* [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
@ 2009-05-09 0:48 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:48 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: Avi Kivity, H. Peter Anvin, Frederic Weisbecker,
Ananth N Mavinakayanahalli, Andrew Morton, Andi Kleen,
Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap, kvm
Hi,
Here are the patches of kprobe-based event tracer for x86, version 5,
which allows you to probe various kernel events through ftrace interface.
This version supports only x86(-32/-64) (but porting it on other arch
just needs kprobes/kretprobes and register and stack access APIs).
This patchset also includes x86(-64) instruction decoder which
supports non-SSE/FP opcodes and includes x86 opcode map. I think
it will be possible to share this opcode map with KVM's decoder.
This series can be applied on the latest linux-2.6-tip tree.
This patchset includes following changes:
- Add x86 instruction decoder [1/7]
- Check insertion point safety in kprobe [2/7]
- Cleanup fix_riprel() with insn decoder [3/7]
- Add kprobe-tracer plugin [4/7]
- Fix kernel_trap_sp() on x86 according to systemtap runtime. [5/7]
- Add arch-dep register and stack fetching functions [6/7]
- Support fetching various status (register/stack/memory/etc.) [7/7]
Future items:
- .init function tracing support.
- Support primitive types(long, ulong, int, uint, etc) for args.
kprobe-based event tracer
---------------------------
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
/debug/tracing/kprobe_events.
Synopsis of kprobe_events:
p SYMBOL[+offs|-offs]|MEMADDR [FETCHARGS] : set a probe
r SYMBOL[+0] [FETCHARGS] : set a return probe
FETCHARGS:
%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.
E.g.
echo p do_sys_open a0 a1 a2 a3 > /debug/tracing/kprobe_events
This sets a kprobe on the top of do_sys_open() function with recording
1st to 4th arguments.
echo r do_sys_open rv rp >> /debug/tracing/kprobe_events
This sets a kretprobe on the return point of do_sys_open() function with
recording return value and return address.
echo > /debug/tracing/kprobe_events
This clears all probe points. and you can see the traced information via
/debug/tracing/trace.
cat /debug/tracing/trace
# tracer: nop
#
# TASK-PID CPU# TIMESTAMP FUNCTION
# | | | | |
<...>-2376 [001] 262.389131: do_sys_open: @do_sys_open+0 0xffffff9c 0x98db83e 0x8880 0x0
<...>-2376 [001] 262.391166: sys_open: <-do_sys_open+0 0x5 0xc06e8ebb
<...>-2376 [001] 264.384876: do_sys_open: @do_sys_open+0 0xffffff9c 0x98db83e 0x8880 0x0
<...>-2376 [001] 264.386880: sys_open: <-do_sys_open+0 0x5 0xc06e8ebb
<...>-2084 [001] 265.380330: do_sys_open: @do_sys_open+0 0xffffff9c 0x804be3e 0x0 0x1b6
<...>-2084 [001] 265.380399: sys_open: <-do_sys_open+0 0x3 0xc06e8ebb
@SYMBOL means that kernel hits a probe, and <-SYMBOL means kernel returns
from SYMBOL(e.g. "sys_open: <-do_sys_open+0" means kernel returns from
do_sys_open to sys_open).
Thank you,
---
Masami Hiramatsu (7):
tracing: add arguments support on kprobe-based event tracer
x86: add pt_regs register and stack access APIs
x86: fix kernel_trap_sp()
tracing: add kprobe-based event tracer
kprobes: cleanup fix_riprel() using insn decoder on x86
kprobes: checks probe address is instruction boudary on x86
x86: instruction decorder API
Documentation/trace/ftrace.txt | 70 +++
arch/x86/include/asm/inat.h | 125 +++++
arch/x86/include/asm/insn.h | 134 +++++
arch/x86/include/asm/ptrace.h | 70 +++
arch/x86/kernel/kprobes.c | 182 +++----
arch/x86/kernel/ptrace.c | 60 ++
arch/x86/lib/Makefile | 9
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 +++++++++++++
kernel/trace/Kconfig | 9
kernel/trace/Makefile | 1
kernel/trace/trace_kprobe.c | 793 ++++++++++++++++++++++++++++++++
14 files changed, 2922 insertions(+), 107 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
create mode 100644 kernel/trace/trace_kprobe.c
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread* [PATCH -tip v5 1/7] x86: instruction decorder API
2009-05-09 0:48 ` Masami Hiramatsu
(?)
@ 2009-05-09 0:48 ` Masami Hiramatsu
2009-05-11 9:27 ` Christoph Hellwig
2009-05-13 8:23 ` Gleb Natapov
-1 siblings, 2 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:48 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: systemtap, kvm, Masami Hiramatsu, Jim Keniston, H. Peter Anvin,
Steven Rostedt, Ananth N Mavinakayanahalli, Ingo Molnar,
Frederic Weisbecker, Andi Kleen, Vegard Nossum, Avi Kivity
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.
Changes from prototype:
- Support movaps which is used by EFI support code
- Support nopw with many prefixes.
TODO:
- Integrate user-space test harness as a build-time test.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Signed-off-by: Jim Keniston <jkenisto@us.ibm.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Vegard Nossum <vegard.nossum@gmail.com>
Cc: Avi Kivity <avi@redhat.com>
---
arch/x86/include/asm/inat.h | 125 ++++++
arch/x86/include/asm/insn.h | 134 ++++++
arch/x86/lib/Makefile | 9
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, 1844 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
diff --git a/arch/x86/include/asm/inat.h b/arch/x86/include/asm/inat.h
new file mode 100644
index 0000000..01e079a
--- /dev/null
+++ b/arch/x86/include/asm/inat.h
@@ -0,0 +1,125 @@
+#ifndef _ASM_INAT_INAT_H
+#define _ASM_INAT_INAT_H
+/*
+ * x86 instruction attributes
+ *
+ * Written by Masami Hiramatsu <mhiramat@redhat.com>
+ *
+ * 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
diff --git a/arch/x86/include/asm/insn.h b/arch/x86/include/asm/insn.h
new file mode 100644
index 0000000..9fdd650
--- /dev/null
+++ b/arch/x86/include/asm/insn.h
@@ -0,0 +1,134 @@
+#ifndef _ASM_X86_INSN_H
+#define _ASM_X86_INSN_H
+/*
+ * x86 instruction analysis
+ *
+ * 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.
+ *
+ * Copyright (C) IBM Corporation, 2009
+ */
+
+#include <linux/types.h>
+/* insn_attr_t is defined in inat.h */
+#include <asm/inat.h>
+
+struct insn_field {
+ union {
+ s32 value;
+ u8 bytes[4];
+ };
+ bool got; /* true if we've run insn_get_xxx() for this field */
+ u8 nbytes;
+};
+
+struct insn {
+ struct insn_field prefixes; /*
+ * Prefixes
+ * prefixes.bytes[3]: last prefix
+ */
+ struct insn_field rex_prefix; /* REX prefix */
+ struct insn_field opcode; /*
+ * opcode.bytes[0]: opcode1
+ * opcode.bytes[1]: opcode2
+ * opcode.bytes[2]: opcode3
+ */
+ struct insn_field modrm;
+ struct insn_field sib;
+ struct insn_field displacement;
+ union {
+ struct insn_field immediate;
+ struct insn_field moffset1; /* for 64bit MOV */
+ struct insn_field immediate1; /* for 64bit imm or off16/32 */
+ };
+ union {
+ struct insn_field moffset2; /* for 64bit MOV */
+ struct insn_field immediate2; /* for 64bit imm or seg16 */
+ };
+
+ insn_attr_t attr;
+ u8 opnd_bytes;
+ u8 addr_bytes;
+ u8 length;
+ bool x86_64;
+
+ const u8 *kaddr; /* kernel address of insn (copy) to analyze */
+ const u8 *next_byte;
+};
+
+#define OPCODE1(insn) ((insn)->opcode.bytes[0])
+#define OPCODE2(insn) ((insn)->opcode.bytes[1])
+#define OPCODE3(insn) ((insn)->opcode.bytes[2])
+
+#define MODRM_MOD(insn) (((insn)->modrm.value & 0xc0) >> 6)
+#define MODRM_REG(insn) (((insn)->modrm.value & 0x38) >> 3)
+#define MODRM_RM(insn) ((insn)->modrm.value & 0x07)
+
+#define SIB_SCALE(insn) (((insn)->sib.value & 0xc0) >> 6)
+#define SIB_INDEX(insn) (((insn)->sib.value & 0x38) >> 3)
+#define SIB_BASE(insn) ((insn)->sib.value & 0x07)
+
+#define REX_W(insn) ((insn)->rex_prefix.value & 8)
+#define REX_R(insn) ((insn)->rex_prefix.value & 4)
+#define REX_X(insn) ((insn)->rex_prefix.value & 2)
+#define REX_B(insn) ((insn)->rex_prefix.value & 1)
+
+/* The last prefix is needed for two-byte and three-byte opcodes */
+#define LAST_PREFIX(insn) ((insn)->prefixes.bytes[3])
+
+#define MOFFSET64(insn) (((u64)((insn)->moffset2.value) << 32) | \
+ (u32)((insn)->moffset1.value))
+
+#define IMMEDIATE64(insn) (((u64)((insn)->immediate2.value) << 32) | \
+ (u32)((insn)->immediate1.value))
+
+extern void insn_init(struct insn *insn, const u8 *kaddr, bool x86_64);
+extern void insn_get_prefixes(struct insn *insn);
+extern void insn_get_opcode(struct insn *insn);
+extern void insn_get_modrm(struct insn *insn);
+extern void insn_get_sib(struct insn *insn);
+extern void insn_get_displacement(struct insn *insn);
+extern void insn_get_immediate(struct insn *insn);
+extern void insn_get_length(struct insn *insn);
+
+/* Attribute will be determined after getting ModRM (for opcode groups) */
+static inline void insn_get_attr(struct insn *insn)
+{
+ insn_get_modrm(insn);
+}
+
+/* Instruction uses RIP-relative addressing */
+extern bool insn_rip_relative(struct insn *insn);
+
+#ifdef CONFIG_X86_64
+/* Init insn for kernel text */
+#define insn_init_kernel(insn, kaddr) insn_init(insn, kaddr, 1)
+#else /* CONFIG_X86_32 */
+#define insn_init_kernel(insn, kaddr) insn_init(insn, kaddr, 0)
+#endif
+
+#define INSN_PREFIXES_OFFS(insn) (0)
+#define INSN_REXPREFIX_OFFS(insn) ((insn)->prefixes.nbytes)
+#define INSN_OPCODE_OFFS(insn) (INSN_REXPREFIX_OFFS(insn) + \
+ ((insn)->rex_prefix.nbytes))
+#define INSN_MODRM_OFFS(insn) (INSN_OPCODE_OFFS(insn) + \
+ ((insn)->opcode.nbytes))
+#define INSN_SIB_OFFS(insn) (INSN_MODRM_OFFS(insn) + \
+ ((insn)->modrm.nbytes))
+#define INSN_DISPLACEMENT_OFFS(insn) (INSN_SIB_OFFS(insn) + \
+ ((insn)->sib.nbytes))
+#define INSN_IMMEDIATE_OFFS(insn) (INSN_DISPLACEMENT_OFFS(insn) + \
+ ((insn)->displacement.nbytes))
+
+#endif /* _ASM_X86_INSN_H */
diff --git a/arch/x86/lib/Makefile b/arch/x86/lib/Makefile
index 55e11aa..2436975 100644
--- a/arch/x86/lib/Makefile
+++ b/arch/x86/lib/Makefile
@@ -2,12 +2,21 @@
# Makefile for x86 specific library files.
#
+quiet_cmd_inat_tables = GEN $@
+ cmd_inat_tables = awk -f $(srctree)/arch/x86/scripts/gen-insn-attr-x86.awk $(srctree)/arch/x86/lib/x86-opcode-map.txt > $@
+
+$(obj)/inat.o: $(obj)/inat-tables.c
+
+$(obj)/inat-tables.c: FORCE
+ $(call cmd,inat_tables)
+
obj-$(CONFIG_SMP) := msr-on-cpu.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
diff --git a/arch/x86/lib/inat.c b/arch/x86/lib/inat.c
new file mode 100644
index 0000000..d6a34be
--- /dev/null
+++ b/arch/x86/lib/inat.c
@@ -0,0 +1,80 @@
+/*
+ * x86 instruction attribute tables
+ *
+ * Written by Masami Hiramatsu <mhiramat@redhat.com>
+ *
+ * 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/module.h>
+#include <asm/insn.h>
+
+/* Attribute tables are generated from opcode map */
+#include "inat-tables.c"
+
+/* Attribute search APIs */
+insn_attr_t inat_get_opcode_attribute(u8 opcode)
+{
+ return inat_primary_table[opcode];
+}
+
+insn_attr_t inat_get_escape_attribute(u8 opcode, u8 last_pfx,
+ insn_attr_t esc_attr)
+{
+ const insn_attr_t *table;
+ insn_attr_t lpfx_attr;
+ int n, m = 0;
+
+ n = INAT_ESCAPE_NUM(esc_attr);
+ if (last_pfx) {
+ lpfx_attr = inat_get_opcode_attribute(last_pfx);
+ m = INAT_LPREFIX_NUM(lpfx_attr);
+ }
+ table = inat_escape_tables[n][0];
+ if (!table)
+ return 0;
+ if (INAT_HAS_VARIANT(table[opcode]) && m) {
+ table = inat_escape_tables[n][m];
+ if (!table)
+ return 0;
+ }
+ return table[opcode];
+}
+
+#define REGBITS(modrm) (((modrm) >> 3) & 0x7)
+
+insn_attr_t inat_get_group_attribute(u8 modrm, u8 last_pfx,
+ insn_attr_t grp_attr)
+{
+ const insn_attr_t *table;
+ insn_attr_t lpfx_attr;
+ int n, m = 0;
+
+ n = INAT_GROUP_NUM(grp_attr);
+ if (last_pfx) {
+ lpfx_attr = inat_get_opcode_attribute(last_pfx);
+ m = INAT_LPREFIX_NUM(lpfx_attr);
+ }
+ table = inat_group_tables[n][0];
+ if (!table)
+ return INAT_GROUP_COMMON(grp_attr);
+ if (INAT_HAS_VARIANT(table[REGBITS(modrm)]) && m) {
+ table = inat_escape_tables[n][m];
+ if (!table)
+ return INAT_GROUP_COMMON(grp_attr);
+ }
+ return table[REGBITS(modrm)] | INAT_GROUP_COMMON(grp_attr);
+}
+
diff --git a/arch/x86/lib/insn.c b/arch/x86/lib/insn.c
new file mode 100644
index 0000000..58de68e
--- /dev/null
+++ b/arch/x86/lib/insn.c
@@ -0,0 +1,471 @@
+/*
+ * x86 instruction analysis
+ *
+ * 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.
+ *
+ * Copyright (C) IBM Corporation, 2002, 2004, 2009
+ */
+
+#include <linux/string.h>
+#include <linux/module.h>
+#include <asm/inat.h>
+#include <asm/insn.h>
+
+#define get_next(t, insn) \
+ ({t r; r = *(t*)insn->next_byte; insn->next_byte += sizeof(t); r; })
+
+#define peek_next(t, insn) \
+ ({t r; r = *(t*)insn->next_byte; r; })
+
+/**
+ * insn_init() - initialize struct insn
+ * @insn: &struct insn to be initialized
+ * @kaddr: address (in kernel memory) of instruction (or copy thereof)
+ * @x86_64: true for 64-bit kernel or 64-bit app
+ */
+void insn_init(struct insn *insn, const u8 *kaddr, bool x86_64)
+{
+ memset(insn, 0, sizeof(*insn));
+ insn->kaddr = kaddr;
+ insn->next_byte = kaddr;
+ insn->x86_64 = x86_64;
+ insn->opnd_bytes = 4;
+ if (x86_64)
+ insn->addr_bytes = 8;
+ else
+ insn->addr_bytes = 4;
+}
+EXPORT_SYMBOL_GPL(insn_init);
+
+/**
+ * insn_get_prefixes - scan x86 instruction prefix bytes
+ * @insn: &struct insn containing instruction
+ *
+ * Populates the @insn->prefixes bitmap, and updates @insn->next_byte
+ * to point to the (first) opcode. No effect if @insn->prefixes.got
+ * is already true.
+ */
+void insn_get_prefixes(struct insn *insn)
+{
+ struct insn_field *prefixes = &insn->prefixes;
+ insn_attr_t attr;
+ u8 b, lb, i, nb;
+
+ if (prefixes->got)
+ return;
+
+ nb = 0;
+ lb = 0;
+ b = peek_next(u8, insn);
+ attr = inat_get_opcode_attribute(b);
+ while (INAT_IS_PREFIX(attr)) {
+ /* Skip if same prefix */
+ for (i = 0; i < nb; i++)
+ if (prefixes->bytes[i] == b)
+ goto found;
+ if (nb == 4)
+ /* Invalid instruction */
+ break;
+ prefixes->bytes[nb++] = b;
+ if (INAT_IS_ADDRSZ(attr)) {
+ /* address size switches 2/4 or 4/8 */
+ if (insn->x86_64)
+ insn->addr_bytes ^= 12;
+ else
+ insn->addr_bytes ^= 6;
+ } else if (INAT_IS_OPNDSZ(attr)) {
+ /* oprand size switches 2/4 */
+ insn->opnd_bytes ^= 6;
+ }
+found:
+ prefixes->nbytes++;
+ insn->next_byte++;
+ lb = b;
+ b = peek_next(u8, insn);
+ attr = inat_get_opcode_attribute(b);
+ }
+ /* Set the last prefix */
+ if (lb && lb != LAST_PREFIX(insn)) {
+ if (LAST_PREFIX(insn)) {
+ /* Swap the last prefix */
+ b = LAST_PREFIX(insn);
+ for (i = 0; i < nb; i++)
+ if (prefixes->bytes[i] == lb)
+ prefixes->bytes[i] = b;
+ }
+ LAST_PREFIX(insn) = lb;
+ }
+
+ if (insn->x86_64) {
+ b = peek_next(u8, insn);
+ attr = inat_get_opcode_attribute(b);
+ if (INAT_IS_REX_PREFIX(attr)) {
+ insn->rex_prefix.value = b;
+ insn->rex_prefix.nbytes = 1;
+ insn->rex_prefix.got = true;
+ insn->next_byte++;
+ if (REX_W(insn))
+ /* REX.W overrides opnd_size */
+ insn->opnd_bytes = 8;
+ }
+ }
+ prefixes->got = true;
+ return;
+}
+EXPORT_SYMBOL_GPL(insn_get_prefixes);
+
+/**
+ * insn_get_opcode - collect opcode(s)
+ * @insn: &struct insn containing instruction
+ *
+ * Populates @insn->opcode, updates @insn->next_byte to point past the
+ * opcode byte(s), and set @insn->attr (except for groups).
+ * If necessary, first collects any preceding (prefix) bytes.
+ * Sets @insn->opcode.value = opcode1. No effect if @insn->opcode.got
+ * is already true.
+ *
+ */
+void insn_get_opcode(struct insn *insn)
+{
+ struct insn_field *opcode = &insn->opcode;
+ u8 op, pfx;
+ if (opcode->got)
+ return;
+ if (!insn->prefixes.got)
+ insn_get_prefixes(insn);
+
+ /* Get first opcode */
+ op = get_next(u8, insn);
+ OPCODE1(insn) = op;
+ opcode->nbytes = 1;
+ insn->attr = inat_get_opcode_attribute(op);
+ while (INAT_IS_ESCAPE(insn->attr)) {
+ /* Get escaped opcode */
+ op = get_next(u8, insn);
+ opcode->bytes[opcode->nbytes++] = op;
+ pfx = LAST_PREFIX(insn);
+ insn->attr = inat_get_escape_attribute(op, pfx, insn->attr);
+ }
+ opcode->got = true;
+}
+EXPORT_SYMBOL_GPL(insn_get_opcode);
+
+/**
+ * insn_get_modrm - collect ModRM byte, if any
+ * @insn: &struct insn containing instruction
+ *
+ * Populates @insn->modrm and updates @insn->next_byte to point past the
+ * ModRM byte, if any. If necessary, first collects the preceding bytes
+ * (prefixes and opcode(s)). No effect if @insn->modrm.got is already true.
+ */
+void insn_get_modrm(struct insn *insn)
+{
+ struct insn_field *modrm = &insn->modrm;
+ u8 pfx, mod;
+ if (modrm->got)
+ return;
+ if (!insn->opcode.got)
+ insn_get_opcode(insn);
+
+ if (INAT_HAS_MODRM(insn->attr)) {
+ mod = get_next(u8, insn);
+ modrm->value = mod;
+ modrm->nbytes = 1;
+ if (INAT_IS_GROUP(insn->attr)) {
+ pfx = LAST_PREFIX(insn);
+ insn->attr = inat_get_group_attribute(mod, pfx,
+ insn->attr);
+ }
+ }
+
+ if (insn->x86_64 && INAT_IS_FORCE64(insn->attr))
+ insn->opnd_bytes = 8;
+ modrm->got = true;
+}
+EXPORT_SYMBOL_GPL(insn_get_modrm);
+
+
+/**
+ * insn_rip_relative() - Does instruction use RIP-relative addressing mode?
+ * @insn: &struct insn containing instruction
+ *
+ * If necessary, first collects the instruction up to and including the
+ * ModRM byte. No effect if @insn->x86_64 is false.
+ */
+bool insn_rip_relative(struct insn *insn)
+{
+ struct insn_field *modrm = &insn->modrm;
+
+ if (!insn->x86_64)
+ return false;
+ if (!modrm->got)
+ insn_get_modrm(insn);
+ /*
+ * For rip-relative instructions, the mod field (top 2 bits)
+ * is zero and the r/m field (bottom 3 bits) is 0x5.
+ */
+ return (modrm->nbytes && (modrm->value & 0xc7) == 0x5);
+}
+EXPORT_SYMBOL_GPL(insn_rip_relative);
+
+/**
+ *
+ * insn_get_sib() - Get the SIB byte of instruction
+ * @insn: &struct insn containing instruction
+ *
+ * If necessary, first collects the instruction up to and including the
+ * ModRM byte.
+ */
+void insn_get_sib(struct insn *insn)
+{
+ if (insn->sib.got)
+ return;
+ if (!insn->modrm.got)
+ insn_get_modrm(insn);
+ if (insn->modrm.nbytes)
+ if (insn->addr_bytes != 2 &&
+ MODRM_MOD(insn) != 3 && MODRM_RM(insn) == 4) {
+ insn->sib.value = get_next(u8, insn);
+ insn->sib.nbytes = 1;
+ }
+ insn->sib.got = true;
+}
+EXPORT_SYMBOL_GPL(insn_get_sib);
+
+
+/**
+ *
+ * insn_get_displacement() - Get the displacement of instruction
+ * @insn: &struct insn containing instruction
+ *
+ * If necessary, first collects the instruction up to and including the
+ * SIB byte.
+ * Displacement value is sign-expanded.
+ */
+void insn_get_displacement(struct insn *insn)
+{
+ u8 mod;
+ if (insn->displacement.got)
+ return;
+ if (!insn->sib.got)
+ insn_get_sib(insn);
+ if (insn->modrm.nbytes) {
+ /*
+ * Interpreting the modrm byte:
+ * mod = 00 - no displacement fields (exceptions below)
+ * mod = 01 - 1-byte displacement field
+ * mod = 10 - displacement field is 4 bytes, or 2 bytes if
+ * address size = 2 (0x67 prefix in 32-bit mode)
+ * mod = 11 - no memory operand
+ *
+ * If address size = 2...
+ * mod = 00, r/m = 110 - displacement field is 2 bytes
+ *
+ * If address size != 2...
+ * mod != 11, r/m = 100 - SIB byte exists
+ * mod = 00, SIB base = 101 - displacement field is 4 bytes
+ * mod = 00, r/m = 101 - rip-relative addressing, displacement
+ * field is 4 bytes
+ */
+ mod = MODRM_MOD(insn);
+ if (mod == 3)
+ goto out;
+ if (mod == 1) {
+ insn->displacement.value = get_next(s8, insn);
+ insn->displacement.nbytes = 1;
+ } else if (insn->addr_bytes == 2) {
+ if ((mod == 0 && MODRM_RM(insn) == 6) || mod == 2) {
+ insn->displacement.value = get_next(s16, insn);
+ insn->displacement.nbytes = 2;
+ }
+ } else {
+ if ((mod == 0 && MODRM_RM(insn) == 5) || mod == 2 ||
+ (mod == 0 && SIB_BASE(insn) == 5)) {
+ insn->displacement.value = get_next(s32, insn);
+ insn->displacement.nbytes = 4;
+ }
+ }
+ }
+out:
+ insn->displacement.got = true;
+}
+EXPORT_SYMBOL_GPL(insn_get_displacement);
+
+/* Decode moffset16/32/64 */
+static void __get_moffset(struct insn *insn)
+{
+ switch (insn->addr_bytes) {
+ case 2:
+ insn->moffset1.value = get_next(s16, insn);
+ insn->moffset1.nbytes = 2;
+ break;
+ case 4:
+ insn->moffset1.value = get_next(s32, insn);
+ insn->moffset1.nbytes = 4;
+ break;
+ case 8:
+ insn->moffset1.value = get_next(s32, insn);
+ insn->moffset1.nbytes = 4;
+ insn->moffset2.value = get_next(s32, insn);
+ insn->moffset2.nbytes = 4;
+ break;
+ }
+ insn->moffset1.got = insn->moffset2.got = true;
+}
+
+/* Decode imm v32(Iz) */
+static void __get_immv32(struct insn *insn)
+{
+ switch (insn->opnd_bytes) {
+ case 2:
+ insn->immediate.value = get_next(s16, insn);
+ insn->immediate.nbytes = 2;
+ break;
+ case 4:
+ case 8:
+ insn->immediate.value = get_next(s32, insn);
+ insn->immediate.nbytes = 4;
+ break;
+ }
+}
+
+/* Decode imm v64(Iv/Ov) */
+static void __get_immv(struct insn *insn)
+{
+ switch (insn->opnd_bytes) {
+ case 2:
+ insn->immediate1.value = get_next(s16, insn);
+ insn->immediate1.nbytes = 2;
+ break;
+ case 4:
+ insn->immediate1.value = get_next(s32, insn);
+ insn->immediate1.nbytes = 4;
+ break;
+ case 8:
+ insn->immediate1.value = get_next(s32, insn);
+ insn->immediate1.nbytes = 4;
+ insn->immediate2.value = get_next(s32, insn);
+ insn->immediate2.nbytes = 4;
+ break;
+ }
+ insn->immediate1.got = insn->immediate2.got = true;
+}
+
+/* Decode ptr16:16/32(Ap) */
+static void __get_immptr(struct insn *insn)
+{
+ switch (insn->opnd_bytes) {
+ case 2:
+ insn->immediate1.value = get_next(s16, insn);
+ insn->immediate1.nbytes = 2;
+ break;
+ case 4:
+ insn->immediate1.value = get_next(s32, insn);
+ insn->immediate1.nbytes = 4;
+ break;
+ case 8:
+ /* ptr16:64 is not supported (no segment) */
+ WARN_ON(1);
+ return;
+ }
+ insn->immediate2.value = get_next(u16, insn);
+ insn->immediate2.nbytes = 2;
+ insn->immediate1.got = insn->immediate2.got = true;
+}
+
+/**
+ *
+ * insn_get_immediate() - Get the immediates of instruction
+ * @insn: &struct insn containing instruction
+ *
+ * If necessary, first collects the instruction up to and including the
+ * displacement bytes.
+ * Basically, most of immediates are sign-expanded. Unsigned-value can be
+ * get by bit masking with ((1 << (nbytes * 8)) - 1)
+ */
+void insn_get_immediate(struct insn *insn)
+{
+ if (insn->immediate.got)
+ return;
+ if (!insn->displacement.got)
+ insn_get_displacement(insn);
+
+ if (INAT_HAS_MOFFSET(insn->attr)) {
+ __get_moffset(insn);
+ goto done;
+ }
+
+ if (!INAT_HAS_IMM(insn->attr))
+ /* no immediates */
+ goto done;
+
+ switch (INAT_IMM_SIZE(insn->attr)) {
+ case INAT_IMM_BYTE:
+ insn->immediate.value = get_next(s8, insn);
+ insn->immediate.nbytes = 1;
+ break;
+ case INAT_IMM_WORD:
+ insn->immediate.value = get_next(s16, insn);
+ insn->immediate.nbytes = 2;
+ break;
+ case INAT_IMM_DWORD:
+ insn->immediate.value = get_next(s32, insn);
+ insn->immediate.nbytes = 4;
+ break;
+ case INAT_IMM_QWORD:
+ insn->immediate1.value = get_next(s32, insn);
+ insn->immediate1.nbytes = 4;
+ insn->immediate2.value = get_next(s32, insn);
+ insn->immediate2.nbytes = 4;
+ break;
+ case INAT_IMM_PTR:
+ __get_immptr(insn);
+ break;
+ case INAT_IMM_VWORD32:
+ __get_immv32(insn);
+ break;
+ case INAT_IMM_VWORD:
+ __get_immv(insn);
+ break;
+ default:
+ break;
+ }
+ if (INAT_HAS_ADDIMM(insn->attr)) {
+ insn->immediate2.value = get_next(s8, insn);
+ insn->immediate2.nbytes = 1;
+ }
+done:
+ insn->immediate.got = true;
+}
+EXPORT_SYMBOL_GPL(insn_get_immediate);
+
+/**
+ *
+ * insn_get_length() - Get the length of instruction
+ * @insn: &struct insn containing instruction
+ *
+ * If necessary, first collects the instruction up to and including the
+ * immediates bytes.
+ */
+void insn_get_length(struct insn *insn)
+{
+ if (insn->length)
+ return;
+ if (!insn->immediate.got)
+ insn_get_immediate(insn);
+ insn->length = (u8)((unsigned long)insn->next_byte
+ - (unsigned long)insn->kaddr);
+}
+EXPORT_SYMBOL_GPL(insn_get_length);
diff --git a/arch/x86/lib/x86-opcode-map.txt b/arch/x86/lib/x86-opcode-map.txt
new file mode 100644
index 0000000..ab2a58d
--- /dev/null
+++ b/arch/x86/lib/x86-opcode-map.txt
@@ -0,0 +1,711 @@
+# x86 Opcode Maps
+#
+#<Opcode maps>
+# Table: table-name
+# Referrer: escaped-name
+# opcode: mnemonic|GrpXXX [operand1[,operand2...]] [(extra1)[,(extra2)...] [| 2nd-mnemonic ...]
+# (or)
+# opcode: escape # escaped-name
+# EndTable
+#
+#<group maps>
+# GrpTable: GrpXXX
+# reg: mnemonic [operand1[,operand2...]] [(extra1)[,(extra2)...] [| 2nd-mnemonic ...]
+# EndTable
+#
+
+Table: one byte opcode
+Referrer:
+# 0x00 - 0x0f
+00: ADD Eb,Gb
+01: ADD Ev,Gv
+02: ADD Gb,Eb
+03: ADD Gv,Ev
+04: ADD AL,Ib
+05: ADD rAX,Iz
+06: PUSH ES (i64)
+07: POP ES (i64)
+08: OR Eb,Gb
+09: OR Ev,Gv
+0a: OR Gb,Eb
+0b: OR Gv,Ev
+0c: OR AL,Ib
+0d: OR rAX,Iz
+0e: PUSH CS (i64)
+0f: escape # 2-byte escape
+# 0x10 - 0x1f
+10: ADC Eb,Gb
+11: ADC Ev,Gv
+12: ADC Gb,Eb
+13: ADC Gv,Ev
+14: ADC AL,Ib
+15: ADC rAX,Iz
+16: PUSH SS (i64)
+17: POP SS (i64)
+18: SBB Eb,Gb
+19: SBB Ev,Gv
+1a: SBB Gb,Eb
+1b: SBB Gv,Ev
+1c: SBB AL,Ib
+1d: SBB rAX,Iz
+1e: PUSH DS (i64)
+1f: POP DS (i64)
+# 0x20 - 0x2f
+20: AND Eb,Gb
+21: AND Ev,Gv
+22: AND Gb,Eb
+23: AND Gv,Ev
+24: AND AL,Ib
+25: AND rAx,Iz
+26: SEG=ES (Prefix)
+27: DAA (i64)
+28: SUB Eb,Gb
+29: SUB Ev,Gv
+2a: SUB Gb,Eb
+2b: SUB Gv,Ev
+2c: SUB AL,Ib
+2d: SUB rAX,Iz
+2e: SEG=CS (Prefix)
+2f: DAS (i64)
+# 0x30 - 0x3f
+30: XOR Eb,Gb
+31: XOR Ev,Gv
+32: XOR Gb,Eb
+33: XOR Gv,Ev
+34: XOR AL,Ib
+35: XOR rAX,Iz
+36: SEG=SS (Prefix)
+37: AAA (i64)
+38: CMP Eb,Gb
+39: CMP Ev,Gv
+3a: CMP Gb,Eb
+3b: CMP Gv,Ev
+3c: CMP AL,Ib
+3d: CMP rAX,Iz
+3e: SEG=DS (Prefix)
+3f: AAS (i64)
+# 0x40 - 0x4f
+40: INC eAX (i64) | REX (o64)
+41: INC eCX (i64) | REX.B (o64)
+42: INC eDX (i64) | REX.X (o64)
+43: INC eBX (i64) | REX.XB (o64)
+44: INC eSP (i64) | REX.R (o64)
+45: INC eBP (i64) | REX.RB (o64)
+46: INC eSI (i64) | REX.RX (o64)
+47: INC eDI (i64) | REX.RXB (o64)
+48: DEC eAX (i64) | REX.W (o64)
+49: DEC eCX (i64) | REX.WB (o64)
+4a: DEC eDX (i64) | REX.WX (o64)
+4b: DEC eBX (i64) | REX.WXB (o64)
+4c: DEC eSP (i64) | REX.WR (o64)
+4d: DEC eBP (i64) | REX.WRB (o64)
+4e: DEC eSI (i64) | REX.WRX (o64)
+4f: DEC eDI (i64) | REX.WRXB (o64)
+# 0x50 - 0x5f
+50: PUSH rAX/r8 (d64)
+51: PUSH rCX/r9 (d64)
+52: PUSH rDX/r10 (d64)
+53: PUSH rBX/r11 (d64)
+54: PUSH rSP/r12 (d64)
+55: PUSH rBP/r13 (d64)
+56: PUSH rSI/r14 (d64)
+57: PUSH rDI/r15 (d64)
+58: POP rAX/r8 (d64)
+59: POP rCX/r9 (d64)
+5a: POP rDX/r10 (d64)
+5b: POP rBX/r11 (d64)
+5c: POP rSP/r12 (d64)
+5d: POP rBP/r13 (d64)
+5e: POP rSI/r14 (d64)
+5f: POP rDI/r15 (d64)
+# 0x60 - 0x6f
+60: PUSHA/PUSHAD (i64)
+61: POPA/POPAD (i64)
+62: BOUND Gv,Ma (i64)
+63: ARPL Ew,Gw (i64) | MOVSXD Gv,Ev (o64)
+64: SEG=FS (Prefix)
+65: SEG=GS (Prefix)
+66: Operand-Size (Prefix)
+67: Address-Size (Prefix)
+68: PUSH Iz (d64)
+69: IMUL Gv,Ev,Iz
+6a: PUSH Ib (d64)
+6b: IMUL Gv,Ev,Ib
+6c: INS/INSB Yb,DX
+6d: INS/INSW/INSD Yz,DX
+6e: OUTS/OUTSB DX,Xb
+6f: OUTS/OUTSW/OUTSD DX,Xz
+# 0x70 - 0x7f
+70: JO Jb
+71: JNO Jb
+72: JB/JNAE/JC Jb
+73: JNB/JAE/JNC Jb
+74: JZ/JE Jb
+75: JNZ/JNE Jb
+76: JBE/JNA Jb
+77: JNBE/JA Jb
+78: JS Jb
+79: JNS Jb
+7a: JP/JPE Jb
+7b: JNP/JPO Jb
+7c: JL/JNGE Jb
+7d: JNL/JGE Jb
+7e: JLE/JNG Jb
+7f: JNLE/JG Jb
+# 0x80 - 0x8f
+80: Grp1 Eb,Ib (1A)
+81: Grp1 Ev,Iz (1A)
+82: Grp1 Eb,Ib (1A),(i64)
+83: Grp1 Ev,Ib (1A)
+84: TEST Eb,Gb
+85: TEST Ev,Gv
+86: XCHG Eb,Gb
+87: XCHG Ev,Gv
+88: MOV Eb,Gb
+89: MOV Ev,Gv
+8a: MOV Gb,Eb
+8b: MOV Gv,Ev
+8c: MOV Ev,Sw
+8d: LEA Gv,M
+8e: MOV Sw,Ew
+8f: Grp1A (1A) | POP Ev (d64)
+# 0x90 - 0x9f
+90: NOP | PAUSE (F3) | XCHG r8,rAX
+91: XCHG rCX/r9,rAX
+92: XCHG rDX/r10,rAX
+93: XCHG rBX/r11,rAX
+94: XCHG rSP/r12,rAX
+95: XCHG rBP/r13,rAX
+96: XCHG rSI/r14,rAX
+97: XCHG rDI/r15,rAX
+98: CBW/CWDE/CDQE
+99: CWD/CDQ/CQO
+9a: CALLF Ap (i64)
+9b: FWAIT/WAIT
+9c: PUSHF/D/Q Fv (d64)
+9d: POPF/D/Q Fv (d64)
+9e: SAHF
+9f: LAHF
+# 0xa0 - 0xaf
+a0: MOV AL,Ob
+a1: MOV rAX,Ov
+a2: MOV Ob,AL
+a3: MOV Ov,rAX
+a4: MOVS/B Xb,Yb
+a5: MOVS/W/D/Q Xv,Yv
+a6: CMPS/B Xb,Yb
+a7: CMPS/W/D Xv,Yv
+a8: TEST AL,Ib
+a9: TEST rAX,Iz
+aa: STOS/B Yb,AL
+ab: STOS/W/D/Q Yv,rAX
+ac: LODS/B AL,Xb
+ad: LODS/W/D/Q rAX,Xv
+ae: SCAS/B AL,Yb
+af: SCAS/W/D/Q rAX,Xv
+# 0xb0 - 0xbf
+b0: MOV AL/R8L,Ib
+b1: MOV CL/R9L,Ib
+b2: MOV DL/R10L,Ib
+b3: MOV BL/R11L,Ib
+b4: MOV AH/R12L,Ib
+b5: MOV CH/R13L,Ib
+b6: MOV DH/R14L,Ib
+b7: MOV BH/R15L,Ib
+b8: MOV rAX/r8,Iv
+b9: MOV rCX/r9,Iv
+ba: MOV rDX/r10,Iv
+bb: MOV rBX/r11,Iv
+bc: MOV rSP/r12,Iv
+bd: MOV rBP/r13,Iv
+be: MOV rSI/r14,Iv
+bf: MOV rDI/r15,Iv
+# 0xc0 - 0xcf
+c0: Grp2 Eb,Ib (1A)
+c1: Grp2 Ev,Ib (1A)
+c2: RETN Iw (f64)
+c3: RETN
+c4: LES Gz,Mp (i64)
+c5: LDS Gz,Mp (i64)
+c6: Grp11 Eb,Ib (1A)
+c7: Grp11 Ev,Iz (1A)
+c8: ENTER Iw,Ib
+c9: LEAVE (d64)
+ca: RETF Iw
+cb: RETF
+cc: INT3
+cd: INT Ib
+ce: INTO (i64)
+cf: IRET/D/Q
+# 0xd0 - 0xdf
+d0: Grp2 Eb,1 (1A)
+d1: Grp2 Ev,1 (1A)
+d2: Grp2 Eb,CL (1A)
+d3: Grp2 Ev,CL (1A)
+d4: AAM Ib (i64)
+d5: AAD Ib (i64)
+d6:
+d7: XLAT/XLATB
+d8: ESC
+d9: ESC
+da: ESC
+db: ESC
+dc: ESC
+dd: ESC
+de: ESC
+df: ESC
+# 0xe0 - 0xef
+e0: LOOPNE/LOOPNZ Jb (f64)
+e1: LOOPE/LOOPZ Jb (f64)
+e2: LOOP Jb (f64)
+e3: JrCXZ Jb (f64)
+e4: IN AL,Ib
+e5: IN eAX,Ib
+e6: OUT Ib,AL
+e7: OUT Ib,eAX
+e8: CALL Jz (f64)
+e9: JMP-near Jz (f64)
+ea: JMP-far Ap (i64)
+eb: JMP-short Jb (f64)
+ec: IN AL,DX
+ed: IN eAX,DX
+ee: OUT DX,AL
+ef: OUT DX,eAX
+# 0xf0 - 0xff
+f0: LOCK (Prefix)
+f1:
+f2: REPNE (Prefix)
+f3: REP/REPE (Prefix)
+f4: HLT
+f5: CMC
+f6: Grp3_1 Eb (1A)
+f7: Grp3_2 Ev (1A)
+f8: CLC
+f9: STC
+fa: CLI
+fb: STI
+fc: CLD
+fd: STD
+fe: Grp4 (1A)
+ff: Grp5 (1A)
+EndTable
+
+Table: 2-byte opcode # First Byte is 0x0f
+Referrer: 2-byte escape
+# 0x0f 0x00-0x0f
+00: Grp6 (1A)
+01: Grp7 (1A)
+02: LAR Gv,Ew
+03: LSL Gv,Ew
+04:
+05: SYSCALL (o64)
+06: CLTS
+07: SYSRET (o64)
+08: INVD
+09: WBINVD
+0a:
+0b: UD2 (1B)
+0c:
+0d: NOP Ev
+0e:
+0f:
+# 0x0f 0x10-0x1f
+10:
+11:
+12:
+13:
+14:
+15:
+16:
+17:
+18: Grp16 (1A)
+19:
+1a:
+1b:
+1c:
+1d:
+1e:
+1f: NOP Ev
+# 0x0f 0x20-0x2f
+20: MOV Rd,Cd
+21: MOV Rd,Dd
+22: MOV Cd,Rd
+23: MOV Dd,Rd
+24:
+25:
+26:
+27:
+28: movaps Vps,Wps | movapd Vpd,Wpd (66)
+29: movaps Wps,Vps | movapd Wpd,Vpd (66)
+2a:
+2b:
+2c:
+2d:
+2e:
+2f:
+# 0x0f 0x30-0x3f
+30: WRMSR
+31: RDTSC
+32: RDMSR
+33: RDPMC
+34: SYSENTER
+35: SYSEXIT
+36:
+37: GETSEC
+38: escape # 3-byte escape 1
+39:
+3a: escape # 3-byte escape 2
+3b:
+3c:
+3d:
+3e:
+3f:
+# 0x0f 0x40-0x4f
+40: CMOVO Gv,Ev
+41: CMOVNO Gv,Ev
+42: CMOVB/C/NAE Gv,Ev
+43: CMOVAE/NB/NC Gv,Ev
+44: CMOVE/Z Gv,Ev
+45: CMOVNE/NZ Gv,Ev
+46: CMOVBE/NA Gv,Ev
+47: CMOVA/NBE Gv,Ev
+48: CMOVS Gv,Ev
+49: CMOVNS Gv,Ev
+4a: CMOVP/PE Gv,Ev
+4b: CMOVNP/PO Gv,Ev
+4c: CMOVL/NGE Gv,Ev
+4d: CMOVNL/GE Gv,Ev
+4e: CMOVLE/NG Gv,Ev
+4f: CMOVNLE/G Gv,Ev
+# 0x0f 0x50-0x5f
+50:
+51:
+52:
+53:
+54:
+55:
+56:
+57:
+58:
+59:
+5a:
+5b:
+5c:
+5d:
+5e:
+5f:
+# 0x0f 0x60-0x6f
+60:
+61:
+62:
+63:
+64:
+65:
+66:
+67:
+68:
+69:
+6a:
+6b:
+6c:
+6d:
+6e:
+6f:
+# 0x0f 0x70-0x7f
+70:
+71: Grp12 (1A)
+72: Grp13 (1A)
+73: Grp14 (1A)
+74:
+75:
+76:
+77:
+78: VMREAD Ed/q,Gd/q
+79: VMWRITE Gd/q,Ed/q
+7a:
+7b:
+7c:
+7d:
+7e:
+7f:
+# 0x0f 0x80-0x8f
+80: JO Jz (f64)
+81: JNO Jz (f64)
+82: JB/JNAE/JC Jz (f64)
+83: JNB/JAE/JNC Jz (f64)
+84: JZ/JE Jz (f64)
+85: JNZ/JNE Jz (f64)
+86: JBE/JNA Jz (f64)
+87: JNBE/JA Jz (f64)
+88: JS Jz (f64)
+89: JNS Jz (f64)
+8a: JP/JPE Jz (f64)
+8b: JNP/JPO Jz (f64)
+8c: JL/JNGE Jz (f64)
+8d: JNL/JGE Jz (f64)
+8e: JLE/JNG Jz (f64)
+8f: JNLE/JG Jz (f64)
+# 0x0f 0x90-0x9f
+90: SETO Eb
+91: SETNO Eb
+92: SETB/C/NAE Eb
+93: SETAE/NB/NC Eb
+94: SETE/Z Eb
+95: SETNE/NZ Eb
+96: SETBE/NA Eb
+97: SETA/NBE Eb
+98: SETS Eb
+99: SETNS Eb
+9a: SETP/PE Eb
+9b: SETNP/PO Eb
+9c: SETL/NGE Eb
+9d: SETNL/GE Eb
+9e: SETLE/NG Eb
+9f: SETNLE/G Eb
+# 0x0f 0xa0-0xaf
+a0: PUSH FS (d64)
+a1: POP FS (d64)
+a2: CPUID
+a3: BT Ev,Gv
+a4: SHLD Ev,Gv,Ib
+a5: SHLD Ev,Gv,CL
+a6:
+a7:
+a8: PUSH GS (d64)
+a9: POP GS (d64)
+aa: RSM
+ab: BTS Ev,Gv
+ac: SHRD Ev,Gv,Ib
+ad: SHRD Ev,Gv,CL
+ae: Grp15 (1A),(1C)
+af: IMUL Gv,Ev
+# 0x0f 0xb0-0xbf
+b0: CMPXCHG Eb,Gb
+b1: CMPXCHG Ev,Gv
+b2: LSS Gv,Mp
+b3: BTR Ev,Gv
+b4: LFS Gv,Mp
+b5: LGS Gv,Mp
+b6: MOVZX Gv,Eb
+b7: MOVZX Gv,Ew
+b8: JMPE | POPCNT Gv,Ev (F3)
+b9: Grp10 (1A)
+ba: Grp8 Ev,Ib (1A)
+bb: BTC Ev,Gv
+bc: BSF Gv,Ev
+bd: BSR Gv,Ev
+be: MOVSX Gv,Eb
+bf: MOVSX Gv,Ew
+# 0x0f 0xc0-0xcf
+c0: XADD Eb,Gb
+c1: XADD Ev,Gv
+c2:
+c3: movnti Md/q,Gd/q
+c4:
+c5:
+c6:
+c7: Grp9 (1A)
+c8: BSWAP RAX/EAX/R8/R8D
+c9: BSWAP RCX/ECX/R9/R9D
+ca: BSWAP RDX/EDX/R10/R10D
+cb: BSWAP RBX/EBX/R11/R11D
+cc: BSWAP RSP/ESP/R12/R12D
+cd: BSWAP RBP/EBP/R13/R13D
+ce: BSWAP RSI/ESI/R14/R14D
+cf: BSWAP RDI/EDI/R15/R15D
+# 0x0f 0xd0-0xdf
+d0:
+d1:
+d2:
+d3:
+d4:
+d5:
+d6:
+d7:
+d8:
+d9:
+da:
+db:
+dc:
+dd:
+de:
+df:
+# 0x0f 0xe0-0xef
+e0:
+e1:
+e2:
+e3:
+e4:
+e5:
+e6:
+e7:
+e8:
+e9:
+ea:
+eb:
+ec:
+ed:
+ee:
+ef:
+# 0x0f 0xf0-0xff
+f0:
+f1:
+f2:
+f3:
+f4:
+f5:
+f6:
+f7:
+f8:
+f9:
+fa:
+fb:
+fc:
+fd:
+fe:
+ff:
+EndTable
+
+Table: 3-byte opcode 1
+Referrer: 3-byte escape 1
+80: INVEPT Gd/q,Mdq (66)
+81: INVPID Gd/q,Mdq (66)
+f0: MOVBE Gv,Mv | CRC32 Gd,Eb (F2)
+f1: MOVBE Mv,Gv | CRC32 Gd,Ev (F2)
+EndTable
+
+Table: 3-byte opcode 2
+Referrer: 3-byte escape 2
+# all opcode is for SSE
+EndTable
+
+GrpTable: Grp1
+0: ADD
+1: OR
+2: ADC
+3: SBB
+4: AND
+5: SUB
+6: XOR
+7: CMP
+EndTable
+
+GrpTable: Grp1A
+0: POP
+EndTable
+
+GrpTable: Grp2
+0: ROL
+1: ROR
+2: RCL
+3: RCR
+4: SHL/SAL
+5: SHR
+6:
+7: SAR
+EndTable
+
+GrpTable: Grp3_1
+0: TEST Eb,Ib
+1:
+2: NOT Eb
+3: NEG Eb
+4: MUL AL,Eb
+5: IMUL AL,Eb
+6: DIV AL,Eb
+7: IDIV AL,Eb
+EndTable
+
+GrpTable: Grp3_2
+0: TEST Ev,Iz
+1:
+2: NOT Ev
+3: NEG Ev
+4: MUL rAX,Ev
+5: IMUL rAX,Ev
+6: DIV rAX,Ev
+7: IDIV rAX,Ev
+EndTable
+
+GrpTable: Grp4
+0: INC Eb
+1: DEC Eb
+EndTable
+
+GrpTable: Grp5
+0: INC Ev
+1: DEC Ev
+2: CALLN Ev (f64)
+3: CALLF Ep
+4: JMPN Ev (f64)
+5: JMPF Ep
+6: PUSH Ev (d64)
+7:
+EndTable
+
+GrpTable: Grp6
+0: SLDT Rv/Mw
+1: STR Rv/Mw
+2: LLDT Ew
+3: LTR Ew
+4: VERR Ew
+5: VERW Ew
+EndTable
+
+GrpTable: Grp7
+0: SGDT Ms | VMCALL (001),(11B) | VMLAUNCH (010),(11B) | VMRESUME (011),(11B) | VMXOFF (100),(11B)
+1: SIDT Ms | MONITOR (000),(11B) | MWAIT (001)
+2: LGDT Ms | XGETBV (000),(11B) | XSETBV (001),(11B)
+3: LIDT Ms
+4: SMSW Mw/Rv
+5:
+6: LMSW Ew
+7: INVLPG Mb | SWAPGS (o64),(000),(11B) | RDTSCP (001),(11B)
+EndTable
+
+GrpTable: Grp8
+4: BT
+5: BTS
+6: BTR
+7: BTC
+EndTable
+
+GrpTable: Grp9
+1: CMPXCHG8B/16B Mq/Mdq
+6: VMPTRLD Mq | VMCLEAR Mq (66) | VMXON Mq (F3)
+7: VMPTRST Mq
+EndTable
+
+GrpTable: Grp10
+EndTable
+
+GrpTable: Grp11
+0: MOV
+EndTable
+
+GrpTable: Grp12
+EndTable
+
+GrpTable: Grp13
+EndTable
+
+GrpTable: Grp14
+EndTable
+
+GrpTable: Grp15
+0: fxsave
+1: fxstor
+2: ldmxcsr
+3: stmxcsr
+4: XSAVE
+5: XRSTOR | lfence (11B)
+6: mfence (11B)
+7: clflush | sfence (11B)
+EndTable
+
+GrpTable: Grp16
+0: prefetch NTA
+1: prefetch T0
+2: prefetch T1
+3: prefetch T2
+EndTable
diff --git a/arch/x86/scripts/gen-insn-attr-x86.awk b/arch/x86/scripts/gen-insn-attr-x86.awk
new file mode 100644
index 0000000..869f3cc
--- /dev/null
+++ b/arch/x86/scripts/gen-insn-attr-x86.awk
@@ -0,0 +1,314 @@
+#!/bin/awk -f
+# gen-insn-attr-x86.awk: Instruction attribute table generator
+# Written by Masami Hiramatsu <mhiramat@redhat.com>
+#
+# Usage: cat x86-opcode-map.txt | ./gen-insn-attr-x86.awk > inat-tables.c
+
+BEGIN {
+ print "/* x86 opcode map generated from x86-opcode-map.txt */"
+ print "/* Do not change this code. */"
+ ggid = 1
+ geid = 1
+
+ opnd_expr = "^[A-Za-z]"
+ ext_expr = "^\\("
+ sep_expr = "^\\|$"
+ group_expr = "^Grp[0-9]+A*"
+
+ imm_expr = "^[IJAO][a-z]"
+ imm_flag["Ib"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)"
+ imm_flag["Jb"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)"
+ imm_flag["Iw"] = "INAT_MAKE_IMM(INAT_IMM_WORD)"
+ imm_flag["Id"] = "INAT_MAKE_IMM(INAT_IMM_DWORD)"
+ imm_flag["Iq"] = "INAT_MAKE_IMM(INAT_IMM_QWORD)"
+ imm_flag["Ap"] = "INAT_MAKE_IMM(INAT_IMM_PTR)"
+ imm_flag["Iz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)"
+ imm_flag["Jz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)"
+ imm_flag["Iv"] = "INAT_MAKE_IMM(INAT_IMM_VWORD)"
+ imm_flag["Ob"] = "INAT_MOFFSET"
+ imm_flag["Ov"] = "INAT_MOFFSET"
+
+ modrm_expr = "^([CDEGMNPQRSUVW][a-z]+|NTA|T[0-2])"
+ force64_expr = "\\([df]64\\)"
+ rex_expr = "^REX(\\.[XRWB]+)*"
+ fpu_expr = "^ESC" # TODO
+
+ lprefix1_expr = "\\(66\\)"
+ delete lptable1
+ lprefix2_expr = "\\(F2\\)"
+ delete lptable2
+ lprefix3_expr = "\\(F3\\)"
+ delete lptable3
+ max_lprefix = 4
+
+ prefix_expr = "\\(Prefix\\)"
+ prefix_num["Operand-Size"] = "INAT_PFX_OPNDSZ"
+ prefix_num["REPNE"] = "INAT_PFX_REPNE"
+ prefix_num["REP/REPE"] = "INAT_PFX_REPE"
+ prefix_num["LOCK"] = "INAT_PFX_LOCK"
+ prefix_num["SEG=CS"] = "INAT_PFX_CS"
+ prefix_num["SEG=DS"] = "INAT_PFX_DS"
+ prefix_num["SEG=ES"] = "INAT_PFX_ES"
+ prefix_num["SEG=FS"] = "INAT_PFX_FS"
+ prefix_num["SEG=GS"] = "INAT_PFX_GS"
+ prefix_num["SEG=SS"] = "INAT_PFX_SS"
+ prefix_num["Address-Size"] = "INAT_PFX_ADDRSZ"
+
+ delete table
+ delete etable
+ delete gtable
+ eid = -1
+ gid = -1
+}
+
+function semantic_error(msg) {
+ print "Semantic error at " NR ": " msg > "/dev/stderr"
+ exit 1
+}
+
+function debug(msg) {
+ print "DEBUG: " msg
+}
+
+function array_size(arr, i,c) {
+ c = 0
+ for (i in arr)
+ c++
+ return c
+}
+
+/^Table:/ {
+ print "/* " $0 " */"
+}
+
+/^Referrer:/ {
+ if (NF == 1) {
+ # primary opcode table
+ tname = "inat_primary_table"
+ eid = -1
+ } else {
+ # escape opcode table
+ ref = ""
+ for (i = 2; i <= NF; i++)
+ ref = ref $i
+ eid = escape[ref]
+ tname = sprintf("inat_escape_table_%d", eid)
+ }
+}
+
+/^GrpTable:/ {
+ print "/* " $0 " */"
+ if (!($2 in group))
+ semantic_error("No group: " $2 )
+ gid = group[$2]
+ tname = "inat_group_table_" gid
+}
+
+function print_table(tbl,name,fmt,n)
+{
+ print "const insn_attr_t " name " = {"
+ for (i = 0; i < n; i++) {
+ id = sprintf(fmt, i)
+ if (tbl[id])
+ print " [" id "] = " tbl[id] ","
+ }
+ print "};"
+}
+
+/^EndTable/ {
+ if (gid != -1) {
+ # print group tables
+ if (array_size(table) != 0) {
+ print_table(table, tname "[INAT_GROUP_TABLE_SIZE]",
+ "0x%x", 8)
+ gtable[gid,0] = tname
+ }
+ if (array_size(lptable1) != 0) {
+ print_table(lptable1, tname "_1[INAT_GROUP_TABLE_SIZE]",
+ "0x%x", 8)
+ gtable[gid,1] = tname "_1"
+ }
+ if (array_size(lptable2) != 0) {
+ print_table(lptable2, tname "_2[INAT_GROUP_TABLE_SIZE]",
+ "0x%x", 8)
+ gtable[gid,2] = tname "_2"
+ }
+ if (array_size(lptable3) != 0) {
+ print_table(lptable3, tname "_3[INAT_GROUP_TABLE_SIZE]",
+ "0x%x", 8)
+ gtable[gid,3] = tname "_3"
+ }
+ } else {
+ # print primary/escaped tables
+ if (array_size(table) != 0) {
+ print_table(table, tname "[INAT_OPCODE_TABLE_SIZE]",
+ "0x%02x", 256)
+ etable[eid,0] = tname
+ }
+ if (array_size(lptable1) != 0) {
+ print_table(lptable1,tname "_1[INAT_OPCODE_TABLE_SIZE]",
+ "0x%02x", 256)
+ etable[eid,1] = tname "_1"
+ }
+ if (array_size(lptable2) != 0) {
+ print_table(lptable2,tname "_2[INAT_OPCODE_TABLE_SIZE]",
+ "0x%02x", 256)
+ etable[eid,2] = tname "_2"
+ }
+ if (array_size(lptable3) != 0) {
+ print_table(lptable3,tname "_3[INAT_OPCODE_TABLE_SIZE]",
+ "0x%02x", 256)
+ etable[eid,3] = tname "_3"
+ }
+ }
+ print ""
+ delete table
+ delete lptable1
+ delete lptable2
+ delete lptable3
+ gid = -1
+ eid = -1
+}
+
+function add_flags(old,new) {
+ if (old && new)
+ return old " | " new
+ else if (old)
+ return old
+ else
+ return new
+}
+
+# convert operands to flags.
+function convert_operands(opnd, i,imm,mod)
+{
+ imm = null
+ mod = null
+ for (i in opnd) {
+ i = opnd[i]
+ if (match(i, imm_expr) == 1) {
+ if (!imm_flag[i])
+ semantic_error("Unknown imm opnd: " i)
+ if (imm) {
+ if (i != "Ib")
+ semantic_error("ADDIMM error")
+ imm = add_flags(imm, "INAT_ADDIMM")
+ } else
+ imm = imm_flag[i]
+ } else if (match(i, modrm_expr))
+ mod = "INAT_MODRM"
+ }
+ return add_flags(imm, mod)
+}
+
+/^[0-9a-f]+\:/ {
+ if (NR == 1)
+ next
+ # get index
+ idx = "0x" substr($1, 1, index($1,":") - 1)
+ if (idx in table)
+ semantic_error("Redefine " idx " in " tname)
+
+ # check if escaped opcode
+ if ("escape" == $2) {
+ if ($3 != "#")
+ semantic_error("No escaped name")
+ ref = ""
+ for (i = 4; i <= NF; i++)
+ ref = ref $i
+ if (ref in escape)
+ semantic_error("Redefine escape (" ref ")")
+ escape[ref] = geid
+ geid++
+ table[idx] = "INAT_MAKE_ESCAPE(" escape[ref] ")"
+ next
+ }
+
+ variant = null
+ # converts
+ i = 2
+ while (i <= NF) {
+ opcode = $(i++)
+ delete opnds
+ ext = null
+ flags = null
+ opnd = null
+ # parse one opcode
+ if (match($i, opnd_expr)) {
+ opnd = $i
+ split($(i++), opnds, ",")
+ flags = convert_operands(opnds)
+ }
+ if (match($i, ext_expr))
+ ext = $(i++)
+ if (match($i, sep_expr))
+ i++
+ else if (i < NF)
+ semantic_error($i " is not a separator")
+
+ # check if group opcode
+ if (match(opcode, group_expr)) {
+ if (!(opcode in group)) {
+ group[opcode] = ggid
+ ggid++
+ }
+ flags = add_flags(flags, "INAT_MAKE_GROUP(" group[opcode] ")")
+ }
+ # check force(or default) 64bit
+ if (match(ext, force64_expr))
+ flags = add_flags(flags, "INAT_FORCE64")
+
+ # check REX prefix
+ if (match(opcode, rex_expr))
+ flags = add_flags(flags, "INAT_REXPFX")
+
+ # check coprocessor escape : TODO
+ if (match(opcode, fpu_expr))
+ flags = add_flags(flags, "INAT_MODRM")
+
+ # check prefixes
+ if (match(ext, prefix_expr)) {
+ if (!prefix_num[opcode])
+ semantic_error("Unknown prefix: " opcode)
+ flags = add_flags(flags, "INAT_MAKE_PREFIX(" prefix_num[opcode] ")")
+ }
+ if (length(flags) == 0)
+ continue
+ # check if last prefix
+ if (match(ext, lprefix1_expr)) {
+ lptable1[idx] = add_flags(lptable1[idx],flags)
+ variant = "INAT_VARIANT"
+ } else if (match(ext, lprefix2_expr)) {
+ lptable2[idx] = add_flags(lptable2[idx],flags)
+ variant = "INAT_VARIANT"
+ } else if (match(ext, lprefix3_expr)) {
+ lptable3[idx] = add_flags(lptable3[idx],flags)
+ variant = "INAT_VARIANT"
+ } else {
+ table[idx] = add_flags(table[idx],flags)
+ }
+ }
+ if (variant)
+ table[idx] = add_flags(table[idx],variant)
+}
+
+END {
+ # print escape opcode map's array
+ print "/* Escape opcode map array */"
+ print "const insn_attr_t const *inat_escape_tables[INAT_ESC_MAX + 1]" \
+ "[INAT_LPREFIX_MAX + 1] = {"
+ for (i = 0; i < geid; i++)
+ for (j = 0; j < max_lprefix; j++)
+ if (etable[i,j])
+ print " ["i"]["j"] = "etable[i,j]","
+ print "};\n"
+ # print group opcode map's array
+ print "/* Group opcode map array */"
+ print "const insn_attr_t const *inat_group_tables[INAT_GRP_MAX + 1]"\
+ "[INAT_LPREFIX_MAX + 1] = {"
+ for (i = 0; i < ggid; i++)
+ for (j = 0; j < max_lprefix; j++)
+ if (gtable[i,j])
+ print " ["i"]["j"] = "gtable[i,j]","
+ print "};"
+}
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply related [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
2009-05-09 0:48 ` [PATCH -tip v5 1/7] x86: instruction decorder API Masami Hiramatsu
@ 2009-05-11 9:27 ` Christoph Hellwig
2009-05-13 8:23 ` Gleb Natapov
1 sibling, 0 replies; 57+ messages in thread
From: Christoph Hellwig @ 2009-05-11 9:27 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm, Jim Keniston,
H. Peter Anvin, Ananth N Mavinakayanahalli, Frederic Weisbecker,
Andi Kleen, Vegard Nossum, Avi Kivity
On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
> 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.
Could also be used to implement a simple disasembler for OOPS output ala
arch/s390/kernel/dis.c?
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
@ 2009-05-11 9:27 ` Christoph Hellwig
0 siblings, 0 replies; 57+ messages in thread
From: Christoph Hellwig @ 2009-05-11 9:27 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm, Jim Keniston,
H. Peter Anvin, Ananth N Mavinakayanahalli, Frederic Weisbecker,
Andi Kleen, Vegard Nossum, Avi Kivity
On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
> 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.
Could also be used to implement a simple disasembler for OOPS output ala
arch/s390/kernel/dis.c?
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
2009-05-11 9:27 ` Christoph Hellwig
@ 2009-05-11 14:36 ` Masami Hiramatsu
-1 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 14:36 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm, Jim Keniston,
H. Peter Anvin, Ananth N Mavinakayanahalli, Frederic Weisbecker,
Andi Kleen, Vegard Nossum, Avi Kivity
Christoph Hellwig wrote:
> On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
>> 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.
>
> Could also be used to implement a simple disasembler for OOPS output ala
> arch/s390/kernel/dis.c?
Yes, but as you may know, it's not so "simple"...especially show it as
"gas-like(a.k.a. at&t)" expression :-)
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
@ 2009-05-11 14:36 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 14:36 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm, Jim Keniston,
H. Peter Anvin, Ananth N Mavinakayanahalli, Frederic Weisbecker,
Andi Kleen, Vegard Nossum, Avi Kivity
Christoph Hellwig wrote:
> On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
>> 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.
>
> Could also be used to implement a simple disasembler for OOPS output ala
> arch/s390/kernel/dis.c?
Yes, but as you may know, it's not so "simple"...especially show it as
"gas-like(a.k.a. at&t)" expression :-)
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
2009-05-09 0:48 ` [PATCH -tip v5 1/7] x86: instruction decorder API Masami Hiramatsu
@ 2009-05-13 8:23 ` Gleb Natapov
2009-05-13 8:23 ` Gleb Natapov
1 sibling, 0 replies; 57+ messages in thread
From: Gleb Natapov @ 2009-05-13 8:23 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm, Jim Keniston,
H. Peter Anvin, Ananth N Mavinakayanahalli, Frederic Weisbecker,
Andi Kleen, Vegard Nossum, Avi Kivity
On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
> +++ b/arch/x86/scripts/gen-insn-attr-x86.awk
> @@ -0,0 +1,314 @@
> +#!/bin/awk -f
On some distributions (debian) it is /usr/bin/awk.
--
Gleb.
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
@ 2009-05-13 8:23 ` Gleb Natapov
0 siblings, 0 replies; 57+ messages in thread
From: Gleb Natapov @ 2009-05-13 8:23 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm, Jim Keniston,
H. Peter Anvin, Ananth N Mavinakayanahalli, Frederic Weisbecker,
Andi Kleen, Vegard Nossum, Avi Kivity
On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
> +++ b/arch/x86/scripts/gen-insn-attr-x86.awk
> @@ -0,0 +1,314 @@
> +#!/bin/awk -f
On some distributions (debian) it is /usr/bin/awk.
--
Gleb.
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
2009-05-13 8:23 ` Gleb Natapov
(?)
@ 2009-05-13 9:35 ` Przemysław Pawełczyk
2009-05-13 9:43 ` Gleb Natapov
-1 siblings, 1 reply; 57+ messages in thread
From: Przemysław Pawełczyk @ 2009-05-13 9:35 UTC (permalink / raw)
To: Gleb Natapov
Cc: Masami Hiramatsu, Ingo Molnar, Steven Rostedt, lkml, systemtap,
kvm, Jim Keniston, H. Peter Anvin, Ananth N Mavinakayanahalli,
Frederic Weisbecker, Andi Kleen, Vegard Nossum, Avi Kivity
On Wed, May 13, 2009 at 10:23, Gleb Natapov <gleb@redhat.com> wrote:
> On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
>> +++ b/arch/x86/scripts/gen-insn-attr-x86.awk
>> @@ -0,0 +1,314 @@
>> +#!/bin/awk -f
> On some distributions (debian) it is /usr/bin/awk.
True, but on most of them (all?) there is also an appropriate link in /bin.
If shebang could have more that one argument, then '/usr/bin/env awk
-f' would be the best solution I think.
--
Przemysław Pawełczyk
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
2009-05-13 9:35 ` Przemysław Pawełczyk
@ 2009-05-13 9:43 ` Gleb Natapov
0 siblings, 0 replies; 57+ messages in thread
From: Gleb Natapov @ 2009-05-13 9:43 UTC (permalink / raw)
To: Przemysssaw Paweeeczyk
Cc: Masami Hiramatsu, Ingo Molnar, Steven Rostedt, lkml, systemtap,
kvm, Jim Keniston, H. Peter Anvin, Ananth N Mavinakayanahalli,
Frederic Weisbecker, Andi Kleen, Vegard Nossum, Avi Kivity
On Wed, May 13, 2009 at 11:35:16AM +0200, Przemysssaw Paweeeczyk wrote:
> On Wed, May 13, 2009 at 10:23, Gleb Natapov <gleb@redhat.com> wrote:
> > On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
> >> +++ b/arch/x86/scripts/gen-insn-attr-x86.awk
> >> @@ -0,0 +1,314 @@
> >> +#!/bin/awk -f
> > On some distributions (debian) it is /usr/bin/awk.
>
> True, but on most of them (all?) there is also an appropriate link in /bin.
Nope, not on debian testing. Although I assume if kernel compilation
will start to fail it will appear :)
> If shebang could have more that one argument, then '/usr/bin/env awk
> -f' would be the best solution I think.
>
--
Gleb.
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
@ 2009-05-13 9:43 ` Gleb Natapov
0 siblings, 0 replies; 57+ messages in thread
From: Gleb Natapov @ 2009-05-13 9:43 UTC (permalink / raw)
To: Przemysssaw Paweeeczyk
Cc: Masami Hiramatsu, Ingo Molnar, Steven Rostedt, lkml, systemtap,
kvm, Jim Keniston, H. Peter Anvin, Ananth N Mavinakayanahalli,
Frederic Weisbecker, Andi Kleen, Vegard Nossum, Avi Kivity
On Wed, May 13, 2009 at 11:35:16AM +0200, Przemysssaw Paweeeczyk wrote:
> On Wed, May 13, 2009 at 10:23, Gleb Natapov <gleb@redhat.com> wrote:
> > On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
> >> +++ b/arch/x86/scripts/gen-insn-attr-x86.awk
> >> @@ -0,0 +1,314 @@
> >> +#!/bin/awk -f
> > On some distributions (debian) it is /usr/bin/awk.
>
> True, but on most of them (all?) there is also an appropriate link in /bin.
Nope, not on debian testing. Although I assume if kernel compilation
will start to fail it will appear :)
> If shebang could have more that one argument, then '/usr/bin/env awk
> -f' would be the best solution I think.
>
--
Gleb.
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
2009-05-13 9:43 ` Gleb Natapov
@ 2009-05-13 14:35 ` Masami Hiramatsu
-1 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-13 14:35 UTC (permalink / raw)
To: Gleb Natapov
Cc: Przemysssaw Paweeeczyk, Ingo Molnar, Steven Rostedt, lkml,
systemtap, kvm, Jim Keniston, H. Peter Anvin,
Ananth N Mavinakayanahalli, Frederic Weisbecker, Andi Kleen,
Vegard Nossum, Avi Kivity
Gleb Natapov wrote:
> On Wed, May 13, 2009 at 11:35:16AM +0200, Przemysssaw Paweeeczyk wrote:
>> On Wed, May 13, 2009 at 10:23, Gleb Natapov <gleb@redhat.com> wrote:
>>> On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
>>>> +++ b/arch/x86/scripts/gen-insn-attr-x86.awk
>>>> @@ -0,0 +1,314 @@
>>>> +#!/bin/awk -f
>>> On some distributions (debian) it is /usr/bin/awk.
>> True, but on most of them (all?) there is also an appropriate link in /bin.
> Nope, not on debian testing. Although I assume if kernel compilation
> will start to fail it will appear :)
>
>> If shebang could have more that one argument, then '/usr/bin/env awk
>> -f' would be the best solution I think.
Ah, I see.
Actually, it will be executed from Makefile with 'awk -f'.
> --- a/arch/x86/lib/Makefile
> +++ b/arch/x86/lib/Makefile
> @@ -2,12 +2,21 @@
> # Makefile for x86 specific library files.
> #
>
> +quiet_cmd_inat_tables = GEN $@
> + cmd_inat_tables = awk -f $(srctree)/arch/x86/scripts/gen-insn-attr-x86.awk $(srctree)/arch/x86/lib/x86-opcode-map.txt > $@
> +
So, if awk is on the PATH, it will pass.
Maybe, I need to add 'HOSTAWK = awk' line in Makefile.
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
@ 2009-05-13 14:35 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-13 14:35 UTC (permalink / raw)
To: Gleb Natapov
Cc: Przemysssaw Paweeeczyk, Ingo Molnar, Steven Rostedt, lkml,
systemtap, kvm, Jim Keniston, H. Peter Anvin,
Ananth N Mavinakayanahalli, Frederic Weisbecker, Andi Kleen,
Vegard Nossum, Avi Kivity
Gleb Natapov wrote:
> On Wed, May 13, 2009 at 11:35:16AM +0200, Przemysssaw Paweeeczyk wrote:
>> On Wed, May 13, 2009 at 10:23, Gleb Natapov <gleb@redhat.com> wrote:
>>> On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
>>>> +++ b/arch/x86/scripts/gen-insn-attr-x86.awk
>>>> @@ -0,0 +1,314 @@
>>>> +#!/bin/awk -f
>>> On some distributions (debian) it is /usr/bin/awk.
>> True, but on most of them (all?) there is also an appropriate link in /bin.
> Nope, not on debian testing. Although I assume if kernel compilation
> will start to fail it will appear :)
>
>> If shebang could have more that one argument, then '/usr/bin/env awk
>> -f' would be the best solution I think.
Ah, I see.
Actually, it will be executed from Makefile with 'awk -f'.
> --- a/arch/x86/lib/Makefile
> +++ b/arch/x86/lib/Makefile
> @@ -2,12 +2,21 @@
> # Makefile for x86 specific library files.
> #
>
> +quiet_cmd_inat_tables = GEN $@
> + cmd_inat_tables = awk -f $(srctree)/arch/x86/scripts/gen-insn-attr-x86.awk $(srctree)/arch/x86/lib/x86-opcode-map.txt > $@
> +
So, if awk is on the PATH, it will pass.
Maybe, I need to add 'HOSTAWK = awk' line in Makefile.
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 1/7] x86: instruction decorder API
2009-05-13 14:35 ` Masami Hiramatsu
(?)
@ 2009-05-13 15:20 ` Gleb Natapov
-1 siblings, 0 replies; 57+ messages in thread
From: Gleb Natapov @ 2009-05-13 15:20 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Przemysssaw Paweeeczyk, Ingo Molnar, Steven Rostedt, lkml,
systemtap, kvm, Jim Keniston, H. Peter Anvin,
Ananth N Mavinakayanahalli, Frederic Weisbecker, Andi Kleen,
Vegard Nossum, Avi Kivity
On Wed, May 13, 2009 at 10:35:55AM -0400, Masami Hiramatsu wrote:
> Gleb Natapov wrote:
> > On Wed, May 13, 2009 at 11:35:16AM +0200, Przemysssaw Paweeeczyk wrote:
> >> On Wed, May 13, 2009 at 10:23, Gleb Natapov <gleb@redhat.com> wrote:
> >>> On Fri, May 08, 2009 at 08:48:42PM -0400, Masami Hiramatsu wrote:
> >>>> +++ b/arch/x86/scripts/gen-insn-attr-x86.awk
> >>>> @@ -0,0 +1,314 @@
> >>>> +#!/bin/awk -f
> >>> On some distributions (debian) it is /usr/bin/awk.
> >> True, but on most of them (all?) there is also an appropriate link in /bin.
> > Nope, not on debian testing. Although I assume if kernel compilation
> > will start to fail it will appear :)
> >
> >> If shebang could have more that one argument, then '/usr/bin/env awk
> >> -f' would be the best solution I think.
>
> Ah, I see.
> Actually, it will be executed from Makefile with 'awk -f'.
>
> > --- a/arch/x86/lib/Makefile
> > +++ b/arch/x86/lib/Makefile
> > @@ -2,12 +2,21 @@
> > # Makefile for x86 specific library files.
> > #
> >
> > +quiet_cmd_inat_tables = GEN $@
> > + cmd_inat_tables = awk -f $(srctree)/arch/x86/scripts/gen-insn-attr-x86.awk $(srctree)/arch/x86/lib/x86-opcode-map.txt > $@
> > +
>
> So, if awk is on the PATH, it will pass.
Ah, that is good enough I thing. I tried to run scrip manually.
> Maybe, I need to add 'HOSTAWK = awk' line in Makefile.
>
--
Gleb.
^ permalink raw reply [flat|nested] 57+ messages in thread
* [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
2009-05-09 0:48 ` Masami Hiramatsu
@ 2009-05-09 0:48 ` Masami Hiramatsu
-1 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:48 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: systemtap, kvm, Masami Hiramatsu, Ananth N Mavinakayanahalli,
Jim Keniston, Ingo Molnar
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.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Jim Keniston <jkenisto@us.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
---
arch/x86/kernel/kprobes.c | 54 +++++++++++++++++++++++++++++++++++++++++++++
1 files changed, 54 insertions(+), 0 deletions(-)
diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
index 7b5169d..3d5e85f 100644
--- a/arch/x86/kernel/kprobes.c
+++ b/arch/x86/kernel/kprobes.c
@@ -48,12 +48,14 @@
#include <linux/preempt.h>
#include <linux/module.h>
#include <linux/kdebug.h>
+#include <linux/kallsyms.h>
#include <asm/cacheflush.h>
#include <asm/desc.h>
#include <asm/pgtable.h>
#include <asm/uaccess.h>
#include <asm/alternative.h>
+#include <asm/insn.h>
void jprobe_return_end(void);
@@ -244,6 +246,56 @@ 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;
+
+ /*
+ * Don't use p->ainsn.insn, which could be modified -- e.g.,
+ * by fix_riprel().
+ */
+ 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];
+
+ /* Lookup symbol including addr */
+ if (!kallsyms_lookup(paddr, NULL, &offset, NULL, __dummy_buf))
+ return 0;
+
+ /* Decode instructions */
+ addr = paddr - offset;
+ while (addr < paddr) {
+ insn_init_kernel(&insn, (void *)addr);
+ insn_get_opcode(&insn);
+ if (OPCODE1(&insn) == BREAKPOINT_INSTRUCTION) {
+ ret = recover_probed_instruction(buf, addr);
+ if (ret)
+ return 0;
+ insn_init_kernel(&insn, buf);
+ }
+ insn_get_length(&insn);
+ addr += insn.length;
+ }
+
+ return (addr == paddr);
+}
+
/*
* Returns non-zero if opcode modifies the interrupt flag.
*/
@@ -359,6 +411,8 @@ 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)
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply related [flat|nested] 57+ messages in thread* [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
@ 2009-05-09 0:48 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:48 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: systemtap, kvm, Masami Hiramatsu, Ananth N Mavinakayanahalli,
Jim Keniston, Ingo Molnar
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.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Jim Keniston <jkenisto@us.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
---
arch/x86/kernel/kprobes.c | 54 +++++++++++++++++++++++++++++++++++++++++++++
1 files changed, 54 insertions(+), 0 deletions(-)
diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
index 7b5169d..3d5e85f 100644
--- a/arch/x86/kernel/kprobes.c
+++ b/arch/x86/kernel/kprobes.c
@@ -48,12 +48,14 @@
#include <linux/preempt.h>
#include <linux/module.h>
#include <linux/kdebug.h>
+#include <linux/kallsyms.h>
#include <asm/cacheflush.h>
#include <asm/desc.h>
#include <asm/pgtable.h>
#include <asm/uaccess.h>
#include <asm/alternative.h>
+#include <asm/insn.h>
void jprobe_return_end(void);
@@ -244,6 +246,56 @@ 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;
+
+ /*
+ * Don't use p->ainsn.insn, which could be modified -- e.g.,
+ * by fix_riprel().
+ */
+ 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];
+
+ /* Lookup symbol including addr */
+ if (!kallsyms_lookup(paddr, NULL, &offset, NULL, __dummy_buf))
+ return 0;
+
+ /* Decode instructions */
+ addr = paddr - offset;
+ while (addr < paddr) {
+ insn_init_kernel(&insn, (void *)addr);
+ insn_get_opcode(&insn);
+ if (OPCODE1(&insn) == BREAKPOINT_INSTRUCTION) {
+ ret = recover_probed_instruction(buf, addr);
+ if (ret)
+ return 0;
+ insn_init_kernel(&insn, buf);
+ }
+ insn_get_length(&insn);
+ addr += insn.length;
+ }
+
+ return (addr == paddr);
+}
+
/*
* Returns non-zero if opcode modifies the interrupt flag.
*/
@@ -359,6 +411,8 @@ 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)
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply related [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
2009-05-09 0:48 ` Masami Hiramatsu
(?)
@ 2009-05-11 14:20 ` Steven Rostedt
2009-05-11 15:01 ` Masami Hiramatsu
-1 siblings, 1 reply; 57+ messages in thread
From: Steven Rostedt @ 2009-05-11 14:20 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Jim Keniston
On Fri, 8 May 2009, Masami Hiramatsu wrote:
> 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.
>
> Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
> Cc: Jim Keniston <jkenisto@us.ibm.com>
> Cc: Ingo Molnar <mingo@elte.hu>
> ---
>
> arch/x86/kernel/kprobes.c | 54 +++++++++++++++++++++++++++++++++++++++++++++
> 1 files changed, 54 insertions(+), 0 deletions(-)
>
> diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
> index 7b5169d..3d5e85f 100644
> --- a/arch/x86/kernel/kprobes.c
> +++ b/arch/x86/kernel/kprobes.c
> @@ -48,12 +48,14 @@
> #include <linux/preempt.h>
> #include <linux/module.h>
> #include <linux/kdebug.h>
> +#include <linux/kallsyms.h>
>
> #include <asm/cacheflush.h>
> #include <asm/desc.h>
> #include <asm/pgtable.h>
> #include <asm/uaccess.h>
> #include <asm/alternative.h>
> +#include <asm/insn.h>
>
> void jprobe_return_end(void);
>
> @@ -244,6 +246,56 @@ 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;
> +
I'm just doing a casual scan of the patch set.
> + /*
> + * Don't use p->ainsn.insn, which could be modified -- e.g.,
This comment talks about "p", what's that? It's not used in this function.
> + * by fix_riprel().
> + */
> + memcpy(buf, kp->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
> + buf[0] = kp->opcode;
Why is it OK to copy addr to "buf" and then rewrite the first character of
buf? Does it have something to do with the above "p"?
I don't mean to be critical here, but I've been doing "Mother Day"
activities all weekend and for some reason that was also the best time for
everyone to Cc me on patches. I'm way behind in my email, and it would be
nice if the comments described why things that "look" wrong are not.
> + 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];
> +
> + /* Lookup symbol including addr */
The above comment is very close to a "add one to i" for i++ type of
comment.
> + if (!kallsyms_lookup(paddr, NULL, &offset, NULL, __dummy_buf))
> + return 0;
> +
> + /* Decode instructions */
> + addr = paddr - offset;
> + while (addr < paddr) {
> + insn_init_kernel(&insn, (void *)addr);
> + insn_get_opcode(&insn);
> + if (OPCODE1(&insn) == BREAKPOINT_INSTRUCTION) {
> + ret = recover_probed_instruction(buf, addr);
Oh, the above puts back the original op code. That is why it is OK?
I'd comment that a little bit more. Just so that reviewers have an easier
idea of what is happening.
> + if (ret)
> + return 0;
> + insn_init_kernel(&insn, buf);
insn_init_kernel? Is that like a text poke or something?
> + }
> + insn_get_length(&insn);
> + addr += insn.length;
> + }
> +
> + return (addr == paddr);
> +}
> +
> /*
> * Returns non-zero if opcode modifies the interrupt flag.
> */
> @@ -359,6 +411,8 @@ 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();
Oh look, I found the phantom "p"!
-- Steve
> if (!p->ainsn.insn)
>
>
> --
> Masami Hiramatsu
>
> Software Engineer
> Hitachi Computer Products (America) Inc.
> Software Solutions Division
>
> e-mail: mhiramat@redhat.com
>
^ permalink raw reply [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
2009-05-11 14:20 ` Steven Rostedt
@ 2009-05-11 15:01 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 15:01 UTC (permalink / raw)
To: Steven Rostedt
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Jim Keniston
Steven Rostedt wrote:
> On Fri, 8 May 2009, Masami Hiramatsu wrote:
>
>> 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.
>>
>> Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
>> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
>> Cc: Jim Keniston <jkenisto@us.ibm.com>
>> Cc: Ingo Molnar <mingo@elte.hu>
>> ---
>>
>> arch/x86/kernel/kprobes.c | 54 +++++++++++++++++++++++++++++++++++++++++++++
>> 1 files changed, 54 insertions(+), 0 deletions(-)
>>
>> diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
>> index 7b5169d..3d5e85f 100644
>> --- a/arch/x86/kernel/kprobes.c
>> +++ b/arch/x86/kernel/kprobes.c
>> @@ -48,12 +48,14 @@
>> #include <linux/preempt.h>
>> #include <linux/module.h>
>> #include <linux/kdebug.h>
>> +#include <linux/kallsyms.h>
>>
>> #include <asm/cacheflush.h>
>> #include <asm/desc.h>
>> #include <asm/pgtable.h>
>> #include <asm/uaccess.h>
>> #include <asm/alternative.h>
>> +#include <asm/insn.h>
>>
>> void jprobe_return_end(void);
>>
>> @@ -244,6 +246,56 @@ 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;
>> +
>
> I'm just doing a casual scan of the patch set.
Thank you!
>
>> + /*
>> + * Don't use p->ainsn.insn, which could be modified -- e.g.,
>
> This comment talks about "p", what's that? It's not used in this function.
oops, this should be kp.
>
>> + * by fix_riprel().
>> + */
>> + memcpy(buf, kp->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
>> + buf[0] = kp->opcode;
>
> Why is it OK to copy addr to "buf" and then rewrite the first character of
> buf? Does it have something to do with the above "p"?
Yes, each kprobe copied probed instruction to kp->ainsn.insn,
which is an executable buffer for single stepping.
So, 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.
> I don't mean to be critical here, but I've been doing "Mother Day"
> activities all weekend and for some reason that was also the best time for
> everyone to Cc me on patches. I'm way behind in my email, and it would be
> nice if the comments described why things that "look" wrong are not.
>
>
>> + 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];
>> +
>> + /* Lookup symbol including addr */
>
> The above comment is very close to a "add one to i" for i++ type of
> comment.
Agreed.
>
>> + if (!kallsyms_lookup(paddr, NULL, &offset, NULL, __dummy_buf))
>> + return 0;
>> +
>> + /* Decode instructions */
>> + addr = paddr - offset;
>> + while (addr < paddr) {
>> + insn_init_kernel(&insn, (void *)addr);
>> + insn_get_opcode(&insn);
>> + if (OPCODE1(&insn) == BREAKPOINT_INSTRUCTION) {
>> + ret = recover_probed_instruction(buf, addr);
>
> Oh, the above puts back the original op code. That is why it is OK?
Oops, no. I have to use get_kprobe() instead. Thanks!
>
> I'd comment that a little bit more. Just so that reviewers have an easier
> idea of what is happening.
>
>> + if (ret)
>> + return 0;
>> + insn_init_kernel(&insn, buf);
>
> insn_init_kernel? Is that like a text poke or something?
it's a wrapper of insn_init() which initialize struct insn.
Thank you,
>> + }
>> + insn_get_length(&insn);
>> + addr += insn.length;
>> + }
>> +
>> + return (addr == paddr);
>> +}
>> +
>> /*
>> * Returns non-zero if opcode modifies the interrupt flag.
>> */
>> @@ -359,6 +411,8 @@ 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();
>
> Oh look, I found the phantom "p"!
>
> -- Steve
>
>> if (!p->ainsn.insn)
>>
>>
>> --
>> Masami Hiramatsu
>>
>> Software Engineer
>> Hitachi Computer Products (America) Inc.
>> Software Solutions Division
>>
>> e-mail: mhiramat@redhat.com
>>
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
@ 2009-05-11 15:01 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 15:01 UTC (permalink / raw)
To: Steven Rostedt
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Jim Keniston
Steven Rostedt wrote:
> On Fri, 8 May 2009, Masami Hiramatsu wrote:
>
>> 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.
>>
>> Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
>> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
>> Cc: Jim Keniston <jkenisto@us.ibm.com>
>> Cc: Ingo Molnar <mingo@elte.hu>
>> ---
>>
>> arch/x86/kernel/kprobes.c | 54 +++++++++++++++++++++++++++++++++++++++++++++
>> 1 files changed, 54 insertions(+), 0 deletions(-)
>>
>> diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
>> index 7b5169d..3d5e85f 100644
>> --- a/arch/x86/kernel/kprobes.c
>> +++ b/arch/x86/kernel/kprobes.c
>> @@ -48,12 +48,14 @@
>> #include <linux/preempt.h>
>> #include <linux/module.h>
>> #include <linux/kdebug.h>
>> +#include <linux/kallsyms.h>
>>
>> #include <asm/cacheflush.h>
>> #include <asm/desc.h>
>> #include <asm/pgtable.h>
>> #include <asm/uaccess.h>
>> #include <asm/alternative.h>
>> +#include <asm/insn.h>
>>
>> void jprobe_return_end(void);
>>
>> @@ -244,6 +246,56 @@ 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;
>> +
>
> I'm just doing a casual scan of the patch set.
Thank you!
>
>> + /*
>> + * Don't use p->ainsn.insn, which could be modified -- e.g.,
>
> This comment talks about "p", what's that? It's not used in this function.
oops, this should be kp.
>
>> + * by fix_riprel().
>> + */
>> + memcpy(buf, kp->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
>> + buf[0] = kp->opcode;
>
> Why is it OK to copy addr to "buf" and then rewrite the first character of
> buf? Does it have something to do with the above "p"?
Yes, each kprobe copied probed instruction to kp->ainsn.insn,
which is an executable buffer for single stepping.
So, 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.
> I don't mean to be critical here, but I've been doing "Mother Day"
> activities all weekend and for some reason that was also the best time for
> everyone to Cc me on patches. I'm way behind in my email, and it would be
> nice if the comments described why things that "look" wrong are not.
>
>
>> + 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];
>> +
>> + /* Lookup symbol including addr */
>
> The above comment is very close to a "add one to i" for i++ type of
> comment.
Agreed.
>
>> + if (!kallsyms_lookup(paddr, NULL, &offset, NULL, __dummy_buf))
>> + return 0;
>> +
>> + /* Decode instructions */
>> + addr = paddr - offset;
>> + while (addr < paddr) {
>> + insn_init_kernel(&insn, (void *)addr);
>> + insn_get_opcode(&insn);
>> + if (OPCODE1(&insn) == BREAKPOINT_INSTRUCTION) {
>> + ret = recover_probed_instruction(buf, addr);
>
> Oh, the above puts back the original op code. That is why it is OK?
Oops, no. I have to use get_kprobe() instead. Thanks!
>
> I'd comment that a little bit more. Just so that reviewers have an easier
> idea of what is happening.
>
>> + if (ret)
>> + return 0;
>> + insn_init_kernel(&insn, buf);
>
> insn_init_kernel? Is that like a text poke or something?
it's a wrapper of insn_init() which initialize struct insn.
Thank you,
>> + }
>> + insn_get_length(&insn);
>> + addr += insn.length;
>> + }
>> +
>> + return (addr == paddr);
>> +}
>> +
>> /*
>> * Returns non-zero if opcode modifies the interrupt flag.
>> */
>> @@ -359,6 +411,8 @@ 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();
>
> Oh look, I found the phantom "p"!
>
> -- Steve
>
>> if (!p->ainsn.insn)
>>
>>
>> --
>> Masami Hiramatsu
>>
>> Software Engineer
>> Hitachi Computer Products (America) Inc.
>> Software Solutions Division
>>
>> e-mail: mhiramat@redhat.com
>>
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
2009-05-11 15:01 ` Masami Hiramatsu
@ 2009-05-11 15:14 ` Masami Hiramatsu
-1 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 15:14 UTC (permalink / raw)
To: Steven Rostedt
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Jim Keniston
Masami Hiramatsu wrote:
>>> + if (!kallsyms_lookup(paddr, NULL, &offset, NULL, __dummy_buf))
>>> + return 0;
>>> +
>>> + /* Decode instructions */
>>> + addr = paddr - offset;
>>> + while (addr < paddr) {
>>> + insn_init_kernel(&insn, (void *)addr);
>>> + insn_get_opcode(&insn);
>>> + if (OPCODE1(&insn) == BREAKPOINT_INSTRUCTION) {
>>> + ret = recover_probed_instruction(buf, addr);
>> Oh, the above puts back the original op code. That is why it is OK?
>
> Oops, no. I have to use get_kprobe() instead. Thanks!
Ah, I forgot another possibility. There might be another subsystem,
like kgdb, will put their break point on the kernel.
In that case, decoder will decode the instruction is a break point
instruction and the first opcode is int3. So, this part is correct.
In the future, we need to add a generic recover_instruction() code
for those text modification subsystems.
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
@ 2009-05-11 15:14 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 15:14 UTC (permalink / raw)
To: Steven Rostedt
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Jim Keniston
Masami Hiramatsu wrote:
>>> + if (!kallsyms_lookup(paddr, NULL, &offset, NULL, __dummy_buf))
>>> + return 0;
>>> +
>>> + /* Decode instructions */
>>> + addr = paddr - offset;
>>> + while (addr < paddr) {
>>> + insn_init_kernel(&insn, (void *)addr);
>>> + insn_get_opcode(&insn);
>>> + if (OPCODE1(&insn) == BREAKPOINT_INSTRUCTION) {
>>> + ret = recover_probed_instruction(buf, addr);
>> Oh, the above puts back the original op code. That is why it is OK?
>
> Oops, no. I have to use get_kprobe() instead. Thanks!
Ah, I forgot another possibility. There might be another subsystem,
like kgdb, will put their break point on the kernel.
In that case, decoder will decode the instruction is a break point
instruction and the first opcode is int3. So, this part is correct.
In the future, we need to add a generic recover_instruction() code
for those text modification subsystems.
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
2009-05-11 15:01 ` Masami Hiramatsu
(?)
(?)
@ 2009-05-11 15:22 ` Steven Rostedt
2009-05-11 18:21 ` Masami Hiramatsu
-1 siblings, 1 reply; 57+ messages in thread
From: Steven Rostedt @ 2009-05-11 15:22 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Jim Keniston
On Mon, 11 May 2009, Masami Hiramatsu wrote:
>
> >
> >> + * by fix_riprel().
> >> + */
> >> + memcpy(buf, kp->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
> >> + buf[0] = kp->opcode;
> >
> > Why is it OK to copy addr to "buf" and then rewrite the first character of
> > buf? Does it have something to do with the above "p"?
>
> Yes, each kprobe copied probed instruction to kp->ainsn.insn,
> which is an executable buffer for single stepping.
> So, 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.
For code that is awkward, complex or non-trivial, don't be afraid to put
in a paragraph explaining the code. The above explanation should be a
comment in the code. Otherwise people like me would just look at it and
say "huh?".
Note, I'm a bit cranky this morning, so I hope I don't offend anyone.
-- Steve
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
2009-05-11 15:22 ` Steven Rostedt
@ 2009-05-11 18:21 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 18:21 UTC (permalink / raw)
To: Steven Rostedt
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Jim Keniston
Steven Rostedt wrote:
> On Mon, 11 May 2009, Masami Hiramatsu wrote:
>>>> + * by fix_riprel().
>>>> + */
>>>> + memcpy(buf, kp->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
>>>> + buf[0] = kp->opcode;
>>> Why is it OK to copy addr to "buf" and then rewrite the first character of
>>> buf? Does it have something to do with the above "p"?
>> Yes, each kprobe copied probed instruction to kp->ainsn.insn,
>> which is an executable buffer for single stepping.
>> So, 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.
>
> For code that is awkward, complex or non-trivial, don't be afraid to put
> in a paragraph explaining the code. The above explanation should be a
> comment in the code. Otherwise people like me would just look at it and
> say "huh?".
>
> Note, I'm a bit cranky this morning, so I hope I don't offend anyone.
No, that's very helpful review for me. Thanks :-)
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 2/7] kprobes: checks probe address is instruction boudary on x86
@ 2009-05-11 18:21 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 18:21 UTC (permalink / raw)
To: Steven Rostedt
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Jim Keniston
Steven Rostedt wrote:
> On Mon, 11 May 2009, Masami Hiramatsu wrote:
>>>> + * by fix_riprel().
>>>> + */
>>>> + memcpy(buf, kp->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
>>>> + buf[0] = kp->opcode;
>>> Why is it OK to copy addr to "buf" and then rewrite the first character of
>>> buf? Does it have something to do with the above "p"?
>> Yes, each kprobe copied probed instruction to kp->ainsn.insn,
>> which is an executable buffer for single stepping.
>> So, 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.
>
> For code that is awkward, complex or non-trivial, don't be afraid to put
> in a paragraph explaining the code. The above explanation should be a
> comment in the code. Otherwise people like me would just look at it and
> say "huh?".
>
> Note, I'm a bit cranky this morning, so I hope I don't offend anyone.
No, that's very helpful review for me. Thanks :-)
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* [PATCH -tip v5 3/7] kprobes: cleanup fix_riprel() using insn decoder on x86
2009-05-09 0:48 ` Masami Hiramatsu
@ 2009-05-09 0:48 ` Masami Hiramatsu
-1 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:48 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: systemtap, kvm, Masami Hiramatsu, Ananth N Mavinakayanahalli,
Jim Keniston, Ingo Molnar
Cleanup fix_riprel() in arch/x86/kernel/kprobes.c by using x86 instruction
decoder.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Jim Keniston <jkenisto@us.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
---
arch/x86/kernel/kprobes.c | 128 ++++++++-------------------------------------
1 files changed, 23 insertions(+), 105 deletions(-)
diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
index 3d5e85f..f33fb5e 100644
--- a/arch/x86/kernel/kprobes.c
+++ b/arch/x86/kernel/kprobes.c
@@ -108,50 +108,6 @@ 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[] = {
@@ -329,68 +285,30 @@ 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;
+ insn_init_kernel(&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
}
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply related [flat|nested] 57+ messages in thread* [PATCH -tip v5 3/7] kprobes: cleanup fix_riprel() using insn decoder on x86
@ 2009-05-09 0:48 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:48 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: systemtap, kvm, Masami Hiramatsu, Ananth N Mavinakayanahalli,
Jim Keniston, Ingo Molnar
Cleanup fix_riprel() in arch/x86/kernel/kprobes.c by using x86 instruction
decoder.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Jim Keniston <jkenisto@us.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
---
arch/x86/kernel/kprobes.c | 128 ++++++++-------------------------------------
1 files changed, 23 insertions(+), 105 deletions(-)
diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c
index 3d5e85f..f33fb5e 100644
--- a/arch/x86/kernel/kprobes.c
+++ b/arch/x86/kernel/kprobes.c
@@ -108,50 +108,6 @@ 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[] = {
@@ -329,68 +285,30 @@ 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;
+ insn_init_kernel(&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
}
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply related [flat|nested] 57+ messages in thread
* [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
2009-05-09 0:48 ` Masami Hiramatsu
` (3 preceding siblings ...)
(?)
@ 2009-05-09 0:48 ` Masami Hiramatsu
2009-05-09 16:36 ` Frédéric Weisbecker
2009-05-11 9:32 ` Christoph Hellwig
-1 siblings, 2 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:48 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: systemtap, kvm, Masami Hiramatsu, Steven Rostedt,
Ananth N Mavinakayanahalli, Ingo Molnar, Frederic Weisbecker
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).
Changes from v4:
- Change interface name from 'kprobe_probes' to 'kprobe_events'
- Skip comments (words after '#') from inputs of 'kprobe_events'.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
---
Documentation/trace/ftrace.txt | 55 +++++
kernel/trace/Kconfig | 9 +
kernel/trace/Makefile | 1
kernel/trace/trace_kprobe.c | 404 ++++++++++++++++++++++++++++++++++++++++
4 files changed, 469 insertions(+), 0 deletions(-)
create mode 100644 kernel/trace/trace_kprobe.c
diff --git a/Documentation/trace/ftrace.txt b/Documentation/trace/ftrace.txt
index fd9a3e6..2b8ead6 100644
--- a/Documentation/trace/ftrace.txt
+++ b/Documentation/trace/ftrace.txt
@@ -1310,6 +1310,61 @@ dereference in a kernel module:
[...]
+kprobe-based event tracer
+---------------------------
+
+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
+/debug/tracing/kprobe_events.
+
+Synopsis of kprobe_events:
+ p SYMBOL[+offs|-offs]|MEMADDR : set a probe
+ r SYMBOL[+0] : set a return probe
+
+E.g.
+ echo p sys_open > /debug/tracing/kprobe_events
+
+ This sets a kprobe on the top of sys_open() function.
+
+ echo r sys_open >> /debug/tracing/kprobe_events
+
+ This sets a kretprobe on the return point of sys_open() function.
+
+ echo > /debug/tracing/kprobe_events
+
+ This clears all probe points. and you can see the traced information via
+/debug/tracing/trace.
+
+ cat /debug/tracing/trace
+# tracer: nop
+#
+# TASK-PID CPU# TIMESTAMP FUNCTION
+# | | | | |
+ <...>-5117 [003] 416.481638: sys_open: @sys_open+0
+ <...>-5117 [003] 416.481662: syscall_call: <-sys_open+0
+ <...>-5117 [003] 416.481739: sys_open: @sys_open+0
+ <...>-5117 [003] 416.481762: sysenter_do_call: <-sys_open+0
+ <...>-5117 [003] 416.481818: sys_open: @sys_open+0
+ <...>-5117 [003] 416.481842: sysenter_do_call: <-sys_open+0
+ <...>-5117 [003] 416.481882: sys_open: @sys_open+0
+ <...>-5117 [003] 416.481905: sysenter_do_call: <-sys_open+0
+
+ @SYMBOL means that kernel hits a probe, and <-SYMBOL means kernel returns
+from SYMBOL(e.g. "sysenter_do_call: <-sys_open+0" means kernel returns from
+sys_open to sysenter_do_call).
+
+
function graph tracer
---------------------------
diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
index 7370253..914df9c 100644
--- a/kernel/trace/Kconfig
+++ b/kernel/trace/Kconfig
@@ -398,6 +398,15 @@ config BLK_DEV_IO_TRACE
If unsure, say N.
+config KPROBE_TRACER
+ depends on KPROBES
+ depends on X86
+ bool "Trace kprobes"
+ select TRACING
+ help
+ This tracer probes everywhere where kprobes can probe it, and
+ records various registers and memories specified by user.
+
config DYNAMIC_FTRACE
bool "enable/disable ftrace tracepoints dynamically"
depends on FUNCTION_TRACER
diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile
index 06b8585..166c859 100644
--- a/kernel/trace/Makefile
+++ b/kernel/trace/Makefile
@@ -51,5 +51,6 @@ obj-$(CONFIG_EVENT_TRACING) += trace_export.o
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_KPROBE_TRACER) += trace_kprobe.o
libftrace-y := ftrace.o
diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
new file mode 100644
index 0000000..8112505
--- /dev/null
+++ b/kernel/trace/trace_kprobe.c
@@ -0,0 +1,404 @@
+/*
+ * kprobe based kernel tracer
+ *
+ * Created by Masami Hiramatsu <mhiramat@redhat.com>
+ *
+ * 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/ftrace.h>
+#include "trace.h"
+
+/**
+ * kprobe_trace_core
+ */
+#define TRACE_MAXARGS 6
+
+struct trace_probe {
+ struct list_head list;
+ union {
+ struct kprobe kp;
+ struct kretprobe rp;
+ };
+ const char *symbol; /* symbol name */
+};
+
+static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs);
+
+static int kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs)
+{
+ struct trace_probe *tp = container_of(kp, struct trace_probe, kp);
+
+ kprobe_trace_record(instruction_pointer(regs), tp, regs);
+ return 0;
+}
+
+static int kretprobe_trace_func(struct kretprobe_instance *ri,
+ struct pt_regs *regs)
+{
+ struct trace_probe *tp = container_of(ri->rp, struct trace_probe, rp);
+
+ kprobe_trace_record((unsigned long)ri->ret_addr, tp, regs);
+ return 0;
+}
+
+static int probe_is_return(struct trace_probe *tp)
+{
+ return (tp->rp.handler == kretprobe_trace_func);
+}
+
+static const char *probe_symbol(struct trace_probe *tp)
+{
+ return tp->symbol ? tp->symbol : "unknown";
+}
+
+static long probe_offset(struct trace_probe *tp)
+{
+ return (probe_is_return(tp)) ? tp->rp.kp.offset : tp->kp.offset;
+}
+
+static void *probe_address(struct trace_probe *tp)
+{
+ return (probe_is_return(tp)) ? tp->rp.kp.addr : tp->kp.addr;
+}
+
+
+static DEFINE_MUTEX(probe_lock);
+static LIST_HEAD(probe_list);
+
+static struct trace_probe *alloc_trace_probe(const char *symbol)
+{
+ 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) {
+ kfree(tp);
+ return ERR_PTR(-ENOMEM);
+ }
+ }
+
+ INIT_LIST_HEAD(&tp->list);
+ return tp;
+}
+
+static void free_trace_probe(struct trace_probe *tp)
+{
+ kfree(tp->symbol);
+ kfree(tp);
+}
+
+static int register_trace_probe(struct trace_probe *tp)
+{
+ int ret;
+
+ mutex_lock(&probe_lock);
+ list_add_tail(&tp->list, &probe_list);
+
+ 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;
+ }
+ list_del(&tp->list);
+ }
+ mutex_unlock(&probe_lock);
+ return ret;
+}
+
+static void unregister_trace_probe(struct trace_probe *tp)
+{
+ if (probe_is_return(tp))
+ unregister_kretprobe(&tp->rp);
+ else
+ unregister_kprobe(&tp->kp);
+ list_del(&tp->list);
+}
+
+static int create_trace_probe(int argc, char **argv)
+{
+ /*
+ * Argument syntax:
+ * - Add kprobe: p SYMBOL[+OFFS|-OFFS]|ADDRESS
+ * - Add kretprobe: r SYMBOL[+0]
+ */
+ struct trace_probe *tp;
+ struct kprobe *kp;
+ char *tmp;
+ int ret = 0;
+ int is_return = 0;
+ char *symbol = 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 (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 */
+ 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';
+ }
+ if (offset && is_return)
+ return -EINVAL;
+ }
+
+ /* setup a probe */
+ tp = alloc_trace_probe(symbol);
+ 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) {
+ /* TODO: check offset is collect by using insn_decoder */
+ kp->symbol_name = tp->symbol;
+ kp->offset = offset;
+ } else
+ kp->addr = addr;
+
+ 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;
+
+ if (tp == NULL)
+ return 0;
+
+ if (tp->symbol)
+ seq_printf(m, "%c %s%+ld\n",
+ probe_is_return(tp) ? 'r' : 'p',
+ probe_symbol(tp), probe_offset(tp));
+ else
+ seq_printf(m, "%c 0x%p\n",
+ probe_is_return(tp) ? 'r' : 'p',
+ probe_address(tp));
+ 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);
+}
+
+
+#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;
+ char **argv = NULL;
+ int argc = 0;
+ 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';
+
+ argv = argv_split(GFP_KERNEL, kbuf, &argc);
+ if (!argv) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ if (argc)
+ ret = create_trace_probe(argc, argv);
+
+ argv_free(argv);
+ if (ret < 0)
+ 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,
+};
+
+/* event recording functions */
+static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs)
+{
+ __trace_bprintk(ip, "%s%s%+ld\n",
+ probe_is_return(tp) ? "<-" : "@",
+ probe_symbol(tp), probe_offset(tp));
+}
+
+/* Make a debugfs interface for controling probe points */
+static __init int init_kprobe_trace(void)
+{
+ struct dentry *d_tracer;
+ struct dentry *entry;
+
+ 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);
+
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply related [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
2009-05-09 0:48 ` [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer Masami Hiramatsu
@ 2009-05-09 16:36 ` Frédéric Weisbecker
2009-05-11 9:32 ` Christoph Hellwig
1 sibling, 0 replies; 57+ messages in thread
From: Frédéric Weisbecker @ 2009-05-09 16:36 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm,
Ananth N Mavinakayanahalli
Hi,
2009/5/9 Masami Hiramatsu <mhiramat@redhat.com>:
> 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).
>
> Changes from v4:
> - Change interface name from 'kprobe_probes' to 'kprobe_events'
> - Skip comments (words after '#') from inputs of 'kprobe_events'.
>
> Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
> Cc: Steven Rostedt <rostedt@goodmis.org>
> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
> Cc: Ingo Molnar <mingo@elte.hu>
> Cc: Frederic Weisbecker <fweisbec@gmail.com>
> ---
>
> Documentation/trace/ftrace.txt | 55 +++++
> kernel/trace/Kconfig | 9 +
> kernel/trace/Makefile | 1
> kernel/trace/trace_kprobe.c | 404 ++++++++++++++++++++++++++++++++++++++++
> 4 files changed, 469 insertions(+), 0 deletions(-)
> create mode 100644 kernel/trace/trace_kprobe.c
>
> diff --git a/Documentation/trace/ftrace.txt b/Documentation/trace/ftrace.txt
> index fd9a3e6..2b8ead6 100644
> --- a/Documentation/trace/ftrace.txt
> +++ b/Documentation/trace/ftrace.txt
> @@ -1310,6 +1310,61 @@ dereference in a kernel module:
> [...]
>
>
> +kprobe-based event tracer
> +---------------------------
> +
> +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
> +/debug/tracing/kprobe_events.
> +
> +Synopsis of kprobe_events:
> + p SYMBOL[+offs|-offs]|MEMADDR : set a probe
> + r SYMBOL[+0] : set a return probe
> +
> +E.g.
> + echo p sys_open > /debug/tracing/kprobe_events
> +
> + This sets a kprobe on the top of sys_open() function.
> +
> + echo r sys_open >> /debug/tracing/kprobe_events
> +
> + This sets a kretprobe on the return point of sys_open() function.
> +
> + echo > /debug/tracing/kprobe_events
> +
> + This clears all probe points. and you can see the traced information via
> +/debug/tracing/trace.
> +
> + cat /debug/tracing/trace
> +# tracer: nop
> +#
> +# TASK-PID CPU# TIMESTAMP FUNCTION
> +# | | | | |
> + <...>-5117 [003] 416.481638: sys_open: @sys_open+0
> + <...>-5117 [003] 416.481662: syscall_call: <-sys_open+0
> + <...>-5117 [003] 416.481739: sys_open: @sys_open+0
> + <...>-5117 [003] 416.481762: sysenter_do_call: <-sys_open+0
> + <...>-5117 [003] 416.481818: sys_open: @sys_open+0
> + <...>-5117 [003] 416.481842: sysenter_do_call: <-sys_open+0
> + <...>-5117 [003] 416.481882: sys_open: @sys_open+0
> + <...>-5117 [003] 416.481905: sysenter_do_call: <-sys_open+0
> +
> + @SYMBOL means that kernel hits a probe, and <-SYMBOL means kernel returns
> +from SYMBOL(e.g. "sysenter_do_call: <-sys_open+0" means kernel returns from
> +sys_open to sysenter_do_call).
> +
> +
> function graph tracer
> ---------------------------
>
> diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
> index 7370253..914df9c 100644
> --- a/kernel/trace/Kconfig
> +++ b/kernel/trace/Kconfig
> @@ -398,6 +398,15 @@ config BLK_DEV_IO_TRACE
>
> If unsure, say N.
>
> +config KPROBE_TRACER
> + depends on KPROBES
> + depends on X86
> + bool "Trace kprobes"
> + select TRACING
> + help
> + This tracer probes everywhere where kprobes can probe it, and
> + records various registers and memories specified by user.
> +
> config DYNAMIC_FTRACE
> bool "enable/disable ftrace tracepoints dynamically"
> depends on FUNCTION_TRACER
> diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile
> index 06b8585..166c859 100644
> --- a/kernel/trace/Makefile
> +++ b/kernel/trace/Makefile
> @@ -51,5 +51,6 @@ obj-$(CONFIG_EVENT_TRACING) += trace_export.o
> obj-$(CONFIG_FTRACE_SYSCALLS) += trace_syscalls.o
> obj-$(CONFIG_EVENT_PROFILE) += trace_event_profile.o
> obj-$(CONFIG_EVENT_TRACING) += trace_events_filter.o
> +obj-$(CONFIG_KPROBE_TRACER) += trace_kprobe.o
>
> libftrace-y := ftrace.o
> diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
> new file mode 100644
> index 0000000..8112505
> --- /dev/null
> +++ b/kernel/trace/trace_kprobe.c
> @@ -0,0 +1,404 @@
> +/*
> + * kprobe based kernel tracer
> + *
> + * Created by Masami Hiramatsu <mhiramat@redhat.com>
> + *
> + * 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/ftrace.h>
> +#include "trace.h"
> +
> +/**
> + * kprobe_trace_core
> + */
> +#define TRACE_MAXARGS 6
> +
> +struct trace_probe {
> + struct list_head list;
> + union {
> + struct kprobe kp;
> + struct kretprobe rp;
> + };
> + const char *symbol; /* symbol name */
> +};
> +
> +static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
> + struct pt_regs *regs);
> +
> +static int kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs)
> +{
> + struct trace_probe *tp = container_of(kp, struct trace_probe, kp);
> +
> + kprobe_trace_record(instruction_pointer(regs), tp, regs);
> + return 0;
> +}
> +
> +static int kretprobe_trace_func(struct kretprobe_instance *ri,
> + struct pt_regs *regs)
> +{
> + struct trace_probe *tp = container_of(ri->rp, struct trace_probe, rp);
> +
> + kprobe_trace_record((unsigned long)ri->ret_addr, tp, regs);
> + return 0;
> +}
> +
> +static int probe_is_return(struct trace_probe *tp)
> +{
> + return (tp->rp.handler == kretprobe_trace_func);
> +}
> +
> +static const char *probe_symbol(struct trace_probe *tp)
> +{
> + return tp->symbol ? tp->symbol : "unknown";
> +}
> +
> +static long probe_offset(struct trace_probe *tp)
> +{
> + return (probe_is_return(tp)) ? tp->rp.kp.offset : tp->kp.offset;
> +}
> +
> +static void *probe_address(struct trace_probe *tp)
> +{
> + return (probe_is_return(tp)) ? tp->rp.kp.addr : tp->kp.addr;
> +}
> +
> +
> +static DEFINE_MUTEX(probe_lock);
> +static LIST_HEAD(probe_list);
> +
> +static struct trace_probe *alloc_trace_probe(const char *symbol)
> +{
> + 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) {
> + kfree(tp);
> + return ERR_PTR(-ENOMEM);
> + }
> + }
> +
> + INIT_LIST_HEAD(&tp->list);
> + return tp;
> +}
> +
> +static void free_trace_probe(struct trace_probe *tp)
> +{
> + kfree(tp->symbol);
> + kfree(tp);
> +}
> +
> +static int register_trace_probe(struct trace_probe *tp)
> +{
> + int ret;
> +
> + mutex_lock(&probe_lock);
> + list_add_tail(&tp->list, &probe_list);
> +
> + 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;
> + }
> + list_del(&tp->list);
> + }
> + mutex_unlock(&probe_lock);
> + return ret;
> +}
> +
> +static void unregister_trace_probe(struct trace_probe *tp)
> +{
> + if (probe_is_return(tp))
> + unregister_kretprobe(&tp->rp);
> + else
> + unregister_kprobe(&tp->kp);
> + list_del(&tp->list);
> +}
> +
> +static int create_trace_probe(int argc, char **argv)
> +{
> + /*
> + * Argument syntax:
> + * - Add kprobe: p SYMBOL[+OFFS|-OFFS]|ADDRESS
> + * - Add kretprobe: r SYMBOL[+0]
> + */
> + struct trace_probe *tp;
> + struct kprobe *kp;
> + char *tmp;
> + int ret = 0;
> + int is_return = 0;
> + char *symbol = 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 (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 */
> + 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';
> + }
> + if (offset && is_return)
> + return -EINVAL;
> + }
> +
> + /* setup a probe */
> + tp = alloc_trace_probe(symbol);
> + 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) {
> + /* TODO: check offset is collect by using insn_decoder */
> + kp->symbol_name = tp->symbol;
> + kp->offset = offset;
> + } else
> + kp->addr = addr;
> +
> + 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;
> +
> + if (tp == NULL)
> + return 0;
> +
> + if (tp->symbol)
> + seq_printf(m, "%c %s%+ld\n",
> + probe_is_return(tp) ? 'r' : 'p',
> + probe_symbol(tp), probe_offset(tp));
> + else
> + seq_printf(m, "%c 0x%p\n",
> + probe_is_return(tp) ? 'r' : 'p',
> + probe_address(tp));
> + 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);
> +}
> +
> +
> +#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;
> + char **argv = NULL;
> + int argc = 0;
> + 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';
> +
> + argv = argv_split(GFP_KERNEL, kbuf, &argc);
> + if (!argv) {
> + ret = -ENOMEM;
> + goto out;
> + }
> +
> + if (argc)
> + ret = create_trace_probe(argc, argv);
> +
> + argv_free(argv);
> + if (ret < 0)
> + 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,
> +};
> +
> +/* event recording functions */
> +static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
> + struct pt_regs *regs)
> +{
> + __trace_bprintk(ip, "%s%s%+ld\n",
> + probe_is_return(tp) ? "<-" : "@",
> + probe_symbol(tp), probe_offset(tp));
> +}
What happens here if you have:
kprobe_trace_record() {
probe_symbol() {
.... probes_open() {
cleanup_all_probes() {
free_trace_probe();
return tp->symbol ? ....; //crack!
I wonder if you shouldn't use a per_cpu list of probes,
spinlocked/irqsaved accessed
and also a kind of prevention against nmi.
Or better, you can use rcu, but we won't be able to probe rcu functions....
Frederic.
> +
> +/* Make a debugfs interface for controling probe points */
> +static __init int init_kprobe_trace(void)
> +{
> + struct dentry *d_tracer;
> + struct dentry *entry;
> +
> + 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);
> +
>
>
> --
> Masami Hiramatsu
>
> Software Engineer
> Hitachi Computer Products (America) Inc.
> Software Solutions Division
>
> e-mail: mhiramat@redhat.com
>
^ permalink raw reply [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
@ 2009-05-09 16:36 ` Frédéric Weisbecker
0 siblings, 0 replies; 57+ messages in thread
From: Frédéric Weisbecker @ 2009-05-09 16:36 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm,
Ananth N Mavinakayanahalli
Hi,
2009/5/9 Masami Hiramatsu <mhiramat@redhat.com>:
> 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).
>
> Changes from v4:
> - Change interface name from 'kprobe_probes' to 'kprobe_events'
> - Skip comments (words after '#') from inputs of 'kprobe_events'.
>
> Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
> Cc: Steven Rostedt <rostedt@goodmis.org>
> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
> Cc: Ingo Molnar <mingo@elte.hu>
> Cc: Frederic Weisbecker <fweisbec@gmail.com>
> ---
>
> Documentation/trace/ftrace.txt | 55 +++++
> kernel/trace/Kconfig | 9 +
> kernel/trace/Makefile | 1
> kernel/trace/trace_kprobe.c | 404 ++++++++++++++++++++++++++++++++++++++++
> 4 files changed, 469 insertions(+), 0 deletions(-)
> create mode 100644 kernel/trace/trace_kprobe.c
>
> diff --git a/Documentation/trace/ftrace.txt b/Documentation/trace/ftrace.txt
> index fd9a3e6..2b8ead6 100644
> --- a/Documentation/trace/ftrace.txt
> +++ b/Documentation/trace/ftrace.txt
> @@ -1310,6 +1310,61 @@ dereference in a kernel module:
> [...]
>
>
> +kprobe-based event tracer
> +---------------------------
> +
> +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
> +/debug/tracing/kprobe_events.
> +
> +Synopsis of kprobe_events:
> + p SYMBOL[+offs|-offs]|MEMADDR : set a probe
> + r SYMBOL[+0] : set a return probe
> +
> +E.g.
> + echo p sys_open > /debug/tracing/kprobe_events
> +
> + This sets a kprobe on the top of sys_open() function.
> +
> + echo r sys_open >> /debug/tracing/kprobe_events
> +
> + This sets a kretprobe on the return point of sys_open() function.
> +
> + echo > /debug/tracing/kprobe_events
> +
> + This clears all probe points. and you can see the traced information via
> +/debug/tracing/trace.
> +
> + cat /debug/tracing/trace
> +# tracer: nop
> +#
> +# TASK-PID CPU# TIMESTAMP FUNCTION
> +# | | | | |
> + <...>-5117 [003] 416.481638: sys_open: @sys_open+0
> + <...>-5117 [003] 416.481662: syscall_call: <-sys_open+0
> + <...>-5117 [003] 416.481739: sys_open: @sys_open+0
> + <...>-5117 [003] 416.481762: sysenter_do_call: <-sys_open+0
> + <...>-5117 [003] 416.481818: sys_open: @sys_open+0
> + <...>-5117 [003] 416.481842: sysenter_do_call: <-sys_open+0
> + <...>-5117 [003] 416.481882: sys_open: @sys_open+0
> + <...>-5117 [003] 416.481905: sysenter_do_call: <-sys_open+0
> +
> + @SYMBOL means that kernel hits a probe, and <-SYMBOL means kernel returns
> +from SYMBOL(e.g. "sysenter_do_call: <-sys_open+0" means kernel returns from
> +sys_open to sysenter_do_call).
> +
> +
> function graph tracer
> ---------------------------
>
> diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
> index 7370253..914df9c 100644
> --- a/kernel/trace/Kconfig
> +++ b/kernel/trace/Kconfig
> @@ -398,6 +398,15 @@ config BLK_DEV_IO_TRACE
>
> If unsure, say N.
>
> +config KPROBE_TRACER
> + depends on KPROBES
> + depends on X86
> + bool "Trace kprobes"
> + select TRACING
> + help
> + This tracer probes everywhere where kprobes can probe it, and
> + records various registers and memories specified by user.
> +
> config DYNAMIC_FTRACE
> bool "enable/disable ftrace tracepoints dynamically"
> depends on FUNCTION_TRACER
> diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile
> index 06b8585..166c859 100644
> --- a/kernel/trace/Makefile
> +++ b/kernel/trace/Makefile
> @@ -51,5 +51,6 @@ obj-$(CONFIG_EVENT_TRACING) += trace_export.o
> 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_KPROBE_TRACER) += trace_kprobe.o
>
> libftrace-y := ftrace.o
> diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
> new file mode 100644
> index 0000000..8112505
> --- /dev/null
> +++ b/kernel/trace/trace_kprobe.c
> @@ -0,0 +1,404 @@
> +/*
> + * kprobe based kernel tracer
> + *
> + * Created by Masami Hiramatsu <mhiramat@redhat.com>
> + *
> + * 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/ftrace.h>
> +#include "trace.h"
> +
> +/**
> + * kprobe_trace_core
> + */
> +#define TRACE_MAXARGS 6
> +
> +struct trace_probe {
> + struct list_head list;
> + union {
> + struct kprobe kp;
> + struct kretprobe rp;
> + };
> + const char *symbol; /* symbol name */
> +};
> +
> +static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
> + struct pt_regs *regs);
> +
> +static int kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs)
> +{
> + struct trace_probe *tp = container_of(kp, struct trace_probe, kp);
> +
> + kprobe_trace_record(instruction_pointer(regs), tp, regs);
> + return 0;
> +}
> +
> +static int kretprobe_trace_func(struct kretprobe_instance *ri,
> + struct pt_regs *regs)
> +{
> + struct trace_probe *tp = container_of(ri->rp, struct trace_probe, rp);
> +
> + kprobe_trace_record((unsigned long)ri->ret_addr, tp, regs);
> + return 0;
> +}
> +
> +static int probe_is_return(struct trace_probe *tp)
> +{
> + return (tp->rp.handler == kretprobe_trace_func);
> +}
> +
> +static const char *probe_symbol(struct trace_probe *tp)
> +{
> + return tp->symbol ? tp->symbol : "unknown";
> +}
> +
> +static long probe_offset(struct trace_probe *tp)
> +{
> + return (probe_is_return(tp)) ? tp->rp.kp.offset : tp->kp.offset;
> +}
> +
> +static void *probe_address(struct trace_probe *tp)
> +{
> + return (probe_is_return(tp)) ? tp->rp.kp.addr : tp->kp.addr;
> +}
> +
> +
> +static DEFINE_MUTEX(probe_lock);
> +static LIST_HEAD(probe_list);
> +
> +static struct trace_probe *alloc_trace_probe(const char *symbol)
> +{
> + 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) {
> + kfree(tp);
> + return ERR_PTR(-ENOMEM);
> + }
> + }
> +
> + INIT_LIST_HEAD(&tp->list);
> + return tp;
> +}
> +
> +static void free_trace_probe(struct trace_probe *tp)
> +{
> + kfree(tp->symbol);
> + kfree(tp);
> +}
> +
> +static int register_trace_probe(struct trace_probe *tp)
> +{
> + int ret;
> +
> + mutex_lock(&probe_lock);
> + list_add_tail(&tp->list, &probe_list);
> +
> + 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;
> + }
> + list_del(&tp->list);
> + }
> + mutex_unlock(&probe_lock);
> + return ret;
> +}
> +
> +static void unregister_trace_probe(struct trace_probe *tp)
> +{
> + if (probe_is_return(tp))
> + unregister_kretprobe(&tp->rp);
> + else
> + unregister_kprobe(&tp->kp);
> + list_del(&tp->list);
> +}
> +
> +static int create_trace_probe(int argc, char **argv)
> +{
> + /*
> + * Argument syntax:
> + * - Add kprobe: p SYMBOL[+OFFS|-OFFS]|ADDRESS
> + * - Add kretprobe: r SYMBOL[+0]
> + */
> + struct trace_probe *tp;
> + struct kprobe *kp;
> + char *tmp;
> + int ret = 0;
> + int is_return = 0;
> + char *symbol = 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 (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 */
> + 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';
> + }
> + if (offset && is_return)
> + return -EINVAL;
> + }
> +
> + /* setup a probe */
> + tp = alloc_trace_probe(symbol);
> + 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) {
> + /* TODO: check offset is collect by using insn_decoder */
> + kp->symbol_name = tp->symbol;
> + kp->offset = offset;
> + } else
> + kp->addr = addr;
> +
> + 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;
> +
> + if (tp == NULL)
> + return 0;
> +
> + if (tp->symbol)
> + seq_printf(m, "%c %s%+ld\n",
> + probe_is_return(tp) ? 'r' : 'p',
> + probe_symbol(tp), probe_offset(tp));
> + else
> + seq_printf(m, "%c 0x%p\n",
> + probe_is_return(tp) ? 'r' : 'p',
> + probe_address(tp));
> + 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);
> +}
> +
> +
> +#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;
> + char **argv = NULL;
> + int argc = 0;
> + 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';
> +
> + argv = argv_split(GFP_KERNEL, kbuf, &argc);
> + if (!argv) {
> + ret = -ENOMEM;
> + goto out;
> + }
> +
> + if (argc)
> + ret = create_trace_probe(argc, argv);
> +
> + argv_free(argv);
> + if (ret < 0)
> + 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,
> +};
> +
> +/* event recording functions */
> +static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
> + struct pt_regs *regs)
> +{
> + __trace_bprintk(ip, "%s%s%+ld\n",
> + probe_is_return(tp) ? "<-" : "@",
> + probe_symbol(tp), probe_offset(tp));
> +}
What happens here if you have:
kprobe_trace_record() {
probe_symbol() {
.... probes_open() {
cleanup_all_probes() {
free_trace_probe();
return tp->symbol ? ....; //crack!
I wonder if you shouldn't use a per_cpu list of probes,
spinlocked/irqsaved accessed
and also a kind of prevention against nmi.
Or better, you can use rcu, but we won't be able to probe rcu functions....
Frederic.
> +
> +/* Make a debugfs interface for controling probe points */
> +static __init int init_kprobe_trace(void)
> +{
> + struct dentry *d_tracer;
> + struct dentry *entry;
> +
> + 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);
> +
>
>
> --
> Masami Hiramatsu
>
> Software Engineer
> Hitachi Computer Products (America) Inc.
> Software Solutions Division
>
> e-mail: mhiramat@redhat.com
>
^ permalink raw reply [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
2009-05-09 16:36 ` Frédéric Weisbecker
(?)
@ 2009-05-09 17:33 ` Masami Hiramatsu
2009-05-11 21:26 ` Frederic Weisbecker
-1 siblings, 1 reply; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 17:33 UTC (permalink / raw)
To: Frédéric Weisbecker
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm,
Ananth N Mavinakayanahalli
Frédéric Weisbecker wrote:
> Hi,
>
> 2009/5/9 Masami Hiramatsu <mhiramat@redhat.com>:
[...]
>> +
>> +/* event recording functions */
>> +static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
>> + struct pt_regs *regs)
>> +{
>> + __trace_bprintk(ip, "%s%s%+ld\n",
>> + probe_is_return(tp) ? "<-" : "@",
>> + probe_symbol(tp), probe_offset(tp));
>> +}
>
>
>
> What happens here if you have:
>
> kprobe_trace_record() {
> probe_symbol() {
> .... probes_open() {
> cleanup_all_probes() {
> free_trace_probe();
> return tp->symbol ? ....; //crack!
>
> I wonder if you shouldn't use a per_cpu list of probes,
> spinlocked/irqsaved accessed
> and also a kind of prevention against nmi.
Sure, cleanup_all_probes() invokes unregister_kprobe() via
unregister_trace_probe(), which waits running probe-handlers by
using synchronize_sched()(because kprobes disables preemption
around its handlers), before free_trace_probe().
So you don't need any locks there :-)
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
2009-05-09 17:33 ` Masami Hiramatsu
@ 2009-05-11 21:26 ` Frederic Weisbecker
0 siblings, 0 replies; 57+ messages in thread
From: Frederic Weisbecker @ 2009-05-11 21:26 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm,
Ananth N Mavinakayanahalli
On Sat, May 09, 2009 at 01:33:53PM -0400, Masami Hiramatsu wrote:
> Frédéric Weisbecker wrote:
> > Hi,
> >
> > 2009/5/9 Masami Hiramatsu <mhiramat@redhat.com>:
> [...]
> >> +
> >> +/* event recording functions */
> >> +static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
> >> + struct pt_regs *regs)
> >> +{
> >> + __trace_bprintk(ip, "%s%s%+ld\n",
> >> + probe_is_return(tp) ? "<-" : "@",
> >> + probe_symbol(tp), probe_offset(tp));
> >> +}
> >
> >
> >
> > What happens here if you have:
> >
> > kprobe_trace_record() {
> > probe_symbol() {
> > .... probes_open() {
> > cleanup_all_probes() {
> > free_trace_probe();
> > return tp->symbol ? ....; //crack!
> >
> > I wonder if you shouldn't use a per_cpu list of probes,
> > spinlocked/irqsaved accessed
> > and also a kind of prevention against nmi.
>
> Sure, cleanup_all_probes() invokes unregister_kprobe() via
> unregister_trace_probe(), which waits running probe-handlers by
> using synchronize_sched()(because kprobes disables preemption
> around its handlers), before free_trace_probe().
>
> So you don't need any locks there :-)
>
> Thank you,
>
>
Aah, ok :)
So this patch looks sane.
Thanks.
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
2009-05-09 0:48 ` [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer Masami Hiramatsu
2009-05-09 16:36 ` Frédéric Weisbecker
@ 2009-05-11 9:32 ` Christoph Hellwig
2009-05-11 10:53 ` Ingo Molnar
2009-05-11 15:28 ` Frank Ch. Eigler
1 sibling, 2 replies; 57+ messages in thread
From: Christoph Hellwig @ 2009-05-11 9:32 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm,
Ananth N Mavinakayanahalli, Frederic Weisbecker, Tom Zanussi
On Fri, May 08, 2009 at 08:48:59PM -0400, Masami Hiramatsu wrote:
> 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).
That's some pretty cool functionality, especially together with patch 7.
But as with so many tracing bits in the kernel it's just lowlevel bits
without a good user interface. We'd really need some high-level way
for sysadmins/developers to use it. E.g. a version of the systemtap
compiler that doesn't build a kernel module but instead uses the event
tracer + the kprobes tracer.
Or a model like Tom's zedtrace where a perl script would do the dwarf
lookups and generates these probes in addition to the filtered event
traces.
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
2009-05-11 9:32 ` Christoph Hellwig
@ 2009-05-11 10:53 ` Ingo Molnar
2009-05-11 15:28 ` Frank Ch. Eigler
1 sibling, 0 replies; 57+ messages in thread
From: Ingo Molnar @ 2009-05-11 10:53 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Masami Hiramatsu, Steven Rostedt, lkml, systemtap, kvm,
Ananth N Mavinakayanahalli, Frederic Weisbecker, Tom Zanussi
* Christoph Hellwig <hch@infradead.org> wrote:
> On Fri, May 08, 2009 at 08:48:59PM -0400, Masami Hiramatsu wrote:
> > 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).
>
> That's some pretty cool functionality, especially together with
> patch 7.
Yes. I insisted on this model, because this is essentially
kprobes-done-right. Exposing unsafe kernel instrumentation APIs was
a big mistake to merge upstream, it delayed the proper design of
this stuff by almost a decade.
There's two more details to be solved before this can go into the
tracing tree.
> But as with so many tracing bits in the kernel it's just lowlevel
> bits without a good user interface. We'd really need some
> high-level way for sysadmins/developers to use it. E.g. a version
> of the systemtap compiler that doesn't build a kernel module but
> instead uses the event tracer + the kprobes tracer.
Yes, exactly.
> Or a model like Tom's zedtrace where a perl script would do the
> dwarf lookups and generates these probes in addition to the
> filtered event traces.
Correct, that's the other, IMHO superior direction that is being
pursued. If you look at the evolution of the filter code it gives
the seeds for safe scripting done in the kernel.
Such filters/scripts can then be reused for a whole lot more stuff,
such as security rules. (netfilters could use it too, etc.)
Ingo
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
@ 2009-05-11 10:53 ` Ingo Molnar
0 siblings, 0 replies; 57+ messages in thread
From: Ingo Molnar @ 2009-05-11 10:53 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Masami Hiramatsu, Steven Rostedt, lkml, systemtap, kvm,
Ananth N Mavinakayanahalli, Frederic Weisbecker, Tom Zanussi
* Christoph Hellwig <hch@infradead.org> wrote:
> On Fri, May 08, 2009 at 08:48:59PM -0400, Masami Hiramatsu wrote:
> > 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).
>
> That's some pretty cool functionality, especially together with
> patch 7.
Yes. I insisted on this model, because this is essentially
kprobes-done-right. Exposing unsafe kernel instrumentation APIs was
a big mistake to merge upstream, it delayed the proper design of
this stuff by almost a decade.
There's two more details to be solved before this can go into the
tracing tree.
> But as with so many tracing bits in the kernel it's just lowlevel
> bits without a good user interface. We'd really need some
> high-level way for sysadmins/developers to use it. E.g. a version
> of the systemtap compiler that doesn't build a kernel module but
> instead uses the event tracer + the kprobes tracer.
Yes, exactly.
> Or a model like Tom's zedtrace where a perl script would do the
> dwarf lookups and generates these probes in addition to the
> filtered event traces.
Correct, that's the other, IMHO superior direction that is being
pursued. If you look at the evolution of the filter code it gives
the seeds for safe scripting done in the kernel.
Such filters/scripts can then be reused for a whole lot more stuff,
such as security rules. (netfilters could use it too, etc.)
Ingo
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
2009-05-11 9:32 ` Christoph Hellwig
@ 2009-05-11 15:28 ` Frank Ch. Eigler
2009-05-11 15:28 ` Frank Ch. Eigler
1 sibling, 0 replies; 57+ messages in thread
From: Frank Ch. Eigler @ 2009-05-11 15:28 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Masami Hiramatsu, Ingo Molnar, Steven Rostedt, lkml, systemtap,
kvm, Ananth N Mavinakayanahalli, Frederic Weisbecker,
Tom Zanussi
Christoph Hellwig <hch@infradead.org> writes:
> [...] But as with so many tracing bits in the kernel it's just
> lowlevel bits without a good user interface. We'd really need some
> high-level way for sysadmins/developers to use it. E.g. a version
> of the systemtap compiler that doesn't build a kernel module but
> instead uses the event tracer + the kprobes tracer. [...]
This (the translator detecting that a particular script is simple
enough to be executed in this manner) is theoretically possible.
- FChE
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 4/7] tracing: add kprobe-based event tracer
@ 2009-05-11 15:28 ` Frank Ch. Eigler
0 siblings, 0 replies; 57+ messages in thread
From: Frank Ch. Eigler @ 2009-05-11 15:28 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Masami Hiramatsu, Ingo Molnar, Steven Rostedt, lkml, systemtap,
kvm, Ananth N Mavinakayanahalli, Frederic Weisbecker,
Tom Zanussi
Christoph Hellwig <hch@infradead.org> writes:
> [...] But as with so many tracing bits in the kernel it's just
> lowlevel bits without a good user interface. We'd really need some
> high-level way for sysadmins/developers to use it. E.g. a version
> of the systemtap compiler that doesn't build a kernel module but
> instead uses the event tracer + the kprobes tracer. [...]
This (the translator detecting that a particular script is simple
enough to be executed in this manner) is theoretically possible.
- FChE
^ permalink raw reply [flat|nested] 57+ messages in thread
* [PATCH -tip v5 5/7] x86: fix kernel_trap_sp()
2009-05-09 0:48 ` Masami Hiramatsu
` (4 preceding siblings ...)
(?)
@ 2009-05-09 0:49 ` Masami Hiramatsu
2009-05-11 9:28 ` Christoph Hellwig
-1 siblings, 1 reply; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:49 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: systemtap, kvm, Masami Hiramatsu, Harvey Harrison, Ingo Molnar,
Thomas Gleixner, Jan Blunck
Use ®s->sp instead of regs for getting the top of stack in kernel mode.
(on x86-64, regs->sp always points the top of stack)
[ impact: Oprofile decodes only stack for backtracing on i386 ]
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jan Blunck <jblunck@suse.de>
---
arch/x86/include/asm/ptrace.h | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/x86/include/asm/ptrace.h b/arch/x86/include/asm/ptrace.h
index 5cdd19f..90b76b3 100644
--- a/arch/x86/include/asm/ptrace.h
+++ b/arch/x86/include/asm/ptrace.h
@@ -187,14 +187,14 @@ static inline int v8086_mode(struct pt_regs *regs)
/*
* X86_32 CPUs don't save ss and esp if the CPU is already in kernel mode
- * when it traps. So regs will be the current sp.
+ * when it traps. So ®s->sp will be the current sp.
*
* This is valid only for kernel mode traps.
*/
static inline unsigned long kernel_trap_sp(struct pt_regs *regs)
{
#ifdef CONFIG_X86_32
- return (unsigned long)regs;
+ return (unsigned long)®s->sp;
#else
return regs->sp;
#endif
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply related [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 5/7] x86: fix kernel_trap_sp()
2009-05-09 0:49 ` [PATCH -tip v5 5/7] x86: fix kernel_trap_sp() Masami Hiramatsu
@ 2009-05-11 9:28 ` Christoph Hellwig
2009-05-11 13:48 ` Masami Hiramatsu
0 siblings, 1 reply; 57+ messages in thread
From: Christoph Hellwig @ 2009-05-11 9:28 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm,
Harvey Harrison, Thomas Gleixner, Jan Blunck
On Fri, May 08, 2009 at 08:49:04PM -0400, Masami Hiramatsu wrote:
> Use ®s->sp instead of regs for getting the top of stack in kernel mode.
> (on x86-64, regs->sp always points the top of stack)
Shouldn't this patch be sent for inclusion ASAP instead of sitting in
this series?
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 5/7] x86: fix kernel_trap_sp()
2009-05-11 9:28 ` Christoph Hellwig
@ 2009-05-11 13:48 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 13:48 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm,
Harvey Harrison, Thomas Gleixner, Jan Blunck
Christoph Hellwig wrote:
> On Fri, May 08, 2009 at 08:49:04PM -0400, Masami Hiramatsu wrote:
>> Use ®s->sp instead of regs for getting the top of stack in kernel mode.
>> (on x86-64, regs->sp always points the top of stack)
>
> Shouldn't this patch be sent for inclusion ASAP instead of sitting in
> this series?
Yes, and it's just for asking a comment from oprofile developers,
because this change will change its behavior a bit.
Anyway, I'll post it separately against upstream.
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 5/7] x86: fix kernel_trap_sp()
@ 2009-05-11 13:48 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 13:48 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Ingo Molnar, Steven Rostedt, lkml, systemtap, kvm,
Harvey Harrison, Thomas Gleixner, Jan Blunck
Christoph Hellwig wrote:
> On Fri, May 08, 2009 at 08:49:04PM -0400, Masami Hiramatsu wrote:
>> Use ®s->sp instead of regs for getting the top of stack in kernel mode.
>> (on x86-64, regs->sp always points the top of stack)
>
> Shouldn't this patch be sent for inclusion ASAP instead of sitting in
> this series?
Yes, and it's just for asking a comment from oprofile developers,
because this change will change its behavior a bit.
Anyway, I'll post it separately against upstream.
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* [PATCH -tip v5 6/7] x86: add pt_regs register and stack access APIs
2009-05-09 0:48 ` Masami Hiramatsu
` (5 preceding siblings ...)
(?)
@ 2009-05-09 0:49 ` Masami Hiramatsu
-1 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:49 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: systemtap, kvm, Masami Hiramatsu, Steven Rostedt,
Ananth N Mavinakayanahalli, Ingo Molnar, Frederic Weisbecker,
Roland McGrath
Add following APIs for accessing registers and stack entries from pt_regs.
- query_register_offset(const char *name)
Query the offset of "name" register.
- query_register_name(unsigned offset)
Query the name of register by its offset.
- get_register(struct pt_regs *regs, unsigned offset)
Get the value of a register by its offset.
- valid_stack_address(struct pt_regs *regs, unsigned long addr)
Check the address is in the stack.
- get_stack_nth(struct pt_regs *reg, unsigned nth)
Get Nth entry of the stack. (N >= 0)
- get_argument_nth(struct pt_regs *reg, unsigned nth)
Get Nth argument at function call. (N >= 0)
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Roland McGrath <roland@redhat.com>
---
arch/x86/include/asm/ptrace.h | 66 +++++++++++++++++++++++++++++++++++++++++
arch/x86/kernel/ptrace.c | 60 +++++++++++++++++++++++++++++++++++++
2 files changed, 126 insertions(+), 0 deletions(-)
diff --git a/arch/x86/include/asm/ptrace.h b/arch/x86/include/asm/ptrace.h
index 90b76b3..c316b85 100644
--- a/arch/x86/include/asm/ptrace.h
+++ b/arch/x86/include/asm/ptrace.h
@@ -7,6 +7,7 @@
#ifdef __KERNEL__
#include <asm/segment.h>
+#include <asm/page_types.h>
#endif
#ifndef __ASSEMBLY__
@@ -215,6 +216,71 @@ 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 query_register_offset(const char *name);
+extern const char *query_register_name(unsigned offset);
+#define MAX_REG_OFFSET (offsetof(struct pt_regs, ss))
+
+/* Get register value from its offset */
+static inline unsigned long get_register(struct pt_regs *regs, unsigned offset)
+{
+ if (unlikely(offset > MAX_REG_OFFSET))
+ return 0;
+ return *(unsigned long *)((unsigned long)regs + offset);
+}
+
+/* Check the address in the stack */
+static inline int valid_stack_address(struct pt_regs *regs, unsigned long addr)
+{
+ return ((addr & ~(THREAD_SIZE - 1)) ==
+ (kernel_trap_sp(regs) & ~(THREAD_SIZE - 1)));
+}
+
+/* Get Nth entry of the stack */
+static inline unsigned long get_stack_nth(struct pt_regs *regs, unsigned n)
+{
+ unsigned long *addr = (unsigned long *)kernel_trap_sp(regs);
+ addr += n;
+ if (valid_stack_address(regs, (unsigned long)addr))
+ return *addr;
+ else
+ return 0;
+}
+
+/* Get Nth argument at function call */
+static inline unsigned long get_argument_nth(struct pt_regs *regs, unsigned n)
+{
+#ifdef CONFIG_X86_32
+#define NR_REGPARMS 3
+ if (n < NR_REGPARMS) {
+ switch (n) {
+ case 0: return regs->ax;
+ case 1: return regs->dx;
+ case 2: return regs->cx;
+ }
+ return 0;
+#else /* CONFIG_X86_64 */
+#define NR_REGPARMS 6
+ 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;
+#endif
+ } else {
+ /*
+ * The typical case: arg n is on the stack.
+ * (Note: stack[0] = return address, so skip it)
+ */
+ return get_stack_nth(regs, 1 + n - NR_REGPARMS);
+ }
+}
+
/*
* These are defined as per linux/ptrace.h, which see.
*/
diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index 09ecbde..00eb9d7 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -48,6 +48,66 @@ enum x86_regset {
REGSET_IOPERM32,
};
+struct pt_regs_offset {
+ const char *name;
+ int offset;
+};
+
+#define REG_OFFSET(r) offsetof(struct pt_regs, r)
+#define REG_OFFSET_NAME(r) {.name = #r, .offset = REG_OFFSET(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,
+};
+
+int 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;
+}
+
+const char *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.
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply related [flat|nested] 57+ messages in thread* [PATCH -tip v5 7/7] tracing: add arguments support on kprobe-based event tracer
2009-05-09 0:48 ` Masami Hiramatsu
` (6 preceding siblings ...)
(?)
@ 2009-05-09 0:49 ` Masami Hiramatsu
2009-05-11 14:35 ` Steven Rostedt
-1 siblings, 1 reply; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-09 0:49 UTC (permalink / raw)
To: Ingo Molnar, Steven Rostedt, lkml
Cc: systemtap, kvm, Masami Hiramatsu, Steven Rostedt,
Ananth N Mavinakayanahalli, Ingo Molnar, Frederic Weisbecker
Support following probe arguments and add fetch functions on kprobe-based
event tracer.
%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.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
---
Documentation/trace/ftrace.txt | 47 +++-
kernel/trace/trace_kprobe.c | 431 ++++++++++++++++++++++++++++++++++++++--
2 files changed, 441 insertions(+), 37 deletions(-)
diff --git a/Documentation/trace/ftrace.txt b/Documentation/trace/ftrace.txt
index 2b8ead6..ce91398 100644
--- a/Documentation/trace/ftrace.txt
+++ b/Documentation/trace/ftrace.txt
@@ -1329,17 +1329,34 @@ current_tracer, instead of that, just set probe points via
/debug/tracing/kprobe_events.
Synopsis of kprobe_events:
- p SYMBOL[+offs|-offs]|MEMADDR : set a probe
- r SYMBOL[+0] : set a return probe
+ p SYMBOL[+offs|-offs]|MEMADDR [FETCHARGS] : set a probe
+ r SYMBOL[+0] [FETCHARGS] : set a return probe
+
+ FETCHARGS:
+ %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.
E.g.
- echo p sys_open > /debug/tracing/kprobe_events
+ echo p do_sys_open a0 a1 a2 a3 > /debug/tracing/kprobe_events
- This sets a kprobe on the top of sys_open() function.
+ This sets a kprobe on the top of do_sys_open() function with recording
+1st to 4th arguments.
- echo r sys_open >> /debug/tracing/kprobe_events
+ echo r do_sys_open rv ra >> /debug/tracing/kprobe_events
- This sets a kretprobe on the return point of sys_open() function.
+ This sets a kretprobe on the return point of do_sys_open() function with
+recording return value and return address.
echo > /debug/tracing/kprobe_events
@@ -1351,18 +1368,16 @@ E.g.
#
# TASK-PID CPU# TIMESTAMP FUNCTION
# | | | | |
- <...>-5117 [003] 416.481638: sys_open: @sys_open+0
- <...>-5117 [003] 416.481662: syscall_call: <-sys_open+0
- <...>-5117 [003] 416.481739: sys_open: @sys_open+0
- <...>-5117 [003] 416.481762: sysenter_do_call: <-sys_open+0
- <...>-5117 [003] 416.481818: sys_open: @sys_open+0
- <...>-5117 [003] 416.481842: sysenter_do_call: <-sys_open+0
- <...>-5117 [003] 416.481882: sys_open: @sys_open+0
- <...>-5117 [003] 416.481905: sysenter_do_call: <-sys_open+0
+ <...>-2376 [001] 262.389131: do_sys_open: @do_sys_open+0 0xffffff9c 0x98db83e 0x8880 0x0
+ <...>-2376 [001] 262.391166: sys_open: <-do_sys_open+0 0x5 0xc06e8ebb
+ <...>-2376 [001] 264.384876: do_sys_open: @do_sys_open+0 0xffffff9c 0x98db83e 0x8880 0x0
+ <...>-2376 [001] 264.386880: sys_open: <-do_sys_open+0 0x5 0xc06e8ebb
+ <...>-2084 [001] 265.380330: do_sys_open: @do_sys_open+0 0xffffff9c 0x804be3e 0x0 0x1b6
+ <...>-2084 [001] 265.380399: sys_open: <-do_sys_open+0 0x3 0xc06e8ebb
@SYMBOL means that kernel hits a probe, and <-SYMBOL means kernel returns
-from SYMBOL(e.g. "sysenter_do_call: <-sys_open+0" means kernel returns from
-sys_open to sysenter_do_call).
+from SYMBOL(e.g. "sys_open: <-do_sys_open+0" means kernel returns from
+do_sys_open to sys_open).
function graph tracer
diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index 8112505..b4f05de 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -27,10 +27,134 @@
#include <linux/types.h>
#include <linux/string.h>
#include <linux/ctype.h>
+#include <linux/ptrace.h>
#include <linux/ftrace.h>
#include "trace.h"
+/* currently, trace_kprobe only supports X86. */
+
+struct fetch_func {
+ unsigned long (*func)(struct pt_regs *, void *);
+ void *data;
+};
+
+static unsigned long call_fetch(struct fetch_func *f, struct pt_regs *regs)
+{
+ return f->func(regs, f->data);
+}
+
+/* fetch handlers */
+static unsigned long fetch_register(struct pt_regs *regs, void *offset)
+{
+ return get_register(regs, (unsigned)((unsigned long)offset));
+}
+
+static unsigned long fetch_stack(struct pt_regs *regs, void *num)
+{
+ return get_stack_nth(regs, (unsigned)((unsigned long)num));
+}
+
+static unsigned long fetch_memory(struct pt_regs *regs, void *addr)
+{
+ unsigned long retval;
+ if (probe_kernel_address(addr, retval))
+ return 0;
+ return retval;
+}
+
+static unsigned long fetch_argument(struct pt_regs *regs, void *num)
+{
+ return get_argument_nth(regs, (unsigned)((unsigned long)num));
+}
+
+static unsigned long fetch_retvalue(struct pt_regs *regs, void *dummy)
+{
+ return regs_return_value(regs);
+}
+
+static 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 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 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 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
*/
@@ -43,6 +167,8 @@ struct trace_probe {
struct kretprobe rp;
};
const char *symbol; /* symbol name */
+ unsigned int nr_args;
+ struct fetch_func args[TRACE_MAXARGS];
};
static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
@@ -111,6 +237,13 @@ static struct trace_probe *alloc_trace_probe(const char *symbol)
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->symbol);
kfree(tp);
}
@@ -150,17 +283,158 @@ static void unregister_trace_probe(struct trace_probe *tp)
list_del(&tp->list);
}
+/* 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, ¶m);
+ 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 = 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, ¶m);
+ 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, ¶m);
+ 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 SYMBOL[+OFFS|-OFFS]|ADDRESS
- * - Add kretprobe: r SYMBOL[+0]
+ * - Add kprobe: p SYMBOL[+OFFS|-OFFS]|ADDRESS [FETCHARGS]
+ * - Add kretprobe: r 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;
- char *tmp;
- int ret = 0;
+ int i, ret = 0;
int is_return = 0;
char *symbol = NULL;
long offset = 0;
@@ -187,19 +461,9 @@ static int create_trace_probe(int argc, char **argv)
/* a symbol specified */
symbol = argv[1];
/* TODO: support .init module functions */
- 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';
- }
+ ret = split_symbol_offset(symbol, &offset);
+ if (ret)
+ return ret;
if (offset && is_return)
return -EINVAL;
}
@@ -224,6 +488,15 @@ static int create_trace_probe(int argc, char **argv)
} else
kp->addr = addr;
+ /* parse arguments */
+ argc -= 2; argv += 2; ret = 0;
+ for (i = 0; i < argc && i < TRACE_MAXARGS; i++) {
+ 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;
@@ -265,21 +538,55 @@ static void probes_seq_stop(struct seq_file *m, void *v)
mutex_unlock(&probe_lock);
}
+static void arg_seq_print(struct seq_file *m, struct fetch_func *ff)
+{
+ if (ff->func == fetch_argument)
+ seq_printf(m, "a%lu", (unsigned long)ff->data);
+ else if (ff->func == fetch_register) {
+ const char *name;
+ name = query_register_name((unsigned)((long)ff->data));
+ seq_printf(m, "%%%s", name);
+ } else if (ff->func == fetch_stack)
+ seq_printf(m, "s%lu", (unsigned long)ff->data);
+ else if (ff->func == fetch_memory)
+ seq_printf(m, "@0x%p", ff->data);
+ else if (ff->func == fetch_symbol) {
+ struct symbol_cache *sc = ff->data;
+ seq_printf(m, "@%s%+ld", sc->symbol, sc->offset);
+ } else if (ff->func == fetch_retvalue)
+ seq_printf(m, "rv");
+ else if (ff->func == fetch_ip)
+ seq_printf(m, "ra");
+ else if (ff->func == fetch_indirect) {
+ struct indirect_fetch_data *id = ff->data;
+ seq_printf(m, "%+ld(", id->offset);
+ arg_seq_print(m, &id->orig);
+ seq_printf(m, ")");
+ }
+}
+
static int probes_seq_show(struct seq_file *m, void *v)
{
struct trace_probe *tp = v;
+ int i;
if (tp == NULL)
return 0;
if (tp->symbol)
- seq_printf(m, "%c %s%+ld\n",
+ seq_printf(m, "%c %s%+ld",
probe_is_return(tp) ? 'r' : 'p',
probe_symbol(tp), probe_offset(tp));
else
- seq_printf(m, "%c 0x%p\n",
+ seq_printf(m, "%c 0x%p",
probe_is_return(tp) ? 'r' : 'p',
probe_address(tp));
+
+ for (i = 0; i < tp->nr_args; i++) {
+ seq_printf(m, " ");
+ arg_seq_print(m, &tp->args[i]);
+ }
+ seq_printf(m, "\n");
return 0;
}
@@ -374,13 +681,95 @@ static const struct file_operations kprobe_events_ops = {
};
/* event recording functions */
-static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
- struct pt_regs *regs)
+/* TODO: rewrite based on trace_vprintk(maybe, trace_vprintk_begin/end?) */
+static void kprobe_trace_printk_0(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs)
{
__trace_bprintk(ip, "%s%s%+ld\n",
probe_is_return(tp) ? "<-" : "@",
probe_symbol(tp), probe_offset(tp));
}
+static void kprobe_trace_printk_1(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs)
+{
+ __trace_bprintk(ip, "%s%s%+ld 0x%lx\n",
+ probe_is_return(tp) ? "<-" : "@",
+ probe_symbol(tp), probe_offset(tp),
+ call_fetch(&tp->args[0], regs));
+}
+static void kprobe_trace_printk_2(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs)
+{
+ __trace_bprintk(ip, "%s%s%+ld 0x%lx 0x%lx\n",
+ probe_is_return(tp) ? "<-" : "@", probe_symbol(tp),
+ probe_offset(tp),
+ call_fetch(&tp->args[0], regs),
+ call_fetch(&tp->args[1], regs));
+}
+static void kprobe_trace_printk_3(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs)
+{
+ __trace_bprintk(ip, "%s%s%+ld 0x%lx 0x%lx 0x%lx\n",
+ probe_is_return(tp) ? "<-" : "@", probe_symbol(tp),
+ probe_offset(tp),
+ call_fetch(&tp->args[0], regs),
+ call_fetch(&tp->args[1], regs),
+ call_fetch(&tp->args[2], regs));
+}
+static void kprobe_trace_printk_4(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs)
+{
+ __trace_bprintk(ip, "%s%s%+ld 0x%lx 0x%lx 0x%lx 0x%lx\n",
+ probe_is_return(tp) ? "<-" : "@", probe_symbol(tp),
+ probe_offset(tp),
+ call_fetch(&tp->args[0], regs),
+ call_fetch(&tp->args[1], regs),
+ call_fetch(&tp->args[2], regs),
+ call_fetch(&tp->args[3], regs));
+}
+static void kprobe_trace_printk_5(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs)
+{
+ __trace_bprintk(ip, "%s%s%+ld 0x%lx 0x%lx 0x%lx 0x%lx 0x%lx\n",
+ probe_is_return(tp) ? "<-" : "@", probe_symbol(tp),
+ probe_offset(tp),
+ call_fetch(&tp->args[0], regs),
+ call_fetch(&tp->args[1], regs),
+ call_fetch(&tp->args[2], regs),
+ call_fetch(&tp->args[3], regs),
+ call_fetch(&tp->args[4], regs));
+}
+static void kprobe_trace_printk_6(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs)
+{
+ __trace_bprintk(ip, "%s%s%+ld 0x%lx 0x%lx 0x%lx 0x%lx 0x%lx 0x%lx\n",
+ probe_is_return(tp) ? "<-" : "@", probe_symbol(tp),
+ probe_offset(tp),
+ call_fetch(&tp->args[0], regs),
+ call_fetch(&tp->args[1], regs),
+ call_fetch(&tp->args[2], regs),
+ call_fetch(&tp->args[3], regs),
+ call_fetch(&tp->args[4], regs),
+ call_fetch(&tp->args[5], regs));
+}
+
+static void (*kprobe_trace_printk_n[TRACE_MAXARGS + 1])(unsigned long ip,
+ struct trace_probe *,
+ struct pt_regs *) = {
+ [0] = kprobe_trace_printk_0,
+ [1] = kprobe_trace_printk_1,
+ [2] = kprobe_trace_printk_2,
+ [3] = kprobe_trace_printk_3,
+ [4] = kprobe_trace_printk_4,
+ [5] = kprobe_trace_printk_5,
+ [6] = kprobe_trace_printk_6,
+};
+
+static void kprobe_trace_record(unsigned long ip, struct trace_probe *tp,
+ struct pt_regs *regs)
+{
+ kprobe_trace_printk_n[tp->nr_args](ip, tp, regs);
+}
/* Make a debugfs interface for controling probe points */
static __init int init_kprobe_trace(void)
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply related [flat|nested] 57+ messages in thread* Re: [PATCH -tip v5 7/7] tracing: add arguments support on kprobe-based event tracer
2009-05-09 0:49 ` [PATCH -tip v5 7/7] tracing: add arguments support on kprobe-based event tracer Masami Hiramatsu
@ 2009-05-11 14:35 ` Steven Rostedt
0 siblings, 0 replies; 57+ messages in thread
From: Steven Rostedt @ 2009-05-11 14:35 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Frederic Weisbecker
On Fri, 8 May 2009, Masami Hiramatsu wrote:
> Support following probe arguments and add fetch functions on kprobe-based
> event tracer.
>
> %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.
>
> Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
> Cc: Steven Rostedt <rostedt@goodmis.org>
> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
> Cc: Ingo Molnar <mingo@elte.hu>
> Cc: Frederic Weisbecker <fweisbec@gmail.com>
> ---
>
> Documentation/trace/ftrace.txt | 47 +++-
> kernel/trace/trace_kprobe.c | 431 ++++++++++++++++++++++++++++++++++++++--
> 2 files changed, 441 insertions(+), 37 deletions(-)
>
> diff --git a/Documentation/trace/ftrace.txt b/Documentation/trace/ftrace.txt
> index 2b8ead6..ce91398 100644
> --- a/Documentation/trace/ftrace.txt
The Documentation/trace/ftrace.txt file is getting too big. Could you make
a separate "Documentation/trace/kprobes.txt file, and split out the
kprobe bits.
Thanks,
-- Steve
> +++ b/Documentation/trace/ftrace.txt
> @@ -1329,17 +1329,34 @@ current_tracer, instead of that, just set probe points via
> /debug/tracing/kprobe_events.
>
> Synopsis of kprobe_events:
> - p SYMBOL[+offs|-offs]|MEMADDR : set a probe
> - r SYMBOL[+0] : set a return probe
> + p SYMBOL[+offs|-offs]|MEMADDR [FETCHARGS] : set a probe
> + r SYMBOL[+0] [FETCHARGS] : set a return probe
> +
> + FETCHARGS:
> + %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.
>
> E.g.
> - echo p sys_open > /debug/tracing/kprobe_events
> + echo p do_sys_open a0 a1 a2 a3 > /debug/tracing/kprobe_events
>
> - This sets a kprobe on the top of sys_open() function.
> + This sets a kprobe on the top of do_sys_open() function with recording
> +1st to 4th arguments.
>
> - echo r sys_open >> /debug/tracing/kprobe_events
> + echo r do_sys_open rv ra >> /debug/tracing/kprobe_events
>
> - This sets a kretprobe on the return point of sys_open() function.
> + This sets a kretprobe on the return point of do_sys_open() function with
> +recording return value and return address.
>
> echo > /debug/tracing/kprobe_events
>
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 7/7] tracing: add arguments support on kprobe-based event tracer
@ 2009-05-11 14:35 ` Steven Rostedt
0 siblings, 0 replies; 57+ messages in thread
From: Steven Rostedt @ 2009-05-11 14:35 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, lkml, systemtap, kvm, Ananth N Mavinakayanahalli,
Frederic Weisbecker
On Fri, 8 May 2009, Masami Hiramatsu wrote:
> Support following probe arguments and add fetch functions on kprobe-based
> event tracer.
>
> %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.
>
> Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
> Cc: Steven Rostedt <rostedt@goodmis.org>
> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
> Cc: Ingo Molnar <mingo@elte.hu>
> Cc: Frederic Weisbecker <fweisbec@gmail.com>
> ---
>
> Documentation/trace/ftrace.txt | 47 +++-
> kernel/trace/trace_kprobe.c | 431 ++++++++++++++++++++++++++++++++++++++--
> 2 files changed, 441 insertions(+), 37 deletions(-)
>
> diff --git a/Documentation/trace/ftrace.txt b/Documentation/trace/ftrace.txt
> index 2b8ead6..ce91398 100644
> --- a/Documentation/trace/ftrace.txt
The Documentation/trace/ftrace.txt file is getting too big. Could you make
a separate "Documentation/trace/kprobes.txt file, and split out the
kprobe bits.
Thanks,
-- Steve
> +++ b/Documentation/trace/ftrace.txt
> @@ -1329,17 +1329,34 @@ current_tracer, instead of that, just set probe points via
> /debug/tracing/kprobe_events.
>
> Synopsis of kprobe_events:
> - p SYMBOL[+offs|-offs]|MEMADDR : set a probe
> - r SYMBOL[+0] : set a return probe
> + p SYMBOL[+offs|-offs]|MEMADDR [FETCHARGS] : set a probe
> + r SYMBOL[+0] [FETCHARGS] : set a return probe
> +
> + FETCHARGS:
> + %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.
>
> E.g.
> - echo p sys_open > /debug/tracing/kprobe_events
> + echo p do_sys_open a0 a1 a2 a3 > /debug/tracing/kprobe_events
>
> - This sets a kprobe on the top of sys_open() function.
> + This sets a kprobe on the top of do_sys_open() function with recording
> +1st to 4th arguments.
>
> - echo r sys_open >> /debug/tracing/kprobe_events
> + echo r do_sys_open rv ra >> /debug/tracing/kprobe_events
>
> - This sets a kretprobe on the return point of sys_open() function.
> + This sets a kretprobe on the return point of do_sys_open() function with
> +recording return value and return address.
>
> echo > /debug/tracing/kprobe_events
>
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
2009-05-09 0:48 ` Masami Hiramatsu
@ 2009-05-09 4:43 ` Ingo Molnar
-1 siblings, 0 replies; 57+ messages in thread
From: Ingo Molnar @ 2009-05-09 4:43 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Steven Rostedt, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm
* Masami Hiramatsu <mhiramat@redhat.com> wrote:
> Hi,
>
> Here are the patches of kprobe-based event tracer for x86, version
> 5, which allows you to probe various kernel events through ftrace
> interface.
>
> This version supports only x86(-32/-64) (but porting it on other
> arch just needs kprobes/kretprobes and register and stack access
> APIs).
>
> This patchset also includes x86(-64) instruction decoder which
> supports non-SSE/FP opcodes and includes x86 opcode map. I think
> it will be possible to share this opcode map with KVM's decoder.
>
> This series can be applied on the latest linux-2.6-tip tree.
>
> This patchset includes following changes:
> - Add x86 instruction decoder [1/7]
> - Check insertion point safety in kprobe [2/7]
> - Cleanup fix_riprel() with insn decoder [3/7]
> - Add kprobe-tracer plugin [4/7]
> - Fix kernel_trap_sp() on x86 according to systemtap runtime. [5/7]
> - Add arch-dep register and stack fetching functions [6/7]
> - Support fetching various status (register/stack/memory/etc.) [7/7]
>
> Future items:
> - .init function tracing support.
> - Support primitive types(long, ulong, int, uint, etc) for args.
Ok, this looks pretty complete already.
Two high-level comments:
- There's no self-test - would it be possible to add one? See
trace_selftest* in kernel/trace/
- No generic integration.
It would be nice if these ops:
> E.g.
> echo p do_sys_open a0 a1 a2 a3 > /debug/tracing/kprobe_events
>
> This sets a kprobe on the top of do_sys_open() function with recording
> 1st to 4th arguments.
>
> echo r do_sys_open rv rp >> /debug/tracing/kprobe_events
were just generally available in just about any other tracer - a bit
like the event tracer.
It would also be nice to use the 'function attributes' facilities of
the function tracer, combined with a new special syntax of the
function-filter regex parser, to enable the recovery of return
values (or the call arguments), for selected set of functions.
For example, today we can already do things like:
echo 'sys_read:traceon:4' > /debug/tracing/set_ftrace_filter
for 'trace triggers': the above will trigger tracing to be enabled
on the entry of sys_read(), 4 times.
Likewise, something like:
echo 'sys_read:args' > /debug/tracing/set_ftrace_filter
echo 'sys_read:return' > /debug/tracing/set_ftrace_filter
Could activate kprobes based argument and return-value tracing.
Ingo
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
@ 2009-05-09 4:43 ` Ingo Molnar
0 siblings, 0 replies; 57+ messages in thread
From: Ingo Molnar @ 2009-05-09 4:43 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Steven Rostedt, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm
* Masami Hiramatsu <mhiramat@redhat.com> wrote:
> Hi,
>
> Here are the patches of kprobe-based event tracer for x86, version
> 5, which allows you to probe various kernel events through ftrace
> interface.
>
> This version supports only x86(-32/-64) (but porting it on other
> arch just needs kprobes/kretprobes and register and stack access
> APIs).
>
> This patchset also includes x86(-64) instruction decoder which
> supports non-SSE/FP opcodes and includes x86 opcode map. I think
> it will be possible to share this opcode map with KVM's decoder.
>
> This series can be applied on the latest linux-2.6-tip tree.
>
> This patchset includes following changes:
> - Add x86 instruction decoder [1/7]
> - Check insertion point safety in kprobe [2/7]
> - Cleanup fix_riprel() with insn decoder [3/7]
> - Add kprobe-tracer plugin [4/7]
> - Fix kernel_trap_sp() on x86 according to systemtap runtime. [5/7]
> - Add arch-dep register and stack fetching functions [6/7]
> - Support fetching various status (register/stack/memory/etc.) [7/7]
>
> Future items:
> - .init function tracing support.
> - Support primitive types(long, ulong, int, uint, etc) for args.
Ok, this looks pretty complete already.
Two high-level comments:
- There's no self-test - would it be possible to add one? See
trace_selftest* in kernel/trace/
- No generic integration.
It would be nice if these ops:
> E.g.
> echo p do_sys_open a0 a1 a2 a3 > /debug/tracing/kprobe_events
>
> This sets a kprobe on the top of do_sys_open() function with recording
> 1st to 4th arguments.
>
> echo r do_sys_open rv rp >> /debug/tracing/kprobe_events
were just generally available in just about any other tracer - a bit
like the event tracer.
It would also be nice to use the 'function attributes' facilities of
the function tracer, combined with a new special syntax of the
function-filter regex parser, to enable the recovery of return
values (or the call arguments), for selected set of functions.
For example, today we can already do things like:
echo 'sys_read:traceon:4' > /debug/tracing/set_ftrace_filter
for 'trace triggers': the above will trigger tracing to be enabled
on the entry of sys_read(), 4 times.
Likewise, something like:
echo 'sys_read:args' > /debug/tracing/set_ftrace_filter
echo 'sys_read:return' > /debug/tracing/set_ftrace_filter
Could activate kprobes based argument and return-value tracing.
Ingo
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
2009-05-09 4:43 ` Ingo Molnar
(?)
@ 2009-05-11 14:40 ` Masami Hiramatsu
2009-05-11 14:56 ` Steven Rostedt
-1 siblings, 1 reply; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 14:40 UTC (permalink / raw)
To: Ingo Molnar
Cc: Steven Rostedt, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm, Tom Zanussi
Ingo Molnar wrote:
> * Masami Hiramatsu <mhiramat@redhat.com> wrote:
>
>> Hi,
>>
>> Here are the patches of kprobe-based event tracer for x86, version
>> 5, which allows you to probe various kernel events through ftrace
>> interface.
>>
>> This version supports only x86(-32/-64) (but porting it on other
>> arch just needs kprobes/kretprobes and register and stack access
>> APIs).
>>
>> This patchset also includes x86(-64) instruction decoder which
>> supports non-SSE/FP opcodes and includes x86 opcode map. I think
>> it will be possible to share this opcode map with KVM's decoder.
>>
>> This series can be applied on the latest linux-2.6-tip tree.
>>
>> This patchset includes following changes:
>> - Add x86 instruction decoder [1/7]
>> - Check insertion point safety in kprobe [2/7]
>> - Cleanup fix_riprel() with insn decoder [3/7]
>> - Add kprobe-tracer plugin [4/7]
>> - Fix kernel_trap_sp() on x86 according to systemtap runtime. [5/7]
>> - Add arch-dep register and stack fetching functions [6/7]
>> - Support fetching various status (register/stack/memory/etc.) [7/7]
>>
>> Future items:
>> - .init function tracing support.
>> - Support primitive types(long, ulong, int, uint, etc) for args.
>
> Ok, this looks pretty complete already.
>
> Two high-level comments:
>
> - There's no self-test - would it be possible to add one? See
> trace_selftest* in kernel/trace/
I'm not so sure. Currently, it seems that those self-tests are
only for tracers which define new event-entry on ring-buffer.
Since this tracer just use ftrace_bprintk, it might need
another kind of selftest. e.g. comparing outputs with
expected patterns.
In that case, would it be better to make a user-space self test
including filters and tracepoints?
> - No generic integration.
Right, this just rides on the ftrace's printk :-)
>
> It would be nice if these ops:
>
>> E.g.
>> echo p do_sys_open a0 a1 a2 a3 > /debug/tracing/kprobe_events
>>
>> This sets a kprobe on the top of do_sys_open() function with recording
>> 1st to 4th arguments.
>>
>> echo r do_sys_open rv rp >> /debug/tracing/kprobe_events
>
> were just generally available in just about any other tracer - a bit
> like the event tracer.
>
> It would also be nice to use the 'function attributes' facilities of
> the function tracer, combined with a new special syntax of the
> function-filter regex parser, to enable the recovery of return
> values (or the call arguments), for selected set of functions.
>
> For example, today we can already do things like:
>
> echo 'sys_read:traceon:4' > /debug/tracing/set_ftrace_filter
>
> for 'trace triggers': the above will trigger tracing to be enabled
> on the entry of sys_read(), 4 times.
>
> Likewise, something like:
>
> echo 'sys_read:args' > /debug/tracing/set_ftrace_filter
> echo 'sys_read:return' > /debug/tracing/set_ftrace_filter
>
> Could activate kprobes based argument and return-value tracing.
Ah, that's a good idea. And also, I have another idea for using
filters with kprobes, which is adding new kprobe-entry as a new
trace-event when user defines a new probe via kprobe_events.
Thus, user can use it as a dynamic-defined tracepoint.
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
2009-05-11 14:40 ` Masami Hiramatsu
@ 2009-05-11 14:56 ` Steven Rostedt
2009-05-11 20:05 ` Masami Hiramatsu
0 siblings, 1 reply; 57+ messages in thread
From: Steven Rostedt @ 2009-05-11 14:56 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Ingo Molnar, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm, Tom Zanussi
On Mon, 11 May 2009, Masami Hiramatsu wrote:
> >
> > Two high-level comments:
> >
> > - There's no self-test - would it be possible to add one? See
> > trace_selftest* in kernel/trace/
>
> I'm not so sure. Currently, it seems that those self-tests are
> only for tracers which define new event-entry on ring-buffer.
> Since this tracer just use ftrace_bprintk, it might need
> another kind of selftest. e.g. comparing outputs with
> expected patterns.
> In that case, would it be better to make a user-space self test
> including filters and tracepoints?
Or have the workings in the selftest in kernel. As if a user started it.
It does not need to write to the ring buffer, that is just what I did. The
event selftests don't check if anything was written to the ring buffer,
they just make sure that the tests don't crash the system.
-- Steve
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
2009-05-11 14:56 ` Steven Rostedt
@ 2009-05-11 20:05 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 20:05 UTC (permalink / raw)
To: Steven Rostedt
Cc: Ingo Molnar, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm, Tom Zanussi
Steven Rostedt wrote:
> On Mon, 11 May 2009, Masami Hiramatsu wrote:
>>> Two high-level comments:
>>>
>>> - There's no self-test - would it be possible to add one? See
>>> trace_selftest* in kernel/trace/
>> I'm not so sure. Currently, it seems that those self-tests are
>> only for tracers which define new event-entry on ring-buffer.
>> Since this tracer just use ftrace_bprintk, it might need
>> another kind of selftest. e.g. comparing outputs with
>> expected patterns.
>> In that case, would it be better to make a user-space self test
>> including filters and tracepoints?
>
> Or have the workings in the selftest in kernel. As if a user started it.
> It does not need to write to the ring buffer, that is just what I did. The
> event selftests don't check if anything was written to the ring buffer,
> they just make sure that the tests don't crash the system.
Would you mean that it is enough to enable some probes and just
see what happened at boot time?
That's so easy to add.
Thank you :-),
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
@ 2009-05-11 20:05 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-11 20:05 UTC (permalink / raw)
To: Steven Rostedt
Cc: Ingo Molnar, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm, Tom Zanussi
Steven Rostedt wrote:
> On Mon, 11 May 2009, Masami Hiramatsu wrote:
>>> Two high-level comments:
>>>
>>> - There's no self-test - would it be possible to add one? See
>>> trace_selftest* in kernel/trace/
>> I'm not so sure. Currently, it seems that those self-tests are
>> only for tracers which define new event-entry on ring-buffer.
>> Since this tracer just use ftrace_bprintk, it might need
>> another kind of selftest. e.g. comparing outputs with
>> expected patterns.
>> In that case, would it be better to make a user-space self test
>> including filters and tracepoints?
>
> Or have the workings in the selftest in kernel. As if a user started it.
> It does not need to write to the ring buffer, that is just what I did. The
> event selftests don't check if anything was written to the ring buffer,
> they just make sure that the tests don't crash the system.
Would you mean that it is enough to enable some probes and just
see what happened at boot time?
That's so easy to add.
Thank you :-),
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
2009-05-11 20:05 ` Masami Hiramatsu
@ 2009-05-11 21:47 ` Ingo Molnar
-1 siblings, 0 replies; 57+ messages in thread
From: Ingo Molnar @ 2009-05-11 21:47 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Steven Rostedt, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm, Tom Zanussi
* Masami Hiramatsu <mhiramat@redhat.com> wrote:
> Steven Rostedt wrote:
> > On Mon, 11 May 2009, Masami Hiramatsu wrote:
> >>> Two high-level comments:
> >>>
> >>> - There's no self-test - would it be possible to add one? See
> >>> trace_selftest* in kernel/trace/
> >> I'm not so sure. Currently, it seems that those self-tests are
> >> only for tracers which define new event-entry on ring-buffer.
> >> Since this tracer just use ftrace_bprintk, it might need
> >> another kind of selftest. e.g. comparing outputs with
> >> expected patterns.
> >> In that case, would it be better to make a user-space self test
> >> including filters and tracepoints?
> >
> > Or have the workings in the selftest in kernel. As if a user started it.
> > It does not need to write to the ring buffer, that is just what I did. The
> > event selftests don't check if anything was written to the ring buffer,
> > they just make sure that the tests don't crash the system.
>
> Would you mean that it is enough to enable some probes and just
> see what happened at boot time?
> That's so easy to add.
Yes, that's the idea!
Try to think of regressions/crashes/misbehavior you generally
trigger while you developed kprobes, and try to add a reasonable set
of probes that test the code from those angles.
It doesnt have to be a full, complex test-suite, but even just 80%
of coverage of functionality keeps 4/5th of all regressions out of
the kernel at a very early stage ...
Ingo
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
@ 2009-05-11 21:47 ` Ingo Molnar
0 siblings, 0 replies; 57+ messages in thread
From: Ingo Molnar @ 2009-05-11 21:47 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Steven Rostedt, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm, Tom Zanussi
* Masami Hiramatsu <mhiramat@redhat.com> wrote:
> Steven Rostedt wrote:
> > On Mon, 11 May 2009, Masami Hiramatsu wrote:
> >>> Two high-level comments:
> >>>
> >>> - There's no self-test - would it be possible to add one? See
> >>> trace_selftest* in kernel/trace/
> >> I'm not so sure. Currently, it seems that those self-tests are
> >> only for tracers which define new event-entry on ring-buffer.
> >> Since this tracer just use ftrace_bprintk, it might need
> >> another kind of selftest. e.g. comparing outputs with
> >> expected patterns.
> >> In that case, would it be better to make a user-space self test
> >> including filters and tracepoints?
> >
> > Or have the workings in the selftest in kernel. As if a user started it.
> > It does not need to write to the ring buffer, that is just what I did. The
> > event selftests don't check if anything was written to the ring buffer,
> > they just make sure that the tests don't crash the system.
>
> Would you mean that it is enough to enable some probes and just
> see what happened at boot time?
> That's so easy to add.
Yes, that's the idea!
Try to think of regressions/crashes/misbehavior you generally
trigger while you developed kprobes, and try to add a reasonable set
of probes that test the code from those angles.
It doesnt have to be a full, complex test-suite, but even just 80%
of coverage of functionality keeps 4/5th of all regressions out of
the kernel at a very early stage ...
Ingo
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
2009-05-09 4:43 ` Ingo Molnar
@ 2009-05-12 22:03 ` Masami Hiramatsu
-1 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-12 22:03 UTC (permalink / raw)
To: Ingo Molnar
Cc: Steven Rostedt, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm
Ingo Molnar wrote:
> * Masami Hiramatsu <mhiramat@redhat.com> wrote:
>
>> Hi,
>>
>> Here are the patches of kprobe-based event tracer for x86, version
>> 5, which allows you to probe various kernel events through ftrace
>> interface.
>>
>> This version supports only x86(-32/-64) (but porting it on other
>> arch just needs kprobes/kretprobes and register and stack access
>> APIs).
>>
>> This patchset also includes x86(-64) instruction decoder which
>> supports non-SSE/FP opcodes and includes x86 opcode map. I think
>> it will be possible to share this opcode map with KVM's decoder.
>>
>> This series can be applied on the latest linux-2.6-tip tree.
>>
>> This patchset includes following changes:
>> - Add x86 instruction decoder [1/7]
>> - Check insertion point safety in kprobe [2/7]
>> - Cleanup fix_riprel() with insn decoder [3/7]
>> - Add kprobe-tracer plugin [4/7]
>> - Fix kernel_trap_sp() on x86 according to systemtap runtime. [5/7]
>> - Add arch-dep register and stack fetching functions [6/7]
>> - Support fetching various status (register/stack/memory/etc.) [7/7]
>>
>> Future items:
>> - .init function tracing support.
>> - Support primitive types(long, ulong, int, uint, etc) for args.
>
> Ok, this looks pretty complete already.
>
> Two high-level comments:
>
> - There's no self-test - would it be possible to add one? See
> trace_selftest* in kernel/trace/
>
> - No generic integration.
Hmm, Ingo, could you tell me what I can do for the integration?
Would you means that I should use filters?
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
@ 2009-05-12 22:03 ` Masami Hiramatsu
0 siblings, 0 replies; 57+ messages in thread
From: Masami Hiramatsu @ 2009-05-12 22:03 UTC (permalink / raw)
To: Ingo Molnar
Cc: Steven Rostedt, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm
Ingo Molnar wrote:
> * Masami Hiramatsu <mhiramat@redhat.com> wrote:
>
>> Hi,
>>
>> Here are the patches of kprobe-based event tracer for x86, version
>> 5, which allows you to probe various kernel events through ftrace
>> interface.
>>
>> This version supports only x86(-32/-64) (but porting it on other
>> arch just needs kprobes/kretprobes and register and stack access
>> APIs).
>>
>> This patchset also includes x86(-64) instruction decoder which
>> supports non-SSE/FP opcodes and includes x86 opcode map. I think
>> it will be possible to share this opcode map with KVM's decoder.
>>
>> This series can be applied on the latest linux-2.6-tip tree.
>>
>> This patchset includes following changes:
>> - Add x86 instruction decoder [1/7]
>> - Check insertion point safety in kprobe [2/7]
>> - Cleanup fix_riprel() with insn decoder [3/7]
>> - Add kprobe-tracer plugin [4/7]
>> - Fix kernel_trap_sp() on x86 according to systemtap runtime. [5/7]
>> - Add arch-dep register and stack fetching functions [6/7]
>> - Support fetching various status (register/stack/memory/etc.) [7/7]
>>
>> Future items:
>> - .init function tracing support.
>> - Support primitive types(long, ulong, int, uint, etc) for args.
>
> Ok, this looks pretty complete already.
>
> Two high-level comments:
>
> - There's no self-test - would it be possible to add one? See
> trace_selftest* in kernel/trace/
>
> - No generic integration.
Hmm, Ingo, could you tell me what I can do for the integration?
Would you means that I should use filters?
Thank you,
--
Masami Hiramatsu
Software Engineer
Hitachi Computer Products (America) Inc.
Software Solutions Division
e-mail: mhiramat@redhat.com
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
2009-05-12 22:03 ` Masami Hiramatsu
@ 2009-05-13 13:21 ` Ingo Molnar
-1 siblings, 0 replies; 57+ messages in thread
From: Ingo Molnar @ 2009-05-13 13:21 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Steven Rostedt, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm
* Masami Hiramatsu <mhiramat@redhat.com> wrote:
> Ingo Molnar wrote:
> > * Masami Hiramatsu <mhiramat@redhat.com> wrote:
> >
> >> Hi,
> >>
> >> Here are the patches of kprobe-based event tracer for x86, version
> >> 5, which allows you to probe various kernel events through ftrace
> >> interface.
> >>
> >> This version supports only x86(-32/-64) (but porting it on other
> >> arch just needs kprobes/kretprobes and register and stack access
> >> APIs).
> >>
> >> This patchset also includes x86(-64) instruction decoder which
> >> supports non-SSE/FP opcodes and includes x86 opcode map. I think
> >> it will be possible to share this opcode map with KVM's decoder.
> >>
> >> This series can be applied on the latest linux-2.6-tip tree.
> >>
> >> This patchset includes following changes:
> >> - Add x86 instruction decoder [1/7]
> >> - Check insertion point safety in kprobe [2/7]
> >> - Cleanup fix_riprel() with insn decoder [3/7]
> >> - Add kprobe-tracer plugin [4/7]
> >> - Fix kernel_trap_sp() on x86 according to systemtap runtime. [5/7]
> >> - Add arch-dep register and stack fetching functions [6/7]
> >> - Support fetching various status (register/stack/memory/etc.) [7/7]
> >>
> >> Future items:
> >> - .init function tracing support.
> >> - Support primitive types(long, ulong, int, uint, etc) for args.
> >
> > Ok, this looks pretty complete already.
> >
> > Two high-level comments:
> >
> > - There's no self-test - would it be possible to add one? See
> > trace_selftest* in kernel/trace/
> >
> > - No generic integration.
>
> Hmm, Ingo, could you tell me what I can do for the integration?
> Would you means that I should use filters?
yeah, that - and for the tracepoints to show up under
/debug/tracing/events/. They'd in essence be 'flexible', dynamic
event tracepoints that extend upon existing, built-in tracepoints.
To user-space tools the two would show up in a very similar way and
with a similar usage (once they are injected).
Ingo
^ permalink raw reply [flat|nested] 57+ messages in thread
* Re: [PATCH -tip v5 0/7] tracing: kprobe-based event tracer and x86 instruction decoder
@ 2009-05-13 13:21 ` Ingo Molnar
0 siblings, 0 replies; 57+ messages in thread
From: Ingo Molnar @ 2009-05-13 13:21 UTC (permalink / raw)
To: Masami Hiramatsu
Cc: Steven Rostedt, lkml, Avi Kivity, H. Peter Anvin,
Frederic Weisbecker, Ananth N Mavinakayanahalli, Andrew Morton,
Andi Kleen, Jim Keniston, K.Prasad, KOSAKI Motohiro, systemtap,
kvm
* Masami Hiramatsu <mhiramat@redhat.com> wrote:
> Ingo Molnar wrote:
> > * Masami Hiramatsu <mhiramat@redhat.com> wrote:
> >
> >> Hi,
> >>
> >> Here are the patches of kprobe-based event tracer for x86, version
> >> 5, which allows you to probe various kernel events through ftrace
> >> interface.
> >>
> >> This version supports only x86(-32/-64) (but porting it on other
> >> arch just needs kprobes/kretprobes and register and stack access
> >> APIs).
> >>
> >> This patchset also includes x86(-64) instruction decoder which
> >> supports non-SSE/FP opcodes and includes x86 opcode map. I think
> >> it will be possible to share this opcode map with KVM's decoder.
> >>
> >> This series can be applied on the latest linux-2.6-tip tree.
> >>
> >> This patchset includes following changes:
> >> - Add x86 instruction decoder [1/7]
> >> - Check insertion point safety in kprobe [2/7]
> >> - Cleanup fix_riprel() with insn decoder [3/7]
> >> - Add kprobe-tracer plugin [4/7]
> >> - Fix kernel_trap_sp() on x86 according to systemtap runtime. [5/7]
> >> - Add arch-dep register and stack fetching functions [6/7]
> >> - Support fetching various status (register/stack/memory/etc.) [7/7]
> >>
> >> Future items:
> >> - .init function tracing support.
> >> - Support primitive types(long, ulong, int, uint, etc) for args.
> >
> > Ok, this looks pretty complete already.
> >
> > Two high-level comments:
> >
> > - There's no self-test - would it be possible to add one? See
> > trace_selftest* in kernel/trace/
> >
> > - No generic integration.
>
> Hmm, Ingo, could you tell me what I can do for the integration?
> Would you means that I should use filters?
yeah, that - and for the tracepoints to show up under
/debug/tracing/events/. They'd in essence be 'flexible', dynamic
event tracepoints that extend upon existing, built-in tracepoints.
To user-space tools the two would show up in a very similar way and
with a similar usage (once they are injected).
Ingo
^ permalink raw reply [flat|nested] 57+ messages in thread