprobes.h>
+/* In IA64, MCOUNT_ADDR is set in link time, so it's not a constant at compile time */
+#define MCOUNT_ADDR (((struct fnptr *)mcount)->ip)
+#define FTRACE_ADDR (((struct fnptr *)ftrace_caller)->ip)
+
+static inline unsigned long ftrace_call_adjust(unsigned long addr)
+{
+ /* second bundle, insn 2 */
+ return addr - 0x12;
+}
+
+struct dyn_arch_ftrace {
+};
+#endif
+
+#endif /* CONFIG_FUNCTION_TRACER */
+
+#endif /* _ASM_IA64_FTRACE_H */
diff --git a/arch/ia64/kernel/Makefile b/arch/ia64/kernel/Makefile
index f2778f2..dc62df0 100644
--- a/arch/ia64/kernel/Makefile
+++ b/arch/ia64/kernel/Makefile
@@ -2,6 +2,10 @@
# Makefile for the linux kernel.
#
+ifdef CONFIG_DYNAMIC_FTRACE
+CFLAGS_REMOVE_ftrace.o = -pg
+endif
+
extra-y := head.o init_task.o vmlinux.lds
obj-y := acpi.o entry.o efi.o efi_stub.o gate-data.o fsys.o ia64_ksyms.o irq.o irq_ia64.o \
@@ -28,6 +32,7 @@ obj-$(CONFIG_IA64_CYCLONE) += cyclone.o
obj-$(CONFIG_CPU_FREQ) += cpufreq/
obj-$(CONFIG_IA64_MCA_RECOVERY) += mca_recovery.o
obj-$(CONFIG_KPROBES) += kprobes.o jprobes.o
+obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o
obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o crash.o
obj-$(CONFIG_CRASH_DUMP) += crash_dump.o
obj-$(CONFIG_IA64_UNCACHED_ALLOCATOR) += uncached.o
diff --git a/arch/ia64/kernel/entry.S b/arch/ia64/kernel/entry.S
index d435f4a..e0be92a 100644
--- a/arch/ia64/kernel/entry.S
+++ b/arch/ia64/kernel/entry.S
@@ -47,6 +47,7 @@
#include <asm/processor.h>
#include <asm/thread_info.h>
#include <asm/unistd.h>
+#include <asm/ftrace.h>
#include "minstate.h"
@@ -1404,6 +1405,105 @@ GLOBAL_ENTRY(unw_init_running)
br.ret.sptk.many rp
END(unw_init_running)
+#ifdef CONFIG_FUNCTION_TRACER
+#ifdef CONFIG_DYNAMIC_FTRACE
+GLOBAL_ENTRY(_mcount)
+ br ftrace_stub
+END(_mcount)
+
+.here:
+ br.ret.sptk.many b0
+
+GLOBAL_ENTRY(ftrace_caller)
+ alloc out0 = ar.pfs, 8, 0, 4, 0
+ mov out3 = r0
+ ;;
+ mov out2 = b0
+ add r3 = 0x20, r3
+ mov out1 = r1;
+ br.call.sptk.many b0 = ftrace_patch_gp
+ //this might be called from module, so we must patch gp
+ftrace_patch_gp:
+ movl gp=__gp
+ mov b0 = r3
+ ;;
+.global ftrace_call;
+ftrace_call:
+{
+ .mlx
+ nop.m 0x0
+ movl r3 = .here;;
+}
+ alloc loc0 = ar.pfs, 4, 4, 2, 0
+ ;;
+ mov loc1 = b0
+ mov out0 = b0
+ mov loc2 = r8
+ mov loc3 = r15
+ ;;
+ adds out0 = -MCOUNT_INSN_SIZE, out0
+ mov out1 = in2
+ mov b6 = r3
+
+ br.call.sptk.many b0 = b6
+ ;;
+ mov ar.pfs = loc0
+ mov b0 = loc1
+ mov r8 = loc2
+ mov r15 = loc3
+ br ftrace_stub
+ ;;
+END(ftrace_caller)
+
+#else
+GLOBAL_ENTRY(_mcount)
+ movl r2 = ftrace_stub
+ movl r3 = ftrace_trace_function;;
+ ld8 r3 = [r3];;
+ ld8 r3 = [r3];;
+ cmp.eq p7,p0 = r2, r3
+(p7) br.sptk.many ftrace_stub
+ ;;
+
+ alloc loc0 = ar.pfs, 4, 4, 2, 0
+ ;;
+ mov loc1 = b0
+ mov out0 = b0
+ mov loc2 = r8
+ mov loc3 = r15
+ ;;
+ adds out0 = -MCOUNT_INSN_SIZE, out0
+ mov out1 = in2
+ mov b6 = r3
+
+ br.call.sptk.many b0 = b6
+ ;;
+ mov ar.pfs = loc0
+ mov b0 = loc1
+ mov r8 = loc2
+ mov r15 = loc3
+ br ftrace_stub
+ ;;
+END(_mcount)
+#endif
+
+GLOBAL_ENTRY(ftrace_stub)
+ mov r3 = b0
+ movl r2 = _mcount_ret_helper
+ ;;
+ mov b6 = r2
+ mov b7 = r3
+ br.ret.sptk.many b6
+
+_mcount_ret_helper:
+ mov b0 = r42
+ mov r1 = r41
+ mov ar.pfs = r40
+ br b7
+END(ftrace_stub)
+
+#endif /* CONFIG_FUNCTION_TRACER */
+
.rodata
.align 8
.globl sys_call_table
diff --git a/arch/ia64/kernel/ftrace.c b/arch/ia64/kernel/ftrace.c
new file mode 100644
index 0000000..7fc8c96
--- /dev/null
+++ b/arch/ia64/kernel/ftrace.c
@@ -0,0 +1,206 @@
+/*
+ * Dynamic function tracing support.
+ *
+ * Copyright (C) 2008 Shaohua Li <shaohua.li@intel.com>
+ *
+ * For licencing details, see COPYING.
+ *
+ * Defines low-level handling of mcount calls when the kernel
+ * is compiled with the -pg flag. When using dynamic ftrace, the
+ * mcount call-sites get patched lazily with NOP till they are
+ * enabled. All code mutation routines here take effect atomically.
+ */
+
+#include <linux/uaccess.h>
+#include <linux/ftrace.h>
+
+#include <asm/cacheflush.h>
+#include <asm/patch.h>
+
+/* In IA64, each function will be added below two bundles with -pg option */
+static unsigned char __attribute__((aligned(8)))
+ftrace_orig_code[MCOUNT_INSN_SIZE] = {
+ 0x02, 0x40, 0x31, 0x10, 0x80, 0x05, /* alloc r40=ar.pfs,12,8,0 */
+ 0xb0, 0x02, 0x00, 0x00, 0x42, 0x40, /* mov r43=r0;; */
+ 0x05, 0x00, 0xc4, 0x00, /* mov r42=b0 */
+ 0x11, 0x48, 0x01, 0x02, 0x00, 0x21, /* mov r41=r1 */
+ 0x00, 0x00, 0x00, 0x02, 0x00, 0x00, /* nop.i 0x0 */
+ 0x08, 0x00, 0x00, 0x50 /* br.call.sptk.many b0 = _mcount;; */
+};
+
+struct ftrace_orig_insn {
+ u64 dummy1, dummy2, dummy3;
+ u64 dummy4:64-41+13;
+ u64 imm20:20;
+ u64 dummy5:3;
+ u64 sign:1;
+ u64 dummy6:4;
+};
+
+/* mcount stub will be converted below for nop */
+static unsigned char ftrace_nop_code[MCOUNT_INSN_SIZE] = {
+ 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, /* [MII] nop.m 0x0 */
+ 0x30, 0x00, 0x00, 0x60, 0x00, 0x00, /* mov r3=ip */
+ 0x00, 0x00, 0x04, 0x00, /* nop.i 0x0 */
+ 0x05, 0x00, 0x00, 0x00, 0x01, 0x00, /* [MLX] nop.m 0x0 */
+ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* nop.x 0x0;; */
+ 0x00, 0x00, 0x04, 0x00
+};
+
+static unsigned char *ftrace_nop_replace(void)
+{
+ return ftrace_nop_code;
+}
+
+/*
+ * mcount stub will be converted below for call
+ * Note: Just the last instruction is changed against nop
+ * */
+static unsigned char __attribute__((aligned(8)))
+ftrace_call_code[MCOUNT_INSN_SIZE] = {
+ 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, /* [MII] nop.m 0x0 */
+ 0x30, 0x00, 0x00, 0x60, 0x00, 0x00, /* mov r3=ip */
+ 0x00, 0x00, 0x04, 0x00, /* nop.i 0x0 */
+ 0x05, 0x00, 0x00, 0x00, 0x01, 0x00, /* [MLX] nop.m 0x0 */
+ 0xff, 0xff, 0xff, 0xff, 0x7f, 0x00, /* brl.many .;;*/
+ 0xf8, 0xff, 0xff, 0xc8
+};
+
+struct ftrace_call_insn {
+ u64 dummy1, dummy2;
+ u64 dummy3:48;
+ u64 imm39_l:16;
+ u64 imm39_h:23;
+ u64 dummy4:13;
+ u64 imm20:20;
+ u64 dummy5:3;
+ u64 i:1;
+ u64 dummy6:4;
+};
+
+static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
+{
+ struct ftrace_call_insn *code = (void *)ftrace_call_code;
+ unsigned long offset = addr - (ip + 0x10);
+
+ code->imm39_l = offset >> 24;
+ code->imm39_h = offset >> 40;
+ code->imm20 = offset >> 4;
+ code->i = offset >> 63;
+ return ftrace_call_code;
+}
+
+static int
+ftrace_modify_code(unsigned long ip, unsigned char *old_code,
+ unsigned char *new_code, int do_check)
+{
+ 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 (!do_check)
+ goto skip_check;
+
+ /* 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;
+
+skip_check:
+ /* 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;
+}
+
+static int ftrace_make_nop_check(struct dyn_ftrace *rec, unsigned long addr)
+{
+ unsigned char __attribute__((aligned(8))) replaced[MCOUNT_INSN_SIZE];
+ unsigned long ip = rec->ip;
+
+ if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE))
+ return -EFAULT;
+ if (rec->flags & FTRACE_FL_CONVERTED) {
+ struct ftrace_call_insn *call_insn, *tmp_call;
+
+ call_insn = (void *)ftrace_call_code;
+ tmp_call = (void *)replaced;
+ call_insn->imm39_l = tmp_call->imm39_l;
+ call_insn->imm39_h = tmp_call->imm39_h;
+ call_insn->imm20 = tmp_call->imm20;
+ call_insn->i = tmp_call->i;
+ if (memcmp(replaced, ftrace_call_code, MCOUNT_INSN_SIZE) != 0)
+ return -EINVAL;
+ return 0;
+ } else {
+ struct ftrace_orig_insn *call_insn, *tmp_call;
+
+ call_insn = (void *)ftrace_orig_code;
+ tmp_call = (void *)replaced;
+ call_insn->sign = tmp_call->sign;
+ call_insn->imm20 = tmp_call->imm20;
+ if (memcmp(replaced, ftrace_orig_code, MCOUNT_INSN_SIZE) != 0)
+ return -EINVAL;
+ return 0;
+ }
+}
+
+int ftrace_make_nop(struct module *mod,
+ struct dyn_ftrace *rec, unsigned long addr)
+{
+ int ret;
+ char *new;
+
+ ret = ftrace_make_nop_check(rec, addr);
+ if (ret)
+ return ret;
+ new = ftrace_nop_replace();
+ return ftrace_modify_code(rec->ip, NULL, new, 0);
+}
+
+int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
+{
+ unsigned long ip = rec->ip;
+ unsigned char *old, *new;
+
+ old= ftrace_nop_replace();
+ new = ftrace_call_replace(ip, addr);
+ return ftrace_modify_code(ip, old, new, 1);
+}
+
+/* in IA64, _mcount can't directly call ftrace_stub. Only jump is ok */
+int ftrace_update_ftrace_func(ftrace_func_t func)
+{
+ unsigned long ip;
+ unsigned long addr = ((struct fnptr *)ftrace_call)->ip;
+
+ if (func == ftrace_stub)
+ return 0;
+ ip = ((struct fnptr *)func)->ip;
+
+ ia64_patch_imm64(addr + 2, ip);
+
+ flush_icache_range(addr, addr + 16);
+ return 0;
+}
+
+/* run from kstop_machine */
+int __init ftrace_dyn_arch_init(void *data)
+{
+ *(unsigned long *)data = 0;
+
+ return 0;
+}
diff --git a/arch/ia64/kernel/ia64_ksyms.c b/arch/ia64/kernel/ia64_ksyms.c
index 6da1f20..2d31186 100644
--- a/arch/ia64/kernel/ia64_ksyms.c
+++ b/arch/ia64/kernel/ia64_ksyms.c
@@ -112,3 +112,9 @@ EXPORT_SYMBOL_GPL(esi_call_phys);
#endif
extern char ia64_ivt[];
EXPORT_SYMBOL(ia64_ivt);
+
+#include <asm/ftrace.h>
+#ifdef CONFIG_FUNCTION_TRACER
+/* mcount is defined in assembly */
+EXPORT_SYMBOL(_mcount);
+#endif
diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
index 677432b..0547214 100644
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h
@@ -126,6 +126,10 @@ extern int ftrace_update_ftrace_func(ftrace_func_t func);
extern void ftrace_caller(void);
extern void ftrace_call(void);
extern void mcount_call(void);
+
+#ifndef FTRACE_ADDR
+#define FTRACE_ADDR ((unsigned long)ftrace_caller)
+#endif
#ifdef CONFIG_FUNCTION_GRAPH_TRACER
extern void ftrace_graph_caller(void);
extern int ftrace_enable_ftrace_graph_caller(void);
diff --git a/include/trace/workqueue.h b/include/trace/workqueue.h
new file mode 100644
index 0000000..867829d
--- /dev/null
+++ b/include/trace/workqueue.h
@@ -0,0 +1,25 @@
+#ifndef __TRACE_WORKQUEUE_H
+#define __TRACE_WORKQUEUE_H
+
+#include <linux/tracepoint.h>
+#include <linux/workqueue.h>
+#include <linux/sched.h>
+
+DECLARE_TRACE(workqueue_insertion,
+ TPPROTO(struct task_struct *wq_thread, struct work_struct *work),
+ TPARGS(wq_thread, work));
+
+DECLARE_TRACE(workqueue_execution,
+ TPPROTO(struct task_struct *wq_thread, struct work_struct *work),
+ TPARGS(wq_thread, work));
+
+/* Trace the creation of one workqueue thread on a cpu */
+DECLARE_TRACE(workqueue_creation,
+ TPPROTO(struct task_struct *wq_thread, int cpu),
+ TPARGS(wq_thread, cpu));
+
+DECLARE_TRACE(workqueue_destruction,
+ TPPROTO(struct task_struct *wq_thread),
+ TPARGS(wq_thread));
+
+#endif /* __TRACE_WORKQUEUE_H */
diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
index 9442392..dde1d46 100644
--- a/kernel/trace/Kconfig
+++ b/kernel/trace/Kconfig
@@ -284,6 +284,17 @@ config KMEMTRACE
If unsure, say N.
+config WORKQUEUE_TRACER
+ bool "Trace workqueues"
+ select TRACING
+ help
+ The workqueue tracer provides some statistical informations
+ about each cpu workqueue thread such as the number of the
+ works inserted and executed since their creation. It can help
+ to evaluate the amount of work each of them have to perform.
+ For example it can help a developer to decide whether he should
+ choose a per cpu workqueue instead of a singlethreaded one.
+
config DYNAMIC_FTRACE
bool "enable/disable ftrace tracepoints dynamically"
diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile
index 05c9182..f76d48f 100644
--- a/kernel/trace/Makefile
+++ b/kernel/trace/Makefile
@@ -36,5 +36,6 @@ obj-$(CONFIG_TRACE_BRANCH_PROFILING) += trace_branch.o
obj-$(CONFIG_HW_BRANCH_TRACER) += trace_hw_branches.o
obj-$(CONFIG_POWER_TRACER) += trace_power.o
obj-$(CONFIG_KMEMTRACE) += kmemtrace.o
+obj-$(CONFIG_WORKQUEUE_TRACER) += trace_workqueue.o
libftrace-y := ftrace.o
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 9e54a6c..8c1c9c0 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -263,14 +263,6 @@ static void ftrace_update_pid_func(void)
# error Dynamic ftrace depends on MCOUNT_RECORD
#endif
-/*
- * Since MCOUNT_ADDR may point to mcount itself, we do not want
- * to get it confused by reading a reference in the code as we
- * are parsing on objcopy output of text. Use a variable for
- * it instead.
- */
-static unsigned long mcount_addr = MCOUNT_ADDR;
-
enum {
FTRACE_ENABLE_CALLS = (1 << 0),
FTRACE_DISABLE_CALLS = (1 << 1),
@@ -463,7 +455,7 @@ __ftrace_replace_code(struct dyn_ftrace *rec, int enable)
unsigned long ip, fl;
unsigned long ftrace_addr;
- ftrace_addr = (unsigned long)ftrace_caller;
+ ftrace_addr = (unsigned long)FTRACE_ADDR;
ip = rec->ip;
@@ -575,7 +567,7 @@ ftrace_code_disable(struct module *mod, struct dyn_ftrace *rec)
ip = rec->ip;
- ret = ftrace_make_nop(mod, rec, mcount_addr);
+ ret = ftrace_make_nop(mod, rec, MCOUNT_ADDR);
if (ret) {
ftrace_bug(ret, ip);
rec->flags |= FTRACE_FL_FAILED;
@@ -1902,7 +1894,7 @@ int register_ftrace_function(struct ftrace_ops *ops)
}
/**
- * unregister_ftrace_function - unresgister a function for profiling.
+ * unregister_ftrace_function - unregister a function for profiling.
* @ops - ops structure that holds the function to unregister
*
* Unregister a function that was added to be called by ftrace profiling.
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 0418fc3..40217fb 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -2353,7 +2353,6 @@ static int tracing_set_tracer(char *buf)
if (ret)
goto out;
}
- init_tracer_stat(t);
trace_branch_enable(tr);
out:
@@ -3218,7 +3217,6 @@ __init static int tracer_alloc_buffers(void)
#else
current_trace = &nop_trace;
#endif
- init_tracer_stat(current_trace);
/* All seems OK, enable tracing */
tracing_disabled = 0;
diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index b3f9ad1..79c8721 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -334,24 +334,6 @@ struct tracer_flags {
/* Makes more easy to define a tracer opt */
#define TRACER_OPT(s, b) .name = #s, .bit = b
-/*
- * If you want to provide a stat file (one-shot statistics), fill
- * an iterator with stat_start/stat_next and a stat_show callbacks.
- * The others callbacks are optional.
- */
-struct tracer_stat {
- /* The name of your stat file */
- const char *name;
- /* Iteration over statistic entries */
- void *(*stat_start)(void);
- void *(*stat_next)(void *prev, int idx);
- /* Compare two entries for sorting (optional) for stats */
- int (*stat_cmp)(void *p1, void *p2);
- /* Print a stat entry */
- int (*stat_show)(struct seq_file *s, void *p);
- /* Print the headers of your stat entries */
- int (*stat_headers)(struct seq_file *s);
-};
/*
* A specific tracer, represented by methods that operate on a trace array:
@@ -466,8 +448,6 @@ void tracing_start_sched_switch_record(void);
int register_tracer(struct tracer *type);
void unregister_tracer(struct tracer *type);
-void init_tracer_stat(struct tracer *trace);
-
extern unsigned long nsecs_to_usecs(unsigned long nsecs);
extern unsigned long tracing_max_latency;
diff --git a/kernel/trace/trace_branch.c b/kernel/trace/trace_branch.c
index da5cf3e..ca017e0 100644
--- a/kernel/trace/trace_branch.c
+++ b/kernel/trace/trace_branch.c
@@ -16,12 +16,12 @@
#include <asm/local.h>
#include "trace.h"
+#include "trace_stat.h"
#include "trace_output.h"
-static struct tracer branch_trace;
-
#ifdef CONFIG_BRANCH_TRACER
+static struct tracer branch_trace;
static int branch_tracing_enabled __read_mostly;
static DEFINE_MUTEX(branch_tracing_mutex);
@@ -191,6 +191,30 @@ static struct trace_event trace_branch_event = {
.binary = trace_nop_print,
};
+static struct tracer branch_trace __read_mostly =
+{
+ .name = "branch",
+ .init = branch_trace_init,
+ .reset = branch_trace_reset,
+#ifdef CONFIG_FTRACE_SELFTEST
+ .selftest = trace_selftest_startup_branch,
+#endif /* CONFIG_FTRACE_SELFTEST */
+};
+
+__init static int init_branch_tracer(void)
+{
+ int ret;
+
+ ret = register_ftrace_event(&trace_branch_event);
+ if (!ret) {
+ printk(KERN_WARNING "Warning: could not register "
+ "branch events\n");
+ return 1;
+ }
+ return register_tracer(&branch_trace);
+}
+device_initcall(init_branch_tracer);
+
#else
static inline
void trace_likely_condition(struct ftrace_branch_data *f, int val, int expect)
@@ -305,6 +329,29 @@ static int annotated_branch_stat_cmp(void *p1, void *p2)
return 0;
}
+static struct tracer_stat annotated_branch_stats = {
+ .name = "branch_annotated",
+ .stat_start = annotated_branch_stat_start,
+ .stat_next = annotated_branch_stat_next,
+ .stat_cmp = annotated_branch_stat_cmp,
+ .stat_headers = annotated_branch_stat_headers,
+ .stat_show = branch_stat_show
+};
+
+__init static int init_annotated_branch_stats(void)
+{
+ int ret;
+
+ ret = register_stat_tracer(&annotated_branch_stats);
+ if (!ret) {
+ printk(KERN_WARNING "Warning: could not register "
+ "annotated branches stats\n");
+ return 1;
+ }
+ return 0;
+}
+fs_initcall(init_annotated_branch_stats);
+
#ifdef CONFIG_PROFILE_ALL_BRANCHES
extern unsigned long __start_branch_profile[];
@@ -339,60 +386,25 @@ all_branch_stat_next(void *v, int idx)
return p;
}
-static struct tracer_stat branch_stats[] = {
- {.name = "annotated",
- .stat_start = annotated_branch_stat_start,
- .stat_next = annotated_branch_stat_next,
- .stat_cmp = annotated_branch_stat_cmp,
- .stat_headers = annotated_branch_stat_headers,
- .stat_show = branch_stat_show},
-
- {.name = "all",
+static struct tracer_stat all_branch_stats = {
+ .name = "branch_all",
.stat_start = all_branch_stat_start,
.stat_next = all_branch_stat_next,
.stat_headers = all_branch_stat_headers,
- .stat_show = branch_stat_show},
-
- { }
-};
-#else
-static struct tracer_stat branch_stats[] = {
- {.name = "annotated",
- .stat_start = annotated_branch_stat_start,
- .stat_next = annotated_branch_stat_next,
- .stat_cmp = annotated_branch_stat_cmp,
- .stat_headers = annotated_branch_stat_headers,
- .stat_show = branch_stat_show},
-
- { }
+ .stat_show = branch_stat_show
};
-#endif /* CONFIG_PROFILE_ALL_BRANCHES */
-
-static struct tracer branch_trace __read_mostly =
+__init static int all_annotated_branch_stats(void)
{
- .name = "branch",
-#ifdef CONFIG_BRANCH_TRACER
- .init = branch_trace_init,
- .reset = branch_trace_reset,
-#ifdef CONFIG_FTRACE_SELFTEST
- .selftest = trace_selftest_startup_branch,
-#endif /* CONFIG_FTRACE_SELFTEST */
-#endif
- .stats = branch_stats
-};
-
-__init static int init_branch_trace(void)
-{
-#ifdef CONFIG_BRANCH_TRACER
int ret;
- ret = register_ftrace_event(&trace_branch_event);
+
+ ret = register_stat_tracer(&all_branch_stats);
if (!ret) {
- printk(KERN_WARNING "Warning: could not register branch events\n");
+ printk(KERN_WARNING "Warning: could not register "
+ "all branches stats\n");
return 1;
}
-#endif
-
- return register_tracer(&branch_trace);
+ return 0;
}
-device_initcall(init_branch_trace);
+fs_initcall(all_annotated_branch_stats);
+#endif /* CONFIG_PROFILE_ALL_BRANCHES */
diff --git a/kernel/trace/trace_stat.c b/kernel/trace/trace_stat.c
index 1515f9e..cb29282 100644
--- a/kernel/trace/trace_stat.c
+++ b/kernel/trace/trace_stat.c
@@ -10,28 +10,32 @@
#include <linux/list.h>
-#include <linux/seq_file.h>
#include <linux/debugfs.h>
+#include "trace_stat.h"
#include "trace.h"
/* List of stat entries from a tracer */
struct trace_stat_list {
- struct list_head list;
- void *stat;
+ struct list_head list;
+ void *stat;
};
/* A stat session is the stats output in one file */
struct tracer_stat_session {
- struct tracer_stat *ts;
- struct list_head stat_list;
- struct mutex stat_mutex;
+ struct list_head session_list;
+ struct tracer_stat *ts;
+ struct list_head stat_list;
+ struct mutex stat_mutex;
+ struct dentry *file;
};
/* All of the sessions currently in use. Each stat file embeed one session */
-static struct tracer_stat_session **all_stat_sessions;
-static int nb_sessions;
-static struct dentry *stat_dir, **stat_files;
+static LIST_HEAD(all_stat_sessions);
+static DEFINE_MUTEX(all_stat_sessions_mutex);
+
+/* The root directory for all stat files */
+static struct dentry *stat_dir;
static void reset_stat_session(struct tracer_stat_session *session)
@@ -44,66 +48,77 @@ static void reset_stat_session(struct tracer_stat_session *session)
INIT_LIST_HEAD(&session->stat_list);
}
-/* Called when a tracer is initialized */
-static int init_all_sessions(int nb, struct tracer_stat *ts)
+static void destroy_session(struct tracer_stat_session *session)
{
- int i, j;
- struct tracer_stat_session *session;
+ debugfs_remove(session->file);
+ reset_stat_session(session);
+ mutex_destroy(&session->stat_mutex);
+ kfree(session);
+}
- nb_sessions = 0;
- if (all_stat_sessions) {
- for (i = 0; i < nb_sessions; i++) {
- session = all_stat_sessions[i];
- reset_stat_session(session);
- mutex_destroy(&session->stat_mutex);
- kfree(session);
- }
- }
- all_stat_sessions = kmalloc(sizeof(struct tracer_stat_session *) * nb,
- GFP_KERNEL);
- if (!all_stat_sessions)
- return -ENOMEM;
+static int init_stat_file(struct tracer_stat_session *session);
- for (i = 0; i < nb; i++) {
- session = kmalloc(sizeof(struct tracer_stat_session) * nb,
- GFP_KERNEL);
- if (!session)
- goto free_sessions;
+int register_stat_tracer(struct tracer_stat *trace)
+{
+ struct tracer_stat_session *session, *node, *tmp;
+ int ret;
+
+ if (!trace)
+ return -EINVAL;
+
+ if (!trace->stat_start || !trace->stat_next || !trace->stat_show)
+ return -EINVAL;
- INIT_LIST_HEAD(&session->stat_list);
- mutex_init(&session->stat_mutex);
- session->ts = &ts[i];
- all_stat_sessions[i] = session;
+ /* Already registered? */
+ mutex_lock(&all_stat_sessions_mutex);
+ list_for_each_entry_safe(node, tmp, &all_stat_sessions, session_list) {
+ if (node->ts == trace)
+ return -EINVAL;
}
- nb_sessions = nb;
- return 0;
+ mutex_unlock(&all_stat_sessions_mutex);
+
+ /* Init the session */
+ session = kmalloc(sizeof(struct tracer_stat_session), GFP_KERNEL);
+ if (!session)
+ return -ENOMEM;
-free_sessions:
+ session->ts = trace;
+ INIT_LIST_HEAD(&session->session_list);
+ INIT_LIST_HEAD(&session->stat_list);
+ mutex_init(&session->stat_mutex);
+ session->file = NULL;
- for (j = 0; j < i; j++)
- kfree(all_stat_sessions[i]);
+ ret = init_stat_file(session);
+ if (ret) {
+ destroy_session(session);
+ return ret;
+ }
- kfree(all_stat_sessions);
- all_stat_sessions = NULL;
+ /* Register */
+ mutex_lock(&all_stat_sessions_mutex);
+ list_add_tail(&session->session_list, &all_stat_sessions);
+ mutex_unlock(&all_stat_sessions_mutex);
- return -ENOMEM;
+ return 0;
}
-static int basic_tracer_stat_checks(struct tracer_stat *ts)
+void unregister_stat_tracer(struct tracer_stat *trace)
{
- int i;
+ struct tracer_stat_session *node, *tmp;
- if (!ts)
- return 0;
-
- for (i = 0; ts[i].name; i++) {
- if (!ts[i].stat_start || !ts[i].stat_next || !ts[i].stat_show)
- return -EBUSY;
+ mutex_lock(&all_stat_sessions_mutex);
+ list_for_each_entry_safe(node, tmp, &all_stat_sessions, session_list) {
+ if (node->ts == trace) {
+ list_del(&node->session_list);
+ destroy_session(node);
+ break;
+ }
}
- return i;
+ mutex_unlock(&all_stat_sessions_mutex);
}
+
/*
* For tracers that don't provide a stat_cmp callback.
* This one will force an immediate insertion on tail of
@@ -280,63 +295,7 @@ static const struct file_operations tracing_stat_fops = {
.release = tracing_stat_release
};
-
-static void destroy_trace_stat_files(void)
-{
- int i;
-
- if (stat_files) {
- for (i = 0; i < nb_sessions; i++)
- debugfs_remove(stat_files[i]);
- kfree(stat_files);
- stat_files = NULL;
- }
-}
-
-static void init_trace_stat_files(void)
-{
- int i;
-
- if (!stat_dir || !nb_sessions)
- return;
-
- stat_files = kmalloc(sizeof(struct dentry *) * nb_sessions, GFP_KERNEL);
-
- if (!stat_files) {
- pr_warning("trace stat: not enough memory\n");
- return;
- }
-
- for (i = 0; i < nb_sessions; i++) {
- struct tracer_stat_session *session = all_stat_sessions[i];
- stat_files[i] = debugfs_create_file(session->ts->name, 0644,
- stat_dir,
- session, &tracing_stat_fops);
- if (!stat_files[i])
- pr_warning("cannot create %s entry\n",
- session->ts->name);
- }
-}
-
-void init_tracer_stat(struct tracer *trace)
-{
- int nb = basic_tracer_stat_checks(trace->stats);
-
- destroy_trace_stat_files();
-
- if (nb < 0) {
- pr_warning("stat tracing: missing stat callback on %s\n",
- trace->name);
- return;
- }
- if (!nb)
- return;
-
- init_all_sessions(nb, trace->stats);
- init_trace_stat_files();
-}
-
-static int __init tracing_stat_init(void)
+static int tracing_stat_init(void)
{
struct dentry *d_tracing;
@@ -348,4 +307,16 @@ static int __init tracing_stat_init(void)
"'trace_stat' entry\n");
return 0;
}
-fs_initcall(tracing_stat_init);
+
+static int init_stat_file(struct tracer_stat_session *session)
+{
+ if (!stat_dir && tracing_stat_init())
+ return -ENODEV;
+
+ session->file = debugfs_create_file(session->ts->name, 0644,
+ stat_dir,
+ session, &tracing_stat_fops);
+ if (!session->file)
+ return -ENOMEM;
+ return 0;
+}
diff --git a/kernel/trace/trace_stat.h b/kernel/trace/trace_stat.h
new file mode 100644
index 0000000..202274c
--- /dev/null
+++ b/kernel/trace/trace_stat.h
@@ -0,0 +1,31 @@
+#ifndef __TRACE_STAT_H
+#define __TRACE_STAT_H
+
+#include <linux/seq_file.h>
+
+/*
+ * If you want to provide a stat file (one-shot statistics), fill
+ * an iterator with stat_start/stat_next and a stat_show callbacks.
+ * The others callbacks are optional.
+ */
+struct tracer_stat {
+ /* The name of your stat file */
+ const char *name;
+ /* Iteration over statistic entries */
+ void *(*stat_start)(void);
+ void *(*stat_next)(void *prev, int idx);
+ /* Compare two entries for stats sorting */
+ int (*stat_cmp)(void *p1, void *p2);
+ /* Print a stat entry */
+ int (*stat_show)(struct seq_file *s, void *p);
+ /* Print the headers of your stat entries */
+ int (*stat_headers)(struct seq_file *s);
+};
+
+/*
+ * Destroy or create a stat file
+ */
+extern int register_stat_tracer(struct tracer_stat *trace);
+extern void unregister_stat_tracer(struct tracer_stat *trace);
+
+#endif /* __TRACE_STAT_H */
diff --git a/kernel/trace/trace_workqueue.c b/kernel/trace/trace_workqueue.c
new file mode 100644
index 0000000..f8118d3
--- /dev/null
+++ b/kernel/trace/trace_workqueue.c
@@ -0,0 +1,287 @@
+/*
+ * Workqueue statistical tracer.
+ *
+ * Copyright (C) 2008 Frederic Weisbecker <fweisbec@gmail.com>
+ *
+ */
+
+
+#include <trace/workqueue.h>
+#include <linux/list.h>
+#include "trace_stat.h"
+#include "trace.h"
+
+
+/* A cpu workqueue thread */
+struct cpu_workqueue_stats {
+ struct list_head list;
+/* Useful to know if we print the cpu headers */
+ bool first_entry;
+ int cpu;
+ pid_t pid;
+/* Can be inserted from interrupt or user context, need to be atomic */
+ atomic_t inserted;
+/*
+ * Don't need to be atomic, works are serialized in a single workqueue thread
+ * on a single CPU.
+ */
+ unsigned int executed;
+};
+
+/* List of workqueue threads on one cpu */
+struct workqueue_global_stats {
+ struct list_head list;
+ spinlock_t lock;
+};
+
+/* Don't need a global lock because allocated before the workqueues, and
+ * never freed.
+ */
+static struct workqueue_global_stats *all_workqueue_stat;
+
+/* Insertion of a work */
+static void
+probe_workqueue_insertion(struct task_struct *wq_thread,
+ struct work_struct *work)
+{
+ int cpu = cpumask_first(&wq_thread->cpus_allowed);
+ struct cpu_workqueue_stats *node, *next;
+ unsigned long flags;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ list_for_each_entry_safe(node, next, &all_workqueue_stat[cpu].list,
+ list) {
+ if (node->pid == wq_thread->pid) {
+ atomic_inc(&node->inserted);
+ goto found;
+ }
+ }
+ pr_debug("trace_workqueue: entry not found\n");
+found:
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+}
+
+/* Execution of a work */
+static void
+probe_workqueue_execution(struct task_struct *wq_thread,
+ struct work_struct *work)
+{
+ int cpu = cpumask_first(&wq_thread->cpus_allowed);
+ struct cpu_workqueue_stats *node, *next;
+ unsigned long flags;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ list_for_each_entry_safe(node, next, &all_workqueue_stat[cpu].list,
+ list) {
+ if (node->pid == wq_thread->pid) {
+ node->executed++;
+ goto found;
+ }
+ }
+ pr_debug("trace_workqueue: entry not found\n");
+found:
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+}
+
+/* Creation of a cpu workqueue thread */
+static void probe_workqueue_creation(struct task_struct *wq_thread, int cpu)
+{
+ struct cpu_workqueue_stats *cws;
+ unsigned long flags;
+
+ WARN_ON(cpu < 0 || cpu >= num_possible_cpus());
+
+ /* Workqueues are sometimes created in atomic context */
+ cws = kzalloc(sizeof(struct cpu_workqueue_stats), GFP_ATOMIC);
+ if (!cws) {
+ pr_warning("trace_workqueue: not enough memory\n");
+ return;
+ }
+ tracing_record_cmdline(wq_thread);
+
+ INIT_LIST_HEAD(&cws->list);
+ cws->cpu = cpu;
+
+ cws->pid = wq_thread->pid;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ if (list_empty(&all_workqueue_stat[cpu].list))
+ cws->first_entry = true;
+ list_add_tail(&cws->list, &all_workqueue_stat[cpu].list);
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+}
+
+/* Destruction of a cpu workqueue thread */
+static void probe_workqueue_destruction(struct task_struct *wq_thread)
+{
+ /* Workqueue only execute on one cpu */
+ int cpu = cpumask_first(&wq_thread->cpus_allowed);
+ struct cpu_workqueue_stats *node, *next;
+ unsigned long flags;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ list_for_each_entry_safe(node, next, &all_workqueue_stat[cpu].list,
+ list) {
+ if (node->pid == wq_thread->pid) {
+ list_del(&node->list);
+ kfree(node);
+ goto found;
+ }
+ }
+
+ pr_debug("trace_workqueue: don't find workqueue to destroy\n");
+found:
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+
+}
+
+static struct cpu_workqueue_stats *workqueue_stat_start_cpu(int cpu)
+{
+ unsigned long flags;
+ struct cpu_workqueue_stats *ret = NULL;
+
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+
+ if (!list_empty(&all_workqueue_stat[cpu].list))
+ ret = list_entry(all_workqueue_stat[cpu].list.next,
+ struct cpu_workqueue_stats, list);
+
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+
+ return ret;
+}
+
+static void *workqueue_stat_start(void)
+{
+ int cpu;
+ void *ret = NULL;
+
+ for_each_possible_cpu(cpu) {
+ ret = workqueue_stat_start_cpu(cpu);
+ if (ret)
+ return ret;
+ }
+ return NULL;
+}
+
+static void *workqueue_stat_next(void *prev, int idx)
+{
+ struct cpu_workqueue_stats *prev_cws = prev;
+ int cpu = prev_cws->cpu;
+ unsigned long flags;
+ void *ret = NULL;
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ if (list_is_last(&prev_cws->list, &all_workqueue_stat[cpu].list)) {
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+ for (++cpu ; cpu < num_possible_cpus(); cpu++) {
+ ret = workqueue_stat_start_cpu(cpu);
+ if (ret)
+ return ret;
+ }
+ return NULL;
+ }
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+
+ return list_entry(prev_cws->list.next, struct cpu_workqueue_stats,
+ list);
+}
+
+static int workqueue_stat_show(struct seq_file *s, void *p)
+{
+ struct cpu_workqueue_stats *cws = p;
+ unsigned long flags;
+ int cpu = cws->cpu;
+
+ seq_printf(s, "%3d %6d %6u %s\n", cws->cpu,
+ atomic_read(&cws->inserted),
+ cws->executed,
+ trace_find_cmdline(cws->pid));
+
+ spin_lock_irqsave(&all_workqueue_stat[cpu].lock, flags);
+ if (&cws->list == all_workqueue_stat[cpu].list.next)
+ seq_printf(s, "\n");
+ spin_unlock_irqrestore(&all_workqueue_stat[cpu].lock, flags);
+
+ return 0;
+}
+
+static int workqueue_stat_headers(struct seq_file *s)
+{
+ seq_printf(s, "# CPU INSERTED EXECUTED NAME\n");
+ seq_printf(s, "# | | | |\n\n");
+ return 0;
+}
+
+struct tracer_stat workqueue_stats __read_mostly = {
+ .name = "workqueues",
+ .stat_start = workqueue_stat_start,
+ .stat_next = workqueue_stat_next,
+ .stat_show = workqueue_stat_show,
+ .stat_headers = workqueue_stat_headers
+};
+
+
+int __init stat_workqueue_init(void)
+{
+ if (register_stat_tracer(&workqueue_stats)) {
+ pr_warning("Unable to register workqueue stat tracer\n");
+ return 1;
+ }
+
+ return 0;
+}
+fs_initcall(stat_workqueue_init);
+
+/*
+ * Workqueues are created very early, just after pre-smp initcalls.
+ * So we must register our tracepoints at this stage.
+ */
+int __init trace_workqueue_early_init(void)
+{
+ int ret, cpu;
+
+ ret = register_trace_workqueue_insertion(probe_workqueue_insertion);
+ if (ret)
+ goto out;
+
+ ret = register_trace_workqueue_execution(probe_workqueue_execution);
+ if (ret)
+ goto no_insertion;
+
+ ret = register_trace_workqueue_creation(probe_workqueue_creation);
+ if (ret)
+ goto no_execution;
+
+ ret = register_trace_workqueue_destruction(probe_workqueue_destruction);
+ if (ret)
+ goto no_creation;
+
+ all_workqueue_stat = kmalloc(sizeof(struct workqueue_global_stats)
+ * num_possible_cpus(), GFP_KERNEL);
+
+ if (!all_workqueue_stat) {
+ pr_warning("trace_workqueue: not enough memory\n");
+ goto no_creation;
+ }
+
+ for_each_possible_cpu(cpu) {
+ spin_lock_init(&all_workqueue_stat[cpu].lock);
+ INIT_LIST_HEAD(&all_workqueue_stat[cpu].list);
+ }
+
+ return 0;
+
+no_creation:
+ unregister_trace_workqueue_creation(probe_workqueue_creation);
+no_execution:
+ unregister_trace_workqueue_execution(probe_workqueue_execution);
+no_insertion:
+ unregister_trace_workqueue_insertion(probe_workqueue_insertion);
+out:
+ pr_warning("trace_workqueue: unable to trace workqueues\n");
+
+ return 1;
+}
+early_initcall(trace_workqueue_early_init);
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 2f44583..1fc2bc2 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -33,6 +33,7 @@
#include <linux/kallsyms.h>
#include <linux/debug_locks.h>
#include <linux/lockdep.h>
+#include <trace/workqueue.h>
/*
* The per-CPU workqueue (if single thread, we always use the first
@@ -125,9 +126,13 @@ struct cpu_workqueue_struct *get_wq_data(struct work_struct *work)
return (void *) (atomic_long_read(&work->data) & WORK_STRUCT_WQ_DATA_MASK);
}
+DEFINE_TRACE(workqueue_insertion);
+
static void insert_work(struct cpu_workqueue_struct *cwq,
struct work_struct *work, struct list_head *head)
{
+ trace_workqueue_insertion(cwq->thread, work);
+
set_wq_data(work, cwq);
/*
* Ensure that we get the right work->data if we see the
@@ -259,6 +264,8 @@ int queue_delayed_work_on(int cpu, struct workqueue_struct *wq,
}
EXPORT_SYMBOL_GPL(queue_delayed_work_on);
+DEFINE_TRACE(workqueue_execution);
+
static void run_workqueue(struct cpu_workqueue_struct *cwq)
{
spin_lock_irq(&cwq->lock);
@@ -284,7 +291,7 @@ static void run_workqueue(struct cpu_workqueue_struct *cwq)
*/
struct lockdep_map lockdep_map = work->lockdep_map;
#endif
-
+ trace_workqueue_execution(cwq->thread, work);
cwq->current_work = work;
list_del_init(cwq->worklist.next);
spin_unlock_irq(&cwq->lock);
@@ -765,6 +772,8 @@ init_cpu_workqueue(struct workqueue_struct *wq, int cpu)
return cwq;
}
+DEFINE_TRACE(workqueue_creation);
+
static int create_workqueue_thread(struct cpu_workqueue_struct *cwq, int cpu)
{
struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
@@ -787,6 +796,8 @@ static int create_workqueue_thread(struct cpu_workqueue_struct *cwq, int cpu)
sched_setscheduler_nocheck(p, SCHED_FIFO, ¶m);
cwq->thread = p;
+ trace_workqueue_creation(cwq->thread, cpu);
+
return 0;
}
@@ -868,6 +879,8 @@ struct workqueue_struct *__create_workqueue_key(const char *name,
}
EXPORT_SYMBOL_GPL(__create_workqueue_key);
+DEFINE_TRACE(workqueue_destruction);
+
static void cleanup_workqueue_thread(struct cpu_workqueue_struct *cwq)
{
/*
@@ -891,6 +904,7 @@ static void cleanup_workqueue_thread(struct cpu_workqueue_struct *cwq)
* checks list_empty(), and a "normal" queue_work() can't use
* a dead CPU.
*/
+ trace_workqueue_destruction(cwq->thread);
kthread_stop(cwq->thread);
cwq->thread = NULL;
}
diff --git a/scripts/Makefile.build b/scripts/Makefile.build
index 5d90030..b5efa98 100644
--- a/scripts/Makefile.build
+++ b/scripts/Makefile.build
@@ -112,13 +112,13 @@ endif
# ---------------------------------------------------------------------------
# Default is built-in, unless we know otherwise
-modkern_cflags := $(CFLAGS_KERNEL)
+modkern_cflags = $(if $(part-of-module), $(CFLAGS_MODULE), $(CFLAGS_KERNEL))
quiet_modtag := $(empty) $(empty)
-$(real-objs-m) : modkern_cflags := $(CFLAGS_MODULE)
-$(real-objs-m:.o=.i) : modkern_cflags := $(CFLAGS_MODULE)
-$(real-objs-m:.o=.s) : modkern_cflags := $(CFLAGS_MODULE)
-$(real-objs-m:.o=.lst): modkern_cflags := $(CFLAGS_MODULE)
+$(real-objs-m) : part-of-module := y
+$(real-objs-m:.o=.i) : part-of-module := y
+$(real-objs-m:.o=.s) : part-of-module := y
+$(real-objs-m:.o=.lst): part-of-module := y
$(real-objs-m) : quiet_modtag := [M]
$(real-objs-m:.o=.i) : quiet_modtag := [M]
@@ -215,7 +215,8 @@ endif
ifdef CONFIG_FTRACE_MCOUNT_RECORD
cmd_record_mcount = perl $(srctree)/scripts/recordmcount.pl "$(ARCH)" \
"$(if $(CONFIG_64BIT),64,32)" \
- "$(OBJDUMP)" "$(OBJCOPY)" "$(CC)" "$(LD)" "$(NM)" "$(RM)" "$(MV)" "$(@)";
+ "$(OBJDUMP)" "$(OBJCOPY)" "$(CC)" "$(LD)" "$(NM)" "$(RM)" "$(MV)" \
+ "$(if $(part-of-module),1,0)" "$(@)";
endif
define rule_cc_o_c
diff --git a/scripts/recordmcount.pl b/scripts/recordmcount.pl
index fe83141..2ded5c8 100755
--- a/scripts/recordmcount.pl
+++ b/scripts/recordmcount.pl
@@ -100,14 +100,19 @@ $P =~ s@.*/@@g;
my $V = '0.1';
-if ($#ARGV < 6) {
- print "usage: $P arch objdump objcopy cc ld nm rm mv inputfile\n";
+if ($#ARGV < 7) {
+ print "usage: $P arch objdump objcopy cc ld nm rm mv is_module inputfile\n";
print "version: $V\n";
exit(1);
}
my ($arch, $bits, $objdump, $objcopy, $cc,
- $ld, $nm, $rm, $mv, $inputfile) = @ARGV;
+ $ld, $nm, $rm, $mv, $is_module, $inputfile) = @ARGV;
+
+# This file refers to mcount and shouldn't be ftraced, so lets' ignore it
+if ($inputfile eq "kernel/trace/ftrace.o") {
+ exit(0);
+}
# Acceptable sections to record.
my %text_sections = (
@@ -201,6 +206,13 @@ if ($arch eq "x86_64") {
$alignment = 2;
$section_type = '%progbits';
+} elsif ($arch eq "ia64") {
+ $mcount_regex = "^\\s*([0-9a-fA-F]+):.*\\s_mcount\$";
+ $type = "data8";
+
+ if ($is_module eq "0") {
+ $cc .= " -mconstant-gp";
+ }
} else {
die "Arch $arch is not supported with CONFIG_FTRACE_MCOUNT_RECORD";
}
^ permalink raw reply related [flat|nested] 15+ messages in thread
* Re: [PATCH 0/9] ftrace: updates for tip
2009-01-13 16:54 ` Steven Rostedt
@ 2009-01-14 11:12 ` Ingo Molnar
0 siblings, 0 replies; 15+ messages in thread
From: Ingo Molnar @ 2009-01-14 11:12 UTC (permalink / raw)
To: Steven Rostedt
Cc: LKML, Andrew Morton, Frederic Weisbecker, Shaohua Li, ia64,
Tony Luck, Sam Ravnborg, u.kleine-koenig
* Steven Rostedt <rostedt@goodmis.org> wrote:
>
> On Tue, 13 Jan 2009, Steven Rostedt wrote:
>
> >
> >
> > On Tue, 13 Jan 2009, Steven Rostedt wrote:
> >
> > >
> > > The following patches are in:
> > >
> > > git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
> > >
> > > branch: tip/devel
> >
> > Ingo,
> >
> > Don't pull any of this yet please. I'm going to rebase some of the patches
> > to fix the change logs (typos). I'll let you know when that's finished.
> >
>
> OK, I rebased in changing the change logs. Here's what I did:
>
> s/destinct/distinct/
> s/K?nig/Koenig/
>
> Here's the changes:
>
>
> The following patches are in:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
>
> branch: tip/devel
>
>
> Frederic Weisbecker (2):
> tracing/ftrace: separate events tracing and stats tracing engine
> tracing: add a new workqueue tracer
>
> Shaohua Li (6):
> ftrace, ia64: make recordmcount distinct module compile
> ftrace, ia64: explictly ignore a file in recordmcount.pl
> ftrace, ia64: Add macro for ftrace_caller
> ftrace, ia64: Add recordmcount for ia64
> ftrace, ia64: IA64 static ftrace support
> ftrace, ia64: IA64 dynamic ftrace support
>
> Uwe Kleine-Koenig (1):
> ftrace, trivial: fix typo "resgister" -> "register"
>
> ----
> arch/ia64/Kconfig | 3 +
> arch/ia64/include/asm/ftrace.h | 28 ++++
> arch/ia64/kernel/Makefile | 5 +
> arch/ia64/kernel/entry.S | 100 ++++++++++++++
> arch/ia64/kernel/ftrace.c | 206 ++++++++++++++++++++++++++++
> arch/ia64/kernel/ia64_ksyms.c | 6 +
> include/linux/ftrace.h | 4 +
> include/trace/workqueue.h | 25 ++++
> kernel/trace/Kconfig | 11 ++
> kernel/trace/Makefile | 1 +
> kernel/trace/ftrace.c | 14 +--
> kernel/trace/trace.c | 2 -
> kernel/trace/trace.h | 20 ---
> kernel/trace/trace_branch.c | 108 +++++++++-------
> kernel/trace/trace_stat.c | 191 +++++++++++---------------
> kernel/trace/trace_stat.h | 31 +++++
> kernel/trace/trace_workqueue.c | 287 ++++++++++++++++++++++++++++++++++++++++
> kernel/workqueue.c | 16 ++-
> scripts/Makefile.build | 13 +-
> scripts/recordmcount.pl | 18 ++-
> 20 files changed, 888 insertions(+), 201 deletions(-)
pulled into tip/tracing/ftrace, thanks Steve!
Ingo
^ permalink raw reply [flat|nested] 15+ messages in thread
end of thread, other threads:[~2009-01-14 11:13 UTC | newest]
Thread overview: 15+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-01-13 6:11 [PATCH 0/9] ftrace: updates for tip Steven Rostedt
2009-01-13 6:11 ` [PATCH 1/9] ftrace, ia64: make recordmcount destinct module compile Steven Rostedt
2009-01-13 6:11 ` [PATCH 2/9] ftrace, ia64: explictly ignore a file in recordmcount.pl Steven Rostedt
2009-01-13 6:11 ` [PATCH 3/9] ftrace, ia64: Add macro for ftrace_caller Steven Rostedt
2009-01-13 6:11 ` [PATCH 4/9] ftrace, ia64: Add recordmcount for ia64 Steven Rostedt
2009-01-13 6:11 ` [PATCH 5/9] ftrace, ia64: IA64 static ftrace support Steven Rostedt
2009-01-13 6:11 ` [PATCH 6/9] ftrace, ia64: IA64 dynamic " Steven Rostedt
2009-01-13 6:11 ` [PATCH 7/9] tracing/ftrace: separate events tracing and stats tracing engine Steven Rostedt
2009-01-13 6:11 ` [PATCH 8/9] tracing: add a new workqueue tracer Steven Rostedt
2009-01-13 6:11 ` [PATCH 9/9] ftrace, trivial: fix typo "resgister" -> "register" Steven Rostedt
2009-01-13 8:03 ` Uwe Kleine-König
2009-01-13 13:04 ` Steven Rostedt
2009-01-13 14:06 ` [PATCH 0/9] ftrace: updates for tip Steven Rostedt
2009-01-13 16:54 ` Steven Rostedt
2009-01-14 11:12 ` Ingo Molnar
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox