Skip to contentSkip to conversation

tracing: kprobe-based event tracer and x86 instruction decoder

PATCH In review Masami Hiramatsu 3 participants 23 messages Jul 1, 2009 – Jul 7, 2009

Subsystems (7)
Documentation x86 KVM Tracing +3 more subsystems
lore
Share & tools
lore
#0 lore raw

cover tracing: kprobe-based event tracer and x86 instruction decoder

Hi,

Here are the v10 patches. I just updated for the latest -tip and fixed typos and Kconfig dependency.

Here are the patches of kprobe-based event tracer for x86, version 10, which allows you to probe various kernel events through ftrace interface. The tracer supports per-probe filtering which allows you to set filters on each probe and shows formats of each probe. I think this is more generic integration with ftrace, especially event-tracer.

This patchset also includes x86(-64) instruction decoder which supports non-SSE/FP opcodes and includes x86 opcode map. The decoder is used for finding the instruction boundaries when inserting new kprobes. I think it will be possible to share this opcode map with KVM's decoder.
The decoder is tested when building kernel, the test compares the results of objdump and the decoder right after building vmlinux. You can enable that test by CONFIG_X86_DECODER_SELFTEST=y.

This series can be applied on the latest linux-2.6-tip tree.

This supports only x86(-32/-64) (but porting it on other arch just needs kprobes/kretprobes and register and stack access APIs).

This patchset includes following changes:
- Add x86 instruction decoder [1/7]
- Add x86 instruction decoder selftest [2/7]
- Check insertion point safety in kprobe [3/7]
- Cleanup fix_riprel() with insn decoder [4/7]
- Add arch-dep register and stack fetching functions [5/7]
- Add dynamic event_call support to ftrace [6/7]
- Add kprobe-based event tracer [7/7]

Enhancement ideas will be added after merging:
- Add profiling interface for each event.
- Make a stress test of kprobes on this tracer.
(see http://sources.redhat.com/ml/systemtap/2009-q2/msg01055.html)
- .init function tracing support.
- Support primitive types(long, ulong, int, uint, etc) for args.

Kprobe-based Event Tracer
=========================

Overview
--------
This tracer is similar to the events tracer which is based on Tracepoint infrastructure. Instead of Tracepoint, this tracer is based on kprobes(kprobe and kretprobe). It probes anywhere where kprobes can probe(this means, all functions body except for __kprobes functions).

Unlike the function tracer, this tracer can probe instructions inside of kernel functions. It allows you to check which instruction has been executed.

Unlike the Tracepoint based events tracer, this tracer can add new probe points on the fly.

Similar to the events tracer, this tracer doesn't need to be activated via current_tracer, instead of that, just set probe points via /sys/kernel/debug/tracing/kprobe_events. And you can set filters on each probe events via /sys/kernel/debug/tracing/events/kprobes/<EVENT>/filter.

Synopsis of kprobe_events
-------------------------
p[:EVENT] SYMBOL[+offs|-offs]|MEMADDR [FETCHARGS] : set a probe
r[:EVENT] SYMBOL[+0] [FETCHARGS] : set a return probe

EVENT : Event name
SYMBOL[+offs|-offs] : Symbol+offset where the probe is inserted
MEMADDR : Address where the probe is inserted

FETCHARGS : Arguments
%REG : Fetch register REG
sN : Fetch Nth entry of stack (N >= 0)
@ADDR : Fetch memory at ADDR (ADDR should be in kernel)
@SYM[+|-offs] : Fetch memory at SYM +|- offs (SYM should be a data symbol)
aN : Fetch function argument. (N >= 0)(*)
rv : Fetch return value.(**)
ra : Fetch return address.(**)
+|-offs(FETCHARG) : fetch memory at FETCHARG +|- offs address.(***)

(*) aN may not correct on asmlinkaged functions and at the middle of
function body.
(**) only for return probe.
(***) this is useful for fetching a field of data structures.

Per-Probe Event Filtering
-------------------------
Per-probe event filtering feature allows you to set different filter on each
probe and gives you what arguments will be shown in trace buffer. If an event name is specified right after 'p:' or 'r:' in kprobe_events, the tracer adds an event under tracing/events/kprobes/<EVENT>, at the directory you can see 'id', 'enabled', 'format' and 'filter'.

enabled:
You can enable/disable the probe by writing 1 or 0 on it.

format:
It shows the format of this probe event. It also shows aliases of arguments
which you specified to kprobe_events.

filter:
You can write filtering rules of this event. And you can use both of aliase
names and field names for describing filters.

Usage examples
--------------
To add a probe as a new event, write a new definition to kprobe_events as below.

echo p:myprobe do_sys_open a0 a1 a2 a3 > /sys/kernel/debug/tracing/kprobe_events

This sets a kprobe on the top of do_sys_open() function with recording
1st to 4th arguments as "myprobe" event.

echo r:myretprobe do_sys_open rv ra >> /sys/kernel/debug/tracing/kprobe_events

This sets a kretprobe on the return point of do_sys_open() function with
recording return value and return address as "myretprobe" event.
You can see the format of these events via
/sys/kernel/debug/tracing/events/kprobes/<EVENT>/format.

cat /sys/kernel/debug/tracing/events/kprobes/myprobe/format
name: myprobe
ID: 23
format:

	field:unsigned short common_type;	offset:0;	size:2;
	field:unsigned char common_flags;	offset:2;	size:1;
	field:unsigned char common_preempt_count;	offset:3;	size:1;
	field:int common_pid;	offset:4;	size:4;
	field:int common_tgid;	offset:8;	size:4;

	field: unsigned long ip;	offset:16;tsize:8;
	field: int nargs;	offset:24;tsize:4;
	field: unsigned long arg0;	offset:32;tsize:8;
	field: unsigned long arg1;	offset:40;tsize:8;
	field: unsigned long arg2;	offset:48;tsize:8;
	field: unsigned long arg3;	offset:56;tsize:8;

	alias: a0;	original: arg0;
	alias: a1;	original: arg1;
	alias: a2;	original: arg2;
	alias: a3;	original: arg3;

print fmt: "%lx: 0x%lx 0x%lx 0x%lx 0x%lx", ip, arg0, arg1, arg2, arg3

You can see that the event has 4 arguments and alias expressions
corresponding to it.

echo > /sys/kernel/debug/tracing/kprobe_events

This clears all probe points. and you can see the traced information via
/sys/kernel/debug/tracing/trace.

cat /sys/kernel/debug/tracing/trace
# tracer: nop
#
# TASK-PID CPU# TIMESTAMP FUNCTION
# | | | | |

           <...>-1447  [001] 1038282.286875: do_sys_open+0x0/0xd6: 0x3 0x7fffd1ec4440 0x8000 0x0
           <...>-1447  [001] 1038282.286878: sys_openat+0xc/0xe <- do_sys_open: 0xfffffffffffffffe 0xffffffff81367a3a
           <...>-1447  [001] 1038282.286885: do_sys_open+0x0/0xd6: 0xffffff9c 0x40413c 0x8000 0x1b6
           <...>-1447  [001] 1038282.286915: sys_open+0x1b/0x1d <- do_sys_open: 0x3 0xffffffff81367a3a
           <...>-1447  [001] 1038282.286969: do_sys_open+0x0/0xd6: 0xffffff9c 0x4041c6 0x98800 0x10
           <...>-1447  [001] 1038282.286976: sys_open+0x1b/0x1d <- do_sys_open: 0x3 0xffffffff81367a3a

Each line shows when the kernel hits a probe, and <- SYMBOL means kernel
returns from SYMBOL(e.g. "sys_open+0x1b/0x1d <- do_sys_open" means kernel returns from do_sys_open to sys_open+0x1b).

Thank you,

---

Masami Hiramatsu (7):
tracing: add kprobe-based event tracer
tracing: ftrace dynamic ftrace_event_call support
x86: add pt_regs register and stack access APIs
kprobes: cleanup fix_riprel() using insn decoder on x86
kprobes: checks probe address is instruction boudary on x86
x86: x86 instruction decoder build-time selftest
x86: instruction decoder API

 Documentation/trace/kprobes.txt        |  138 ++++
 arch/x86/Kconfig.debug                 |    9 
 arch/x86/Makefile                      |    3 
 arch/x86/include/asm/inat.h            |  127 +++
 arch/x86/include/asm/insn.h            |  136 ++++
 arch/x86/include/asm/ptrace.h          |  122 +++
 arch/x86/kernel/kprobes.c              |  197 ++---
 arch/x86/kernel/ptrace.c               |   73 ++
 arch/x86/lib/Makefile                  |   13 
 arch/x86/lib/inat.c                    |   82 ++
 arch/x86/lib/insn.c                    |  473 +++++++++++++
 arch/x86/lib/x86-opcode-map.txt        |  711 +++++++++++++++++++
 arch/x86/scripts/Makefile              |   19 +
 arch/x86/scripts/distill.awk           |   42 +
 arch/x86/scripts/gen-insn-attr-x86.awk |  314 ++++++++
 arch/x86/scripts/test_get_len.c        |   99 +++
 arch/x86/scripts/user_include.h        |   49 +
 include/linux/ftrace_event.h           |   13 
 include/trace/ftrace.h                 |   22 -
 kernel/trace/Kconfig                   |   12 
 kernel/trace/Makefile                  |    1 
 kernel/trace/trace.h                   |   22 +
 kernel/trace/trace_event_types.h       |   20 +
 kernel/trace/trace_events.c            |   70 +-
 kernel/trace/trace_export.c            |   27 -
 kernel/trace/trace_kprobe.c            | 1183 ++++++++++++++++++++++++++++++++
 26 files changed, 3825 insertions(+), 152 deletions(-)
 create mode 100644 Documentation/trace/kprobes.txt
 create mode 100644 arch/x86/include/asm/inat.h
 create mode 100644 arch/x86/include/asm/insn.h
 create mode 100644 arch/x86/lib/inat.c
 create mode 100644 arch/x86/lib/insn.c
 create mode 100644 arch/x86/lib/x86-opcode-map.txt
 create mode 100644 arch/x86/scripts/Makefile
 create mode 100644 arch/x86/scripts/distill.awk
 create mode 100644 arch/x86/scripts/gen-insn-attr-x86.awk
 create mode 100644 arch/x86/scripts/test_get_len.c
 create mode 100644 arch/x86/scripts/user_include.h
 create mode 100644 kernel/trace/trace_kprobe.c
Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#1 lore raw

PATCH 1/7 x86: instruction decoder API

Add x86 instruction decoder to arch-specific libraries. This decoder can decode x86 instructions used in kernel into prefix, opcode, modrm, sib, displacement and immediates. This can also show the length of instructions.

This version introduces instruction attributes for decoding instructions. The instruction attribute tables are generated from the opcode map file (x86-opcode-map.txt) by the generator script(gen-insn-attr-x86.awk).

Currently, the opcode maps are based on opcode maps in Intel(R) 64 and IA-32 Architectures Software Developers Manual Vol.2: Appendix.A, and consist of below two types of opcode tables.

1-byte/2-bytes/3-bytes opcodes, which has 256 elements, are written as below;

Table: table-name
Referrer: escaped-name
opcode: mnemonic|GrpXXX [operand1[,operand2...]] [(extra1)[,(extra2)...] [| 2nd-mnemonic ...]
(or)
opcode: escape # escaped-name
EndTable

Group opcodes, which has 8 elements, are written as below;

GrpTable: GrpXXX
reg: mnemonic [operand1[,operand2...]] [(extra1)[,(extra2)...] [| 2nd-mnemonic ...] EndTable

These opcode maps do NOT include most of SSE and FP opcodes, because those opcodes are not used in the kernel.

Cc Steven Rostedt <[email protected]>
Cc Ananth N Mavinakayanahalli <[email protected]>
Cc Srikar Dronamraju <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Andi Kleen <[email protected]>
Cc Vegard Nossum <[email protected]>
Cc Avi Kivity <[email protected]>
Cc Przemysław Pawełczyk <[email protected]>

---

 arch/x86/include/asm/inat.h            |  125 ++++++
 arch/x86/include/asm/insn.h            |  134 ++++++
 arch/x86/lib/Makefile                  |   13 +
 arch/x86/lib/inat.c                    |   80 ++++
 arch/x86/lib/insn.c                    |  471 +++++++++++++++++++++
 arch/x86/lib/x86-opcode-map.txt        |  711 ++++++++++++++++++++++++++++++++
 arch/x86/scripts/gen-insn-attr-x86.awk |  314 ++++++++++++++
 7 files changed, 1848 insertions(+), 0 deletions(-)
 create mode 100644 arch/x86/include/asm/inat.h
 create mode 100644 arch/x86/include/asm/insn.h
 create mode 100644 arch/x86/lib/inat.c
 create mode 100644 arch/x86/lib/insn.c
 create mode 100644 arch/x86/lib/x86-opcode-map.txt
 create mode 100644 arch/x86/scripts/gen-insn-attr-x86.awk
arch/x86/include/asm/inat.h new file +125 -0
new file mode 100644
index 0000000..01e079a
@@ -0,0 +1,125 @@
+#ifndef _ASM_INAT_INAT_H
+#define _ASM_INAT_INAT_H
+/*
+ * x86 instruction attributes
+ *
+ * Written by Masami Hiramatsu <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ */
+#include <linux/types.h>
+
+/* Instruction attributes */
+typedef u32 insn_attr_t;
+
+/*
+ * Internal bits. Don't use bitmasks directly, because these bits are
+ * unstable. You should add checking macros and use that macro in
+ * your code.
+ */
+
+#define INAT_OPCODE_TABLE_SIZE 256
+#define INAT_GROUP_TABLE_SIZE 8
+
+/* Legacy instruction prefixes */
+#define INAT_PFX_OPNDSZ	1	/* 0x66 */ /* LPFX1 */
+#define INAT_PFX_REPNE	2	/* 0xF2 */ /* LPFX2 */
+#define INAT_PFX_REPE	3	/* 0xF3 */ /* LPFX3 */
+#define INAT_PFX_LOCK	4	/* 0xF0 */
+#define INAT_PFX_CS	5	/* 0x2E */
+#define INAT_PFX_DS	6	/* 0x3E */
+#define INAT_PFX_ES	7	/* 0x26 */
+#define INAT_PFX_FS	8	/* 0x64 */
+#define INAT_PFX_GS	9	/* 0x65 */
+#define INAT_PFX_SS	10	/* 0x36 */
+#define INAT_PFX_ADDRSZ	11	/* 0x67 */
+
+#define INAT_LPREFIX_MAX	3
+
+/* Immediate size */
+#define INAT_IMM_BYTE		1
+#define INAT_IMM_WORD		2
+#define INAT_IMM_DWORD		3
+#define INAT_IMM_QWORD		4
+#define INAT_IMM_PTR		5
+#define INAT_IMM_VWORD32	6
+#define INAT_IMM_VWORD		7
+
+/* Legacy prefix */
+#define INAT_PFX_OFFS	0
+#define INAT_PFX_BITS	4
+#define INAT_PFX_MAX    ((1 << INAT_PFX_BITS) - 1)
+#define INAT_PFX_MASK	(INAT_PFX_MAX << INAT_PFX_OFFS)
+/* Escape opcodes */
+#define INAT_ESC_OFFS	(INAT_PFX_OFFS + INAT_PFX_BITS)
+#define INAT_ESC_BITS	2
+#define INAT_ESC_MAX	((1 << INAT_ESC_BITS) - 1)
+#define INAT_ESC_MASK	(INAT_ESC_MAX << INAT_ESC_OFFS)
+/* Group opcodes (1-16) */
+#define INAT_GRP_OFFS	(INAT_ESC_OFFS + INAT_ESC_BITS)
+#define INAT_GRP_BITS	5
+#define INAT_GRP_MAX	((1 << INAT_GRP_BITS) - 1)
+#define INAT_GRP_MASK	(INAT_GRP_MAX << INAT_GRP_OFFS)
+/* Immediates */
+#define INAT_IMM_OFFS	(INAT_GRP_OFFS + INAT_GRP_BITS)
+#define INAT_IMM_BITS	3
+#define INAT_IMM_MASK	(((1 << INAT_IMM_BITS) - 1) << INAT_IMM_OFFS)
+/* Flags */
+#define INAT_FLAG_OFFS	(INAT_IMM_OFFS + INAT_IMM_BITS)
+#define INAT_REXPFX	(1 << INAT_FLAG_OFFS)
+#define INAT_MODRM	(1 << (INAT_FLAG_OFFS + 1))
+#define INAT_FORCE64	(1 << (INAT_FLAG_OFFS + 2))
+#define INAT_ADDIMM	(1 << (INAT_FLAG_OFFS + 3))
+#define INAT_MOFFSET	(1 << (INAT_FLAG_OFFS + 4))
+#define INAT_VARIANT	(1 << (INAT_FLAG_OFFS + 5))
+
+/* Attribute search APIs */
+extern insn_attr_t inat_get_opcode_attribute(u8 opcode);
+extern insn_attr_t inat_get_escape_attribute(u8 opcode, u8 last_pfx,
+					     insn_attr_t esc_attr);
+extern insn_attr_t inat_get_group_attribute(u8 modrm, u8 last_pfx,
+					    insn_attr_t esc_attr);
+
+/* Attribute checking macros. Use these macros in your code */
+#define INAT_IS_PREFIX(attr)	(attr & INAT_PFX_MASK)
+#define INAT_IS_ADDRSZ(attr)	((attr & INAT_PFX_MASK) == INAT_PFX_ADDRSZ)
+#define INAT_IS_OPNDSZ(attr)	((attr & INAT_PFX_MASK) == INAT_PFX_OPNDSZ)
+#define INAT_LPREFIX_NUM(attr)	\
+	(((attr & INAT_PFX_MASK) > INAT_LPREFIX_MAX) ? 0 :\
+	 (attr & INAT_PFX_MASK))
+#define INAT_MAKE_PREFIX(pfx)	(pfx << INAT_PFX_OFFS)
+
+#define INAT_IS_ESCAPE(attr)	(attr & INAT_ESC_MASK)
+#define INAT_ESCAPE_NUM(attr)	((attr & INAT_ESC_MASK) >> INAT_ESC_OFFS)
+#define INAT_MAKE_ESCAPE(esc)	(esc << INAT_ESC_OFFS)
+
+#define INAT_IS_GROUP(attr)	(attr & INAT_GRP_MASK)
+#define INAT_GROUP_NUM(attr)	((attr & INAT_GRP_MASK) >> INAT_GRP_OFFS)
+#define INAT_GROUP_COMMON(attr)	(attr & ~INAT_GRP_MASK)
+#define INAT_MAKE_GROUP(grp)	((grp << INAT_GRP_OFFS) | INAT_MODRM)
+
+#define INAT_HAS_IMM(attr)	(attr & INAT_IMM_MASK)
+#define INAT_IMM_SIZE(attr)	((attr & INAT_IMM_MASK) >> INAT_IMM_OFFS)
+#define INAT_MAKE_IMM(imm)	(imm << INAT_IMM_OFFS)
+
+#define INAT_IS_REX_PREFIX(attr)	(attr & INAT_REXPFX)
+#define INAT_HAS_MODRM(attr)	(attr & INAT_MODRM)
+#define INAT_IS_FORCE64(attr)	(attr & INAT_FORCE64)
+#define INAT_HAS_ADDIMM(attr)	(attr & INAT_ADDIMM)
+#define INAT_HAS_MOFFSET(attr)	(attr & INAT_MOFFSET)
+#define INAT_HAS_VARIANT(attr)	(attr & INAT_VARIANT)
+
+#endif
arch/x86/include/asm/insn.h new file +134 -0

134 lines not on this page · raw message

arch/x86/lib/Makefile +13 -0
index f9d3563..ac4d666 100644
@@ -2,12 +2,25 @@
 # Makefile for x86 specific library files.
 #
 
+inat_tables_script = $(srctree)/arch/x86/scripts/gen-insn-attr-x86.awk
+inat_tables_maps = $(srctree)/arch/x86/lib/x86-opcode-map.txt
+quiet_cmd_inat_tables = GEN     $@
+      cmd_inat_tables = $(AWK) -f $(inat_tables_script) $(inat_tables_maps) > $@
+
+$(obj)/inat-tables.c: $(inat_tables_script) $(inat_tables_maps)
+	$(call cmd,inat_tables)
+
+$(obj)/inat.o: $(obj)/inat-tables.c
+
+clean-files := inat-tables.c
+
 obj-$(CONFIG_SMP) := msr.o
 
 lib-y := delay.o
 lib-y += thunk_$(BITS).o
 lib-y += usercopy_$(BITS).o getuser.o putuser.o
 lib-y += memcpy_$(BITS).o
+lib-y += insn.o inat.o
 
 ifeq ($(CONFIG_X86_32),y)
         lib-y += checksum_32.o
arch/x86/lib/inat.c new file +80 -0

80 lines not on this page · raw message

arch/x86/lib/insn.c new file +471 -0

471 lines not on this page · raw message

arch/x86/lib/x86-opcode-map.txt new file +711 -0

711 lines not on this page · raw message

arch/x86/scripts/gen-insn-attr-x86.awk new file +314 -0

314 lines not on this page · raw message

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#2 lore raw

PATCH 2/7 x86: x86 instruction decoder build-time selftest

In reply to Masami Hiramatsu

Add a user-space selftest of x86 instruction decoder at kernel build time. When CONFIG_X86_DECODER_SELFTEST=y, Kbuild builds a test harness of x86 instruction decoder and performs it after building vmlinux. The test compares the results of objdump and x86 instruction decoder code and check there are no differences.

Cc H. Peter Anvin <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc Ananth N Mavinakayanahalli <[email protected]>
Cc Srikar Dronamraju <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Andi Kleen <[email protected]>
Cc Vegard Nossum <[email protected]>
Cc Avi Kivity <[email protected]>
Cc Przemysław Pawełczyk <[email protected]>
Cc Sam Ravnborg <[email protected]>

---

 arch/x86/Kconfig.debug          |    9 ++++
 arch/x86/Makefile               |    3 +
 arch/x86/include/asm/inat.h     |    2 +
 arch/x86/include/asm/insn.h     |    2 +
 arch/x86/lib/inat.c             |    2 +
 arch/x86/lib/insn.c             |    2 +
 arch/x86/scripts/Makefile       |   19 +++++++
 arch/x86/scripts/distill.awk    |   42 +++++++++++++++++
 arch/x86/scripts/test_get_len.c |   99 +++++++++++++++++++++++++++++++++++++++
 arch/x86/scripts/user_include.h |   49 +++++++++++++++++++
 10 files changed, 229 insertions(+), 0 deletions(-)
 create mode 100644 arch/x86/scripts/Makefile
 create mode 100644 arch/x86/scripts/distill.awk
 create mode 100644 arch/x86/scripts/test_get_len.c
 create mode 100644 arch/x86/scripts/user_include.h
arch/x86/Kconfig.debug +9 -0
index d105f29..7d0b681 100644
@@ -186,6 +186,15 @@ config X86_DS_SELFTEST
 config HAVE_MMIOTRACE_SUPPORT
 	def_bool y
 
+config X86_DECODER_SELFTEST
+     bool "x86 instruction decoder selftest"
+     depends on DEBUG_KERNEL
+	---help---
+	 Perform x86 instruction decoder selftests at build time.
+	 This option is useful for checking the sanity of x86 instruction
+	 decoder code.
+	 If unsure, say "N".
+
 #
 # IO delay types:
 #
arch/x86/Makefile +3 -0
index 1b68659..7046556 100644
@@ -154,6 +154,9 @@ all: bzImage
 KBUILD_IMAGE := $(boot)/bzImage
 
 bzImage: vmlinux
+ifeq ($(CONFIG_X86_DECODER_SELFTEST),y)
+	$(Q)$(MAKE) $(build)=arch/x86/scripts posttest
+endif
 	$(Q)$(MAKE) $(build)=$(boot) $(KBUILD_IMAGE)
 	$(Q)mkdir -p $(objtree)/arch/$(UTS_MACHINE)/boot
 	$(Q)ln -fsn ../../x86/boot/bzImage $(objtree)/arch/$(UTS_MACHINE)/boot/$@
arch/x86/include/asm/inat.h +2 -0
index 01e079a..9090665 100644
@@ -20,7 +20,9 @@
  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  *
  */
+#ifdef __KERNEL__
 #include <linux/types.h>
+#endif
 
 /* Instruction attributes */
 typedef u32 insn_attr_t;
arch/x86/include/asm/insn.h +2 -0
index 5b50fa3..5736404 100644
@@ -20,7 +20,9 @@
  * Copyright (C) IBM Corporation, 2009
  */
 
+#ifdef __KERNEL__
 #include <linux/types.h>
+#endif
 /* insn_attr_t is defined in inat.h */
 #include <asm/inat.h>
 
arch/x86/lib/inat.c +2 -0
index d6a34be..564ecbd 100644
@@ -18,7 +18,9 @@
  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  *
  */
+#ifdef __KERNEL__
 #include <linux/module.h>
+#endif
 #include <asm/insn.h>
 
 /* Attribute tables are generated from opcode map */
arch/x86/lib/insn.c +2 -0
index 254c848..3b9451a 100644
@@ -18,8 +18,10 @@
  * Copyright (C) IBM Corporation, 2002, 2004, 2009
  */
 
+#ifdef __KERNEL__
 #include <linux/string.h>
 #include <linux/module.h>
+#endif
 #include <asm/inat.h>
 #include <asm/insn.h>
 
arch/x86/scripts/Makefile new file +19 -0
new file mode 100644
index 0000000..f08859e
@@ -0,0 +1,19 @@
+PHONY += posttest
+quiet_cmd_posttest = TEST    $@
+      cmd_posttest = objdump -d $(objtree)/vmlinux | awk -f $(srctree)/arch/x86/scripts/distill.awk | $(obj)/test_get_len
+
+posttest: $(obj)/test_get_len vmlinux
+	$(call cmd,posttest)
+
+test_get_len_SRC = $(srctree)/arch/x86/scripts/test_get_len.c $(srctree)/arch/x86/lib/insn.c $(srctree)/arch/x86/lib/inat.c
+test_get_len_INC = $(srctree)/arch/x86/include/asm/inat.h $(srctree)/arch/x86/include/asm/insn.h $(objtree)/arch/x86/lib/inat-tables.c
+
+quiet_cmd_test_get_len = CC      $@
+      cmd_test_get_len = $(CC) -Wall $(test_get_len_SRC) -I$(objtree)/arch/x86/lib/ -I$(srctree)/arch/x86/include -include $(srctree)/arch/x86/scripts/user_include.h -o $@
+
+
+$(obj)/test_get_len: $(test_get_len_SRC) $(test_get_len_INC)
+	$(call cmd,test_get_len)
+
+clean-files := test_get_len
+
arch/x86/scripts/distill.awk new file +42 -0

42 lines not on this page · raw message

arch/x86/scripts/test_get_len.c new file +99 -0

99 lines not on this page · raw message

arch/x86/scripts/user_include.h new file +49 -0

49 lines not on this page · raw message

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#3 lore raw

PATCH 3/7 kprobes: checks probe address is instruction boudary on x86

In reply to Masami Hiramatsu

Ensure safeness of inserting kprobes by checking whether the specified address is at the first byte of a instruction on x86.
This is done by decoding probed function from its head to the probe point.

---

 arch/x86/kernel/kprobes.c |   69 +++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 69 insertions(+), 0 deletions(-)
arch/x86/kernel/kprobes.c +69 -0
index b5b1848..5341842 100644
@@ -48,6 +48,7 @@
 #include <linux/preempt.h>
 #include <linux/module.h>
 #include <linux/kdebug.h>
+#include <linux/kallsyms.h>
 
 #include <asm/cacheflush.h>
 #include <asm/desc.h>
@@ -55,6 +56,7 @@
 #include <asm/uaccess.h>
 #include <asm/alternative.h>
 #include <asm/debugreg.h>
+#include <asm/insn.h>
 
 void jprobe_return_end(void);
 
@@ -245,6 +247,71 @@ retry:
 	}
 }
 
+/* Recover the probed instruction at addr for further analysis. */
+static int recover_probed_instruction(kprobe_opcode_t *buf, unsigned long addr)
+{
+	struct kprobe *kp;
+	kp = get_kprobe((void *)addr);
+	if (!kp)
+		return -EINVAL;
+
+	/*
+	 *  Basically, kp->ainsn.insn has an original instruction.
+	 *  However, RIP-relative instruction can not do single-stepping
+	 *  at different place, fix_riprel() tweaks the displacement of
+	 *  that instruction. In that case, we can't recover the instruction
+	 *  from the kp->ainsn.insn.
+	 *
+	 *  On the other hand, kp->opcode has a copy of the first byte of
+	 *  the probed instruction, which is overwritten by int3. And
+	 *  the instruction at kp->addr is not modified by kprobes except
+	 *  for the first byte, we can recover the original instruction
+	 *  from it and kp->opcode.
+	 */
+	memcpy(buf, kp->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
+	buf[0] = kp->opcode;
+	return 0;
+}
+
+/* Dummy buffers for kallsyms_lookup */
+static char __dummy_buf[KSYM_NAME_LEN];
+
+/* Check if paddr is at an instruction boundary */
+static int __kprobes can_probe(unsigned long paddr)
+{
+	int ret;
+	unsigned long addr, offset = 0;
+	struct insn insn;
+	kprobe_opcode_t buf[MAX_INSN_SIZE];
+
+	if (!kallsyms_lookup(paddr, NULL, &offset, NULL, __dummy_buf))
+		return 0;
+
+	/* Decode instructions */
+	addr = paddr - offset;
+	while (addr < paddr) {
+		kernel_insn_init(&insn, (void *)addr);
+		insn_get_opcode(&insn);
+
+		/* Check if the instruction has been modified. */
+		if (OPCODE1(&insn) == BREAKPOINT_INSTRUCTION) {
+			ret = recover_probed_instruction(buf, addr);
+			if (ret)
+				/*
+				 * Another debugging subsystem might insert
+				 * this breakpoint. In that case, we can't
+				 * recover it.
+				 */
+				return 0;
+			kernel_insn_init(&insn, buf);
+		}
+		insn_get_length(&insn);
+		addr += insn.length;
+	}
+
+	return (addr == paddr);
+}
+
 /*
  * Returns non-zero if opcode modifies the interrupt flag.
  */
@@ -360,6 +427,8 @@ static void __kprobes arch_copy_kprobe(struct kprobe *p)
 
 int __kprobes arch_prepare_kprobe(struct kprobe *p)
 {
+	if (!can_probe((unsigned long)p->addr))
+		return -EILSEQ;
 	/* insn: must be on special executable page on x86. */
 	p->ainsn.insn = get_insn_slot();
 	if (!p->ainsn.insn)
Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#4 lore raw

PATCH 4/7 kprobes: cleanup fix_riprel() using insn decoder on x86

In reply to Masami Hiramatsu

Cleanup fix_riprel() in arch/x86/kernel/kprobes.c by using x86 instruction decoder.

Cc Ananth N Mavinakayanahalli <[email protected]>
Cc Jim Keniston <[email protected]>
Cc Ingo Molnar <[email protected]>

---

 arch/x86/kernel/kprobes.c |  128 ++++++++-------------------------------------
 1 files changed, 23 insertions(+), 105 deletions(-)
arch/x86/kernel/kprobes.c +23 -105
index 5341842..b77e050 100644
@@ -109,50 +109,6 @@ static const u32 twobyte_is_boostable[256 / 32] = {
 	/*      -----------------------------------------------         */
 	/*      0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f          */
 };
-static const u32 onebyte_has_modrm[256 / 32] = {
-	/*      0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f          */
-	/*      -----------------------------------------------         */
-	W(0x00, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0) | /* 00 */
-	W(0x10, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0) , /* 10 */
-	W(0x20, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0) | /* 20 */
-	W(0x30, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0) , /* 30 */
-	W(0x40, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* 40 */
-	W(0x50, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 50 */
-	W(0x60, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0) | /* 60 */
-	W(0x70, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 70 */
-	W(0x80, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* 80 */
-	W(0x90, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 90 */
-	W(0xa0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* a0 */
-	W(0xb0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* b0 */
-	W(0xc0, 1, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0) | /* c0 */
-	W(0xd0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1) , /* d0 */
-	W(0xe0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* e0 */
-	W(0xf0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1)   /* f0 */
-	/*      -----------------------------------------------         */
-	/*      0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f          */
-};
-static const u32 twobyte_has_modrm[256 / 32] = {
-	/*      0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f          */
-	/*      -----------------------------------------------         */
-	W(0x00, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1) | /* 0f */
-	W(0x10, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0) , /* 1f */
-	W(0x20, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1) | /* 2f */
-	W(0x30, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 3f */
-	W(0x40, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* 4f */
-	W(0x50, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 5f */
-	W(0x60, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* 6f */
-	W(0x70, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1) , /* 7f */
-	W(0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* 8f */
-	W(0x90, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 9f */
-	W(0xa0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1) | /* af */
-	W(0xb0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1) , /* bf */
-	W(0xc0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0) | /* cf */
-	W(0xd0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* df */
-	W(0xe0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* ef */
-	W(0xf0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0)   /* ff */
-	/*      -----------------------------------------------         */
-	/*      0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f          */
-};
 #undef W
 
 struct kretprobe_blackpoint kretprobe_blacklist[] = {
@@ -345,68 +301,30 @@ static int __kprobes is_IF_modifier(kprobe_opcode_t *insn)
 static void __kprobes fix_riprel(struct kprobe *p)
 {
 #ifdef CONFIG_X86_64
-	u8 *insn = p->ainsn.insn;
-	s64 disp;
-	int need_modrm;
-
-	/* Skip legacy instruction prefixes.  */
-	while (1) {
-		switch (*insn) {
-		case 0x66:
-		case 0x67:
-		case 0x2e:
-		case 0x3e:
-		case 0x26:
-		case 0x64:
-		case 0x65:
-		case 0x36:
-		case 0xf0:
-		case 0xf3:
-		case 0xf2:
-			++insn;
-			continue;
-		}
-		break;
-	}
+	struct insn insn;
+	kernel_insn_init(&insn, p->ainsn.insn);
 
-	/* Skip REX instruction prefix.  */
-	if (is_REX_prefix(insn))
-		++insn;
-
-	if (*insn == 0x0f) {
-		/* Two-byte opcode.  */
-		++insn;
-		need_modrm = test_bit(*insn,
-				      (unsigned long *)twobyte_has_modrm);
-	} else
-		/* One-byte opcode.  */
-		need_modrm = test_bit(*insn,
-				      (unsigned long *)onebyte_has_modrm);
-
-	if (need_modrm) {
-		u8 modrm = *++insn;
-		if ((modrm & 0xc7) == 0x05) {
-			/* %rip+disp32 addressing mode */
-			/* Displacement follows ModRM byte.  */
-			++insn;
-			/*
-			 * The copied instruction uses the %rip-relative
-			 * addressing mode.  Adjust the displacement for the
-			 * difference between the original location of this
-			 * instruction and the location of the copy that will
-			 * actually be run.  The tricky bit here is making sure
-			 * that the sign extension happens correctly in this
-			 * calculation, since we need a signed 32-bit result to
-			 * be sign-extended to 64 bits when it's added to the
-			 * %rip value and yield the same 64-bit result that the
-			 * sign-extension of the original signed 32-bit
-			 * displacement would have given.
-			 */
-			disp = (u8 *) p->addr + *((s32 *) insn) -
-			       (u8 *) p->ainsn.insn;
-			BUG_ON((s64) (s32) disp != disp); /* Sanity check.  */
-			*(s32 *)insn = (s32) disp;
-		}
+	if (insn_rip_relative(&insn)) {
+		s64 newdisp;
+		u8 *disp;
+		insn_get_displacement(&insn);
+		/*
+		 * The copied instruction uses the %rip-relative addressing
+		 * mode.  Adjust the displacement for the difference between
+		 * the original location of this instruction and the location
+		 * of the copy that will actually be run.  The tricky bit here
+		 * is making sure that the sign extension happens correctly in
+		 * this calculation, since we need a signed 32-bit result to
+		 * be sign-extended to 64 bits when it's added to the %rip
+		 * value and yield the same 64-bit result that the sign-
+		 * extension of the original signed 32-bit displacement would
+		 * have given.
+		 */
+		newdisp = (u8 *) p->addr + (s64) insn.displacement.value -
+			  (u8 *) p->ainsn.insn;
+		BUG_ON((s64) (s32) newdisp != newdisp); /* Sanity check.  */
+		disp = (u8 *) p->ainsn.insn + INSN_DISPLACEMENT_OFFS(&insn);
+		*(s32 *) disp = (s32) newdisp;
 	}
 #endif
 }
Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#5 lore raw

PATCH 5/7 x86: add pt_regs register and stack access APIs

In reply to Masami Hiramatsu

Add following APIs for accessing registers and stack entries from pt_regs.

- regs_query_register_offset(const char *name)
Query the offset of "name" register.

- regs_query_register_name(unsigned offset)
Query the name of register by its offset.

- regs_get_register(struct pt_regs *regs, unsigned offset) Get the value of a register by its offset.

- regs_within_kernel_stack(struct pt_regs *regs, unsigned long addr) Check the address is in the kernel stack.

- regs_get_kernel_stack_nth(struct pt_regs *reg, unsigned nth) Get Nth entry of the kernel stack. (N >= 0)

- regs_get_argument_nth(struct pt_regs *reg, unsigned nth) Get Nth argument at function call. (N >= 0)

Changes from v9:
-Fix a typo in a comment.

Cc Christoph Hellwig <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc Ananth N Mavinakayanahalli <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Roland McGrath <[email protected]>
Cc Srikar Dronamraju <[email protected]>

---

 arch/x86/include/asm/ptrace.h |  122 +++++++++++++++++++++++++++++++++++++++++
 arch/x86/kernel/ptrace.c      |   73 +++++++++++++++++++++++++
 2 files changed, 195 insertions(+), 0 deletions(-)
arch/x86/include/asm/ptrace.h +122 -0
index 0f0d908..d5e3b3b 100644
@@ -7,6 +7,7 @@
 
 #ifdef __KERNEL__
 #include <asm/segment.h>
+#include <asm/page_types.h>
 #endif
 
 #ifndef __ASSEMBLY__
@@ -216,6 +217,127 @@ static inline unsigned long user_stack_pointer(struct pt_regs *regs)
 	return regs->sp;
 }
 
+/* Query offset/name of register from its name/offset */
+extern int regs_query_register_offset(const char *name);
+extern const char *regs_query_register_name(unsigned offset);
+#define MAX_REG_OFFSET (offsetof(struct pt_regs, ss))
+
+/**
+ * regs_get_register() - get register value from its offset
+ * @regs:	pt_regs from which register value is gotten.
+ * @offset:	offset number of the register.
+ *
+ * regs_get_register returns the value of a register whose offset from @regs
+ * is @offset. The @offset is the offset of the register in struct pt_regs.
+ * If @offset is bigger than MAX_REG_OFFSET, this returns 0.
+ */
+static inline unsigned long regs_get_register(struct pt_regs *regs,
+					      unsigned offset)
+{
+	if (unlikely(offset > MAX_REG_OFFSET))
+		return 0;
+	return *(unsigned long *)((unsigned long)regs + offset);
+}
+
+/**
+ * regs_within_kernel_stack() - check the address in the stack
+ * @regs:	pt_regs which contains kernel stack pointer.
+ * @addr:	address which is checked.
+ *
+ * regs_within_kenel_stack() checks @addr is within the kernel stack page(s).
+ * If @addr is within the kernel stack, it returns true. If not, returns false.
+ */
+static inline int regs_within_kernel_stack(struct pt_regs *regs,
+					   unsigned long addr)
+{
+	return ((addr & ~(THREAD_SIZE - 1))  ==
+		(kernel_stack_pointer(regs) & ~(THREAD_SIZE - 1)));
+}
+
+/**
+ * regs_get_kernel_stack_nth() - get Nth entry of the stack
+ * @regs:	pt_regs which contains kernel stack pointer.
+ * @n:		stack entry number.
+ *
+ * regs_get_kernel_stack_nth() returns @n th entry of the kernel stack which
+ * is specifined by @regs. If the @n th entry is NOT in the kernel stack,
+ * this returns 0.
+ */
+static inline unsigned long regs_get_kernel_stack_nth(struct pt_regs *regs,
+						      unsigned n)
+{
+	unsigned long *addr = (unsigned long *)kernel_stack_pointer(regs);
+	addr += n;
+	if (regs_within_kernel_stack(regs, (unsigned long)addr))
+		return *addr;
+	else
+		return 0;
+}
+
+/**
+ * regs_get_argument_nth() - get Nth argument at function call
+ * @regs:	pt_regs which contains registers at function entry.
+ * @n:		argument number.
+ *
+ * regs_get_argument_nth() returns @n th argument of a function call.
+ * Since usually the kernel stack will be changed right after function entry,
+ * you must use this at function entry. If the @n th entry is NOT in the
+ * kernel stack or pt_regs, this returns 0.
+ */
+#ifdef CONFIG_X86_32
+#define NR_REGPARMS 3
+static inline unsigned long regs_get_argument_nth(struct pt_regs *regs,
+						  unsigned n)
+{
+	if (n < NR_REGPARMS) {
+		switch (n) {
+		case 0:
+			return regs->ax;
+		case 1:
+			return regs->dx;
+		case 2:
+			return regs->cx;
+		}
+		return 0;
+	} else {
+		/*
+		 * The typical case: arg n is on the stack.
+		 * (Note: stack[0] = return address, so skip it)
+		 */
+		return regs_get_kernel_stack_nth(regs, 1 + n - NR_REGPARMS);
+	}
+}
+#else /* CONFIG_X86_64 */
+#define NR_REGPARMS 6
+static inline unsigned long regs_get_argument_nth(struct pt_regs *regs,
+						  unsigned n)
+{
+	if (n < NR_REGPARMS) {
+		switch (n) {
+		case 0:
+			return regs->di;
+		case 1:
+			return regs->si;
+		case 2:
+			return regs->dx;
+		case 3:
+			return regs->cx;
+		case 4:
+			return regs->r8;
+		case 5:
+			return regs->r9;
+		}
+		return 0;
+	} else {
+		/*
+		 * The typical case: arg n is on the stack.
+		 * (Note: stack[0] = return address, so skip it)
+		 */
+		return regs_get_kernel_stack_nth(regs, 1 + n - NR_REGPARMS);
+	}
+}
+#endif
+
 /*
  * These are defined as per linux/ptrace.h, which see.
  */
arch/x86/kernel/ptrace.c +73 -0

79 lines not on this page · raw message

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
Reviewed-by on v1 5/7
#6 lore raw

Re: [PATCH -tip -v10 5/7] x86: add pt_regs register and stack access APIs

Quoted 23 lines · Masami Hiramatsu

Add following APIs for accessing registers and stack entries from pt_regs.

- regs_query_register_offset(const char *name)
Query the offset of "name" register.

- regs_query_register_name(unsigned offset)
Query the name of register by its offset.

- regs_get_register(struct pt_regs *regs, unsigned offset) Get the value of a register by its offset.

- regs_within_kernel_stack(struct pt_regs *regs, unsigned long addr) Check the address is in the kernel stack.

- regs_get_kernel_stack_nth(struct pt_regs *reg, unsigned nth) Get Nth entry of the kernel stack. (N >= 0)

- regs_get_argument_nth(struct pt_regs *reg, unsigned nth) Get Nth argument at function call. (N >= 0)

Changes from v9:
-Fix a typo in a comment.

Quoted 9 lines
Cc Christoph Hellwig <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc Ananth N Mavinakayanahalli <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Roland McGrath <[email protected]>
Cc Srikar Dronamraju <[email protected]>

Looks good!

Frederic.

Quoted 240 lines

---

 arch/x86/include/asm/ptrace.h |  122 +++++++++++++++++++++++++++++++++++++++++
 arch/x86/kernel/ptrace.c      |   73 +++++++++++++++++++++++++
 2 files changed, 195 insertions(+), 0 deletions(-)
arch/x86/include/asm/ptrace.h +122 -0
index 0f0d908..d5e3b3b 100644
@@ -7,6 +7,7 @@
 
 #ifdef __KERNEL__
 #include <asm/segment.h>
+#include <asm/page_types.h>
 #endif
 
 #ifndef __ASSEMBLY__
@@ -216,6 +217,127 @@ static inline unsigned long user_stack_pointer(struct pt_regs *regs)
 	return regs->sp;
 }
 
+/* Query offset/name of register from its name/offset */
+extern int regs_query_register_offset(const char *name);
+extern const char *regs_query_register_name(unsigned offset);
+#define MAX_REG_OFFSET (offsetof(struct pt_regs, ss))
+
+/**
+ * regs_get_register() - get register value from its offset
+ * @regs:	pt_regs from which register value is gotten.
+ * @offset:	offset number of the register.
+ *
+ * regs_get_register returns the value of a register whose offset from @regs
+ * is @offset. The @offset is the offset of the register in struct pt_regs.
+ * If @offset is bigger than MAX_REG_OFFSET, this returns 0.
+ */
+static inline unsigned long regs_get_register(struct pt_regs *regs,
+					      unsigned offset)
+{
+	if (unlikely(offset > MAX_REG_OFFSET))
+		return 0;
+	return *(unsigned long *)((unsigned long)regs + offset);
+}
+
+/**
+ * regs_within_kernel_stack() - check the address in the stack
+ * @regs:	pt_regs which contains kernel stack pointer.
+ * @addr:	address which is checked.
+ *
+ * regs_within_kenel_stack() checks @addr is within the kernel stack page(s).
+ * If @addr is within the kernel stack, it returns true. If not, returns false.
+ */
+static inline int regs_within_kernel_stack(struct pt_regs *regs,
+					   unsigned long addr)
+{
+	return ((addr & ~(THREAD_SIZE - 1))  ==
+		(kernel_stack_pointer(regs) & ~(THREAD_SIZE - 1)));
+}
+
+/**
+ * regs_get_kernel_stack_nth() - get Nth entry of the stack
+ * @regs:	pt_regs which contains kernel stack pointer.
+ * @n:		stack entry number.
+ *
+ * regs_get_kernel_stack_nth() returns @n th entry of the kernel stack which
+ * is specifined by @regs. If the @n th entry is NOT in the kernel stack,
+ * this returns 0.
+ */
+static inline unsigned long regs_get_kernel_stack_nth(struct pt_regs *regs,
+						      unsigned n)
+{
+	unsigned long *addr = (unsigned long *)kernel_stack_pointer(regs);
+	addr += n;
+	if (regs_within_kernel_stack(regs, (unsigned long)addr))
+		return *addr;
+	else
+		return 0;
+}
+
+/**
+ * regs_get_argument_nth() - get Nth argument at function call
+ * @regs:	pt_regs which contains registers at function entry.
+ * @n:		argument number.
+ *
+ * regs_get_argument_nth() returns @n th argument of a function call.
+ * Since usually the kernel stack will be changed right after function entry,
+ * you must use this at function entry. If the @n th entry is NOT in the
+ * kernel stack or pt_regs, this returns 0.
+ */
+#ifdef CONFIG_X86_32
+#define NR_REGPARMS 3
+static inline unsigned long regs_get_argument_nth(struct pt_regs *regs,
+						  unsigned n)
+{
+	if (n < NR_REGPARMS) {
+		switch (n) {
+		case 0:
+			return regs->ax;
+		case 1:
+			return regs->dx;
+		case 2:
+			return regs->cx;
+		}
+		return 0;
+	} else {
+		/*
+		 * The typical case: arg n is on the stack.
+		 * (Note: stack[0] = return address, so skip it)
+		 */
+		return regs_get_kernel_stack_nth(regs, 1 + n - NR_REGPARMS);
+	}
+}
+#else /* CONFIG_X86_64 */
+#define NR_REGPARMS 6
+static inline unsigned long regs_get_argument_nth(struct pt_regs *regs,
+						  unsigned n)
+{
+	if (n < NR_REGPARMS) {
+		switch (n) {
+		case 0:
+			return regs->di;
+		case 1:
+			return regs->si;
+		case 2:
+			return regs->dx;
+		case 3:
+			return regs->cx;
+		case 4:
+			return regs->r8;
+		case 5:
+			return regs->r9;
+		}
+		return 0;
+	} else {
+		/*
+		 * The typical case: arg n is on the stack.
+		 * (Note: stack[0] = return address, so skip it)
+		 */
+		return regs_get_kernel_stack_nth(regs, 1 + n - NR_REGPARMS);
+	}
+}
+#endif
+
 /*
  * These are defined as per linux/ptrace.h, which see.
  */
arch/x86/kernel/ptrace.c +73 -0
index b457f78..2944d3a 100644
@@ -49,6 +49,79 @@ enum x86_regset {
 	REGSET_IOPERM32,
 };
 
+struct pt_regs_offset {
+	const char *name;
+	int offset;
+};
+
+#define REG_OFFSET_NAME(r) {.name = #r, .offset = offsetof(struct pt_regs, r)}
+#define REG_OFFSET_END {.name = NULL, .offset = 0}
+
+static const struct pt_regs_offset regoffset_table[] = {
+#ifdef CONFIG_X86_64
+	REG_OFFSET_NAME(r15),
+	REG_OFFSET_NAME(r14),
+	REG_OFFSET_NAME(r13),
+	REG_OFFSET_NAME(r12),
+	REG_OFFSET_NAME(r11),
+	REG_OFFSET_NAME(r10),
+	REG_OFFSET_NAME(r9),
+	REG_OFFSET_NAME(r8),
+#endif
+	REG_OFFSET_NAME(bx),
+	REG_OFFSET_NAME(cx),
+	REG_OFFSET_NAME(dx),
+	REG_OFFSET_NAME(si),
+	REG_OFFSET_NAME(di),
+	REG_OFFSET_NAME(bp),
+	REG_OFFSET_NAME(ax),
+#ifdef CONFIG_X86_32
+	REG_OFFSET_NAME(ds),
+	REG_OFFSET_NAME(es),
+	REG_OFFSET_NAME(fs),
+	REG_OFFSET_NAME(gs),
+#endif
+	REG_OFFSET_NAME(orig_ax),
+	REG_OFFSET_NAME(ip),
+	REG_OFFSET_NAME(cs),
+	REG_OFFSET_NAME(flags),
+	REG_OFFSET_NAME(sp),
+	REG_OFFSET_NAME(ss),
+	REG_OFFSET_END,
+};
+
+/**
+ * regs_query_register_offset() - query register offset from its name
+ * @name:	the name of a register
+ *
+ * regs_query_register_offset() returns the offset of a register in struct
+ * pt_regs from its name. If the name is invalid, this returns -EINVAL;
+ */
+int regs_query_register_offset(const char *name)
+{
+	const struct pt_regs_offset *roff;
+	for (roff = regoffset_table; roff->name != NULL; roff++)
+		if (!strcmp(roff->name, name))
+			return roff->offset;
+	return -EINVAL;
+}
+
+/**
+ * regs_query_register_name() - query register name from its offset
+ * @offset:	the offset of a register in struct pt_regs.
+ *
+ * regs_query_register_name() returns the name of a register from its
+ * offset in struct pt_regs. If the @offset is invalid, this returns NULL;
+ */
+const char *regs_query_register_name(unsigned offset)
+{
+	const struct pt_regs_offset *roff;
+	for (roff = regoffset_table; roff->name != NULL; roff++)
+		if (roff->offset == offset)
+			return roff->name;
+	return NULL;
+}
+
 /*
  * does not yet catch signals sent when the child dies.
  * in exit.c or in signal.c.
Signature · 8 lines
-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#7 lore raw

Re: [PATCH -tip -v10 5/7] x86: add pt_regs register and stack access APIs

In reply to Masami Hiramatsu
Quoted 1 line · Masami Hiramatsu

Add following APIs for accessing registers and stack entries from pt_regs.

You forgot to state who calls these functions/why are they added? Who only has strings for registers?

I can see the point of having a function for nth argument though, that's useful.

arch/x86/include/asm/ptrace.h static inline unsigned long user_stack_pointer(struct pt_regs *regs) · line 299

Quoted 11 lines
arch/x86/include/asm/ptrace.h +11 -0
+static inline unsigned long regs_get_argument_nth(struct pt_regs *regs,
+						  unsigned n)
+{
+	if (n < NR_REGPARMS) {
+		switch (n) {
+		case 0:
+			return regs->ax;
+		case 1:
+			return regs->dx;
+		case 2:
+			return regs->cx;

[....]

That could be done shorter with a offsetof table.

Quoted 4 lines
arch/x86/include/asm/ptrace.h +4 -0
+	if (n < NR_REGPARMS) {
+		switch (n) {
+		case 0:
+			return regs->di;

arch/x86/include/asm/ptrace.h static inline unsigned long user_stack_pointer(struct pt_regs *regs) · line 328

Quoted 10 lines
arch/x86/include/asm/ptrace.h +10 -0
+		case 1:
+			return regs->si;
+		case 2:
+			return regs->dx;
+		case 3:
+			return regs->cx;
+		case 4:
+			return regs->r8;
+		case 5:
+			return regs->r9;

and that too.

-Andi

Signature · 1 line
[email protected] -- Speaking for myself only.
#8 lore raw

Re: [PATCH -tip -v10 5/7] x86: add pt_regs register and stack access APIs

Quoted 6 lines · Andi Kleen
Quoted 1 line · Masami Hiramatsu

Add following APIs for accessing registers and stack entries from pt_regs.

You forgot to state who calls these functions/why are they added? Who only has strings for registers?

Oh, yes. This patch is needed for kprobes based event tracer on ftrace. Some other debugging tools might be able to use it.

Quoted 20 lines

I can see the point of having a function for nth argument though, that's useful.

arch/x86/include/asm/ptrace.h static inline unsigned long user_stack_pointer(struct pt_regs *regs) · line 299

Quoted 11 lines
arch/x86/include/asm/ptrace.h +11 -0
+static inline unsigned long regs_get_argument_nth(struct pt_regs *regs,
+						  unsigned n)
+{
+	if (n < NR_REGPARMS) {
+		switch (n) {
+		case 0:
+			return regs->ax;
+		case 1:
+			return regs->dx;
+		case 2:
+			return regs->cx;

[....]

That could be done shorter with a offsetof table.

Quoted 16 lines
Quoted 4 lines
arch/x86/include/asm/ptrace.h +4 -0
+	if (n < NR_REGPARMS) {
+		switch (n) {
+		case 0:
+			return regs->di;

arch/x86/include/asm/ptrace.h static inline unsigned long user_stack_pointer(struct pt_regs *regs) · line 328

Quoted 10 lines
arch/x86/include/asm/ptrace.h +10 -0
+		case 1:
+			return regs->si;
+		case 2:
+			return regs->dx;
+		case 3:
+			return regs->cx;
+		case 4:
+			return regs->r8;
+		case 5:
+			return regs->r9;

and that too.

I'm not so sure about your idea.
Would you mean below code?

int offs_table[NR_REGPARMS] = {
	[0] = offsetof(struct pt_regs, di),
	...
};
if (n < NR_REGPARMS)
	return *((unsigned long *)regs + offs_table[n]);

Thank you,

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#9 lore raw

Re: [PATCH -tip -v10 5/7] x86: add pt_regs register and stack access APIs

Quoted 4 lines · Masami Hiramatsu

I'm not so sure about your idea.
Would you mean below code?

int offs_table[NR_REGPARMS] = {

not REGPARMS of course

Quoted 5 lines
	[0] = offsetof(struct pt_regs, di),
	...
};
if (n < NR_REGPARMS)
	return *((unsigned long *)regs + offs_table[n]);

Yes.

-Andi

Signature · 1 line
[email protected] -- Speaking for myself only.
#10 lore raw

PATCH 5/7 x86: add pt_regs register and stack access APIs

Quoted 8 lines · Andi Kleen
Quoted 4 lines · Masami Hiramatsu

I'm not so sure about your idea.
Would you mean below code?

int offs_table[NR_REGPARMS] = {

not REGPARMS of course

Quoted 7 lines
Quoted 5 lines
	[0] = offsetof(struct pt_regs, di),
	...
};
if (n < NR_REGPARMS)
	return *((unsigned long *)regs + offs_table[n]);

Yes.

OK, here, I updated my patch.

Thank you,

x86: add pt_regs register and stack access APIs

From: Masami Hiramatsu <[email protected]>

Add following APIs for accessing registers and stack entries from pt_regs. These APIs are required by kprobes-based event tracer on ftrace. Some other debugging tools might be able to use it too.

- regs_query_register_offset(const char *name)
Query the offset of "name" register.

- regs_query_register_name(unsigned offset)
Query the name of register by its offset.

- regs_get_register(struct pt_regs *regs, unsigned offset) Get the value of a register by its offset.

- regs_within_kernel_stack(struct pt_regs *regs, unsigned long addr) Check the address is in the kernel stack.

- regs_get_kernel_stack_nth(struct pt_regs *reg, unsigned nth) Get Nth entry of the kernel stack. (N >= 0)

- regs_get_argument_nth(struct pt_regs *reg, unsigned nth) Get Nth argument at function call. (N >= 0)

Changes from v10:
- Use an offsetof table in regs_get_argument_nth().

Cc Andi Kleen <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc Ananth N Mavinakayanahalli <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Roland McGrath <[email protected]>
Cc Srikar Dronamraju <[email protected]>

---

 arch/x86/include/asm/ptrace.h |   61 ++++++++++++++++++++++
 arch/x86/kernel/ptrace.c      |  112 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 173 insertions(+), 0 deletions(-)
arch/x86/include/asm/ptrace.h +61 -0
index 0f0d908..a9b7e2d 100644
@@ -7,6 +7,7 @@
 
 #ifdef __KERNEL__
 #include <asm/segment.h>
+#include <asm/page_types.h>
 #endif
 
 #ifndef __ASSEMBLY__
@@ -216,6 +217,66 @@ static inline unsigned long user_stack_pointer(struct pt_regs *regs)
 	return regs->sp;
 }
 
+/* Query offset/name of register from its name/offset */
+extern int regs_query_register_offset(const char *name);
+extern const char *regs_query_register_name(unsigned offset);
+#define MAX_REG_OFFSET (offsetof(struct pt_regs, ss))
+
+/**
+ * regs_get_register() - get register value from its offset
+ * @regs:	pt_regs from which register value is gotten.
+ * @offset:	offset number of the register.
+ *
+ * regs_get_register returns the value of a register whose offset from @regs
+ * is @offset. The @offset is the offset of the register in struct pt_regs.
+ * If @offset is bigger than MAX_REG_OFFSET, this returns 0.
+ */
+static inline unsigned long regs_get_register(struct pt_regs *regs,
+					      unsigned offset)
+{
+	if (unlikely(offset > MAX_REG_OFFSET))
+		return 0;
+	return *(unsigned long *)((unsigned long)regs + offset);
+}
+
+/**
+ * regs_within_kernel_stack() - check the address in the stack
+ * @regs:	pt_regs which contains kernel stack pointer.
+ * @addr:	address which is checked.
+ *
+ * regs_within_kenel_stack() checks @addr is within the kernel stack page(s).
+ * If @addr is within the kernel stack, it returns true. If not, returns false.
+ */
+static inline int regs_within_kernel_stack(struct pt_regs *regs,
+					   unsigned long addr)
+{
+	return ((addr & ~(THREAD_SIZE - 1))  ==
+		(kernel_stack_pointer(regs) & ~(THREAD_SIZE - 1)));
+}
+
+/**
+ * regs_get_kernel_stack_nth() - get Nth entry of the stack
+ * @regs:	pt_regs which contains kernel stack pointer.
+ * @n:		stack entry number.
+ *
+ * regs_get_kernel_stack_nth() returns @n th entry of the kernel stack which
+ * is specifined by @regs. If the @n th entry is NOT in the kernel stack,
+ * this returns 0.
+ */
+static inline unsigned long regs_get_kernel_stack_nth(struct pt_regs *regs,
+						      unsigned n)
+{
+	unsigned long *addr = (unsigned long *)kernel_stack_pointer(regs);
+	addr += n;
+	if (regs_within_kernel_stack(regs, (unsigned long)addr))
+		return *addr;
+	else
+		return 0;
+}
+
+/* Get Nth argument at function call */
+extern unsigned long regs_get_argument_nth(struct pt_regs *regs, unsigned n);
+
 /*
  * These are defined as per linux/ptrace.h, which see.
  */
arch/x86/kernel/ptrace.c +112 -0

118 lines not on this page · raw message

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#11 lore raw

PATCH 6/7 tracing: ftrace dynamic ftrace_event_call support

In reply to Masami Hiramatsu

Add dynamic ftrace_event_call support to ftrace. Trace engines can adds new ftrace_event_call to ftrace on the fly. Each operator functions of the call takes a ftrace_event_call data structure as an argument, because these functions may be shared among several ftrace_event_calls.

Cc Steven Rostedt <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Tom Zanussi <[email protected]>
Cc Frederic Weisbecker <[email protected]>

---

 include/linux/ftrace_event.h |   13 +++++---
 include/trace/ftrace.h       |   22 +++++++------
 kernel/trace/trace_events.c  |   70 ++++++++++++++++++++++++++++++++----------
 kernel/trace/trace_export.c  |   27 ++++++++--------
 4 files changed, 85 insertions(+), 47 deletions(-)
include/linux/ftrace_event.h +8 -5
index 5c093ff..f7733b6 100644
@@ -108,12 +108,13 @@ struct ftrace_event_call {
 	struct dentry		*dir;
 	struct trace_event	*event;
 	int			enabled;
-	int			(*regfunc)(void);
-	void			(*unregfunc)(void);
+	int			(*regfunc)(struct ftrace_event_call *);
+	void			(*unregfunc)(struct ftrace_event_call *);
 	int			id;
-	int			(*raw_init)(void);
-	int			(*show_format)(struct trace_seq *s);
-	int			(*define_fields)(void);
+	int			(*raw_init)(struct ftrace_event_call *);
+	int			(*show_format)(struct ftrace_event_call *,
+					       struct trace_seq *);
+	int			(*define_fields)(struct ftrace_event_call *);
 	struct list_head	fields;
 	int			filter_active;
 	void			*filter;
@@ -138,6 +139,8 @@ extern int filter_current_check_discard(struct ftrace_event_call *call,
 
 extern int trace_define_field(struct ftrace_event_call *call, char *type,
 			      char *name, int offset, int size, int is_signed);
+extern int trace_add_event_call(struct ftrace_event_call *call);
+extern void trace_remove_event_call(struct ftrace_event_call *call);
 
 #define is_signed_type(type)	(((type)(-1)) < 0)
 
include/trace/ftrace.h +11 -11
index 1867553..d696580 100644
@@ -147,7 +147,8 @@
 #undef TRACE_EVENT
 #define TRACE_EVENT(call, proto, args, tstruct, func, print)		\
 static int								\
-ftrace_format_##call(struct trace_seq *s)				\
+ftrace_format_##call(struct ftrace_event_call *event_call,		\
+		     struct trace_seq *s)				\
 {									\
 	struct ftrace_raw_##call field __attribute__((unused));		\
 	int ret = 0;							\
@@ -289,10 +290,9 @@ ftrace_raw_output_##call(struct trace_iterator *iter, int flags)	\
 #undef TRACE_EVENT
 #define TRACE_EVENT(call, proto, args, tstruct, func, print)		\
 int									\
-ftrace_define_fields_##call(void)					\
+ftrace_define_fields_##call(struct ftrace_event_call *event_call)	\
 {									\
 	struct ftrace_raw_##call field;					\
-	struct ftrace_event_call *event_call = &event_##call;		\
 	int ret;							\
 									\
 	__common_field(int, type, 1);					\
@@ -355,7 +355,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	event_trace_printk(_RET_IP_, "<call>: " <fmt>);
  * }
  *
- * static int ftrace_reg_event_<call>(void)
+ * static int ftrace_reg_event_<call>(struct ftrace_event_call *unused)
  * {
  *	int ret;
  *
@@ -366,7 +366,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	return ret;
  * }
  *
- * static void ftrace_unreg_event_<call>(void)
+ * static void ftrace_unreg_event_<call>(struct ftrace_event_call *unused)
  * {
  *	unregister_trace_<call>(ftrace_event_<call>);
  * }
@@ -399,7 +399,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	trace_current_buffer_unlock_commit(event, irq_flags, pc);
  * }
  *
- * static int ftrace_raw_reg_event_<call>(void)
+ * static int ftrace_raw_reg_event_<call>(struct ftrace_event_call *unused)
  * {
  *	int ret;
  *
@@ -410,7 +410,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	return ret;
  * }
  *
- * static void ftrace_unreg_event_<call>(void)
+ * static void ftrace_unreg_event_<call>(struct ftrace_event_call *unused)
  * {
  *	unregister_trace_<call>(ftrace_raw_event_<call>);
  * }
@@ -419,7 +419,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	.trace			= ftrace_raw_output_<call>, <-- stage 2
  * };
  *
- * static int ftrace_raw_init_event_<call>(void)
+ * static int ftrace_raw_init_event_<call>(struct ftrace_event_call *unused)
  * {
  *	int id;
  *
@@ -537,7 +537,7 @@ static void ftrace_raw_event_##call(proto)				\
 		trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
 }									\
 									\
-static int ftrace_raw_reg_event_##call(void)				\
+static int ftrace_raw_reg_event_##call(struct ftrace_event_call *unused)\
 {									\
 	int ret;							\
 									\
@@ -548,7 +548,7 @@ static int ftrace_raw_reg_event_##call(void)				\
 	return ret;							\
 }									\
 									\
-static void ftrace_raw_unreg_event_##call(void)				\
+static void ftrace_raw_unreg_event_##call(struct ftrace_event_call *unused)\
 {									\
 	unregister_trace_##call(ftrace_raw_event_##call);		\
 }									\
@@ -557,7 +557,7 @@ static struct trace_event ftrace_event_type_##call = {			\
 	.trace			= ftrace_raw_output_##call,		\
 };									\
 									\
-static int ftrace_raw_init_event_##call(void)				\
+static int ftrace_raw_init_event_##call(struct ftrace_event_call *unused)\
 {									\
 	int id;								\
 									\
kernel/trace/trace_events.c +53 -17

124 lines not on this page · raw message

kernel/trace/trace_export.c +13 -14
index d06cf89..7cee79d 100644
@@ -60,7 +60,7 @@ extern void __bad_type_size(void);
 #undef TRACE_EVENT_FORMAT
 #define TRACE_EVENT_FORMAT(call, proto, args, fmt, tstruct, tpfmt)	\
 static int								\
-ftrace_format_##call(struct trace_seq *s)				\
+ftrace_format_##call(struct ftrace_event_call *dummy, struct trace_seq *s)\
 {									\
 	struct args field;						\
 	int ret;							\
@@ -76,7 +76,7 @@ ftrace_format_##call(struct trace_seq *s)				\
 #define TRACE_EVENT_FORMAT_NOFILTER(call, proto, args, fmt, tstruct,	\
 				    tpfmt)				\
 static int								\
-ftrace_format_##call(struct trace_seq *s)				\
+ftrace_format_##call(struct ftrace_event_call *dummy, struct trace_seq *s)\
 {									\
 	struct args field;						\
 	int ret;							\
@@ -115,10 +115,16 @@ ftrace_format_##call(struct trace_seq *s)				\
 #define TRACE_FIELD_SPECIAL(type_item, item, len, cmd)	\
 	cmd;
 
+static int ftrace_raw_init_event(struct ftrace_event_call *event_call)
+{
+	INIT_LIST_HEAD(&event_call->fields);
+	init_preds(event_call);
+	return 0;
+}
+
 #undef TRACE_EVENT_FORMAT
 #define TRACE_EVENT_FORMAT(call, proto, args, fmt, tstruct, tpfmt)	\
-int ftrace_define_fields_##call(void);					\
-static int ftrace_raw_init_event_##call(void);				\
+int ftrace_define_fields_##call(struct ftrace_event_call *c);		\
 									\
 struct ftrace_event_call __used						\
 __attribute__((__aligned__(4)))						\
@@ -126,16 +132,10 @@ __attribute__((section("_ftrace_events"))) event_##call = {		\
 	.name			= #call,				\
 	.id			= proto,				\
 	.system			= __stringify(TRACE_SYSTEM),		\
-	.raw_init		= ftrace_raw_init_event_##call,		\
+	.raw_init		= ftrace_raw_init_event,		\
 	.show_format		= ftrace_format_##call,			\
 	.define_fields		= ftrace_define_fields_##call,		\
-};									\
-static int ftrace_raw_init_event_##call(void)				\
-{									\
-	INIT_LIST_HEAD(&event_##call.fields);				\
-	init_preds(&event_##call);					\
-	return 0;							\
-}									\
+};
 
 #undef TRACE_EVENT_FORMAT_NOFILTER
 #define TRACE_EVENT_FORMAT_NOFILTER(call, proto, args, fmt, tstruct,	\
@@ -182,9 +182,8 @@ __attribute__((section("_ftrace_events"))) event_##call = {		\
 #undef TRACE_EVENT_FORMAT
 #define TRACE_EVENT_FORMAT(call, proto, args, fmt, tstruct, tpfmt)	\
 int									\
-ftrace_define_fields_##call(void)					\
+ftrace_define_fields_##call(struct ftrace_event_call *event_call)	\
 {									\
-	struct ftrace_event_call *event_call = &event_##call;		\
 	struct args field;						\
 	int ret;							\
 									\
Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
Acked-by on v1 6/7
#12 lore raw

Re: [PATCH -tip -v10 6/7] tracing: ftrace dynamic ftrace_event_call support

Quoted 10 lines · Masami Hiramatsu

Add dynamic ftrace_event_call support to ftrace. Trace engines can adds new ftrace_event_call to ftrace on the fly. Each operator functions of the call takes a ftrace_event_call data structure as an argument, because these functions may be shared among several ftrace_event_calls.

Cc Steven Rostedt <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Tom Zanussi <[email protected]>
Cc Frederic Weisbecker <[email protected]>

Looks good too.

Quoted 355 lines

---

 include/linux/ftrace_event.h |   13 +++++---
 include/trace/ftrace.h       |   22 +++++++------
 kernel/trace/trace_events.c  |   70 ++++++++++++++++++++++++++++++++----------
 kernel/trace/trace_export.c  |   27 ++++++++--------
 4 files changed, 85 insertions(+), 47 deletions(-)
include/linux/ftrace_event.h +8 -5
index 5c093ff..f7733b6 100644
@@ -108,12 +108,13 @@ struct ftrace_event_call {
 	struct dentry		*dir;
 	struct trace_event	*event;
 	int			enabled;
-	int			(*regfunc)(void);
-	void			(*unregfunc)(void);
+	int			(*regfunc)(struct ftrace_event_call *);
+	void			(*unregfunc)(struct ftrace_event_call *);
 	int			id;
-	int			(*raw_init)(void);
-	int			(*show_format)(struct trace_seq *s);
-	int			(*define_fields)(void);
+	int			(*raw_init)(struct ftrace_event_call *);
+	int			(*show_format)(struct ftrace_event_call *,
+					       struct trace_seq *);
+	int			(*define_fields)(struct ftrace_event_call *);
 	struct list_head	fields;
 	int			filter_active;
 	void			*filter;
@@ -138,6 +139,8 @@ extern int filter_current_check_discard(struct ftrace_event_call *call,
 
 extern int trace_define_field(struct ftrace_event_call *call, char *type,
 			      char *name, int offset, int size, int is_signed);
+extern int trace_add_event_call(struct ftrace_event_call *call);
+extern void trace_remove_event_call(struct ftrace_event_call *call);
 
 #define is_signed_type(type)	(((type)(-1)) < 0)
 
include/trace/ftrace.h +11 -11
index 1867553..d696580 100644
@@ -147,7 +147,8 @@
 #undef TRACE_EVENT
 #define TRACE_EVENT(call, proto, args, tstruct, func, print)		\
 static int								\
-ftrace_format_##call(struct trace_seq *s)				\
+ftrace_format_##call(struct ftrace_event_call *event_call,		\
+		     struct trace_seq *s)				\
 {									\
 	struct ftrace_raw_##call field __attribute__((unused));		\
 	int ret = 0;							\
@@ -289,10 +290,9 @@ ftrace_raw_output_##call(struct trace_iterator *iter, int flags)	\
 #undef TRACE_EVENT
 #define TRACE_EVENT(call, proto, args, tstruct, func, print)		\
 int									\
-ftrace_define_fields_##call(void)					\
+ftrace_define_fields_##call(struct ftrace_event_call *event_call)	\
 {									\
 	struct ftrace_raw_##call field;					\
-	struct ftrace_event_call *event_call = &event_##call;		\
 	int ret;							\
 									\
 	__common_field(int, type, 1);					\
@@ -355,7 +355,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	event_trace_printk(_RET_IP_, "<call>: " <fmt>);
  * }
  *
- * static int ftrace_reg_event_<call>(void)
+ * static int ftrace_reg_event_<call>(struct ftrace_event_call *unused)
  * {
  *	int ret;
  *
@@ -366,7 +366,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	return ret;
  * }
  *
- * static void ftrace_unreg_event_<call>(void)
+ * static void ftrace_unreg_event_<call>(struct ftrace_event_call *unused)
  * {
  *	unregister_trace_<call>(ftrace_event_<call>);
  * }
@@ -399,7 +399,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	trace_current_buffer_unlock_commit(event, irq_flags, pc);
  * }
  *
- * static int ftrace_raw_reg_event_<call>(void)
+ * static int ftrace_raw_reg_event_<call>(struct ftrace_event_call *unused)
  * {
  *	int ret;
  *
@@ -410,7 +410,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	return ret;
  * }
  *
- * static void ftrace_unreg_event_<call>(void)
+ * static void ftrace_unreg_event_<call>(struct ftrace_event_call *unused)
  * {
  *	unregister_trace_<call>(ftrace_raw_event_<call>);
  * }
@@ -419,7 +419,7 @@ static inline int ftrace_get_offsets_##call(				\
  *	.trace			= ftrace_raw_output_<call>, <-- stage 2
  * };
  *
- * static int ftrace_raw_init_event_<call>(void)
+ * static int ftrace_raw_init_event_<call>(struct ftrace_event_call *unused)
  * {
  *	int id;
  *
@@ -537,7 +537,7 @@ static void ftrace_raw_event_##call(proto)				\
 		trace_nowake_buffer_unlock_commit(event, irq_flags, pc); \
 }									\
 									\
-static int ftrace_raw_reg_event_##call(void)				\
+static int ftrace_raw_reg_event_##call(struct ftrace_event_call *unused)\
 {									\
 	int ret;							\
 									\
@@ -548,7 +548,7 @@ static int ftrace_raw_reg_event_##call(void)				\
 	return ret;							\
 }									\
 									\
-static void ftrace_raw_unreg_event_##call(void)				\
+static void ftrace_raw_unreg_event_##call(struct ftrace_event_call *unused)\
 {									\
 	unregister_trace_##call(ftrace_raw_event_##call);		\
 }									\
@@ -557,7 +557,7 @@ static struct trace_event ftrace_event_type_##call = {			\
 	.trace			= ftrace_raw_output_##call,		\
 };									\
 									\
-static int ftrace_raw_init_event_##call(void)				\
+static int ftrace_raw_init_event_##call(struct ftrace_event_call *unused)\
 {									\
 	int id;								\
 									\
kernel/trace/trace_events.c +53 -17
index 53c8fd3..94ff41e 100644
@@ -60,9 +60,7 @@ err:
 }
 EXPORT_SYMBOL_GPL(trace_define_field);
 
-#ifdef CONFIG_MODULES
-
-static void trace_destroy_fields(struct ftrace_event_call *call)
+void trace_destroy_fields(struct ftrace_event_call *call)
 {
 	struct ftrace_event_field *field, *next;
 
@@ -74,8 +72,6 @@ static void trace_destroy_fields(struct ftrace_event_call *call)
 	}
 }
 
-#endif /* CONFIG_MODULES */
-
 static void ftrace_event_enable_disable(struct ftrace_event_call *call,
 					int enable)
 {
@@ -84,14 +80,14 @@ static void ftrace_event_enable_disable(struct ftrace_event_call *call,
 		if (call->enabled) {
 			call->enabled = 0;
 			tracing_stop_cmdline_record();
-			call->unregfunc();
+			call->unregfunc(call);
 		}
 		break;
 	case 1:
 		if (!call->enabled) {
 			call->enabled = 1;
 			tracing_start_cmdline_record();
-			call->regfunc();
+			call->regfunc(call);
 		}
 		break;
 	}
@@ -574,7 +570,7 @@ event_format_read(struct file *filp, char __user *ubuf, size_t cnt,
 	trace_seq_printf(s, "format:\n");
 	trace_write_header(s);
 
-	r = call->show_format(s);
+	r = call->show_format(call, s);
 	if (!r) {
 		/*
 		 * ug!  The format output is bigger than a PAGE!!
@@ -921,7 +917,7 @@ event_create_dir(struct ftrace_event_call *call, struct dentry *d_events,
 		d_events = event_subsystem_dir(call->system, d_events);
 
 	if (call->raw_init) {
-		ret = call->raw_init();
+		ret = call->raw_init(call);
 		if (ret < 0) {
 			pr_warning("Could not initialize trace point"
 				   " events/%s\n", call->name);
@@ -945,7 +941,7 @@ event_create_dir(struct ftrace_event_call *call, struct dentry *d_events,
 					  id);
 
 	if (call->define_fields) {
-		ret = call->define_fields();
+		ret = call->define_fields(call);
 		if (ret < 0) {
 			pr_warning("Could not initialize trace point"
 				   " events/%s\n", call->name);
@@ -965,6 +961,52 @@ event_create_dir(struct ftrace_event_call *call, struct dentry *d_events,
 	return 0;
 }
 
+static int __trace_add_event_call(struct ftrace_event_call *call)
+{
+	struct dentry *d_events;
+
+	if (!call->name)
+		return -EINVAL;
+
+	d_events = event_trace_events_dir();
+	if (!d_events)
+		return -ENOENT;
+
+	list_add(&call->list, &ftrace_events);
+	return event_create_dir(call, d_events, &ftrace_event_id_fops,
+				&ftrace_enable_fops, &ftrace_event_filter_fops,
+				&ftrace_event_format_fops);
+}
+
+/* Add an additional event_call dynamically */
+int trace_add_event_call(struct ftrace_event_call *call)
+{
+	int ret;
+	mutex_lock(&event_mutex);
+	ret = __trace_add_event_call(call);
+	mutex_unlock(&event_mutex);
+	return ret;
+}
+
+static void __trace_remove_event_call(struct ftrace_event_call *call)
+{
+	ftrace_event_enable_disable(call, 0);
+	if (call->event)
+		__unregister_ftrace_event(call->event);
+	debugfs_remove_recursive(call->dir);
+	list_del(&call->list);
+	trace_destroy_fields(call);
+	destroy_preds(call);
+}
+
+/* Remove an event_call */
+void trace_remove_event_call(struct ftrace_event_call *call)
+{
+	mutex_lock(&event_mutex);
+	__trace_remove_event_call(call);
+	mutex_unlock(&event_mutex);
+}
+
 #define for_each_event(event, start, end)			\
 	for (event = start;					\
 	     (unsigned long)event < (unsigned long)end;		\
@@ -1070,13 +1112,7 @@ static void trace_module_remove_events(struct module *mod)
 	list_for_each_entry_safe(call, p, &ftrace_events, list) {
 		if (call->mod == mod) {
 			found = true;
-			ftrace_event_enable_disable(call, 0);
-			if (call->event)
-				__unregister_ftrace_event(call->event);
-			debugfs_remove_recursive(call->dir);
-			list_del(&call->list);
-			trace_destroy_fields(call);
-			destroy_preds(call);
+			__trace_remove_event_call(call);
 		}
 	}
 
kernel/trace/trace_export.c +13 -14
index d06cf89..7cee79d 100644
@@ -60,7 +60,7 @@ extern void __bad_type_size(void);
 #undef TRACE_EVENT_FORMAT
 #define TRACE_EVENT_FORMAT(call, proto, args, fmt, tstruct, tpfmt)	\
 static int								\
-ftrace_format_##call(struct trace_seq *s)				\
+ftrace_format_##call(struct ftrace_event_call *dummy, struct trace_seq *s)\
 {									\
 	struct args field;						\
 	int ret;							\
@@ -76,7 +76,7 @@ ftrace_format_##call(struct trace_seq *s)				\
 #define TRACE_EVENT_FORMAT_NOFILTER(call, proto, args, fmt, tstruct,	\
 				    tpfmt)				\
 static int								\
-ftrace_format_##call(struct trace_seq *s)				\
+ftrace_format_##call(struct ftrace_event_call *dummy, struct trace_seq *s)\
 {									\
 	struct args field;						\
 	int ret;							\
@@ -115,10 +115,16 @@ ftrace_format_##call(struct trace_seq *s)				\
 #define TRACE_FIELD_SPECIAL(type_item, item, len, cmd)	\
 	cmd;
 
+static int ftrace_raw_init_event(struct ftrace_event_call *event_call)
+{
+	INIT_LIST_HEAD(&event_call->fields);
+	init_preds(event_call);
+	return 0;
+}
+
 #undef TRACE_EVENT_FORMAT
 #define TRACE_EVENT_FORMAT(call, proto, args, fmt, tstruct, tpfmt)	\
-int ftrace_define_fields_##call(void);					\
-static int ftrace_raw_init_event_##call(void);				\
+int ftrace_define_fields_##call(struct ftrace_event_call *c);		\
 									\
 struct ftrace_event_call __used						\
 __attribute__((__aligned__(4)))						\
@@ -126,16 +132,10 @@ __attribute__((section("_ftrace_events"))) event_##call = {		\
 	.name			= #call,				\
 	.id			= proto,				\
 	.system			= __stringify(TRACE_SYSTEM),		\
-	.raw_init		= ftrace_raw_init_event_##call,		\
+	.raw_init		= ftrace_raw_init_event,		\
 	.show_format		= ftrace_format_##call,			\
 	.define_fields		= ftrace_define_fields_##call,		\
-};									\
-static int ftrace_raw_init_event_##call(void)				\
-{									\
-	INIT_LIST_HEAD(&event_##call.fields);				\
-	init_preds(&event_##call);					\
-	return 0;							\
-}									\
+};
 
 #undef TRACE_EVENT_FORMAT_NOFILTER
 #define TRACE_EVENT_FORMAT_NOFILTER(call, proto, args, fmt, tstruct,	\
@@ -182,9 +182,8 @@ __attribute__((section("_ftrace_events"))) event_##call = {		\
 #undef TRACE_EVENT_FORMAT
 #define TRACE_EVENT_FORMAT(call, proto, args, fmt, tstruct, tpfmt)	\
 int									\
-ftrace_define_fields_##call(void)					\
+ftrace_define_fields_##call(struct ftrace_event_call *event_call)	\
 {									\
-	struct ftrace_event_call *event_call = &event_##call;		\
 	struct args field;						\
 	int ret;							\
 									\
Signature · 8 lines
-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#13 lore raw

PATCH 7/7 tracing: add kprobe-based event tracer

In reply to Masami Hiramatsu

Add kprobes-based event tracer on ftrace.

This tracer is similar to the events tracer which is based on Tracepoint infrastructure. Instead of Tracepoint, this tracer is based on kprobes(kprobe and kretprobe). It probes anywhere where kprobes can probe(this means, all functions body except for __kprobes functions).

Similar to the events tracer, this tracer doesn't need to be activated via current_tracer, instead of that, just set probe points via /sys/kernel/debug/tracing/kprobe_events. And you can set filters on each probe events via /sys/kernel/debug/tracing/events/kprobes/<EVENT>/filter.

This tracer supports following probe arguments for each probe.

%REG : Fetch register REG
sN : Fetch Nth entry of stack (N >= 0)
@ADDR : Fetch memory at ADDR (ADDR should be in kernel)

  @SYM[+|-offs] : Fetch memory at SYM +|- offs (SYM should be a data symbol)
  aN    : Fetch function argument. (N >= 0)
  rv    : Fetch return value.
  ra    : Fetch return address.

+|-offs(FETCHARG) : fetch memory at FETCHARG +|- offs address.

See Documentation/trace/kprobes.txt for details.

Changes from v9:
- Select CONFIG_GENERIC_TRACER when CONFIG_KPROBE_TRACER=y.

Cc Christoph Hellwig <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Tom Zanussi <[email protected]>

---

 Documentation/trace/kprobes.txt  |  138 ++++
 kernel/trace/Kconfig             |   12 
 kernel/trace/Makefile            |    1 
 kernel/trace/trace.h             |   22 +
 kernel/trace/trace_event_types.h |   20 +
 kernel/trace/trace_kprobe.c      | 1183 ++++++++++++++++++++++++++++++++++++++
 6 files changed, 1376 insertions(+), 0 deletions(-)
 create mode 100644 Documentation/trace/kprobes.txt
 create mode 100644 kernel/trace/trace_kprobe.c
Documentation/trace/kprobes.txt new file +138 -0
new file mode 100644
index 0000000..3a90ebb
@@ -0,0 +1,138 @@
+                         Kprobe-based Event Tracer
+                         =========================
+
+                 Documentation is written by Masami Hiramatsu
+
+
+Overview
+--------
+This tracer is similar to the events tracer which is based on Tracepoint
+infrastructure. Instead of Tracepoint, this tracer is based on kprobes(kprobe
+and kretprobe). It probes anywhere where kprobes can probe(this means, all
+functions body except for __kprobes functions).
+
+Unlike the function tracer, this tracer can probe instructions inside of
+kernel functions. It allows you to check which instruction has been executed.
+
+Unlike the Tracepoint based events tracer, this tracer can add and remove
+probe points on the fly.
+
+Similar to the events tracer, this tracer doesn't need to be activated via
+current_tracer, instead of that, just set probe points via
+/sys/kernel/debug/tracing/kprobe_events. And you can set filters on each
+probe events via /sys/kernel/debug/tracing/events/kprobes/<EVENT>/filter.
+
+
+Synopsis of kprobe_events
+-------------------------
+  p[:EVENT] SYMBOL[+offs|-offs]|MEMADDR [FETCHARGS]	: set a probe
+  r[:EVENT] SYMBOL[+0] [FETCHARGS]			: set a return probe
+
+ EVENT			: Event name
+ SYMBOL[+offs|-offs]	: Symbol+offset where the probe is inserted
+ MEMADDR		: Address where the probe is inserted
+
+ FETCHARGS		: Arguments
+  %REG	: Fetch register REG
+  sN	: Fetch Nth entry of stack (N >= 0)
+  @ADDR	: Fetch memory at ADDR (ADDR should be in kernel)
+  @SYM[+|-offs]	: Fetch memory at SYM +|- offs (SYM should be a data symbol)
+  aN	: Fetch function argument. (N >= 0)(*)
+  rv	: Fetch return value.(**)
+  ra	: Fetch return address.(**)
+  +|-offs(FETCHARG) : fetch memory at FETCHARG +|- offs address.(***)
+
+  (*) aN may not correct on asmlinkaged functions and at the middle of
+      function body.
+  (**) only for return probe.
+  (***) this is useful for fetching a field of data structures.
+
+
+Per-Probe Event Filtering
+-------------------------
+ Per-probe event filtering feature allows you to set different filter on each
+probe and gives you what arguments will be shown in trace buffer. If an event
+name is specified right after 'p:' or 'r:' in kprobe_events, the tracer adds
+an event under tracing/events/kprobes/<EVENT>, at the directory you can see
+'id', 'enabled', 'format' and 'filter'.
+
+enabled:
+  You can enable/disable the probe by writing 1 or 0 on it.
+
+format:
+  It shows the format of this probe event. It also shows aliases of arguments
+ which you specified to kprobe_events.
+
+filter:
+  You can write filtering rules of this event. And you can use both of aliase
+ names and field names for describing filters.
+
+
+Usage examples
+--------------
+To add a probe as a new event, write a new definition to kprobe_events
+as below.
+
+  echo p:myprobe do_sys_open a0 a1 a2 a3 > /sys/kernel/debug/tracing/kprobe_events
+
+ This sets a kprobe on the top of do_sys_open() function with recording
+1st to 4th arguments as "myprobe" event.
+
+  echo r:myretprobe do_sys_open rv ra >> /sys/kernel/debug/tracing/kprobe_events
+
+ This sets a kretprobe on the return point of do_sys_open() function with
+recording return value and return address as "myretprobe" event.
+ You can see the format of these events via
+/sys/kernel/debug/tracing/events/kprobes/<EVENT>/format.
+
+  cat /sys/kernel/debug/tracing/events/kprobes/myprobe/format
+name: myprobe
+ID: 23
+format:
+	field:unsigned short common_type;	offset:0;	size:2;
+	field:unsigned char common_flags;	offset:2;	size:1;
+	field:unsigned char common_preempt_count;	offset:3;	size:1;
+	field:int common_pid;	offset:4;	size:4;
+	field:int common_tgid;	offset:8;	size:4;
+
+	field: unsigned long ip;	offset:16;tsize:8;
+	field: int nargs;	offset:24;tsize:4;
+	field: unsigned long arg0;	offset:32;tsize:8;
+	field: unsigned long arg1;	offset:40;tsize:8;
+	field: unsigned long arg2;	offset:48;tsize:8;
+	field: unsigned long arg3;	offset:56;tsize:8;
+
+	alias: a0;	original: arg0;
+	alias: a1;	original: arg1;
+	alias: a2;	original: arg2;
+	alias: a3;	original: arg3;
+
+print fmt: "%lx: 0x%lx 0x%lx 0x%lx 0x%lx", ip, arg0, arg1, arg2, arg3
+
+
+ You can see that the event has 4 arguments and alias expressions
+corresponding to it.
+
+  echo > /sys/kernel/debug/tracing/kprobe_events
+
+ This clears all probe points. and you can see the traced information via
+/sys/kernel/debug/tracing/trace.
+
+  cat /sys/kernel/debug/tracing/trace
+# tracer: nop
+#
+#           TASK-PID    CPU#    TIMESTAMP  FUNCTION
+#              | |       |          |         |
+           <...>-1447  [001] 1038282.286875: do_sys_open+0x0/0xd6: 0x3 0x7fffd1ec4440 0x8000 0x0
+           <...>-1447  [001] 1038282.286878: sys_openat+0xc/0xe <- do_sys_open: 0xfffffffffffffffe 0xffffffff81367a3a
+           <...>-1447  [001] 1038282.286885: do_sys_open+0x0/0xd6: 0xffffff9c 0x40413c 0x8000 0x1b6
+           <...>-1447  [001] 1038282.286915: sys_open+0x1b/0x1d <- do_sys_open: 0x3 0xffffffff81367a3a
+           <...>-1447  [001] 1038282.286969: do_sys_open+0x0/0xd6: 0xffffff9c 0x4041c6 0x98800 0x10
+           <...>-1447  [001] 1038282.286976: sys_open+0x1b/0x1d <- do_sys_open: 0x3 0xffffffff81367a3a
+
+
+ Each line shows when the kernel hits a probe, and <- SYMBOL means kernel
+returns from SYMBOL(e.g. "sys_open+0x1b/0x1d <- do_sys_open" means kernel
+returns from do_sys_open to sys_open+0x1b).
+
+
kernel/trace/Kconfig +12 -0
index 860c712..60f3401 100644
@@ -445,6 +445,18 @@ config BLK_DEV_IO_TRACE
 
 	  If unsure, say N.
 
+config KPROBE_TRACER
+	depends on KPROBES
+	depends on X86
+	bool "Trace kprobes"
+	select TRACING
+	select GENERIC_TRACER
+	help
+	  This tracer probes everywhere where kprobes can probe it, and
+	  records various registers and memories specified by user.
+	  This also allows you to trace kprobe probe points as a dynamic
+	  defined events. It provides per-probe event filtering interface.
+
 config DYNAMIC_FTRACE
 	bool "enable/disable ftrace tracepoints dynamically"
 	depends on FUNCTION_TRACER
kernel/trace/Makefile +1 -0
index ce3b1cd..8e6884d 100644
@@ -55,5 +55,6 @@ obj-$(CONFIG_FTRACE_SYSCALLS) += trace_syscalls.o
 obj-$(CONFIG_EVENT_PROFILE) += trace_event_profile.o
 obj-$(CONFIG_EVENT_TRACING) += trace_events_filter.o
 obj-$(CONFIG_KSYM_TRACER) += trace_ksym.o
+obj-$(CONFIG_KPROBE_TRACER) += trace_kprobe.o
 
 libftrace-y := ftrace.o
kernel/trace/trace.h +22 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];
+};
+
+struct kretprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		func;
+	unsigned long		ret_ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];
+};
 
 /*
  * trace_flag_type is an enumeration that holds different
@@ -344,6 +362,10 @@ extern void __ftrace_bad_type(void);
 		IF_ASSIGN(var, ent, struct syscall_trace_exit,		\
 			  TRACE_SYSCALL_EXIT);				\
 		IF_ASSIGN(var, ent, struct trace_ksym, TRACE_KSYM);	\
+		IF_ASSIGN(var, ent, struct kprobe_trace_entry,		\
+			  TRACE_KPROBE);				\
+		IF_ASSIGN(var, ent, struct kretprobe_trace_entry,	\
+			  TRACE_KRETPROBE);				\
 		__ftrace_bad_type();					\
 	} while (0)
 
kernel/trace/trace_event_types.h +20 -0
index 6db005e..ec2e6f3 100644
@@ -175,4 +175,24 @@ TRACE_EVENT_FORMAT(kmem_free, TRACE_KMEM_FREE, kmemtrace_free_entry, ignore,
 	TP_RAW_FMT("type:%u call_site:%lx ptr:%p")
 );
 
+TRACE_EVENT_FORMAT(kprobe, TRACE_KPROBE, kprobe_trace_entry, ignore,
+	TRACE_STRUCT(
+		TRACE_FIELD(unsigned long, ip, ip)
+		TRACE_FIELD(int, nargs, nargs)
+		TRACE_FIELD_SPECIAL(unsigned long args[TRACE_KPROBE_ARGS],
+				    args, TRACE_KPROBE_ARGS, args)
+	),
+	TP_RAW_FMT("%08lx: args:0x%lx ...")
+);
+
+TRACE_EVENT_FORMAT(kretprobe, TRACE_KRETPROBE, kretprobe_trace_entry, ignore,
+	TRACE_STRUCT(
+		TRACE_FIELD(unsigned long, func, func)
+		TRACE_FIELD(unsigned long, ret_ip, ret_ip)
+		TRACE_FIELD(int, nargs, nargs)
+		TRACE_FIELD_SPECIAL(unsigned long args[TRACE_KPROBE_ARGS],
+				    args, TRACE_KPROBE_ARGS, args)
+	),
+	TP_RAW_FMT("%08lx <- %08lx: args:0x%lx ...")
+);
 #undef TRACE_SYSTEM
kernel/trace/trace_kprobe.c new file +1183 -0

1,183 lines not on this page · raw message

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#14 lore raw

Re: [PATCH -tip -v10 7/7] tracing: add kprobe-based event tracer

Quoted 180 lines · Masami Hiramatsu

Add kprobes-based event tracer on ftrace.

This tracer is similar to the events tracer which is based on Tracepoint infrastructure. Instead of Tracepoint, this tracer is based on kprobes(kprobe and kretprobe). It probes anywhere where kprobes can probe(this means, all functions body except for __kprobes functions).

Similar to the events tracer, this tracer doesn't need to be activated via current_tracer, instead of that, just set probe points via /sys/kernel/debug/tracing/kprobe_events. And you can set filters on each probe events via /sys/kernel/debug/tracing/events/kprobes/<EVENT>/filter.

This tracer supports following probe arguments for each probe.

%REG : Fetch register REG
sN : Fetch Nth entry of stack (N >= 0)
@ADDR : Fetch memory at ADDR (ADDR should be in kernel)

  @SYM[+|-offs] : Fetch memory at SYM +|- offs (SYM should be a data symbol)
  aN    : Fetch function argument. (N >= 0)
  rv    : Fetch return value.
  ra    : Fetch return address.

+|-offs(FETCHARG) : fetch memory at FETCHARG +|- offs address.

See Documentation/trace/kprobes.txt for details.

Changes from v9:
- Select CONFIG_GENERIC_TRACER when CONFIG_KPROBE_TRACER=y.

Cc Christoph Hellwig <[email protected]>
Cc Steven Rostedt <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Frederic Weisbecker <[email protected]>
Cc Tom Zanussi <[email protected]>

---

 Documentation/trace/kprobes.txt  |  138 ++++
 kernel/trace/Kconfig             |   12 
 kernel/trace/Makefile            |    1 
 kernel/trace/trace.h             |   22 +
 kernel/trace/trace_event_types.h |   20 +
 kernel/trace/trace_kprobe.c      | 1183 ++++++++++++++++++++++++++++++++++++++
 6 files changed, 1376 insertions(+), 0 deletions(-)
 create mode 100644 Documentation/trace/kprobes.txt
 create mode 100644 kernel/trace/trace_kprobe.c
Documentation/trace/kprobes.txt new file +127 -0
new file mode 100644
index 0000000..3a90ebb
@@ -0,0 +1,138 @@
+                         Kprobe-based Event Tracer
+                         =========================
+
+                 Documentation is written by Masami Hiramatsu
+
+
+Overview
+--------
+This tracer is similar to the events tracer which is based on Tracepoint
+infrastructure. Instead of Tracepoint, this tracer is based on kprobes(kprobe
+and kretprobe). It probes anywhere where kprobes can probe(this means, all
+functions body except for __kprobes functions).
+
+Unlike the function tracer, this tracer can probe instructions inside of
+kernel functions. It allows you to check which instruction has been executed.
+
+Unlike the Tracepoint based events tracer, this tracer can add and remove
+probe points on the fly.
+
+Similar to the events tracer, this tracer doesn't need to be activated via
+current_tracer, instead of that, just set probe points via
+/sys/kernel/debug/tracing/kprobe_events. And you can set filters on each
+probe events via /sys/kernel/debug/tracing/events/kprobes/<EVENT>/filter.
+
+
+Synopsis of kprobe_events
+-------------------------
+  p[:EVENT] SYMBOL[+offs|-offs]|MEMADDR [FETCHARGS]	: set a probe
+  r[:EVENT] SYMBOL[+0] [FETCHARGS]			: set a return probe
+
+ EVENT			: Event name
+ SYMBOL[+offs|-offs]	: Symbol+offset where the probe is inserted
+ MEMADDR		: Address where the probe is inserted
+
+ FETCHARGS		: Arguments
+  %REG	: Fetch register REG
+  sN	: Fetch Nth entry of stack (N >= 0)
+  @ADDR	: Fetch memory at ADDR (ADDR should be in kernel)
+  @SYM[+|-offs]	: Fetch memory at SYM +|- offs (SYM should be a data symbol)
+  aN	: Fetch function argument. (N >= 0)(*)
+  rv	: Fetch return value.(**)
+  ra	: Fetch return address.(**)
+  +|-offs(FETCHARG) : fetch memory at FETCHARG +|- offs address.(***)
+
+  (*) aN may not correct on asmlinkaged functions and at the middle of
+      function body.
+  (**) only for return probe.
+  (***) this is useful for fetching a field of data structures.
+
+
+Per-Probe Event Filtering
+-------------------------
+ Per-probe event filtering feature allows you to set different filter on each
+probe and gives you what arguments will be shown in trace buffer. If an event
+name is specified right after 'p:' or 'r:' in kprobe_events, the tracer adds
+an event under tracing/events/kprobes/<EVENT>, at the directory you can see
+'id', 'enabled', 'format' and 'filter'.
+
+enabled:
+  You can enable/disable the probe by writing 1 or 0 on it.
+
+format:
+  It shows the format of this probe event. It also shows aliases of arguments
+ which you specified to kprobe_events.
+
+filter:
+  You can write filtering rules of this event. And you can use both of aliase
+ names and field names for describing filters.
+
+
+Usage examples
+--------------
+To add a probe as a new event, write a new definition to kprobe_events
+as below.
+
+  echo p:myprobe do_sys_open a0 a1 a2 a3 > /sys/kernel/debug/tracing/kprobe_events
+
+ This sets a kprobe on the top of do_sys_open() function with recording
+1st to 4th arguments as "myprobe" event.
+
+  echo r:myretprobe do_sys_open rv ra >> /sys/kernel/debug/tracing/kprobe_events
+
+ This sets a kretprobe on the return point of do_sys_open() function with
+recording return value and return address as "myretprobe" event.
+ You can see the format of these events via
+/sys/kernel/debug/tracing/events/kprobes/<EVENT>/format.
+
+  cat /sys/kernel/debug/tracing/events/kprobes/myprobe/format
+name: myprobe
+ID: 23
+format:
+	field:unsigned short common_type;	offset:0;	size:2;
+	field:unsigned char common_flags;	offset:2;	size:1;
+	field:unsigned char common_preempt_count;	offset:3;	size:1;
+	field:int common_pid;	offset:4;	size:4;
+	field:int common_tgid;	offset:8;	size:4;
+
+	field: unsigned long ip;	offset:16;tsize:8;
+	field: int nargs;	offset:24;tsize:4;
+	field: unsigned long arg0;	offset:32;tsize:8;
+	field: unsigned long arg1;	offset:40;tsize:8;
+	field: unsigned long arg2;	offset:48;tsize:8;
+	field: unsigned long arg3;	offset:56;tsize:8;
+
+	alias: a0;	original: arg0;
+	alias: a1;	original: arg1;
+	alias: a2;	original: arg2;
+	alias: a3;	original: arg3;
+
+print fmt: "%lx: 0x%lx 0x%lx 0x%lx 0x%lx", ip, arg0, arg1, arg2, arg3
+
+
+ You can see that the event has 4 arguments and alias expressions
+corresponding to it.
+
+  echo > /sys/kernel/debug/tracing/kprobe_events
+
+ This clears all probe points. and you can see the traced information via
+/sys/kernel/debug/tracing/trace.
+
+  cat /sys/kernel/debug/tracing/trace
+# tracer: nop
+#
+#           TASK-PID    CPU#    TIMESTAMP  FUNCTION
+#              | |       |          |         |
+           <...>-1447  [001] 1038282.286875: do_sys_open+0x0/0xd6: 0x3 0x7fffd1ec4440 0x8000 0x0
+           <...>-1447  [001] 1038282.286878: sys_openat+0xc/0xe <- do_sys_open: 0xfffffffffffffffe 0xffffffff81367a3a

Documentation/trace/kprobes.txt line 137

Quoted 10 lines
Documentation/trace/kprobes.txt new file +10 -0
+           <...>-1447  [001] 1038282.286885: do_sys_open+0x0/0xd6: 0xffffff9c 0x40413c 0x8000 0x1b6
+           <...>-1447  [001] 1038282.286915: sys_open+0x1b/0x1d <- do_sys_open: 0x3 0xffffffff81367a3a
+           <...>-1447  [001] 1038282.286969: do_sys_open+0x0/0xd6: 0xffffff9c 0x4041c6 0x98800 0x10
+           <...>-1447  [001] 1038282.286976: sys_open+0x1b/0x1d <- do_sys_open: 0x3 0xffffffff81367a3a
+
+
+ Each line shows when the kernel hits a probe, and <- SYMBOL means kernel
+returns from SYMBOL(e.g. "sys_open+0x1b/0x1d <- do_sys_open" means kernel
+returns from do_sys_open to sys_open+0x1b).
+

I'm looking forward to use it for debugging :)

Quoted 48 lines
kernel/trace/Kconfig +12 -0
index 860c712..60f3401 100644
@@ -445,6 +445,18 @@ config BLK_DEV_IO_TRACE
 
 	  If unsure, say N.
 
+config KPROBE_TRACER
+	depends on KPROBES
+	depends on X86
+	bool "Trace kprobes"
+	select TRACING
+	select GENERIC_TRACER
+	help
+	  This tracer probes everywhere where kprobes can probe it, and
+	  records various registers and memories specified by user.
+	  This also allows you to trace kprobe probe points as a dynamic
+	  defined events. It provides per-probe event filtering interface.
+
 config DYNAMIC_FTRACE
 	bool "enable/disable ftrace tracepoints dynamically"
 	depends on FUNCTION_TRACER
kernel/trace/Makefile +1 -0
index ce3b1cd..8e6884d 100644
@@ -55,5 +55,6 @@ obj-$(CONFIG_FTRACE_SYSCALLS) += trace_syscalls.o
 obj-$(CONFIG_EVENT_PROFILE) += trace_event_profile.o
 obj-$(CONFIG_EVENT_TRACING) += trace_events_filter.o
 obj-$(CONFIG_KSYM_TRACER) += trace_ksym.o
+obj-$(CONFIG_KPROBE_TRACER) += trace_kprobe.o
 
 libftrace-y := ftrace.o
kernel/trace/trace.h +2 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };

kernel/trace/trace.h struct trace_ksym { · line 238

Quoted 10 lines
kernel/trace/trace.h +7 -0
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];

I see that you actually make use of arg as a dynamic sizeable array.
For clarity, args[TRACE_KPROBE_ARGS] could be args[0].

It's just a neat and wouldn't affect the code nor the data but would be clearer for readers of that code.

kernel/trace/trace.h struct trace_ksym { · line 247

Quoted 9 lines
kernel/trace/trace.h +9 -0
+};
+
+struct kretprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		func;
+	unsigned long		ret_ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];
+};

ditto

Quoted 98 lines
kernel/trace/trace.h +4 -0
 /*
  * trace_flag_type is an enumeration that holds different
@@ -344,6 +362,10 @@ extern void __ftrace_bad_type(void);
 		IF_ASSIGN(var, ent, struct syscall_trace_exit,		\
 			  TRACE_SYSCALL_EXIT);				\
 		IF_ASSIGN(var, ent, struct trace_ksym, TRACE_KSYM);	\
+		IF_ASSIGN(var, ent, struct kprobe_trace_entry,		\
+			  TRACE_KPROBE);				\
+		IF_ASSIGN(var, ent, struct kretprobe_trace_entry,	\
+			  TRACE_KRETPROBE);				\
 		__ftrace_bad_type();					\
 	} while (0)
 
kernel/trace/trace_event_types.h +20 -0
index 6db005e..ec2e6f3 100644
@@ -175,4 +175,24 @@ TRACE_EVENT_FORMAT(kmem_free, TRACE_KMEM_FREE, kmemtrace_free_entry, ignore,
 	TP_RAW_FMT("type:%u call_site:%lx ptr:%p")
 );
 
+TRACE_EVENT_FORMAT(kprobe, TRACE_KPROBE, kprobe_trace_entry, ignore,
+	TRACE_STRUCT(
+		TRACE_FIELD(unsigned long, ip, ip)
+		TRACE_FIELD(int, nargs, nargs)
+		TRACE_FIELD_SPECIAL(unsigned long args[TRACE_KPROBE_ARGS],
+				    args, TRACE_KPROBE_ARGS, args)
+	),
+	TP_RAW_FMT("%08lx: args:0x%lx ...")
+);
+
+TRACE_EVENT_FORMAT(kretprobe, TRACE_KRETPROBE, kretprobe_trace_entry, ignore,
+	TRACE_STRUCT(
+		TRACE_FIELD(unsigned long, func, func)
+		TRACE_FIELD(unsigned long, ret_ip, ret_ip)
+		TRACE_FIELD(int, nargs, nargs)
+		TRACE_FIELD_SPECIAL(unsigned long args[TRACE_KPROBE_ARGS],
+				    args, TRACE_KPROBE_ARGS, args)
+	),
+	TP_RAW_FMT("%08lx <- %08lx: args:0x%lx ...")
+);
 #undef TRACE_SYSTEM
kernel/trace/trace_kprobe.c new file +50 -0
new file mode 100644
index 0000000..0951512
@@ -0,0 +1,1183 @@
+/*
+ * kprobe based kernel tracer
+ *
+ * Created by Masami Hiramatsu <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include <linux/module.h>
+#include <linux/uaccess.h>
+#include <linux/kprobes.h>
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/debugfs.h>
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/ctype.h>
+#include <linux/ptrace.h>
+
+#include "trace.h"
+#include "trace_output.h"
+
+#define MAX_ARGSTR_LEN 63
+
+/* currently, trace_kprobe only supports X86. */
+
+struct fetch_func {
+	unsigned long (*func)(struct pt_regs *, void *);
+	void *data;
+};
+
+static __kprobes unsigned long call_fetch(struct fetch_func *f,
+					  struct pt_regs *regs)
+{
+	return f->func(regs, f->data);
+}
+
+/* fetch handlers */

kernel/trace/trace_kprobe.c line 60

Quoted 10 lines
kernel/trace/trace_kprobe.c new file +10 -0
+static __kprobes unsigned long fetch_register(struct pt_regs *regs,
+					      void *offset)
+{
+	return regs_get_register(regs, (unsigned)((unsigned long)offset));
+}
+
+static __kprobes unsigned long fetch_stack(struct pt_regs *regs,
+					   void *num)
+{
+	return regs_get_kernel_stack_nth(regs, (unsigned)((unsigned long)num));

You seem to often use unsigned as an implicit type. Would be better to explicitly use unsigned int.

Anyway, I guess we can merge it in -tip and give it a try.

Thanks,
Frederic.

Quoted 1,133 lines
kernel/trace/trace_kprobe.c new file +1123 -0
+}
+
+static __kprobes unsigned long fetch_memory(struct pt_regs *regs, void *addr)
+{
+	unsigned long retval;
+	if (probe_kernel_address(addr, retval))
+		return 0;
+	return retval;
+}
+
+static __kprobes unsigned long fetch_argument(struct pt_regs *regs, void *num)
+{
+	return regs_get_argument_nth(regs, (unsigned)((unsigned long)num));
+}
+
+static __kprobes unsigned long fetch_retvalue(struct pt_regs *regs,
+					      void *dummy)
+{
+	return regs_return_value(regs);
+}
+
+static __kprobes unsigned long fetch_ip(struct pt_regs *regs, void *dummy)
+{
+	return instruction_pointer(regs);
+}
+
+/* Memory fetching by symbol */
+struct symbol_cache {
+	char *symbol;
+	long offset;
+	unsigned long addr;
+};
+
+static unsigned long update_symbol_cache(struct symbol_cache *sc)
+{
+	sc->addr = (unsigned long)kallsyms_lookup_name(sc->symbol);
+	if (sc->addr)
+		sc->addr += sc->offset;
+	return sc->addr;
+}
+
+static void free_symbol_cache(struct symbol_cache *sc)
+{
+	kfree(sc->symbol);
+	kfree(sc);
+}
+
+static struct symbol_cache *alloc_symbol_cache(const char *sym, long offset)
+{
+	struct symbol_cache *sc;
+	if (!sym || strlen(sym) == 0)
+		return NULL;
+	sc = kzalloc(sizeof(struct symbol_cache), GFP_KERNEL);
+	if (!sc)
+		return NULL;
+
+	sc->symbol = kstrdup(sym, GFP_KERNEL);
+	if (!sc->symbol) {
+		kfree(sc);
+		return NULL;
+	}
+	sc->offset = offset;
+
+	update_symbol_cache(sc);
+	return sc;
+}
+
+static __kprobes unsigned long fetch_symbol(struct pt_regs *regs, void *data)
+{
+	struct symbol_cache *sc = data;
+	if (sc->addr)
+		return fetch_memory(regs, (void *)sc->addr);
+	else
+		return 0;
+}
+
+/* Special indirect memory access interface */
+struct indirect_fetch_data {
+	struct fetch_func orig;
+	long offset;
+};
+
+static __kprobes unsigned long fetch_indirect(struct pt_regs *regs, void *data)
+{
+	struct indirect_fetch_data *ind = data;
+	unsigned long addr;
+	addr = call_fetch(&ind->orig, regs);
+	if (addr) {
+		addr += ind->offset;
+		return fetch_memory(regs, (void *)addr);
+	} else
+		return 0;
+}
+
+static __kprobes void free_indirect_fetch_data(struct indirect_fetch_data *data)
+{
+	if (data->orig.func == fetch_indirect)
+		free_indirect_fetch_data(data->orig.data);
+	else if (data->orig.func == fetch_symbol)
+		free_symbol_cache(data->orig.data);
+	kfree(data);
+}
+
+/**
+ * kprobe_trace_core
+ */
+
+struct trace_probe {
+	struct list_head	list;
+	union {
+		struct kprobe		kp;
+		struct kretprobe	rp;
+	};
+	const char		*symbol;	/* symbol name */
+	unsigned int		nr_args;
+	struct fetch_func	args[TRACE_KPROBE_ARGS];
+	struct ftrace_event_call	call;
+};
+
+static int kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs);
+static int kretprobe_trace_func(struct kretprobe_instance *ri,
+				struct pt_regs *regs);
+
+static __kprobes int probe_is_return(struct trace_probe *tp)
+{
+	return (tp->rp.handler == kretprobe_trace_func);
+}
+
+static __kprobes const char *probe_symbol(struct trace_probe *tp)
+{
+	return tp->symbol ? tp->symbol : "unknown";
+}
+
+static __kprobes long probe_offset(struct trace_probe *tp)
+{
+	return (probe_is_return(tp)) ? tp->rp.kp.offset : tp->kp.offset;
+}
+
+static __kprobes void *probe_address(struct trace_probe *tp)
+{
+	return (probe_is_return(tp)) ? tp->rp.kp.addr : tp->kp.addr;
+}
+
+static int trace_arg_string(char *buf, size_t n, struct fetch_func *ff)
+{
+	int ret = -EINVAL;
+	if (ff->func == fetch_argument)
+		ret = snprintf(buf, n, "a%lu", (unsigned long)ff->data);
+	else if (ff->func == fetch_register) {
+		const char *name;
+		name = regs_query_register_name((unsigned)((long)ff->data));
+		ret = snprintf(buf, n, "%%%s", name);
+	} else if (ff->func == fetch_stack)
+		ret = snprintf(buf, n, "s%lu", (unsigned long)ff->data);
+	else if (ff->func == fetch_memory)
+		ret = snprintf(buf, n, "@0x%p", ff->data);
+	else if (ff->func == fetch_symbol) {
+		struct symbol_cache *sc = ff->data;
+		ret = snprintf(buf, n, "@%s%+ld", sc->symbol, sc->offset);
+	} else if (ff->func == fetch_retvalue)
+		ret = snprintf(buf, n, "rv");
+	else if (ff->func == fetch_ip)
+		ret = snprintf(buf, n, "ra");
+	else if (ff->func == fetch_indirect) {
+		struct indirect_fetch_data *id = ff->data;
+		ret = snprintf(buf, n, "%+ld(", id->offset);
+		if (ret > n)
+			goto end;
+		n -= ret;
+		ret = trace_arg_string(buf, n, &id->orig);
+		if (ret > n)
+			goto end;
+		n -= ret;
+		ret = snprintf(buf, n, ")");
+	}
+end:
+	if (ret > n)
+		return -ENOSPC;
+	return 0;
+}
+
+static int register_probe_event(struct trace_probe *tp);
+static void unregister_probe_event(struct trace_probe *tp);
+
+static DEFINE_MUTEX(probe_lock);
+static LIST_HEAD(probe_list);
+
+static struct trace_probe *alloc_trace_probe(const char *symbol,
+					     const char *event)
+{
+	struct trace_probe *tp;
+
+	tp = kzalloc(sizeof(struct trace_probe), GFP_KERNEL);
+	if (!tp)
+		return ERR_PTR(-ENOMEM);
+
+	if (symbol) {
+		tp->symbol = kstrdup(symbol, GFP_KERNEL);
+		if (!tp->symbol)
+			goto error;
+	}
+	if (event) {
+		tp->call.name = kstrdup(event, GFP_KERNEL);
+		if (!tp->call.name)
+			goto error;
+	}
+
+	INIT_LIST_HEAD(&tp->list);
+	return tp;
+error:
+	kfree(tp->symbol);
+	kfree(tp);
+	return ERR_PTR(-ENOMEM);
+}
+
+static void free_trace_probe(struct trace_probe *tp)
+{
+	int i;
+	for (i = 0; i < tp->nr_args; i++)
+		if (tp->args[i].func == fetch_symbol)
+			free_symbol_cache(tp->args[i].data);
+		else if (tp->args[i].func == fetch_indirect)
+			free_indirect_fetch_data(tp->args[i].data);
+
+	kfree(tp->call.name);
+	kfree(tp->symbol);
+	kfree(tp);
+}
+
+static struct trace_probe *find_probe_event(const char *event)
+{
+	struct trace_probe *tp;
+	list_for_each_entry(tp, &probe_list, list)
+		if (tp->call.name && !strcmp(tp->call.name, event))
+			return tp;
+	return NULL;
+}
+
+static void __unregister_trace_probe(struct trace_probe *tp)
+{
+	if (probe_is_return(tp))
+		unregister_kretprobe(&tp->rp);
+	else
+		unregister_kprobe(&tp->kp);
+}
+
+/* Unregister a trace_probe and probe_event: call with locking probe_lock */
+static void unregister_trace_probe(struct trace_probe *tp)
+{
+	if (tp->call.name)
+		unregister_probe_event(tp);
+	__unregister_trace_probe(tp);
+	list_del(&tp->list);
+}
+
+/* Register a trace_probe and probe_event */
+static int register_trace_probe(struct trace_probe *tp)
+{
+	struct trace_probe *old_tp;
+	int ret;
+
+	mutex_lock(&probe_lock);
+
+	if (probe_is_return(tp))
+		ret = register_kretprobe(&tp->rp);
+	else
+		ret = register_kprobe(&tp->kp);
+
+	if (ret) {
+		pr_warning("Could not insert probe(%d)\n", ret);
+		if (ret == -EILSEQ) {
+			pr_warning("Probing address(0x%p) is not an "
+				   "instruction boundary.\n",
+				   probe_address(tp));
+			ret = -EINVAL;
+		}
+		goto end;
+	}
+	/* register as an event */
+	if (tp->call.name) {
+		old_tp = find_probe_event(tp->call.name);
+		if (old_tp) {
+			/* delete old event */
+			unregister_trace_probe(old_tp);
+			free_trace_probe(old_tp);
+		}
+		ret = register_probe_event(tp);
+		if (ret) {
+			pr_warning("Faild to register probe event(%d)\n", ret);
+			__unregister_trace_probe(tp);
+		}
+	}
+	list_add_tail(&tp->list, &probe_list);
+end:
+	mutex_unlock(&probe_lock);
+	return ret;
+}
+
+/* Split symbol and offset. */
+static int split_symbol_offset(char *symbol, long *offset)
+{
+	char *tmp;
+	int ret;
+
+	if (!offset)
+		return -EINVAL;
+
+	tmp = strchr(symbol, '+');
+	if (!tmp)
+		tmp = strchr(symbol, '-');
+
+	if (tmp) {
+		/* skip sign because strict_strtol doesn't accept '+' */
+		ret = strict_strtol(tmp + 1, 0, offset);
+		if (ret)
+			return ret;
+		if (*tmp == '-')
+			*offset = -(*offset);
+		*tmp = '\0';
+	} else
+		*offset = 0;
+	return 0;
+}
+
+#define PARAM_MAX_ARGS 16
+#define PARAM_MAX_STACK (THREAD_SIZE / sizeof(unsigned long))
+
+static int parse_trace_arg(char *arg, struct fetch_func *ff, int is_return)
+{
+	int ret = 0;
+	unsigned long param;
+	long offset;
+	char *tmp;
+
+	switch (arg[0]) {
+	case 'a':	/* argument */
+		ret = strict_strtoul(arg + 1, 10, &param);
+		if (ret || param > PARAM_MAX_ARGS)
+			ret = -EINVAL;
+		else {
+			ff->func = fetch_argument;
+			ff->data = (void *)param;
+		}
+		break;
+	case 'r':	/* retval or retaddr */
+		if (is_return && arg[1] == 'v') {
+			ff->func = fetch_retvalue;
+			ff->data = NULL;
+		} else if (is_return && arg[1] == 'a') {
+			ff->func = fetch_ip;
+			ff->data = NULL;
+		} else
+			ret = -EINVAL;
+		break;
+	case '%':	/* named register */
+		ret = regs_query_register_offset(arg + 1);
+		if (ret >= 0) {
+			ff->func = fetch_register;
+			ff->data = (void *)(unsigned long)ret;
+			ret = 0;
+		}
+		break;
+	case 's':	/* stack */
+		ret = strict_strtoul(arg + 1, 10, &param);
+		if (ret || param > PARAM_MAX_STACK)
+			ret = -EINVAL;
+		else {
+			ff->func = fetch_stack;
+			ff->data = (void *)param;
+		}
+		break;
+	case '@':	/* memory or symbol */
+		if (isdigit(arg[1])) {
+			ret = strict_strtoul(arg + 1, 0, &param);
+			if (ret)
+				break;
+			ff->func = fetch_memory;
+			ff->data = (void *)param;
+		} else {
+			ret = split_symbol_offset(arg + 1, &offset);
+			if (ret)
+				break;
+			ff->data = alloc_symbol_cache(arg + 1,
+							      offset);
+			if (ff->data)
+				ff->func = fetch_symbol;
+			else
+				ret = -EINVAL;
+		}
+		break;
+	case '+':	/* indirect memory */
+	case '-':
+		tmp = strchr(arg, '(');
+		if (!tmp) {
+			ret = -EINVAL;
+			break;
+		}
+		*tmp = '\0';
+		ret = strict_strtol(arg + 1, 0, &offset);
+		if (ret)
+			break;
+		if (arg[0] == '-')
+			offset = -offset;
+		arg = tmp + 1;
+		tmp = strrchr(arg, ')');
+		if (tmp) {
+			struct indirect_fetch_data *id;
+			*tmp = '\0';
+			id = kzalloc(sizeof(struct indirect_fetch_data),
+				     GFP_KERNEL);
+			if (!id)
+				return -ENOMEM;
+			id->offset = offset;
+			ret = parse_trace_arg(arg, &id->orig, is_return);
+			if (ret)
+				kfree(id);
+			else {
+				ff->func = fetch_indirect;
+				ff->data = (void *)id;
+			}
+		} else
+			ret = -EINVAL;
+		break;
+	default:
+		/* TODO: support custom handler */
+		ret = -EINVAL;
+	}
+	return ret;
+}
+
+static int create_trace_probe(int argc, char **argv)
+{
+	/*
+	 * Argument syntax:
+	 *  - Add kprobe: p[:EVENT] SYMBOL[+OFFS|-OFFS]|ADDRESS [FETCHARGS]
+	 *  - Add kretprobe: r[:EVENT] SYMBOL[+0] [FETCHARGS]
+	 * Fetch args:
+	 *  aN	: fetch Nth of function argument. (N:0-)
+	 *  rv	: fetch return value
+	 *  ra	: fetch return address
+	 *  sN	: fetch Nth of stack (N:0-)
+	 *  @ADDR	: fetch memory at ADDR (ADDR should be in kernel)
+	 *  @SYM[+|-offs] : fetch memory at SYM +|- offs (SYM is a data symbol)
+	 *  %REG	: fetch register REG
+	 * Indirect memory fetch:
+	 *  +|-offs(ARG) : fetch memory at ARG +|- offs address.
+	 */
+	struct trace_probe *tp;
+	struct kprobe *kp;
+	int i, ret = 0;
+	int is_return = 0;
+	char *symbol = NULL, *event = NULL;
+	long offset = 0;
+	void *addr = NULL;
+
+	if (argc < 2)
+		return -EINVAL;
+
+	if (argv[0][0] == 'p')
+		is_return = 0;
+	else if (argv[0][0] == 'r')
+		is_return = 1;
+	else
+		return -EINVAL;
+
+	if (argv[0][1] == ':') {
+		event = &argv[0][2];
+		if (strlen(event) == 0) {
+			pr_info("Event name is not specifiled\n");
+			return -EINVAL;
+		}
+	}
+
+	if (isdigit(argv[1][0])) {
+		if (is_return)
+			return -EINVAL;
+		/* an address specified */
+		ret = strict_strtoul(&argv[0][2], 0, (unsigned long *)&addr);
+		if (ret)
+			return ret;
+	} else {
+		/* a symbol specified */
+		symbol = argv[1];
+		/* TODO: support .init module functions */
+		ret = split_symbol_offset(symbol, &offset);
+		if (ret)
+			return ret;
+		if (offset && is_return)
+			return -EINVAL;
+	}
+
+	/* setup a probe */
+	tp = alloc_trace_probe(symbol, event);
+	if (IS_ERR(tp))
+		return PTR_ERR(tp);
+
+	if (is_return) {
+		kp = &tp->rp.kp;
+		tp->rp.handler = kretprobe_trace_func;
+	} else {
+		kp = &tp->kp;
+		tp->kp.pre_handler = kprobe_trace_func;
+	}
+
+	if (tp->symbol) {
+		kp->symbol_name = tp->symbol;
+		kp->offset = offset;
+	} else
+		kp->addr = addr;
+
+	/* parse arguments */
+	argc -= 2; argv += 2; ret = 0;
+	for (i = 0; i < argc && i < TRACE_KPROBE_ARGS; i++) {
+		if (strlen(argv[i]) > MAX_ARGSTR_LEN) {
+			pr_info("Argument%d(%s) is too long.\n", i, argv[i]);
+			ret = -ENOSPC;
+			goto error;
+		}
+		ret = parse_trace_arg(argv[i], &tp->args[i], is_return);
+		if (ret)
+			goto error;
+	}
+	tp->nr_args = i;
+
+	ret = register_trace_probe(tp);
+	if (ret)
+		goto error;
+	return 0;
+
+error:
+	free_trace_probe(tp);
+	return ret;
+}
+
+static void cleanup_all_probes(void)
+{
+	struct trace_probe *tp;
+	mutex_lock(&probe_lock);
+	/* TODO: Use batch unregistration */
+	while (!list_empty(&probe_list)) {
+		tp = list_entry(probe_list.next, struct trace_probe, list);
+		unregister_trace_probe(tp);
+		free_trace_probe(tp);
+	}
+	mutex_unlock(&probe_lock);
+}
+
+
+/* Probes listing interfaces */
+static void *probes_seq_start(struct seq_file *m, loff_t *pos)
+{
+	mutex_lock(&probe_lock);
+	return seq_list_start(&probe_list, *pos);
+}
+
+static void *probes_seq_next(struct seq_file *m, void *v, loff_t *pos)
+{
+	return seq_list_next(v, &probe_list, pos);
+}
+
+static void probes_seq_stop(struct seq_file *m, void *v)
+{
+	mutex_unlock(&probe_lock);
+}
+
+static int probes_seq_show(struct seq_file *m, void *v)
+{
+	struct trace_probe *tp = v;
+	int i, ret;
+	char buf[MAX_ARGSTR_LEN + 1];
+
+	if (tp == NULL)
+		return 0;
+
+	seq_printf(m, "%c", probe_is_return(tp) ? 'r' : 'p');
+	if (tp->call.name)
+		seq_printf(m, ":%s", tp->call.name);
+
+	if (tp->symbol)
+		seq_printf(m, " %s%+ld", probe_symbol(tp), probe_offset(tp));
+	else
+		seq_printf(m, " 0x%p", probe_address(tp));
+
+	for (i = 0; i < tp->nr_args; i++) {
+		ret = trace_arg_string(buf, MAX_ARGSTR_LEN, &tp->args[i]);
+		if (ret) {
+			pr_warning("Argument%d is too long.\n", i);
+			break;
+		}
+		seq_printf(m, " %s", buf);
+	}
+	seq_printf(m, "\n");
+	return 0;
+}
+
+static const struct seq_operations probes_seq_op = {
+	.start  = probes_seq_start,
+	.next   = probes_seq_next,
+	.stop   = probes_seq_stop,
+	.show   = probes_seq_show
+};
+
+static int probes_open(struct inode *inode, struct file *file)
+{
+	if ((file->f_mode & FMODE_WRITE) &&
+	    !(file->f_flags & O_APPEND))
+		cleanup_all_probes();
+
+	return seq_open(file, &probes_seq_op);
+}
+
+static int command_trace_probe(const char *buf)
+{
+	char **argv;
+	int argc = 0, ret = 0;
+
+	argv = argv_split(GFP_KERNEL, buf, &argc);
+	if (!argv)
+		return -ENOMEM;
+
+	if (argc)
+		ret = create_trace_probe(argc, argv);
+
+	argv_free(argv);
+	return ret;
+}
+
+#define WRITE_BUFSIZE 128
+
+static ssize_t probes_write(struct file *file, const char __user *buffer,
+			    size_t count, loff_t *ppos)
+{
+	char *kbuf, *tmp;
+	int ret;
+	size_t done;
+	size_t size;
+
+	if (!count || count < 0)
+		return 0;
+
+	kbuf = kmalloc(WRITE_BUFSIZE, GFP_KERNEL);
+	if (!kbuf)
+		return -ENOMEM;
+
+	ret = done = 0;
+	do {
+		size = count - done;
+		if (size > WRITE_BUFSIZE)
+			size = WRITE_BUFSIZE;
+		if (copy_from_user(kbuf, buffer + done, size)) {
+			ret = -EFAULT;
+			goto out;
+		}
+		kbuf[size] = '\0';
+		tmp = strchr(kbuf, '\n');
+		if (!tmp) {
+			pr_warning("Line length is too long: "
+				   "Should be less than %d.", WRITE_BUFSIZE);
+			ret = -EINVAL;
+			goto out;
+		}
+		*tmp = '\0';
+		size = tmp - kbuf + 1;
+		done += size;
+		/* Remove comments */
+		tmp = strchr(kbuf, '#');
+		if (tmp)
+			*tmp = '\0';
+
+		ret = command_trace_probe(kbuf);
+		if (ret)
+			goto out;
+
+	} while (done < count);
+	ret = done;
+out:
+	kfree(kbuf);
+	return ret;
+}
+
+static const struct file_operations kprobe_events_ops = {
+	.owner          = THIS_MODULE,
+	.open           = probes_open,
+	.read           = seq_read,
+	.llseek         = seq_lseek,
+	.release        = seq_release,
+	.write		= probes_write,
+};
+
+/* Kprobe handler */
+static __kprobes int kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs)
+{
+	struct trace_probe *tp = container_of(kp, struct trace_probe, kp);
+	struct kprobe_trace_entry *entry;
+	struct ring_buffer_event *event;
+	int size, i, pc;
+	unsigned long irq_flags;
+	struct ftrace_event_call *call = &event_kprobe;
+	if (&tp->call.name)
+		call = &tp->call;
+
+	local_save_flags(irq_flags);
+	pc = preempt_count();
+
+	size = sizeof(struct kprobe_trace_entry) -
+	       (sizeof(unsigned long) * (TRACE_KPROBE_ARGS - tp->nr_args));
+
+	event = trace_current_buffer_lock_reserve(TRACE_KPROBE, size,
+						  irq_flags, pc);
+	if (!event)
+		return 0;
+
+	entry = ring_buffer_event_data(event);
+	entry->nargs = tp->nr_args;
+	entry->ip = (unsigned long)kp->addr;
+	for (i = 0; i < tp->nr_args; i++)
+		entry->args[i] = call_fetch(&tp->args[i], regs);
+
+	if (!filter_current_check_discard(call, entry, event))
+		trace_nowake_buffer_unlock_commit(event, irq_flags, pc);
+	return 0;
+}
+
+/* Kretprobe handler */
+static __kprobes int kretprobe_trace_func(struct kretprobe_instance *ri,
+					  struct pt_regs *regs)
+{
+	struct trace_probe *tp = container_of(ri->rp, struct trace_probe, rp);
+	struct kretprobe_trace_entry *entry;
+	struct ring_buffer_event *event;
+	int size, i, pc;
+	unsigned long irq_flags;
+	struct ftrace_event_call *call = &event_kretprobe;
+	if (&tp->call.name)
+		call = &tp->call;
+
+	local_save_flags(irq_flags);
+	pc = preempt_count();
+
+	size = sizeof(struct kretprobe_trace_entry) -
+	       (sizeof(unsigned long) * (TRACE_KPROBE_ARGS - tp->nr_args));
+
+	event = trace_current_buffer_lock_reserve(TRACE_KRETPROBE, size,
+						  irq_flags, pc);
+	if (!event)
+		return 0;
+
+	entry = ring_buffer_event_data(event);
+	entry->nargs = tp->nr_args;
+	entry->func = (unsigned long)probe_address(tp);
+	entry->ret_ip = (unsigned long)ri->ret_addr;
+	for (i = 0; i < tp->nr_args; i++)
+		entry->args[i] = call_fetch(&tp->args[i], regs);
+
+	if (!filter_current_check_discard(call, entry, event))
+		trace_nowake_buffer_unlock_commit(event, irq_flags, pc);
+
+	return 0;
+}
+
+/* Event entry printers */
+enum print_line_t
+print_kprobe_event(struct trace_iterator *iter, int flags)
+{
+	struct kprobe_trace_entry *field;
+	struct trace_seq *s = &iter->seq;
+	int i;
+
+	trace_assign_type(field, iter->ent);
+
+	if (!seq_print_ip_sym(s, field->ip, flags | TRACE_ITER_SYM_OFFSET))
+		goto partial;
+
+	if (!trace_seq_puts(s, ":"))
+		goto partial;
+
+	for (i = 0; i < field->nargs; i++)
+		if (!trace_seq_printf(s, " 0x%lx", field->args[i]))
+			goto partial;
+
+	if (!trace_seq_puts(s, "\n"))
+		goto partial;
+
+	return TRACE_TYPE_HANDLED;
+partial:
+	return TRACE_TYPE_PARTIAL_LINE;
+}
+
+enum print_line_t
+print_kretprobe_event(struct trace_iterator *iter, int flags)
+{
+	struct kretprobe_trace_entry *field;
+	struct trace_seq *s = &iter->seq;
+	int i;
+
+	trace_assign_type(field, iter->ent);
+
+	if (!seq_print_ip_sym(s, field->ret_ip, flags | TRACE_ITER_SYM_OFFSET))
+		goto partial;
+
+	if (!trace_seq_puts(s, " <- "))
+		goto partial;
+
+	if (!seq_print_ip_sym(s, field->func, flags & ~TRACE_ITER_SYM_OFFSET))
+		goto partial;
+
+	if (!trace_seq_puts(s, ":"))
+		goto partial;
+
+	for (i = 0; i < field->nargs; i++)
+		if (!trace_seq_printf(s, " 0x%lx", field->args[i]))
+			goto partial;
+
+	if (!trace_seq_puts(s, "\n"))
+		goto partial;
+
+	return TRACE_TYPE_HANDLED;
+partial:
+	return TRACE_TYPE_PARTIAL_LINE;
+}
+
+static struct trace_event kprobe_trace_event = {
+	.type	 	= TRACE_KPROBE,
+	.trace		= print_kprobe_event,
+};
+
+static struct trace_event kretprobe_trace_event = {
+	.type	 	= TRACE_KRETPROBE,
+	.trace		= print_kretprobe_event,
+};
+
+static int probe_event_enable(struct ftrace_event_call *call)
+{
+	struct trace_probe *tp = container_of(call, struct trace_probe, call);
+	if (probe_is_return(tp))
+		return enable_kretprobe(&tp->rp);
+	else
+		return enable_kprobe(&tp->kp);
+}
+
+static void probe_event_disable(struct ftrace_event_call *call)
+{
+	struct trace_probe *tp = container_of(call, struct trace_probe, call);
+	if (probe_is_return(tp))
+		disable_kretprobe(&tp->rp);
+	else
+		disable_kprobe(&tp->kp);
+}
+
+static int probe_event_raw_init(struct ftrace_event_call *event_call)
+{
+	INIT_LIST_HEAD(&event_call->fields);
+	init_preds(event_call);
+	return 0;
+}
+
+#undef DEFINE_FIELD
+#define DEFINE_FIELD(type, item, name, is_signed)			\
+	do {								\
+		ret = trace_define_field(event_call, #type, name,	\
+					 offsetof(typeof(field), item),	\
+					 sizeof(field.item), is_signed);\
+		if (ret)						\
+			return ret;					\
+	} while (0)
+
+static int kprobe_event_define_fields(struct ftrace_event_call *event_call)
+{
+	int ret, i;
+	struct kprobe_trace_entry field;
+	char buf[MAX_ARGSTR_LEN + 1];
+	struct trace_probe *tp = container_of(event_call,
+					      struct trace_probe, call);
+
+	__common_field(int, type, 1);
+	__common_field(unsigned char, flags, 0);
+	__common_field(unsigned char, preempt_count, 0);
+	__common_field(int, pid, 1);
+	__common_field(int, tgid, 1);
+
+	DEFINE_FIELD(unsigned long, ip, "ip", 0);
+	DEFINE_FIELD(int, nargs, "nargs", 1);
+	for (i = 0; i < tp->nr_args; i++) {
+		/* Set argN as a field */
+		sprintf(buf, "arg%d", i);
+		DEFINE_FIELD(unsigned long, args[i], buf, 0);
+		/* Set argument string as an alias field */
+		ret = trace_arg_string(buf, MAX_ARGSTR_LEN, &tp->args[i]);
+		if (ret)
+			return ret;
+		DEFINE_FIELD(unsigned long, args[i], buf, 0);
+	}
+	return 0;
+}
+
+static int kretprobe_event_define_fields(struct ftrace_event_call *event_call)
+{
+	int ret, i;
+	struct kretprobe_trace_entry field;
+	char buf[MAX_ARGSTR_LEN + 1];
+	struct trace_probe *tp = container_of(event_call,
+					      struct trace_probe, call);
+
+	__common_field(int, type, 1);
+	__common_field(unsigned char, flags, 0);
+	__common_field(unsigned char, preempt_count, 0);
+	__common_field(int, pid, 1);
+	__common_field(int, tgid, 1);
+
+	DEFINE_FIELD(unsigned long, func, "func", 0);
+	DEFINE_FIELD(unsigned long, ret_ip, "ret_ip", 0);
+	DEFINE_FIELD(int, nargs, "nargs", 1);
+	for (i = 0; i < tp->nr_args; i++) {
+		/* Set argN as a field */
+		sprintf(buf, "arg%d", i);
+		DEFINE_FIELD(unsigned long, args[i], buf, 0);
+		/* Set argument string as an alias field */
+		ret = trace_arg_string(buf, MAX_ARGSTR_LEN, &tp->args[i]);
+		if (ret)
+			return ret;
+		DEFINE_FIELD(unsigned long, args[i], buf, 0);
+	}
+	return 0;
+}
+
+static int __probe_event_show_format(struct ftrace_event_call *call,
+				     struct trace_seq *s, const char *fmt,
+				     const char *arg)
+{
+	int i;
+	char buf[MAX_ARGSTR_LEN + 1];
+	struct trace_probe *tp = container_of(call, struct trace_probe, call);
+
+	/* Show aliases */
+	for (i = 0; i < tp->nr_args; i++) {
+		if (trace_arg_string(buf, MAX_ARGSTR_LEN, &tp->args[i]))
+			return 0;
+		if (!trace_seq_printf(s, "\talias: %s;\toriginal: arg%d;\n",
+				      buf, i))
+			return 0;
+	}
+	/* Show format */
+	if (!trace_seq_printf(s, "\nprint fmt: \"%s", fmt))
+		return 0;
+
+	for (i = 0; i < tp->nr_args; i++)
+		if (!trace_seq_puts(s, " 0x%lx"))
+			return 0;
+
+	if (!trace_seq_printf(s, "\", %s", arg))
+		return 0;
+
+	for (i = 0; i < tp->nr_args; i++)
+		if (!trace_seq_printf(s, ", arg%d", i))
+			return 0;
+
+	return trace_seq_puts(s, "\n");
+}
+
+#undef SHOW_FIELD
+#define SHOW_FIELD(type, item, name)					\
+	do {								\
+		ret = trace_seq_printf(s, "\tfield: " #type " %s;\t"	\
+				"offset:%u;tsize:%u;\n", name,		\
+				(unsigned)offsetof(typeof(field), item),\
+				(unsigned)sizeof(type));		\
+		if (!ret)						\
+			return 0;					\
+	} while (0)
+
+static int kprobe_event_show_format(struct ftrace_event_call *call,
+				    struct trace_seq *s)
+{
+	struct kprobe_trace_entry field __attribute__((unused));
+	int ret, i;
+	char buf[8];
+	struct trace_probe *tp = container_of(call, struct trace_probe, call);
+
+	SHOW_FIELD(unsigned long, ip, "ip");
+	SHOW_FIELD(int, nargs, "nargs");
+
+	/* Show fields */
+	for (i = 0; i < tp->nr_args; i++) {
+		sprintf(buf, "arg%d", i);
+		SHOW_FIELD(unsigned long, args[i], buf);
+	}
+	trace_seq_puts(s, "\n");
+
+	return __probe_event_show_format(call, s, "%lx:", "ip");
+}
+
+static int kretprobe_event_show_format(struct ftrace_event_call *call,
+				       struct trace_seq *s)
+{
+	struct kretprobe_trace_entry field __attribute__((unused));
+	int ret, i;
+	char buf[8];
+	struct trace_probe *tp = container_of(call, struct trace_probe, call);
+
+	SHOW_FIELD(unsigned long, func, "func");
+	SHOW_FIELD(unsigned long, ret_ip, "ret_ip");
+	SHOW_FIELD(int, nargs, "nargs");
+
+	/* Show fields */
+	for (i = 0; i < tp->nr_args; i++) {
+		sprintf(buf, "arg%d", i);
+		SHOW_FIELD(unsigned long, args[i], buf);
+	}
+	trace_seq_puts(s, "\n");
+
+	return __probe_event_show_format(call, s, "%lx <- %lx:",
+					  "func, ret_ip");
+}
+
+static int register_probe_event(struct trace_probe *tp)
+{
+	struct ftrace_event_call *call = &tp->call;
+	int ret;
+
+	/* Initialize ftrace_event_call */
+	call->system = "kprobes";
+	if (probe_is_return(tp)) {
+		call->event = &kretprobe_trace_event;
+		call->id = TRACE_KRETPROBE;
+		call->raw_init = probe_event_raw_init;
+		call->show_format = kretprobe_event_show_format;
+		call->define_fields = kretprobe_event_define_fields;
+	} else {
+		call->event = &kprobe_trace_event;
+		call->id = TRACE_KPROBE;
+		call->raw_init = probe_event_raw_init;
+		call->show_format = kprobe_event_show_format;
+		call->define_fields = kprobe_event_define_fields;
+	}
+	call->enabled = 1;
+	call->regfunc = probe_event_enable;
+	call->unregfunc = probe_event_disable;
+	ret = trace_add_event_call(call);
+	if (ret)
+		pr_info("Failed to register kprobe event: %s\n", call->name);
+	return ret;
+}
+
+static void unregister_probe_event(struct trace_probe *tp)
+{
+	/*
+	 * Prevent to unregister event itself because the event is shared
+	 * among other probes.
+	 */
+	tp->call.event = NULL;
+	trace_remove_event_call(&tp->call);
+}
+
+/* Make a debugfs interface for controling probe points */
+static __init int init_kprobe_trace(void)
+{
+	struct dentry *d_tracer;
+	struct dentry *entry;
+	int ret;
+
+	ret = register_ftrace_event(&kprobe_trace_event);
+	if (!ret) {
+		pr_warning("Could not register kprobe_trace_event type.\n");
+		return 0;
+	}
+	ret = register_ftrace_event(&kretprobe_trace_event);
+	if (!ret) {
+		pr_warning("Could not register kretprobe_trace_event type.\n");
+		return 0;
+	}
+
+	d_tracer = tracing_init_dentry();
+	if (!d_tracer)
+		return 0;
+
+	entry = debugfs_create_file("kprobe_events", 0644, d_tracer,
+				    NULL, &kprobe_events_ops);
+
+	if (!entry)
+		pr_warning("Could not create debugfs "
+			   "'kprobe_events' entry\n");
+	return 0;
+}
+fs_initcall(init_kprobe_trace);
+
+
+#ifdef CONFIG_FTRACE_STARTUP_TEST
+
+static int kprobe_trace_selftest_target(int a1, int a2, int a3,
+					int a4, int a5, int a6)
+{
+	return a1 + a2 + a3 + a4 + a5 + a6;
+}
+
+static __init int kprobe_trace_self_tests_init(void)
+{
+	int ret;
+	int (*target)(int, int, int, int, int, int);
+	target = kprobe_trace_selftest_target;
+
+	pr_info("Testing kprobe tracing: ");
+
+	ret = command_trace_probe("p:testprobe kprobe_trace_selftest_target "
+				  "a1 a2 a3 a4 a5 a6");
+	if (WARN_ON_ONCE(ret))
+		pr_warning("error enabling function entry\n");
+
+	ret = command_trace_probe("r:testprobe2 kprobe_trace_selftest_target "
+				  "ra rv");
+	if (WARN_ON_ONCE(ret))
+		pr_warning("error enabling function return\n");
+
+	ret = target(1, 2, 3, 4, 5, 6);
+
+	cleanup_all_probes();
+
+	pr_cont("OK\n");
+	return 0;
+}
+
+late_initcall(kprobe_trace_self_tests_init);
+
+#endif
Signature · 8 lines
-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#15 lore raw

Re: [PATCH -tip -v10 7/7] tracing: add kprobe-based event tracer

Quoted 31 lines · Frederic Weisbecker
Quoted 14 lines
kernel/trace/trace.h +2 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };

kernel/trace/trace.h struct trace_ksym { · line 238

Quoted 10 lines
kernel/trace/trace.h +7 -0
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];

I see that you actually make use of arg as a dynamic sizeable array.
For clarity, args[TRACE_KPROBE_ARGS] could be args[0].

Quoted 2 lines

It's just a neat and wouldn't affect the code nor the data but would be clearer for readers of that code.

Hmm. In that case, I think we'll need a new macro for field definition, like TRACE_FIELD_ZERO(type, item).

Quoted 119 lines

kernel/trace/trace.h struct trace_ksym { · line 247

Quoted 9 lines
kernel/trace/trace.h +9 -0
+};
+
+struct kretprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		func;
+	unsigned long		ret_ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];
+};

ditto

Quoted 104 lines
kernel/trace/trace.h +4 -0
 /*
  * trace_flag_type is an enumeration that holds different
@@ -344,6 +362,10 @@ extern void __ftrace_bad_type(void);
 		IF_ASSIGN(var, ent, struct syscall_trace_exit,		\
 			  TRACE_SYSCALL_EXIT);				\
 		IF_ASSIGN(var, ent, struct trace_ksym, TRACE_KSYM);	\
+		IF_ASSIGN(var, ent, struct kprobe_trace_entry,		\
+			  TRACE_KPROBE);				\
+		IF_ASSIGN(var, ent, struct kretprobe_trace_entry,	\
+			  TRACE_KRETPROBE);				\
 		__ftrace_bad_type();					\
 	} while (0)
 
kernel/trace/trace_event_types.h +20 -0
index 6db005e..ec2e6f3 100644
@@ -175,4 +175,24 @@ TRACE_EVENT_FORMAT(kmem_free, TRACE_KMEM_FREE, kmemtrace_free_entry, ignore,
 	TP_RAW_FMT("type:%u call_site:%lx ptr:%p")
 );
 
+TRACE_EVENT_FORMAT(kprobe, TRACE_KPROBE, kprobe_trace_entry, ignore,
+	TRACE_STRUCT(
+		TRACE_FIELD(unsigned long, ip, ip)
+		TRACE_FIELD(int, nargs, nargs)
+		TRACE_FIELD_SPECIAL(unsigned long args[TRACE_KPROBE_ARGS],
+				    args, TRACE_KPROBE_ARGS, args)
+	),
+	TP_RAW_FMT("%08lx: args:0x%lx ...")
+);
+
+TRACE_EVENT_FORMAT(kretprobe, TRACE_KRETPROBE, kretprobe_trace_entry, ignore,
+	TRACE_STRUCT(
+		TRACE_FIELD(unsigned long, func, func)
+		TRACE_FIELD(unsigned long, ret_ip, ret_ip)
+		TRACE_FIELD(int, nargs, nargs)
+		TRACE_FIELD_SPECIAL(unsigned long args[TRACE_KPROBE_ARGS],
+				    args, TRACE_KPROBE_ARGS, args)
+	),
+	TP_RAW_FMT("%08lx <- %08lx: args:0x%lx ...")
+);
 #undef TRACE_SYSTEM
kernel/trace/trace_kprobe.c new file +56 -0
new file mode 100644
index 0000000..0951512
@@ -0,0 +1,1183 @@
+/*
+ * kprobe based kernel tracer
+ *
+ * Created by Masami Hiramatsu <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include <linux/module.h>
+#include <linux/uaccess.h>
+#include <linux/kprobes.h>
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/debugfs.h>
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/ctype.h>
+#include <linux/ptrace.h>
+
+#include "trace.h"
+#include "trace_output.h"
+
+#define MAX_ARGSTR_LEN 63
+
+/* currently, trace_kprobe only supports X86. */
+
+struct fetch_func {
+	unsigned long (*func)(struct pt_regs *, void *);
+	void *data;
+};
+
+static __kprobes unsigned long call_fetch(struct fetch_func *f,
+					  struct pt_regs *regs)
+{
+	return f->func(regs, f->data);
+}
+
+/* fetch handlers */
+static __kprobes unsigned long fetch_register(struct pt_regs *regs,
+					      void *offset)
+{
+	return regs_get_register(regs, (unsigned)((unsigned long)offset));
+}
+
Quoted 8 lines

kernel/trace/trace_kprobe.c line 60

Quoted 4 lines
kernel/trace/trace_kprobe.c new file +4 -0
+static __kprobes unsigned long fetch_stack(struct pt_regs *regs,
+					   void *num)
+{
+	return regs_get_kernel_stack_nth(regs, (unsigned)((unsigned long)num));

You seem to often use unsigned as an implicit type. Would be better to explicitly use unsigned int.

Agreed.

Quoted 1 line

Anyway, I guess we can merge it in -tip and give it a try.

Thank you!

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#16 lore raw

Re: [PATCH -tip -v10 7/7] tracing: add kprobe-based event tracer

Quoted 32 lines · Masami Hiramatsu
Quoted 31 lines · Frederic Weisbecker
Quoted 14 lines
kernel/trace/trace.h +2 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };

kernel/trace/trace.h struct trace_ksym { · line 238

Quoted 10 lines
kernel/trace/trace.h +7 -0
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];

I see that you actually make use of arg as a dynamic sizeable array.
For clarity, args[TRACE_KPROBE_ARGS] could be args[0].

Quoted 5 lines
Quoted 2 lines

It's just a neat and wouldn't affect the code nor the data but would be clearer for readers of that code.

Hmm. In that case, I think we'll need a new macro for field definition, like TRACE_FIELD_ZERO(type, item).

You mean that for trace_define_field() to describe fields of events? Actually the fields should be defined dynamically depending on how is built the kprobe event (which arguments are requested, how many, etc..).

Frederic.

Quoted 144 lines
Quoted 119 lines

kernel/trace/trace.h struct trace_ksym { · line 247

Quoted 9 lines
kernel/trace/trace.h +9 -0
+};
+
+struct kretprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		func;
+	unsigned long		ret_ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];
+};

ditto

Quoted 104 lines
kernel/trace/trace.h +4 -0
 /*
  * trace_flag_type is an enumeration that holds different
@@ -344,6 +362,10 @@ extern void __ftrace_bad_type(void);
 		IF_ASSIGN(var, ent, struct syscall_trace_exit,		\
 			  TRACE_SYSCALL_EXIT);				\
 		IF_ASSIGN(var, ent, struct trace_ksym, TRACE_KSYM);	\
+		IF_ASSIGN(var, ent, struct kprobe_trace_entry,		\
+			  TRACE_KPROBE);				\
+		IF_ASSIGN(var, ent, struct kretprobe_trace_entry,	\
+			  TRACE_KRETPROBE);				\
 		__ftrace_bad_type();					\
 	} while (0)
 
kernel/trace/trace_event_types.h +20 -0
index 6db005e..ec2e6f3 100644
@@ -175,4 +175,24 @@ TRACE_EVENT_FORMAT(kmem_free, TRACE_KMEM_FREE, kmemtrace_free_entry, ignore,
 	TP_RAW_FMT("type:%u call_site:%lx ptr:%p")
 );
 
+TRACE_EVENT_FORMAT(kprobe, TRACE_KPROBE, kprobe_trace_entry, ignore,
+	TRACE_STRUCT(
+		TRACE_FIELD(unsigned long, ip, ip)
+		TRACE_FIELD(int, nargs, nargs)
+		TRACE_FIELD_SPECIAL(unsigned long args[TRACE_KPROBE_ARGS],
+				    args, TRACE_KPROBE_ARGS, args)
+	),
+	TP_RAW_FMT("%08lx: args:0x%lx ...")
+);
+
+TRACE_EVENT_FORMAT(kretprobe, TRACE_KRETPROBE, kretprobe_trace_entry, ignore,
+	TRACE_STRUCT(
+		TRACE_FIELD(unsigned long, func, func)
+		TRACE_FIELD(unsigned long, ret_ip, ret_ip)
+		TRACE_FIELD(int, nargs, nargs)
+		TRACE_FIELD_SPECIAL(unsigned long args[TRACE_KPROBE_ARGS],
+				    args, TRACE_KPROBE_ARGS, args)
+	),
+	TP_RAW_FMT("%08lx <- %08lx: args:0x%lx ...")
+);
 #undef TRACE_SYSTEM
kernel/trace/trace_kprobe.c new file +56 -0
new file mode 100644
index 0000000..0951512
@@ -0,0 +1,1183 @@
+/*
+ * kprobe based kernel tracer
+ *
+ * Created by Masami Hiramatsu <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include <linux/module.h>
+#include <linux/uaccess.h>
+#include <linux/kprobes.h>
+#include <linux/seq_file.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/debugfs.h>
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/ctype.h>
+#include <linux/ptrace.h>
+
+#include "trace.h"
+#include "trace_output.h"
+
+#define MAX_ARGSTR_LEN 63
+
+/* currently, trace_kprobe only supports X86. */
+
+struct fetch_func {
+	unsigned long (*func)(struct pt_regs *, void *);
+	void *data;
+};
+
+static __kprobes unsigned long call_fetch(struct fetch_func *f,
+					  struct pt_regs *regs)
+{
+	return f->func(regs, f->data);
+}
+
+/* fetch handlers */
+static __kprobes unsigned long fetch_register(struct pt_regs *regs,
+					      void *offset)
+{
+	return regs_get_register(regs, (unsigned)((unsigned long)offset));
+}
+
Quoted 8 lines

kernel/trace/trace_kprobe.c line 60

Quoted 4 lines
kernel/trace/trace_kprobe.c new file +4 -0
+static __kprobes unsigned long fetch_stack(struct pt_regs *regs,
+					   void *num)
+{
+	return regs_get_kernel_stack_nth(regs, (unsigned)((unsigned long)num));

You seem to often use unsigned as an implicit type. Would be better to explicitly use unsigned int.

Agreed.

Quoted 1 line

Anyway, I guess we can merge it in -tip and give it a try.

Thank you!

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#17 lore raw

Re: [PATCH -tip -v10 7/7] tracing: add kprobe-based event tracer

Quoted 39 lines · Frederic Weisbecker
Quoted 31 lines · Masami Hiramatsu
Quoted 30 lines · Frederic Weisbecker
Quoted 14 lines
kernel/trace/trace.h +2 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };

kernel/trace/trace.h struct trace_ksym { · line 238

Quoted 10 lines
kernel/trace/trace.h +7 -0
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];

I see that you actually make use of arg as a dynamic sizeable array.
For clarity, args[TRACE_KPROBE_ARGS] could be args[0].

Quoted 4 lines
Quoted 2 lines

It's just a neat and wouldn't affect the code nor the data but would be clearer for readers of that code.

Hmm. In that case, I think we'll need a new macro for field definition, like TRACE_FIELD_ZERO(type, item).

Quoted 4 lines

You mean that for trace_define_field() to describe fields of events? Actually the fields should be defined dynamically depending on how is built the kprobe event (which arguments are requested, how many, etc..).

Yeah, if you specified a probe point with its event name, the tracer will make a corresponding event dynamically. There are also anonymous probes which don't have corresponding events. For those anonymous probes, I need to define two generic event types(kprobe and kretprobe).

Thank you,

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#18 lore raw

Re: [PATCH -tip -v10 7/7] tracing: add kprobe-based event tracer

Quoted 45 lines · Masami Hiramatsu
Quoted 39 lines · Frederic Weisbecker
Quoted 31 lines · Masami Hiramatsu
Quoted 30 lines · Frederic Weisbecker
Quoted 14 lines
kernel/trace/trace.h +2 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };

kernel/trace/trace.h struct trace_ksym { · line 238

Quoted 10 lines
kernel/trace/trace.h +7 -0
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];

I see that you actually make use of arg as a dynamic sizeable array.
For clarity, args[TRACE_KPROBE_ARGS] could be args[0].

Quoted 4 lines
Quoted 2 lines

It's just a neat and wouldn't affect the code nor the data but would be clearer for readers of that code.

Hmm. In that case, I think we'll need a new macro for field definition, like TRACE_FIELD_ZERO(type, item).

Quoted 4 lines

You mean that for trace_define_field() to describe fields of events? Actually the fields should be defined dynamically depending on how is built the kprobe event (which arguments are requested, how many, etc..).

Quoted 6 lines

Yeah, if you specified a probe point with its event name, the tracer will make a corresponding event dynamically. There are also anonymous probes which don't have corresponding events. For those anonymous probes, I need to define two generic event types(kprobe and kretprobe).

Thank you,

Ok. Btw, why do you need to define those two anonymous events? Actually your event types are always dynamically created. Those you defined through TRACE_FORMAT_EVENT are only "ghost events", they only stand there as a abstract pattern, right?

#19 lore raw

Re: [PATCH -tip -v10 7/7] tracing: add kprobe-based event tracer

Quoted 48 lines · Frederic Weisbecker
Quoted 47 lines · Masami Hiramatsu
Quoted 33 lines · Frederic Weisbecker
Quoted 32 lines · Masami Hiramatsu
Quoted 31 lines · Frederic Weisbecker
Quoted 14 lines
kernel/trace/trace.h +2 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };

kernel/trace/trace.h struct trace_ksym { · line 238

Quoted 10 lines
kernel/trace/trace.h +7 -0
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];

I see that you actually make use of arg as a dynamic sizeable array.
For clarity, args[TRACE_KPROBE_ARGS] could be args[0].

It's just a neat and wouldn't affect the code nor the data but would be clearer for readers of that code.

Quoted 8 lines
Quoted 2 lines

Hmm. In that case, I think we'll need a new macro for field definition, like TRACE_FIELD_ZERO(type, item).

You mean that for trace_define_field() to describe fields of events? Actually the fields should be defined dynamically depending on how is built the kprobe event (which arguments are requested, how many, etc..).

Yeah, if you specified a probe point with its event name, the tracer will make a corresponding event dynamically. There are also anonymous probes which don't have corresponding events. For those anonymous probes, I need to define two generic event types(kprobe and kretprobe).

Quoted 8 lines
Quoted 1 line

Thank you,

Ok. Btw, why do you need to define those two anonymous events? Actually your event types are always dynamically created. Those you defined through TRACE_FORMAT_EVENT are only "ghost events", they only stand there as a abstract pattern, right?

Not always created.

Below command will create an event "event1";
p probe_point:event1 a1 a2 a3 ... > /debug/tracing/kprobe_events

But next command doesn't create.
p probe_point a1 a2 a3 ... > /debug/tracing/kprobe_events

This just inserts a kprobe to probe_point. the advantage of this "simple" command is that you never be annoyed by making different name for new events :-)

Thank you,

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#20 lore raw

Re: [PATCH -tip -v10 7/7] tracing: add kprobe-based event tracer

Quoted 58 lines · Masami Hiramatsu
Quoted 48 lines · Frederic Weisbecker
Quoted 47 lines · Masami Hiramatsu
Quoted 33 lines · Frederic Weisbecker
Quoted 32 lines · Masami Hiramatsu
Quoted 31 lines · Frederic Weisbecker
Quoted 14 lines
kernel/trace/trace.h +2 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };

kernel/trace/trace.h struct trace_ksym { · line 238

Quoted 10 lines
kernel/trace/trace.h +7 -0
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];

I see that you actually make use of arg as a dynamic sizeable array.
For clarity, args[TRACE_KPROBE_ARGS] could be args[0].

It's just a neat and wouldn't affect the code nor the data but would be clearer for readers of that code.

Quoted 8 lines
Quoted 2 lines

Hmm. In that case, I think we'll need a new macro for field definition, like TRACE_FIELD_ZERO(type, item).

You mean that for trace_define_field() to describe fields of events? Actually the fields should be defined dynamically depending on how is built the kprobe event (which arguments are requested, how many, etc..).

Yeah, if you specified a probe point with its event name, the tracer will make a corresponding event dynamically. There are also anonymous probes which don't have corresponding events. For those anonymous probes, I need to define two generic event types(kprobe and kretprobe).

Quoted 8 lines
Quoted 1 line

Thank you,

Ok. Btw, why do you need to define those two anonymous events? Actually your event types are always dynamically created. Those you defined through TRACE_FORMAT_EVENT are only "ghost events", they only stand there as a abstract pattern, right?

Quoted 7 lines

Not always created.

Below command will create an event "event1";
p probe_point:event1 a1 a2 a3 ... > /debug/tracing/kprobe_events

But next command doesn't create.
p probe_point a1 a2 a3 ... > /debug/tracing/kprobe_events

Aah, ok.

Quoted 3 lines

This just inserts a kprobe to probe_point. the advantage of this "simple" command is that you never be annoyed by making different name for new events :-)

Indeed.
But speaking about that, may be you could dynamically create a name following this simple model: func+offset
Unless we can set several kprobes on the exact same address?

#21 lore raw

Re: [PATCH -tip -v10 7/7] tracing: add kprobe-based event tracer

Quoted 72 lines · Frederic Weisbecker
Quoted 54 lines · Masami Hiramatsu
Quoted 46 lines · Frederic Weisbecker
Quoted 45 lines · Masami Hiramatsu
Quoted 32 lines · Frederic Weisbecker
Quoted 31 lines · Masami Hiramatsu
Quoted 30 lines · Frederic Weisbecker
Quoted 14 lines
kernel/trace/trace.h +2 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };

kernel/trace/trace.h struct trace_ksym { · line 238

Quoted 10 lines
kernel/trace/trace.h +7 -0
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];

I see that you actually make use of arg as a dynamic sizeable array.
For clarity, args[TRACE_KPROBE_ARGS] could be args[0].

It's just a neat and wouldn't affect the code nor the data but would be clearer for readers of that code.

Quoted 7 lines
Quoted 2 lines

Hmm. In that case, I think we'll need a new macro for field definition, like TRACE_FIELD_ZERO(type, item).

You mean that for trace_define_field() to describe fields of events? Actually the fields should be defined dynamically depending on how is built the kprobe event (which arguments are requested, how many, etc..).

Yeah, if you specified a probe point with its event name, the tracer will make a corresponding event dynamically. There are also anonymous probes which don't have corresponding events. For those anonymous probes, I need to define two generic event types(kprobe and kretprobe).

Quoted 7 lines
Quoted 1 line

Thank you,

Ok. Btw, why do you need to define those two anonymous events? Actually your event types are always dynamically created. Those you defined through TRACE_FORMAT_EVENT are only "ghost events", they only stand there as a abstract pattern, right?

Quoted 7 lines

Not always created.

Below command will create an event "event1";
p probe_point:event1 a1 a2 a3 ... > /debug/tracing/kprobe_events

But next command doesn't create.
p probe_point a1 a2 a3 ... > /debug/tracing/kprobe_events

Aah, ok.

Quoted 3 lines

This just inserts a kprobe to probe_point. the advantage of this "simple" command is that you never be annoyed by making different name for new events :-)

Quoted 4 lines

Indeed.
But speaking about that, may be you could dynamically create a name following this simple model: func+offset
Unless we can set several kprobes on the exact same address?

Actually, we can...
I thought that someone might want to insert events in the same address for retrieving more than 6 arguments.

Thanks,

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]
#22 lore raw

Re: [PATCH -tip -v10 7/7] tracing: add kprobe-based event tracer

Quoted 65 lines · Masami Hiramatsu
Quoted 64 lines · Frederic Weisbecker
Quoted 52 lines · Masami Hiramatsu
Quoted 45 lines · Frederic Weisbecker
Quoted 44 lines · Masami Hiramatsu
Quoted 30 lines · Frederic Weisbecker
Quoted 29 lines · Masami Hiramatsu
Quoted 28 lines · Frederic Weisbecker
Quoted 14 lines
kernel/trace/trace.h +2 -0
index 206cb7d..65945eb 100644
@@ -45,6 +45,8 @@ enum trace_type {
 	TRACE_POWER,
 	TRACE_BLK,
 	TRACE_KSYM,
+	TRACE_KPROBE,
+	TRACE_KRETPROBE,
 
 	__TRACE_LAST_TYPE,
 };

kernel/trace/trace.h struct trace_ksym { · line 238

Quoted 10 lines
kernel/trace/trace.h +7 -0
@@ -227,6 +229,22 @@ struct trace_ksym {
 	char			ksym_name[KSYM_NAME_LEN];
 	char			p_name[TASK_COMM_LEN];
 };
+#define TRACE_KPROBE_ARGS 6
+
+struct kprobe_trace_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	int			nargs;
+	unsigned long		args[TRACE_KPROBE_ARGS];

I see that you actually make use of arg as a dynamic sizeable array.
For clarity, args[TRACE_KPROBE_ARGS] could be args[0].

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

It's just a neat and wouldn't affect the code nor the data but would be clearer for readers of that code.

Hmm. In that case, I think we'll need a new macro for field definition, like TRACE_FIELD_ZERO(type, item).

You mean that for trace_define_field() to describe fields of events? Actually the fields should be defined dynamically depending on how is built the kprobe event (which arguments are requested, how many, etc..).

Yeah, if you specified a probe point with its event name, the tracer will make a corresponding event dynamically. There are also anonymous probes which don't have corresponding events. For those anonymous probes, I need to define two generic event types(kprobe and kretprobe).

Quoted 6 lines
Quoted 1 line

Thank you,

Ok. Btw, why do you need to define those two anonymous events? Actually your event types are always dynamically created. Those you defined through TRACE_FORMAT_EVENT are only "ghost events", they only stand there as a abstract pattern, right?

Quoted 7 lines

Not always created.

Below command will create an event "event1";
p probe_point:event1 a1 a2 a3 ... > /debug/tracing/kprobe_events

But next command doesn't create.
p probe_point a1 a2 a3 ... > /debug/tracing/kprobe_events

Aah, ok.

Quoted 7 lines
Quoted 7 lines
Quoted 3 lines

This just inserts a kprobe to probe_point. the advantage of this "simple" command is that you never be annoyed by making different name for new events :-)

Indeed.
But speaking about that, may be you could dynamically create a name following this simple model: func+offset

hmm, and we have two probe types, p(robe) and r(et probe). so, event name should be type@func+offset or type@address.

Quoted 5 lines
Quoted 1 line

Unless we can set several kprobes on the exact same address?

Actually, we can...
I thought that someone might want to insert events in the same address for retrieving more than 6 arguments.

Anyway, I can improve the interface according to user's voice. If you have good idea, I'm happy to hear that:-)

Thank you,

Signature · 7 lines
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: [email protected]