openkylin
/
linux
mirror of https://gitee.com/openkylin/linux.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity

x86: prepare kprobes code for x86 unification 

This patch cleanup kprobes code on x86 for unification.
This patch is based on Arjan's previous work.

- Remove spurious whitespace changes
- Add harmless includes
- Make the 32/64 files more identical
 - Generalize structure fields' and local variable name.
 - Wrap accessing to stack address by macros.
 - Modify bitmap making macro.
 - Merge fixup code into is_riprel() and change its name to fix_riprel().
 - Set MAX_INSN_SIZE to 16 on both arch.
 - Use u32 for bitmaps on both architectures.
 - Clarify some comments.

Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Jim Keniston <jkenisto@us.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
This commit is contained in:
Masami Hiramatsu 2008-01-30 13:31:21 +01:00 committed by Ingo Molnar
parent da07ab0375
commit 8533bbe9f8
6 changed files with 457 additions and 357 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

341
arch/x86/kernel/kprobes_32.c
View File

@ -29,10 +29,15 @@
#include <linux/kprobes.h> #include <linux/kprobes.h>
#include <linux/ptrace.h> #include <linux/ptrace.h>
#include <linux/string.h>
#include <linux/slab.h>
#include <linux/preempt.h> #include <linux/preempt.h>
#include <linux/module.h>
#include <linux/kdebug.h> #include <linux/kdebug.h>
#include <asm/cacheflush.h> #include <asm/cacheflush.h>
#include <asm/desc.h> #include <asm/desc.h>
#include <asm/pgtable.h>
#include <asm/uaccess.h> #include <asm/uaccess.h>
#include <asm/alternative.h> #include <asm/alternative.h>
@ -41,31 +46,17 @@ void jprobe_return_end(void);
DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL; DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk); DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
struct kretprobe_blackpoint kretprobe_blacklist[] = {
{"__switch_to", }, /* This function switches only current task, but
doesn't switch kernel stack.*/
{NULL, NULL} /* Terminator */
};
const int kretprobe_blacklist_size = ARRAY_SIZE(kretprobe_blacklist);
/* insert a jmp code */
static __always_inline void set_jmp_op(void *from, void *to)
{
struct __arch_jmp_op {
char op;
long raddr;
} __attribute__((packed)) *jop;
jop = (struct __arch_jmp_op *)from;
jop->raddr = (long)(to) - ((long)(from) + 5);
jop->op = RELATIVEJUMP_INSTRUCTION;
}
/* /*
* returns non-zero if opcodes can be boosted. * "&regs->sp" looks wrong, but it's correct for x86_32. x86_32 CPUs
* don't save the ss and esp registers if the CPU is already in kernel
* mode when it traps. So for kprobes, regs->sp and regs->ss are not
* the [nonexistent] saved stack pointer and ss register, but rather
* the top 8 bytes of the pre-int3 stack. So &regs->sp happens to
* point to the top of the pre-int3 stack.
*/ */
static __always_inline int can_boost(kprobe_opcode_t *opcodes) #define stack_addr(regs) ((unsigned long *)&regs->sp)
{
#define W(row,b0,b1,b2,b3,b4,b5,b6,b7,b8,b9,ba,bb,bc,bd,be,bf) \ #define W(row, b0, b1, b2, b3, b4, b5, b6, b7, b8, b9, ba, bb, bc, bd, be, bf)\
(((b0##UL << 0x0)|(b1##UL << 0x1)|(b2##UL << 0x2)|(b3##UL << 0x3) | \ (((b0##UL << 0x0)|(b1##UL << 0x1)|(b2##UL << 0x2)|(b3##UL << 0x3) | \
(b4##UL << 0x4)|(b5##UL << 0x5)|(b6##UL << 0x6)|(b7##UL << 0x7) | \ (b4##UL << 0x4)|(b5##UL << 0x5)|(b6##UL << 0x6)|(b7##UL << 0x7) | \
(b8##UL << 0x8)|(b9##UL << 0x9)|(ba##UL << 0xa)|(bb##UL << 0xb) | \ (b8##UL << 0x8)|(b9##UL << 0x9)|(ba##UL << 0xa)|(bb##UL << 0xb) | \
@ -73,33 +64,103 @@ static __always_inline int can_boost(kprobe_opcode_t *opcodes)
<< (row % 32)) << (row % 32))
/* /*
* Undefined/reserved opcodes, conditional jump, Opcode Extension * Undefined/reserved opcodes, conditional jump, Opcode Extension
* Groups, and some special opcodes can not be boost. * Groups, and some special opcodes can not boost.
*/ */
static const unsigned long twobyte_is_boostable[256 / 32] = { static const u32 twobyte_is_boostable[256 / 32] = {
/* 0 1 2 3 4 5 6 7 8 9 a b c d e f */ /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
/* ------------------------------- */ /* ---------------------------------------------- */
W(0x00, 0,0,1,1,0,0,1,0,1,1,0,0,0,0,0,0)| /* 00 */ W(0x00, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0) | /* 00 */
W(0x10, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), /* 10 */ W(0x10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 10 */
W(0x20, 1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0)| /* 20 */ W(0x20, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* 20 */
W(0x30, 0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0), /* 30 */ W(0x30, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 30 */
W(0x40, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)| /* 40 */ W(0x40, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* 40 */
W(0x50, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), /* 50 */ W(0x50, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 50 */
W(0x60, 1,1,1,1,1,1,1,1,1,1,1,1,0,0,1,1)| /* 60 */ W(0x60, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1) | /* 60 */
W(0x70, 0,0,0,0,1,1,1,1,0,0,0,0,0,0,1,1), /* 70 */ W(0x70, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1) , /* 70 */
W(0x80, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)| /* 80 */ W(0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* 80 */
W(0x90, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1), /* 90 */ W(0x90, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 90 */
W(0xa0, 1,1,0,1,1,1,0,0,1,1,0,1,1,1,0,1)| /* a0 */ W(0xa0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1) | /* a0 */
W(0xb0, 1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,1), /* b0 */ W(0xb0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1) , /* b0 */
W(0xc0, 1,1,0,0,0,0,0,0,1,1,1,1,1,1,1,1)| /* c0 */ W(0xc0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1) | /* c0 */
W(0xd0, 0,1,1,1,0,1,0,0,1,1,0,1,1,1,0,1), /* d0 */ W(0xd0, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1) , /* d0 */
W(0xe0, 0,1,1,0,0,1,0,0,1,1,0,1,1,1,0,1)| /* e0 */ W(0xe0, 0, 1, 1, 0, 0, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1) | /* e0 */
W(0xf0, 0,1,1,1,0,1,0,0,1,1,1,0,1,1,1,0) /* f0 */ W(0xf0, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 0) /* f0 */
/* ------------------------------- */ /* ----------------------------------------------- */
/* 0 1 2 3 4 5 6 7 8 9 a b c d e f */ /* 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 #undef W
struct kretprobe_blackpoint kretprobe_blacklist[] = {
{"__switch_to", }, /* This function switches only current task, but
doesn't switch kernel stack.*/
{NULL, NULL} /* Terminator */
};
const int kretprobe_blacklist_size = ARRAY_SIZE(kretprobe_blacklist);
/* Insert a jump instruction at address 'from', which jumps to address 'to'.*/
static __always_inline void set_jmp_op(void *from, void *to)
{
struct __arch_jmp_op {
char op;
s32 raddr;
} __attribute__((packed)) * jop;
jop = (struct __arch_jmp_op *)from;
jop->raddr = (s32)((long)(to) - ((long)(from) + 5));
jop->op = RELATIVEJUMP_INSTRUCTION;
}
/*
* returns non-zero if opcode is boostable.
*/
static __always_inline int can_boost(kprobe_opcode_t *opcodes)
{
kprobe_opcode_t opcode; kprobe_opcode_t opcode;
kprobe_opcode_t *orig_opcodes = opcodes; kprobe_opcode_t *orig_opcodes = opcodes;
retry: retry:
if (opcodes - orig_opcodes > MAX_INSN_SIZE - 1) if (opcodes - orig_opcodes > MAX_INSN_SIZE - 1)
return 0; return 0;
@ -109,7 +170,8 @@ static __always_inline int can_boost(kprobe_opcode_t *opcodes)
if (opcode == 0x0f) { if (opcode == 0x0f) {
if (opcodes - orig_opcodes > MAX_INSN_SIZE - 1) if (opcodes - orig_opcodes > MAX_INSN_SIZE - 1)
return 0; return 0;
return test_bit(*opcodes, twobyte_is_boostable); return test_bit(*opcodes,
(unsigned long *)twobyte_is_boostable);
} }
switch (opcode & 0xf0) { switch (opcode & 0xf0) {
@ -132,12 +194,13 @@ static __always_inline int can_boost(kprobe_opcode_t *opcodes)
case 0xf0: case 0xf0:
if ((opcode & 0x0c) == 0 && opcode != 0xf1) if ((opcode & 0x0c) == 0 && opcode != 0xf1)
goto retry; /* lock/rep(ne) prefix */ goto retry; /* lock/rep(ne) prefix */
/* clear and set flags can be boost */ /* clear and set flags are boostable */
return (opcode == 0xf5 || (0xf7 < opcode && opcode < 0xfe)); return (opcode == 0xf5 || (0xf7 < opcode && opcode < 0xfe));
default: default:
/* segment override prefixes are boostable */
if (opcode == 0x26 || opcode == 0x36 || opcode == 0x3e) if (opcode == 0x26 || opcode == 0x36 || opcode == 0x3e)
goto retry; /* prefixes */ goto retry; /* prefixes */
/* can't boost CS override and call */ /* CS override prefix and call are not boostable */
return (opcode != 0x2e && opcode != 0x9a); return (opcode != 0x2e && opcode != 0x9a);
} }
} }
@ -145,9 +208,9 @@ static __always_inline int can_boost(kprobe_opcode_t *opcodes)
/* /*
* returns non-zero if opcode modifies the interrupt flag. * returns non-zero if opcode modifies the interrupt flag.
*/ */
static int __kprobes is_IF_modifier(kprobe_opcode_t opcode) static int __kprobes is_IF_modifier(kprobe_opcode_t *insn)
{ {
switch (opcode) { switch (*insn) {
case 0xfa: /* cli */ case 0xfa: /* cli */
case 0xfb: /* sti */ case 0xfb: /* sti */
case 0xcf: /* iret/iretd */ case 0xcf: /* iret/iretd */
@ -157,20 +220,24 @@ static int __kprobes is_IF_modifier(kprobe_opcode_t opcode)
return 0; return 0;
} }
static void __kprobes arch_copy_kprobe(struct kprobe *p)
{
memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
if (can_boost(p->addr))
p->ainsn.boostable = 0;
else
p->ainsn.boostable = -1;
p->opcode = *p->addr;
}
int __kprobes arch_prepare_kprobe(struct kprobe *p) int __kprobes arch_prepare_kprobe(struct kprobe *p)
{ {
/* insn: must be on special executable page on i386. */ /* insn: must be on special executable page on x86. */
p->ainsn.insn = get_insn_slot(); p->ainsn.insn = get_insn_slot();
if (!p->ainsn.insn) if (!p->ainsn.insn)
return -ENOMEM; return -ENOMEM;
arch_copy_kprobe(p);
memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
p->opcode = *p->addr;
if (can_boost(p->addr)) {
p->ainsn.boostable = 0;
} else {
p->ainsn.boostable = -1;
}
return 0; return 0;
} }
@ -195,26 +262,26 @@ static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb)
{ {
kcb->prev_kprobe.kp = kprobe_running(); kcb->prev_kprobe.kp = kprobe_running();
kcb->prev_kprobe.status = kcb->kprobe_status; kcb->prev_kprobe.status = kcb->kprobe_status;
kcb->prev_kprobe.old_eflags = kcb->kprobe_old_eflags; kcb->prev_kprobe.old_flags = kcb->kprobe_old_flags;
kcb->prev_kprobe.saved_eflags = kcb->kprobe_saved_eflags; kcb->prev_kprobe.saved_flags = kcb->kprobe_saved_flags;
} }
static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb) static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb)
{ {
__get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp; __get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
kcb->kprobe_status = kcb->prev_kprobe.status; kcb->kprobe_status = kcb->prev_kprobe.status;
kcb->kprobe_old_eflags = kcb->prev_kprobe.old_eflags; kcb->kprobe_old_flags = kcb->prev_kprobe.old_flags;
kcb->kprobe_saved_eflags = kcb->prev_kprobe.saved_eflags; kcb->kprobe_saved_flags = kcb->prev_kprobe.saved_flags;
} }
static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs, static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs,
struct kprobe_ctlblk *kcb) struct kprobe_ctlblk *kcb)
{ {
__get_cpu_var(current_kprobe) = p; __get_cpu_var(current_kprobe) = p;
kcb->kprobe_saved_eflags = kcb->kprobe_old_eflags kcb->kprobe_saved_flags = kcb->kprobe_old_flags
= (regs->flags & (TF_MASK | IF_MASK)); = (regs->flags & (TF_MASK | IF_MASK));
if (is_IF_modifier(p->opcode)) if (is_IF_modifier(p->ainsn.insn))
kcb->kprobe_saved_eflags &= ~IF_MASK; kcb->kprobe_saved_flags &= ~IF_MASK;
} }
static __always_inline void clear_btf(void) static __always_inline void clear_btf(void)
@ -245,7 +312,7 @@ static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs)
void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri, void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
struct pt_regs *regs) struct pt_regs *regs)
{ {
unsigned long *sara = (unsigned long *)&regs->sp; unsigned long *sara = stack_addr(regs);
ri->ret_addr = (kprobe_opcode_t *) *sara; ri->ret_addr = (kprobe_opcode_t *) *sara;
@ -280,7 +347,7 @@ static int __kprobes kprobe_handler(struct pt_regs *regs)
if (kcb->kprobe_status == KPROBE_HIT_SS && if (kcb->kprobe_status == KPROBE_HIT_SS &&
*p->ainsn.insn == BREAKPOINT_INSTRUCTION) { *p->ainsn.insn == BREAKPOINT_INSTRUCTION) {
regs->flags &= ~TF_MASK; regs->flags &= ~TF_MASK;
regs->flags |= kcb->kprobe_saved_eflags; regs->flags |= kcb->kprobe_saved_flags;
goto no_kprobe; goto no_kprobe;
} }
/* We have reentered the kprobe_handler(), since /* We have reentered the kprobe_handler(), since
@ -301,7 +368,7 @@ static int __kprobes kprobe_handler(struct pt_regs *regs)
* another cpu right after we hit, no further * another cpu right after we hit, no further
* handling of this interrupt is appropriate * handling of this interrupt is appropriate
*/ */
regs->ip -= sizeof(kprobe_opcode_t); regs->ip = (unsigned long)addr;
ret = 1; ret = 1;
goto no_kprobe; goto no_kprobe;
} }
@ -325,7 +392,7 @@ static int __kprobes kprobe_handler(struct pt_regs *regs)
* Back up over the (now missing) int3 and run * Back up over the (now missing) int3 and run
* the original instruction. * the original instruction.
*/ */
regs->ip -= sizeof(kprobe_opcode_t); regs->ip = (unsigned long)addr;
ret = 1; ret = 1;
} }
/* Not one of ours: let kernel handle it */ /* Not one of ours: let kernel handle it */
@ -341,7 +408,7 @@ static int __kprobes kprobe_handler(struct pt_regs *regs)
ss_probe: ss_probe:
#if !defined(CONFIG_PREEMPT) || defined(CONFIG_PM) #if !defined(CONFIG_PREEMPT) || defined(CONFIG_PM)
if (p->ainsn.boostable == 1 && !p->post_handler){ if (p->ainsn.boostable == 1 && !p->post_handler) {
/* Boost up -- we can execute copied instructions directly */ /* Boost up -- we can execute copied instructions directly */
reset_current_kprobe(); reset_current_kprobe();
regs->ip = (unsigned long)p->ainsn.insn; regs->ip = (unsigned long)p->ainsn.insn;
@ -359,16 +426,18 @@ static int __kprobes kprobe_handler(struct pt_regs *regs)
} }
/* /*
* For function-return probes, init_kprobes() establishes a probepoint * When a retprobed function returns, this code saves registers and
* here. When a retprobed function returns, this probe is hit and * calls trampoline_handler() runs, which calls the kretprobe's handler.
* trampoline_probe_handler() runs, calling the kretprobe's handler.
*/ */
void __kprobes kretprobe_trampoline_holder(void) void __kprobes kretprobe_trampoline_holder(void)
{ {
asm volatile ( ".global kretprobe_trampoline\n" asm volatile ( ".global kretprobe_trampoline\n"
"kretprobe_trampoline: \n" "kretprobe_trampoline: \n"
" pushf\n" " pushf\n"
/* skip cs, ip, orig_ax */ /*
* Skip cs, ip, orig_ax.
* trampoline_handler() will plug in these values
*/
" subl $12, %esp\n" " subl $12, %esp\n"
" pushl %fs\n" " pushl %fs\n"
" pushl %ds\n" " pushl %ds\n"
@ -382,10 +451,10 @@ static int __kprobes kprobe_handler(struct pt_regs *regs)
" pushl %ebx\n" " pushl %ebx\n"
" movl %esp, %eax\n" " movl %esp, %eax\n"
" call trampoline_handler\n" " call trampoline_handler\n"
/* move flags to cs */ /* Move flags to cs */
" movl 52(%esp), %edx\n" " movl 52(%esp), %edx\n"
" movl %edx, 48(%esp)\n" " movl %edx, 48(%esp)\n"
/* save true return address on flags */ /* Replace saved flags with true return address. */
" movl %eax, 52(%esp)\n" " movl %eax, 52(%esp)\n"
" popl %ebx\n" " popl %ebx\n"
" popl %ecx\n" " popl %ecx\n"
@ -394,16 +463,16 @@ static int __kprobes kprobe_handler(struct pt_regs *regs)
" popl %edi\n" " popl %edi\n"
" popl %ebp\n" " popl %ebp\n"
" popl %eax\n" " popl %eax\n"
/* skip ip, orig_ax, es, ds, fs */ /* Skip ip, orig_ax, es, ds, fs */
" addl $20, %esp\n" " addl $20, %esp\n"
" popf\n" " popf\n"
" ret\n"); " ret\n");
} }
/* /*
* Called from kretprobe_trampoline * Called from kretprobe_trampoline
*/ */
void *__kprobes trampoline_handler(struct pt_regs *regs) void * __kprobes trampoline_handler(struct pt_regs *regs)
{ {
struct kretprobe_instance *ri = NULL; struct kretprobe_instance *ri = NULL;
struct hlist_head *head, empty_rp; struct hlist_head *head, empty_rp;
@ -417,27 +486,27 @@ void *__kprobes trampoline_handler(struct pt_regs *regs)
/* fixup registers */ /* fixup registers */
regs->cs = __KERNEL_CS | get_kernel_rpl(); regs->cs = __KERNEL_CS | get_kernel_rpl();
regs->ip = trampoline_address; regs->ip = trampoline_address;
regs->orig_ax = 0xffffffff; regs->orig_ax = ~0UL;
/* /*
* It is possible to have multiple instances associated with a given * It is possible to have multiple instances associated with a given
* task either because an multiple functions in the call path * task either because multiple functions in the call path have
* have a return probe installed on them, and/or more then one return * return probes installed on them, and/or more then one
* return probe was registered for a target function. * return probe was registered for a target function.
* *
* We can handle this because: * We can handle this because:
* - instances are always inserted at the head of the list * - instances are always pushed into the head of the list
* - when multiple return probes are registered for the same * - when multiple return probes are registered for the same
* function, the first instance's ret_addr will point to the * function, the (chronologically) first instance's ret_addr
* real return address, and all the rest will point to * will be the real return address, and all the rest will
* kretprobe_trampoline * point to kretprobe_trampoline.
*/ */
hlist_for_each_entry_safe(ri, node, tmp, head, hlist) { hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
if (ri->task != current) if (ri->task != current)
/* another task is sharing our hash bucket */ /* another task is sharing our hash bucket */
continue; continue;
if (ri->rp && ri->rp->handler){ if (ri->rp && ri->rp->handler) {
__get_cpu_var(current_kprobe) = &ri->rp->kp; __get_cpu_var(current_kprobe) = &ri->rp->kp;
get_kprobe_ctlblk()->kprobe_status = KPROBE_HIT_ACTIVE; get_kprobe_ctlblk()->kprobe_status = KPROBE_HIT_ACTIVE;
ri->rp->handler(ri, regs); ri->rp->handler(ri, regs);
@ -457,13 +526,14 @@ void *__kprobes trampoline_handler(struct pt_regs *regs)
} }
kretprobe_assert(ri, orig_ret_address, trampoline_address); kretprobe_assert(ri, orig_ret_address, trampoline_address);
spin_unlock_irqrestore(&kretprobe_lock, flags); spin_unlock_irqrestore(&kretprobe_lock, flags);
hlist_for_each_entry_safe(ri, node, tmp, &empty_rp, hlist) { hlist_for_each_entry_safe(ri, node, tmp, &empty_rp, hlist) {
hlist_del(&ri->hlist); hlist_del(&ri->hlist);
kfree(ri); kfree(ri);
} }
return (void*)orig_ret_address; return (void *)orig_ret_address;
} }
/* /*
@ -488,48 +558,55 @@ void *__kprobes trampoline_handler(struct pt_regs *regs)
* that is atop the stack is the address following the copied instruction. * that is atop the stack is the address following the copied instruction.
* We need to make it the address following the original instruction. * We need to make it the address following the original instruction.
* *
* This function also checks instruction size for preparing direct execution. * If this is the first time we've single-stepped the instruction at
* this probepoint, and the instruction is boostable, boost it: add a
* jump instruction after the copied instruction, that jumps to the next
* instruction after the probepoint.
*/ */
static void __kprobes resume_execution(struct kprobe *p, static void __kprobes resume_execution(struct kprobe *p,
struct pt_regs *regs, struct kprobe_ctlblk *kcb) struct pt_regs *regs, struct kprobe_ctlblk *kcb)
{ {
unsigned long *tos = (unsigned long *)&regs->sp; unsigned long *tos = stack_addr(regs);
unsigned long copy_eip = (unsigned long)p->ainsn.insn; unsigned long copy_ip = (unsigned long)p->ainsn.insn;
unsigned long orig_eip = (unsigned long)p->addr; unsigned long orig_ip = (unsigned long)p->addr;
kprobe_opcode_t *insn = p->ainsn.insn;
regs->flags &= ~TF_MASK; regs->flags &= ~TF_MASK;
switch (p->ainsn.insn[0]) { switch (*insn) {
case 0x9c: /* pushfl */ case 0x9c: /* pushfl */
*tos &= ~(TF_MASK | IF_MASK); *tos &= ~(TF_MASK | IF_MASK);
*tos |= kcb->kprobe_old_eflags; *tos |= kcb->kprobe_old_flags;
break; break;
case 0xc2: /* iret/ret/lret */ case 0xc2: /* iret/ret/lret */
case 0xc3: case 0xc3:
case 0xca: case 0xca:
case 0xcb: case 0xcb:
case 0xcf: case 0xcf:
case 0xea: /* jmp absolute -- ip is correct */ case 0xea: /* jmp absolute -- ip is correct */
/* ip is already adjusted, no more changes required */ /* ip is already adjusted, no more changes required */
p->ainsn.boostable = 1; p->ainsn.boostable = 1;
goto no_change; goto no_change;
case 0xe8: /* call relative - Fix return addr */ case 0xe8: /* call relative - Fix return addr */
*tos = orig_eip + (*tos - copy_eip); *tos = orig_ip + (*tos - copy_ip);
break; break;
case 0x9a: /* call absolute -- same as call absolute, indirect */ case 0x9a: /* call absolute -- same as call absolute, indirect */
*tos = orig_eip + (*tos - copy_eip); *tos = orig_ip + (*tos - copy_ip);
goto no_change; goto no_change;
case 0xff: case 0xff:
if ((p->ainsn.insn[1] & 0x30) == 0x10) { if ((insn[1] & 0x30) == 0x10) {
/* /*
* call absolute, indirect * call absolute, indirect
* Fix return addr; ip is correct. * Fix return addr; ip is correct.
* But this is not boostable * But this is not boostable
*/ */
*tos = orig_eip + (*tos - copy_eip); *tos = orig_ip + (*tos - copy_ip);
goto no_change; goto no_change;
} else if (((p->ainsn.insn[1] & 0x31) == 0x20) || /* jmp near, absolute indirect */ } else if (((insn[1] & 0x31) == 0x20) ||
((p->ainsn.insn[1] & 0x31) == 0x21)) { /* jmp far, absolute indirect */ ((insn[1] & 0x31) == 0x21)) {
/* ip is correct. And this is boostable */ /*
* jmp near and far, absolute indirect
* ip is correct. And this is boostable
*/
p->ainsn.boostable = 1; p->ainsn.boostable = 1;
goto no_change; goto no_change;
} }
@ -538,21 +615,21 @@ static void __kprobes resume_execution(struct kprobe *p,
} }
if (p->ainsn.boostable == 0) { if (p->ainsn.boostable == 0) {
if ((regs->ip > copy_eip) && if ((regs->ip > copy_ip) &&
(regs->ip - copy_eip) + 5 < MAX_INSN_SIZE) { (regs->ip - copy_ip) + 5 < MAX_INSN_SIZE) {
/* /*
* These instructions can be executed directly if it * These instructions can be executed directly if it
* jumps back to correct address. * jumps back to correct address.
*/ */
set_jmp_op((void *)regs->ip, set_jmp_op((void *)regs->ip,
(void *)orig_eip + (regs->ip - copy_eip)); (void *)orig_ip + (regs->ip - copy_ip));
p->ainsn.boostable = 1; p->ainsn.boostable = 1;
} else { } else {
p->ainsn.boostable = -1; p->ainsn.boostable = -1;
} }
} }
regs->ip = orig_eip + (regs->ip - copy_eip); regs->ip += orig_ip - copy_ip;
no_change: no_change:
restore_btf(); restore_btf();
@ -578,10 +655,10 @@ static int __kprobes post_kprobe_handler(struct pt_regs *regs)
} }
resume_execution(cur, regs, kcb); resume_execution(cur, regs, kcb);
regs->flags |= kcb->kprobe_saved_eflags; regs->flags |= kcb->kprobe_saved_flags;
trace_hardirqs_fixup_flags(regs->flags); trace_hardirqs_fixup_flags(regs->flags);
/*Restore back the original saved kprobes variables and continue. */ /* Restore back the original saved kprobes variables and continue. */
if (kcb->kprobe_status == KPROBE_REENTER) { if (kcb->kprobe_status == KPROBE_REENTER) {
restore_previous_kprobe(kcb); restore_previous_kprobe(kcb);
goto out; goto out;
@ -617,7 +694,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
* normal page fault. * normal page fault.
*/ */
regs->ip = (unsigned long)cur->addr; regs->ip = (unsigned long)cur->addr;
regs->flags |= kcb->kprobe_old_eflags; regs->flags |= kcb->kprobe_old_flags;
if (kcb->kprobe_status == KPROBE_REENTER) if (kcb->kprobe_status == KPROBE_REENTER)
restore_previous_kprobe(kcb); restore_previous_kprobe(kcb);
else else
@ -628,7 +705,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
case KPROBE_HIT_SSDONE: case KPROBE_HIT_SSDONE:
/* /*
* We increment the nmissed count for accounting, * We increment the nmissed count for accounting,
* we can also use npre/npostfault count for accouting * we can also use npre/npostfault count for accounting
* these specific fault cases. * these specific fault cases.
*/ */
kprobes_inc_nmissed_count(cur); kprobes_inc_nmissed_count(cur);
@ -651,7 +728,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
return 1; return 1;
/* /*
* fixup_exception() could not handle it, * fixup routine could not handle it,
* Let do_page_fault() fix it. * Let do_page_fault() fix it.
*/ */
break; break;
@ -662,7 +739,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
} }
/* /*
* Wrapper routine to for handling exceptions. * Wrapper routine for handling exceptions.
*/ */
int __kprobes kprobe_exceptions_notify(struct notifier_block *self, int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
unsigned long val, void *data) unsigned long val, void *data)
@ -703,11 +780,11 @@ int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs)
struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
kcb->jprobe_saved_regs = *regs; kcb->jprobe_saved_regs = *regs;
kcb->jprobe_saved_esp = &regs->sp; kcb->jprobe_saved_sp = stack_addr(regs);
addr = (unsigned long)(kcb->jprobe_saved_esp); addr = (unsigned long)(kcb->jprobe_saved_sp);
/* /*
* TBD: As Linus pointed out, gcc assumes that the callee * As Linus pointed out, gcc assumes that the callee
* owns the argument space and could overwrite it, e.g. * owns the argument space and could overwrite it, e.g.
* tailcall optimization. So, to be absolutely safe * tailcall optimization. So, to be absolutely safe
* we also save and restore enough stack bytes to cover * we also save and restore enough stack bytes to cover
@ -730,21 +807,20 @@ void __kprobes jprobe_return(void)
" .globl jprobe_return_end \n" " .globl jprobe_return_end \n"
" jprobe_return_end: \n" " jprobe_return_end: \n"
" nop \n"::"b" " nop \n"::"b"
(kcb->jprobe_saved_esp):"memory"); (kcb->jprobe_saved_sp):"memory");
} }
int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
{ {
struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
u8 *addr = (u8 *) (regs->ip - 1); u8 *addr = (u8 *) (regs->ip - 1);
unsigned long stack_addr = (unsigned long)(kcb->jprobe_saved_esp);
struct jprobe *jp = container_of(p, struct jprobe, kp); struct jprobe *jp = container_of(p, struct jprobe, kp);
if ((addr > (u8 *) jprobe_return) && (addr < (u8 *) jprobe_return_end)) { if ((addr > (u8 *) jprobe_return) && (addr < (u8 *) jprobe_return_end)) {
if (&regs->sp != kcb->jprobe_saved_esp) { if (stack_addr(regs) != kcb->jprobe_saved_sp) {
struct pt_regs *saved_regs = &kcb->jprobe_saved_regs; struct pt_regs *saved_regs = &kcb->jprobe_saved_regs;
printk("current sp %p does not match saved sp %p\n", printk("current sp %p does not match saved sp %p\n",
&regs->sp, kcb->jprobe_saved_esp); stack_addr(regs), kcb->jprobe_saved_sp);
printk("Saved registers for jprobe %p\n", jp); printk("Saved registers for jprobe %p\n", jp);
show_registers(saved_regs); show_registers(saved_regs);
printk("Current registers\n"); printk("Current registers\n");
@ -752,20 +828,21 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
BUG(); BUG();
} }
*regs = kcb->jprobe_saved_regs; *regs = kcb->jprobe_saved_regs;
memcpy((kprobe_opcode_t *) stack_addr, kcb->jprobes_stack, memcpy((kprobe_opcode_t *)(kcb->jprobe_saved_sp),
MIN_STACK_SIZE(stack_addr)); kcb->jprobes_stack,
MIN_STACK_SIZE(kcb->jprobe_saved_sp));
preempt_enable_no_resched(); preempt_enable_no_resched();
return 1; return 1;
} }
return 0; return 0;
} }
int __kprobes arch_trampoline_kprobe(struct kprobe *p)
{
return 0;
}
int __init arch_init_kprobes(void) int __init arch_init_kprobes(void)
{ {
return 0; return 0;
} }
int __kprobes arch_trampoline_kprobe(struct kprobe *p)
{
return 0;
}

409
arch/x86/kernel/kprobes_64.c
View File

@ -40,16 +40,97 @@
#include <linux/module.h> #include <linux/module.h>
#include <linux/kdebug.h> #include <linux/kdebug.h>
#include <asm/cacheflush.h>
#include <asm/desc.h>
#include <asm/pgtable.h> #include <asm/pgtable.h>
#include <asm/uaccess.h> #include <asm/uaccess.h>
#include <asm/alternative.h> #include <asm/alternative.h>
void jprobe_return_end(void); void jprobe_return_end(void);
static void __kprobes arch_copy_kprobe(struct kprobe *p);
DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL; DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk); DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
#define stack_addr(regs) ((unsigned long *)regs->sp)
#define W(row, b0, b1, b2, b3, b4, b5, b6, b7, b8, b9, ba, bb, bc, bd, be, bf)\
(((b0##UL << 0x0)|(b1##UL << 0x1)|(b2##UL << 0x2)|(b3##UL << 0x3) | \
(b4##UL << 0x4)|(b5##UL << 0x5)|(b6##UL << 0x6)|(b7##UL << 0x7) | \
(b8##UL << 0x8)|(b9##UL << 0x9)|(ba##UL << 0xa)|(bb##UL << 0xb) | \
(bc##UL << 0xc)|(bd##UL << 0xd)|(be##UL << 0xe)|(bf##UL << 0xf)) \
<< (row % 32))
/*
* Undefined/reserved opcodes, conditional jump, Opcode Extension
* Groups, and some special opcodes can not boost.
*/
static const u32 twobyte_is_boostable[256 / 32] = {
/* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
/* ---------------------------------------------- */
W(0x00, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0) | /* 00 */
W(0x10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 10 */
W(0x20, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* 20 */
W(0x30, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 30 */
W(0x40, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) | /* 40 */
W(0x50, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) , /* 50 */
W(0x60, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1) | /* 60 */
W(0x70, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1) , /* 70 */
W(0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) | /* 80 */
W(0x90, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1) , /* 90 */
W(0xa0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1) | /* a0 */
W(0xb0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1) , /* b0 */
W(0xc0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1) | /* c0 */
W(0xd0, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1) , /* d0 */
W(0xe0, 0, 1, 1, 0, 0, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1) | /* e0 */
W(0xf0, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 0) /* f0 */
/* ----------------------------------------------- */
/* 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[] = { struct kretprobe_blackpoint kretprobe_blacklist[] = {
{"__switch_to", }, /* This function switches only current task, but {"__switch_to", }, /* This function switches only current task, but
doesn't switch kernel stack.*/ doesn't switch kernel stack.*/
@ -70,44 +151,11 @@ static __always_inline void set_jmp_op(void *from, void *to)
} }
/* /*
* returns non-zero if opcode is boostable * returns non-zero if opcode is boostable.
* RIP relative instructions are adjusted at copying time * RIP relative instructions are adjusted at copying time
*/ */
static __always_inline int can_boost(kprobe_opcode_t *opcodes) static __always_inline int can_boost(kprobe_opcode_t *opcodes)
{ {
#define W(row, b0, b1, b2, b3, b4, b5, b6, b7, b8, b9, ba, bb, bc, bd, be, bf)\
(((b0##UL << 0x0)|(b1##UL << 0x1)|(b2##UL << 0x2)|(b3##UL << 0x3) | \
(b4##UL << 0x4)|(b5##UL << 0x5)|(b6##UL << 0x6)|(b7##UL << 0x7) | \
(b8##UL << 0x8)|(b9##UL << 0x9)|(ba##UL << 0xa)|(bb##UL << 0xb) | \
(bc##UL << 0xc)|(bd##UL << 0xd)|(be##UL << 0xe)|(bf##UL << 0xf)) \
<< (row % 64))
/*
* Undefined/reserved opcodes, conditional jump, Opcode Extension
* Groups, and some special opcodes can not boost.
*/
static const unsigned long twobyte_is_boostable[256 / 64] = {
/* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
/* ---------------------------------------------- */
W(0x00, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0)|/* 00 */
W(0x10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)|/* 10 */
W(0x20, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)|/* 20 */
W(0x30, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0),/* 30 */
W(0x40, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1)|/* 40 */
W(0x50, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)|/* 50 */
W(0x60, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1)|/* 60 */
W(0x70, 0, 0, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1),/* 70 */
W(0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)|/* 80 */
W(0x90, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1)|/* 90 */
W(0xa0, 1, 1, 0, 1, 1, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1)|/* a0 */
W(0xb0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1),/* b0 */
W(0xc0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1)|/* c0 */
W(0xd0, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1)|/* d0 */
W(0xe0, 0, 1, 1, 0, 0, 1, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1)|/* e0 */
W(0xf0, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 0) /* f0 */
/* ----------------------------------------------- */
/* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
};
#undef W
kprobe_opcode_t opcode; kprobe_opcode_t opcode;
kprobe_opcode_t *orig_opcodes = opcodes; kprobe_opcode_t *orig_opcodes = opcodes;
@ -120,7 +168,8 @@ static __always_inline int can_boost(kprobe_opcode_t *opcodes)
if (opcode == 0x0f) { if (opcode == 0x0f) {
if (opcodes - orig_opcodes > MAX_INSN_SIZE - 1) if (opcodes - orig_opcodes > MAX_INSN_SIZE - 1)
return 0; return 0;
return test_bit(*opcodes, twobyte_is_boostable); return test_bit(*opcodes,
(unsigned long *)twobyte_is_boostable);
} }
switch (opcode & 0xf0) { switch (opcode & 0xf0) {
@ -169,80 +218,25 @@ static int __kprobes is_IF_modifier(kprobe_opcode_t *insn)
return 1; return 1;
} }
if (*insn >= 0x40 && *insn <= 0x4f && *++insn == 0xcf) /*
return 1; * on 64 bit x86, 0x40-0x4f are prefixes so we need to look
return 0; * at the next byte instead.. but of course not recurse infinitely
} */
if (*insn >= 0x40 && *insn <= 0x4f)
int __kprobes arch_prepare_kprobe(struct kprobe *p) return is_IF_modifier(++insn);
{
/* insn: must be on special executable page on x86_64. */
p->ainsn.insn = get_insn_slot();
if (!p->ainsn.insn) {
return -ENOMEM;
}
arch_copy_kprobe(p);
return 0; return 0;
} }
/* /*
* Determine if the instruction uses the %rip-relative addressing mode. * Adjust the displacement if the instruction uses the %rip-relative
* addressing mode.
* If it does, Return the address of the 32-bit displacement word. * If it does, Return the address of the 32-bit displacement word.
* If not, return null. * If not, return null.
*/ */
static s32 __kprobes *is_riprel(u8 *insn) static void __kprobes fix_riprel(struct kprobe *p)
{ {
#define W(row,b0,b1,b2,b3,b4,b5,b6,b7,b8,b9,ba,bb,bc,bd,be,bf) \ u8 *insn = p->ainsn.insn;
(((b0##UL << 0x0)|(b1##UL << 0x1)|(b2##UL << 0x2)|(b3##UL << 0x3) | \ s64 disp;
(b4##UL << 0x4)|(b5##UL << 0x5)|(b6##UL << 0x6)|(b7##UL << 0x7) | \
(b8##UL << 0x8)|(b9##UL << 0x9)|(ba##UL << 0xa)|(bb##UL << 0xb) | \
(bc##UL << 0xc)|(bd##UL << 0xd)|(be##UL << 0xe)|(bf##UL << 0xf)) \
<< (row % 64))
static const u64 onebyte_has_modrm[256 / 64] = {
/* 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 u64 twobyte_has_modrm[256 / 64] = {
/* 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
int need_modrm; int need_modrm;
/* Skip legacy instruction prefixes. */ /* Skip legacy instruction prefixes. */
@ -271,54 +265,60 @@ static s32 __kprobes *is_riprel(u8 *insn)
if (*insn == 0x0f) { /* Two-byte opcode. */ if (*insn == 0x0f) { /* Two-byte opcode. */
++insn; ++insn;
need_modrm = test_bit(*insn, twobyte_has_modrm); need_modrm = test_bit(*insn,
} else { /* One-byte opcode. */ (unsigned long *)twobyte_has_modrm);
need_modrm = test_bit(*insn, onebyte_has_modrm); } else /* One-byte opcode. */
} need_modrm = test_bit(*insn,
(unsigned long *)onebyte_has_modrm);
if (need_modrm) { if (need_modrm) {
u8 modrm = *++insn; u8 modrm = *++insn;
if ((modrm & 0xc7) == 0x05) { /* %rip+disp32 addressing mode */ if ((modrm & 0xc7) == 0x05) { /* %rip+disp32 addressing mode */
/* Displacement follows ModRM byte. */ /* Displacement follows ModRM byte. */
return (s32 *) ++insn; ++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;
} }
} }
/* No %rip-relative addressing mode here. */
return NULL;
} }
static void __kprobes arch_copy_kprobe(struct kprobe *p) static void __kprobes arch_copy_kprobe(struct kprobe *p)
{ {
s32 *ripdisp; memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE); fix_riprel(p);
ripdisp = is_riprel(p->ainsn.insn); if (can_boost(p->addr))
if (ripdisp) {
/*
* 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.
*/
s64 disp = (u8 *) p->addr + *ripdisp - (u8 *) p->ainsn.insn;
BUG_ON((s64) (s32) disp != disp); /* Sanity check. */
*ripdisp = disp;
}
if (can_boost(p->addr)) {
p->ainsn.boostable = 0; p->ainsn.boostable = 0;
} else { else
p->ainsn.boostable = -1; p->ainsn.boostable = -1;
}
p->opcode = *p->addr; p->opcode = *p->addr;
} }
int __kprobes arch_prepare_kprobe(struct kprobe *p)
{
/* insn: must be on special executable page on x86. */
p->ainsn.insn = get_insn_slot();
if (!p->ainsn.insn)
return -ENOMEM;
arch_copy_kprobe(p);
return 0;
}
void __kprobes arch_arm_kprobe(struct kprobe *p) void __kprobes arch_arm_kprobe(struct kprobe *p)
{ {
text_poke(p->addr, ((unsigned char []){BREAKPOINT_INSTRUCTION}), 1); text_poke(p->addr, ((unsigned char []){BREAKPOINT_INSTRUCTION}), 1);
@ -340,26 +340,26 @@ static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb)
{ {
kcb->prev_kprobe.kp = kprobe_running(); kcb->prev_kprobe.kp = kprobe_running();
kcb->prev_kprobe.status = kcb->kprobe_status; kcb->prev_kprobe.status = kcb->kprobe_status;
kcb->prev_kprobe.old_rflags = kcb->kprobe_old_rflags; kcb->prev_kprobe.old_flags = kcb->kprobe_old_flags;
kcb->prev_kprobe.saved_rflags = kcb->kprobe_saved_rflags; kcb->prev_kprobe.saved_flags = kcb->kprobe_saved_flags;
} }
static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb) static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb)
{ {
__get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp; __get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
kcb->kprobe_status = kcb->prev_kprobe.status; kcb->kprobe_status = kcb->prev_kprobe.status;
kcb->kprobe_old_rflags = kcb->prev_kprobe.old_rflags; kcb->kprobe_old_flags = kcb->prev_kprobe.old_flags;
kcb->kprobe_saved_rflags = kcb->prev_kprobe.saved_rflags; kcb->kprobe_saved_flags = kcb->prev_kprobe.saved_flags;
} }
static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs, static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs,
struct kprobe_ctlblk *kcb) struct kprobe_ctlblk *kcb)
{ {
__get_cpu_var(current_kprobe) = p; __get_cpu_var(current_kprobe) = p;
kcb->kprobe_saved_rflags = kcb->kprobe_old_rflags kcb->kprobe_saved_flags = kcb->kprobe_old_flags
= (regs->flags & (TF_MASK | IF_MASK)); = (regs->flags & (TF_MASK | IF_MASK));
if (is_IF_modifier(p->ainsn.insn)) if (is_IF_modifier(p->ainsn.insn))
kcb->kprobe_saved_rflags &= ~IF_MASK; kcb->kprobe_saved_flags &= ~IF_MASK;
} }
static __always_inline void clear_btf(void) static __always_inline void clear_btf(void)
@ -390,20 +390,27 @@ static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs)
void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri, void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
struct pt_regs *regs) struct pt_regs *regs)
{ {
unsigned long *sara = (unsigned long *)regs->sp; unsigned long *sara = stack_addr(regs);
ri->ret_addr = (kprobe_opcode_t *) *sara; ri->ret_addr = (kprobe_opcode_t *) *sara;
/* Replace the return addr with trampoline addr */ /* Replace the return addr with trampoline addr */
*sara = (unsigned long) &kretprobe_trampoline; *sara = (unsigned long) &kretprobe_trampoline;
} }
int __kprobes kprobe_handler(struct pt_regs *regs) /*
* Interrupts are disabled on entry as trap3 is an interrupt gate and they
* remain disabled thorough out this function.
*/
static int __kprobes kprobe_handler(struct pt_regs *regs)
{ {
struct kprobe *p; struct kprobe *p;
int ret = 0; int ret = 0;
kprobe_opcode_t *addr = (kprobe_opcode_t *)(regs->ip - sizeof(kprobe_opcode_t)); kprobe_opcode_t *addr;
struct kprobe_ctlblk *kcb; struct kprobe_ctlblk *kcb;
addr = (kprobe_opcode_t *)(regs->ip - sizeof(kprobe_opcode_t));
/* /*
* We don't want to be preempted for the entire * We don't want to be preempted for the entire
* duration of kprobe processing * duration of kprobe processing
@ -418,7 +425,7 @@ int __kprobes kprobe_handler(struct pt_regs *regs)
if (kcb->kprobe_status == KPROBE_HIT_SS && if (kcb->kprobe_status == KPROBE_HIT_SS &&
*p->ainsn.insn == BREAKPOINT_INSTRUCTION) { *p->ainsn.insn == BREAKPOINT_INSTRUCTION) {
regs->flags &= ~TF_MASK; regs->flags &= ~TF_MASK;
regs->flags |= kcb->kprobe_saved_rflags; regs->flags |= kcb->kprobe_saved_flags;
goto no_kprobe; goto no_kprobe;
} else if (kcb->kprobe_status == KPROBE_HIT_SSDONE) { } else if (kcb->kprobe_status == KPROBE_HIT_SSDONE) {
/* TODO: Provide re-entrancy from /* TODO: Provide re-entrancy from
@ -429,22 +436,20 @@ int __kprobes kprobe_handler(struct pt_regs *regs)
arch_disarm_kprobe(p); arch_disarm_kprobe(p);
regs->ip = (unsigned long)p->addr; regs->ip = (unsigned long)p->addr;
reset_current_kprobe(); reset_current_kprobe();
ret = 1;
} else {
/* We have reentered the kprobe_handler(), since
* another probe was hit while within the
* handler. We here save the original kprobe
* variables and just single step on instruction
* of the new probe without calling any user
* handlers.
*/
save_previous_kprobe(kcb);
set_current_kprobe(p, regs, kcb);
kprobes_inc_nmissed_count(p);
prepare_singlestep(p, regs);
kcb->kprobe_status = KPROBE_REENTER;
return 1; return 1;
} }
/* We have reentered the kprobe_handler(), since
* another probe was hit while within the handler.
* We here save the original kprobes variables and
* just single step on the instruction of the new probe
* without calling any user handlers.
*/
save_previous_kprobe(kcb);
set_current_kprobe(p, regs, kcb);
kprobes_inc_nmissed_count(p);
prepare_singlestep(p, regs);
kcb->kprobe_status = KPROBE_REENTER;
return 1;
} else { } else {
if (*addr != BREAKPOINT_INSTRUCTION) { if (*addr != BREAKPOINT_INSTRUCTION) {
/* The breakpoint instruction was removed by /* The breakpoint instruction was removed by
@ -578,23 +583,23 @@ fastcall void * __kprobes trampoline_handler(struct pt_regs *regs)
INIT_HLIST_HEAD(&empty_rp); INIT_HLIST_HEAD(&empty_rp);
spin_lock_irqsave(&kretprobe_lock, flags); spin_lock_irqsave(&kretprobe_lock, flags);
head = kretprobe_inst_table_head(current); head = kretprobe_inst_table_head(current);
/* fixup rt_regs */ /* fixup registers */
regs->cs = __KERNEL_CS; regs->cs = __KERNEL_CS;
regs->ip = trampoline_address; regs->ip = trampoline_address;
regs->orig_ax = 0xffffffffffffffff; regs->orig_ax = ~0UL;
/* /*
* It is possible to have multiple instances associated with a given * It is possible to have multiple instances associated with a given
* task either because an multiple functions in the call path * task either because multiple functions in the call path have
* have a return probe installed on them, and/or more then one return * return probes installed on them, and/or more then one
* return probe was registered for a target function. * return probe was registered for a target function.
* *
* We can handle this because: * We can handle this because:
* - instances are always inserted at the head of the list * - instances are always pushed into the head of the list
* - when multiple return probes are registered for the same * - when multiple return probes are registered for the same
* function, the first instance's ret_addr will point to the * function, the (chronologically) first instance's ret_addr
* real return address, and all the rest will point to * will be the real return address, and all the rest will
* kretprobe_trampoline * point to kretprobe_trampoline.
*/ */
hlist_for_each_entry_safe(ri, node, tmp, head, hlist) { hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
if (ri->task != current) if (ri->task != current)
@ -661,9 +666,9 @@ fastcall void * __kprobes trampoline_handler(struct pt_regs *regs)
static void __kprobes resume_execution(struct kprobe *p, static void __kprobes resume_execution(struct kprobe *p,
struct pt_regs *regs, struct kprobe_ctlblk *kcb) struct pt_regs *regs, struct kprobe_ctlblk *kcb)
{ {
unsigned long *tos = (unsigned long *)regs->sp; unsigned long *tos = stack_addr(regs);
unsigned long copy_rip = (unsigned long)p->ainsn.insn; unsigned long copy_ip = (unsigned long)p->ainsn.insn;
unsigned long orig_rip = (unsigned long)p->addr; unsigned long orig_ip = (unsigned long)p->addr;
kprobe_opcode_t *insn = p->ainsn.insn; kprobe_opcode_t *insn = p->ainsn.insn;
/*skip the REX prefix*/ /*skip the REX prefix*/
@ -674,7 +679,7 @@ static void __kprobes resume_execution(struct kprobe *p,
switch (*insn) { switch (*insn) {
case 0x9c: /* pushfl */ case 0x9c: /* pushfl */
*tos &= ~(TF_MASK | IF_MASK); *tos &= ~(TF_MASK | IF_MASK);
*tos |= kcb->kprobe_old_rflags; *tos |= kcb->kprobe_old_flags;
break; break;
case 0xc2: /* iret/ret/lret */ case 0xc2: /* iret/ret/lret */
case 0xc3: case 0xc3:
@ -686,18 +691,23 @@ static void __kprobes resume_execution(struct kprobe *p,
p->ainsn.boostable = 1; p->ainsn.boostable = 1;
goto no_change; goto no_change;
case 0xe8: /* call relative - Fix return addr */ case 0xe8: /* call relative - Fix return addr */
*tos = orig_rip + (*tos - copy_rip); *tos = orig_ip + (*tos - copy_ip);
break; break;
case 0xff: case 0xff:
if ((insn[1] & 0x30) == 0x10) { if ((insn[1] & 0x30) == 0x10) {
/* call absolute, indirect */ /*
/* Fix return addr; ip is correct. */ * call absolute, indirect
/* not boostable */ * Fix return addr; ip is correct.
*tos = orig_rip + (*tos - copy_rip); * But this is not boostable
*/
*tos = orig_ip + (*tos - copy_ip);
goto no_change; goto no_change;
} else if (((insn[1] & 0x31) == 0x20) || /* jmp near, absolute indirect */ } else if (((insn[1] & 0x31) == 0x20) ||
((insn[1] & 0x31) == 0x21)) { /* jmp far, absolute indirect */ ((insn[1] & 0x31) == 0x21)) {
/* ip is correct. And this is boostable */ /*
* jmp near and far, absolute indirect
* ip is correct. And this is boostable
*/
p->ainsn.boostable = 1; p->ainsn.boostable = 1;
goto no_change; goto no_change;
} }
@ -706,21 +716,21 @@ static void __kprobes resume_execution(struct kprobe *p,
} }
if (p->ainsn.boostable == 0) { if (p->ainsn.boostable == 0) {
if ((regs->ip > copy_rip) && if ((regs->ip > copy_ip) &&
(regs->ip - copy_rip) + 5 < MAX_INSN_SIZE) { (regs->ip - copy_ip) + 5 < MAX_INSN_SIZE) {
/* /*
* These instructions can be executed directly if it * These instructions can be executed directly if it
* jumps back to correct address. * jumps back to correct address.
*/ */
set_jmp_op((void *)regs->ip, set_jmp_op((void *)regs->ip,
(void *)orig_rip + (regs->ip - copy_rip)); (void *)orig_ip + (regs->ip - copy_ip));
p->ainsn.boostable = 1; p->ainsn.boostable = 1;
} else { } else {
p->ainsn.boostable = -1; p->ainsn.boostable = -1;
} }
} }
regs->ip = orig_rip + (regs->ip - copy_rip); regs->ip += orig_ip - copy_ip;
no_change: no_change:
restore_btf(); restore_btf();
@ -728,7 +738,11 @@ static void __kprobes resume_execution(struct kprobe *p,
return; return;
} }
int __kprobes post_kprobe_handler(struct pt_regs *regs) /*
* Interrupts are disabled on entry as trap1 is an interrupt gate and they
* remain disabled thoroughout this function.
*/
static int __kprobes post_kprobe_handler(struct pt_regs *regs)
{ {
struct kprobe *cur = kprobe_running(); struct kprobe *cur = kprobe_running();
struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
@ -742,10 +756,10 @@ int __kprobes post_kprobe_handler(struct pt_regs *regs)
} }
resume_execution(cur, regs, kcb); resume_execution(cur, regs, kcb);
regs->flags |= kcb->kprobe_saved_rflags; regs->flags |= kcb->kprobe_saved_flags;
trace_hardirqs_fixup_flags(regs->flags); trace_hardirqs_fixup_flags(regs->flags);
/* Restore the original saved kprobes variables and continue. */ /* Restore back the original saved kprobes variables and continue. */
if (kcb->kprobe_status == KPROBE_REENTER) { if (kcb->kprobe_status == KPROBE_REENTER) {
restore_previous_kprobe(kcb); restore_previous_kprobe(kcb);
goto out; goto out;
@ -782,7 +796,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
* normal page fault. * normal page fault.
*/ */
regs->ip = (unsigned long)cur->addr; regs->ip = (unsigned long)cur->addr;
regs->flags |= kcb->kprobe_old_rflags; regs->flags |= kcb->kprobe_old_flags;
if (kcb->kprobe_status == KPROBE_REENTER) if (kcb->kprobe_status == KPROBE_REENTER)
restore_previous_kprobe(kcb); restore_previous_kprobe(kcb);
else else
@ -793,7 +807,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
case KPROBE_HIT_SSDONE: case KPROBE_HIT_SSDONE:
/* /*
* We increment the nmissed count for accounting, * We increment the nmissed count for accounting,
* we can also use npre/npostfault count for accouting * we can also use npre/npostfault count for accounting
* these specific fault cases. * these specific fault cases.
*/ */
kprobes_inc_nmissed_count(cur); kprobes_inc_nmissed_count(cur);
@ -819,7 +833,7 @@ int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
} }
/* /*
* fixup() could not handle it, * fixup routine could not handle it,
* Let do_page_fault() fix it. * Let do_page_fault() fix it.
*/ */
break; break;
@ -838,7 +852,7 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
struct die_args *args = (struct die_args *)data; struct die_args *args = (struct die_args *)data;
int ret = NOTIFY_DONE; int ret = NOTIFY_DONE;
if (args->regs && user_mode(args->regs)) if (args->regs && user_mode_vm(args->regs))
return ret; return ret;
switch (val) { switch (val) {
@ -871,8 +885,9 @@ int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs)
struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
kcb->jprobe_saved_regs = *regs; kcb->jprobe_saved_regs = *regs;
kcb->jprobe_saved_rsp = (long *) regs->sp; kcb->jprobe_saved_sp = stack_addr(regs);
addr = (unsigned long)(kcb->jprobe_saved_rsp); addr = (unsigned long)(kcb->jprobe_saved_sp);
/* /*
* As Linus pointed out, gcc assumes that the callee * As Linus pointed out, gcc assumes that the callee
* owns the argument space and could overwrite it, e.g. * owns the argument space and could overwrite it, e.g.
@ -897,21 +912,20 @@ void __kprobes jprobe_return(void)
" .globl jprobe_return_end \n" " .globl jprobe_return_end \n"
" jprobe_return_end: \n" " jprobe_return_end: \n"
" nop \n"::"b" " nop \n"::"b"
(kcb->jprobe_saved_rsp):"memory"); (kcb->jprobe_saved_sp):"memory");
} }
int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
{ {
struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
u8 *addr = (u8 *) (regs->ip - 1); u8 *addr = (u8 *) (regs->ip - 1);
unsigned long stack_addr = (unsigned long)(kcb->jprobe_saved_rsp);
struct jprobe *jp = container_of(p, struct jprobe, kp); struct jprobe *jp = container_of(p, struct jprobe, kp);
if ((addr > (u8 *) jprobe_return) && (addr < (u8 *) jprobe_return_end)) { if ((addr > (u8 *) jprobe_return) && (addr < (u8 *) jprobe_return_end)) {
if ((unsigned long *)regs->sp != kcb->jprobe_saved_rsp) { if (stack_addr(regs) != kcb->jprobe_saved_sp) {
struct pt_regs *saved_regs = &kcb->jprobe_saved_regs; struct pt_regs *saved_regs = &kcb->jprobe_saved_regs;
printk("current sp %p does not match saved sp %p\n", printk("current sp %p does not match saved sp %p\n",
(long *)regs->sp, kcb->jprobe_saved_rsp); stack_addr(regs), kcb->jprobe_saved_sp);
printk("Saved registers for jprobe %p\n", jp); printk("Saved registers for jprobe %p\n", jp);
show_registers(saved_regs); show_registers(saved_regs);
printk("Current registers\n"); printk("Current registers\n");
@ -919,8 +933,9 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
BUG(); BUG();
} }
*regs = kcb->jprobe_saved_regs; *regs = kcb->jprobe_saved_regs;
memcpy((kprobe_opcode_t *) stack_addr, kcb->jprobes_stack, memcpy((kprobe_opcode_t *)(kcb->jprobe_saved_sp),
MIN_STACK_SIZE(stack_addr)); kcb->jprobes_stack,
MIN_STACK_SIZE(kcb->jprobe_saved_sp));
preempt_enable_no_resched(); preempt_enable_no_resched();
return 1; return 1;
} }

1
arch/x86/mm/fault_32.c
View File

@ -25,7 +25,6 @@
#include <linux/kprobes.h> #include <linux/kprobes.h>
#include <linux/uaccess.h> #include <linux/uaccess.h>
#include <linux/kdebug.h> #include <linux/kdebug.h>
#include <linux/kprobes.h>
#include <asm/system.h> #include <asm/system.h>
#include <asm/desc.h> #include <asm/desc.h>

1
arch/x86/mm/fault_64.c
View File

@ -25,7 +25,6 @@
#include <linux/kprobes.h> #include <linux/kprobes.h>
#include <linux/uaccess.h> #include <linux/uaccess.h>
#include <linux/kdebug.h> #include <linux/kdebug.h>
#include <linux/kprobes.h>
#include <asm/system.h> #include <asm/system.h>
#include <asm/pgalloc.h> #include <asm/pgalloc.h>

32
include/asm-x86/kprobes_32.h
View File

@ -2,7 +2,6 @@
#define _ASM_KPROBES_H #define _ASM_KPROBES_H
/* /*
* Kernel Probes (KProbes) * Kernel Probes (KProbes)
* include/asm-i386/kprobes.h
* *
* This program is free software; you can redistribute it and/or modify * 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 * it under the terms of the GNU General Public License as published by
@ -23,14 +22,17 @@
* 2002-Oct Created by Vamsi Krishna S <vamsi_krishna@in.ibm.com> Kernel * 2002-Oct Created by Vamsi Krishna S <vamsi_krishna@in.ibm.com> Kernel
* Probes initial implementation ( includes suggestions from * Probes initial implementation ( includes suggestions from
* Rusty Russell). * Rusty Russell).
* 2004-Oct Prasanna S Panchamukhi <prasanna@in.ibm.com> and Jim Keniston
* kenistoj@us.ibm.com adopted from i386.
*/ */
#include <linux/types.h> #include <linux/types.h>
#include <linux/ptrace.h> #include <linux/ptrace.h>
#include <linux/percpu.h>
#define __ARCH_WANT_KPROBES_INSN_SLOT #define __ARCH_WANT_KPROBES_INSN_SLOT
struct kprobe;
struct pt_regs; struct pt_regs;
struct kprobe;
typedef u8 kprobe_opcode_t; typedef u8 kprobe_opcode_t;
#define BREAKPOINT_INSTRUCTION 0xcc #define BREAKPOINT_INSTRUCTION 0xcc
@ -38,9 +40,11 @@ typedef u8 kprobe_opcode_t;
#define MAX_INSN_SIZE 16 #define MAX_INSN_SIZE 16
#define MAX_STACK_SIZE 64 #define MAX_STACK_SIZE 64
#define MIN_STACK_SIZE(ADDR) (((MAX_STACK_SIZE) < \ #define MIN_STACK_SIZE(ADDR) (((MAX_STACK_SIZE) < \
(((unsigned long)current_thread_info()) + THREAD_SIZE - (ADDR))) \ (((unsigned long)current_thread_info()) + THREAD_SIZE \
- (unsigned long)(ADDR))) \
? (MAX_STACK_SIZE) \ ? (MAX_STACK_SIZE) \
: (((unsigned long)current_thread_info()) + THREAD_SIZE - (ADDR))) : (((unsigned long)current_thread_info()) + THREAD_SIZE \
- (unsigned long)(ADDR)))
#define ARCH_SUPPORTS_KRETPROBES #define ARCH_SUPPORTS_KRETPROBES
#define flush_insn_slot(p) do { } while (0) #define flush_insn_slot(p) do { } while (0)
@ -55,8 +59,12 @@ struct arch_specific_insn {
/* copy of the original instruction */ /* copy of the original instruction */
kprobe_opcode_t *insn; kprobe_opcode_t *insn;
/* /*
* If this flag is not 0, this kprobe can be boost when its * boostable = -1: This instruction type is not boostable.
* post_handler and break_handler is not set. * boostable = 0: This instruction type is boostable.
* boostable = 1: This instruction has been boosted: we have
* added a relative jump after the instruction copy in insn,
* so no single-step and fixup are needed (unless there's
* a post_handler or break_handler).
*/ */
int boostable; int boostable;
}; };
@ -64,16 +72,16 @@ struct arch_specific_insn {
struct prev_kprobe { struct prev_kprobe {
struct kprobe *kp; struct kprobe *kp;
unsigned long status; unsigned long status;
unsigned long old_eflags; unsigned long old_flags;
unsigned long saved_eflags; unsigned long saved_flags;
}; };
/* per-cpu kprobe control block */ /* per-cpu kprobe control block */
struct kprobe_ctlblk { struct kprobe_ctlblk {
unsigned long kprobe_status; unsigned long kprobe_status;
unsigned long kprobe_old_eflags; unsigned long kprobe_old_flags;
unsigned long kprobe_saved_eflags; unsigned long kprobe_saved_flags;
unsigned long *jprobe_saved_esp; unsigned long *jprobe_saved_sp;
struct pt_regs jprobe_saved_regs; struct pt_regs jprobe_saved_regs;
kprobe_opcode_t jprobes_stack[MAX_STACK_SIZE]; kprobe_opcode_t jprobes_stack[MAX_STACK_SIZE];
struct prev_kprobe prev_kprobe; struct prev_kprobe prev_kprobe;
@ -88,7 +96,7 @@ static inline void restore_interrupts(struct pt_regs *regs)
local_irq_enable(); local_irq_enable();
} }
extern int kprobe_fault_handler(struct pt_regs *regs, int trapnr);
extern int kprobe_exceptions_notify(struct notifier_block *self, extern int kprobe_exceptions_notify(struct notifier_block *self,
unsigned long val, void *data); unsigned long val, void *data);
extern int kprobe_fault_handler(struct pt_regs *regs, int trapnr);
#endif /* _ASM_KPROBES_H */ #endif /* _ASM_KPROBES_H */

30
include/asm-x86/kprobes_64.h
View File

@ -2,7 +2,6 @@
#define _ASM_KPROBES_H #define _ASM_KPROBES_H
/* /*
* Kernel Probes (KProbes) * Kernel Probes (KProbes)
* include/asm-x86_64/kprobes.h
* *
* This program is free software; you can redistribute it and/or modify * 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 * it under the terms of the GNU General Public License as published by
@ -20,6 +19,9 @@
* *
* Copyright (C) IBM Corporation, 2002, 2004 * Copyright (C) IBM Corporation, 2002, 2004
* *
* 2002-Oct Created by Vamsi Krishna S <vamsi_krishna@in.ibm.com> Kernel
* Probes initial implementation ( includes suggestions from
* Rusty Russell).
* 2004-Oct Prasanna S Panchamukhi <prasanna@in.ibm.com> and Jim Keniston * 2004-Oct Prasanna S Panchamukhi <prasanna@in.ibm.com> and Jim Keniston
* kenistoj@us.ibm.com adopted from i386. * kenistoj@us.ibm.com adopted from i386.
*/ */
@ -35,19 +37,22 @@ struct kprobe;
typedef u8 kprobe_opcode_t; typedef u8 kprobe_opcode_t;
#define BREAKPOINT_INSTRUCTION 0xcc #define BREAKPOINT_INSTRUCTION 0xcc
#define RELATIVEJUMP_INSTRUCTION 0xe9 #define RELATIVEJUMP_INSTRUCTION 0xe9
#define MAX_INSN_SIZE 15 #define MAX_INSN_SIZE 16
#define MAX_STACK_SIZE 64 #define MAX_STACK_SIZE 64
#define MIN_STACK_SIZE(ADDR) (((MAX_STACK_SIZE) < \ #define MIN_STACK_SIZE(ADDR) (((MAX_STACK_SIZE) < \
(((unsigned long)current_thread_info()) + THREAD_SIZE - (ADDR))) \ (((unsigned long)current_thread_info()) + THREAD_SIZE \
- (unsigned long)(ADDR))) \
? (MAX_STACK_SIZE) \ ? (MAX_STACK_SIZE) \
: (((unsigned long)current_thread_info()) + THREAD_SIZE - (ADDR))) : (((unsigned long)current_thread_info()) + THREAD_SIZE \
- (unsigned long)(ADDR)))
#define ARCH_SUPPORTS_KRETPROBES #define ARCH_SUPPORTS_KRETPROBES
#define flush_insn_slot(p) do { } while (0)
extern const int kretprobe_blacklist_size; extern const int kretprobe_blacklist_size;
void arch_remove_kprobe(struct kprobe *p);
void kretprobe_trampoline(void); void kretprobe_trampoline(void);
extern void arch_remove_kprobe(struct kprobe *p);
#define flush_insn_slot(p) do { } while (0)
/* Architecture specific copy of original instruction*/ /* Architecture specific copy of original instruction*/
struct arch_specific_insn { struct arch_specific_insn {
@ -67,16 +72,16 @@ struct arch_specific_insn {
struct prev_kprobe { struct prev_kprobe {
struct kprobe *kp; struct kprobe *kp;
unsigned long status; unsigned long status;
unsigned long old_rflags; unsigned long old_flags;
unsigned long saved_rflags; unsigned long saved_flags;
}; };
/* per-cpu kprobe control block */ /* per-cpu kprobe control block */
struct kprobe_ctlblk { struct kprobe_ctlblk {
unsigned long kprobe_status; unsigned long kprobe_status;
unsigned long kprobe_old_rflags; unsigned long kprobe_old_flags;
unsigned long kprobe_saved_rflags; unsigned long kprobe_saved_flags;
unsigned long *jprobe_saved_rsp; unsigned long *jprobe_saved_sp;
struct pt_regs jprobe_saved_regs; struct pt_regs jprobe_saved_regs;
kprobe_opcode_t jprobes_stack[MAX_STACK_SIZE]; kprobe_opcode_t jprobes_stack[MAX_STACK_SIZE];
struct prev_kprobe prev_kprobe; struct prev_kprobe prev_kprobe;
@ -91,10 +96,7 @@ static inline void restore_interrupts(struct pt_regs *regs)
local_irq_enable(); local_irq_enable();
} }
extern int post_kprobe_handler(struct pt_regs *regs);
extern int kprobe_fault_handler(struct pt_regs *regs, int trapnr); extern int kprobe_fault_handler(struct pt_regs *regs, int trapnr);
extern int kprobe_handler(struct pt_regs *regs);
extern int kprobe_exceptions_notify(struct notifier_block *self, extern int kprobe_exceptions_notify(struct notifier_block *self,
unsigned long val, void *data); unsigned long val, void *data);
#endif /* _ASM_KPROBES_H */ #endif /* _ASM_KPROBES_H */