2 * Kernel Probes (KProbes)
3 * arch/x86_64/kernel/kprobes.c
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
19 * Copyright (C) IBM Corporation, 2002, 2004
21 * 2002-Oct Created by Vamsi Krishna S <vamsi_krishna@in.ibm.com> Kernel
22 * Probes initial implementation ( includes contributions from
24 * 2004-July Suparna Bhattacharya <suparna@in.ibm.com> added jumper probes
25 * interface to access function arguments.
26 * 2004-Oct Jim Keniston <kenistoj@us.ibm.com> and Prasanna S Panchamukhi
27 * <prasanna@in.ibm.com> adapted for x86_64
28 * 2005-Mar Roland McGrath <roland@redhat.com>
29 * Fixed to handle %rip-relative addressing mode correctly.
30 * 2005-May Rusty Lynch <rusty.lynch@intel.com>
31 * Added function return probes functionality
34 #include <linux/config.h>
35 #include <linux/kprobes.h>
36 #include <linux/ptrace.h>
37 #include <linux/string.h>
38 #include <linux/slab.h>
39 #include <linux/preempt.h>
41 #include <asm/cacheflush.h>
42 #include <asm/pgtable.h>
43 #include <asm/kdebug.h>
45 void jprobe_return_end(void);
46 static void __kprobes arch_copy_kprobe(struct kprobe *p);
48 DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
49 DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
52 * returns non-zero if opcode modifies the interrupt flag.
54 static inline int is_IF_modifier(kprobe_opcode_t *insn)
59 case 0xcf: /* iret/iretd */
60 case 0x9d: /* popf/popfd */
64 if (*insn >= 0x40 && *insn <= 0x4f && *++insn == 0xcf)
69 int __kprobes arch_prepare_kprobe(struct kprobe *p)
71 /* insn: must be on special executable page on x86_64. */
72 p->ainsn.insn = get_insn_slot();
81 * Determine if the instruction uses the %rip-relative addressing mode.
82 * If it does, return the address of the 32-bit displacement word.
83 * If not, return null.
85 static inline s32 *is_riprel(u8 *insn)
87 #define W(row,b0,b1,b2,b3,b4,b5,b6,b7,b8,b9,ba,bb,bc,bd,be,bf) \
88 (((b0##UL << 0x0)|(b1##UL << 0x1)|(b2##UL << 0x2)|(b3##UL << 0x3) | \
89 (b4##UL << 0x4)|(b5##UL << 0x5)|(b6##UL << 0x6)|(b7##UL << 0x7) | \
90 (b8##UL << 0x8)|(b9##UL << 0x9)|(ba##UL << 0xa)|(bb##UL << 0xb) | \
91 (bc##UL << 0xc)|(bd##UL << 0xd)|(be##UL << 0xe)|(bf##UL << 0xf)) \
93 static const u64 onebyte_has_modrm[256 / 64] = {
94 /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
95 /* ------------------------------- */
96 W(0x00, 1,1,1,1,0,0,0,0,1,1,1,1,0,0,0,0)| /* 00 */
97 W(0x10, 1,1,1,1,0,0,0,0,1,1,1,1,0,0,0,0)| /* 10 */
98 W(0x20, 1,1,1,1,0,0,0,0,1,1,1,1,0,0,0,0)| /* 20 */
99 W(0x30, 1,1,1,1,0,0,0,0,1,1,1,1,0,0,0,0), /* 30 */
100 W(0x40, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)| /* 40 */
101 W(0x50, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)| /* 50 */
102 W(0x60, 0,0,1,1,0,0,0,0,0,1,0,1,0,0,0,0)| /* 60 */
103 W(0x70, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), /* 70 */
104 W(0x80, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)| /* 80 */
105 W(0x90, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)| /* 90 */
106 W(0xa0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)| /* a0 */
107 W(0xb0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), /* b0 */
108 W(0xc0, 1,1,0,0,1,1,1,1,0,0,0,0,0,0,0,0)| /* c0 */
109 W(0xd0, 1,1,1,1,0,0,0,0,1,1,1,1,1,1,1,1)| /* d0 */
110 W(0xe0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)| /* e0 */
111 W(0xf0, 0,0,0,0,0,0,1,1,0,0,0,0,0,0,1,1) /* f0 */
112 /* ------------------------------- */
113 /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
115 static const u64 twobyte_has_modrm[256 / 64] = {
116 /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
117 /* ------------------------------- */
118 W(0x00, 1,1,1,1,0,0,0,0,0,0,0,0,0,1,0,1)| /* 0f */
119 W(0x10, 1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0)| /* 1f */
120 W(0x20, 1,1,1,1,1,0,1,0,1,1,1,1,1,1,1,1)| /* 2f */
121 W(0x30, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), /* 3f */
122 W(0x40, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)| /* 4f */
123 W(0x50, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)| /* 5f */
124 W(0x60, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)| /* 6f */
125 W(0x70, 1,1,1,1,1,1,1,0,0,0,0,0,1,1,1,1), /* 7f */
126 W(0x80, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)| /* 8f */
127 W(0x90, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)| /* 9f */
128 W(0xa0, 0,0,0,1,1,1,1,1,0,0,0,1,1,1,1,1)| /* af */
129 W(0xb0, 1,1,1,1,1,1,1,1,0,0,1,1,1,1,1,1), /* bf */
130 W(0xc0, 1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0)| /* cf */
131 W(0xd0, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)| /* df */
132 W(0xe0, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1)| /* ef */
133 W(0xf0, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0) /* ff */
134 /* ------------------------------- */
135 /* 0 1 2 3 4 5 6 7 8 9 a b c d e f */
140 /* Skip legacy instruction prefixes. */
160 /* Skip REX instruction prefix. */
161 if ((*insn & 0xf0) == 0x40)
164 if (*insn == 0x0f) { /* Two-byte opcode. */
166 need_modrm = test_bit(*insn, twobyte_has_modrm);
167 } else { /* One-byte opcode. */
168 need_modrm = test_bit(*insn, onebyte_has_modrm);
173 if ((modrm & 0xc7) == 0x05) { /* %rip+disp32 addressing mode */
174 /* Displacement follows ModRM byte. */
175 return (s32 *) ++insn;
179 /* No %rip-relative addressing mode here. */
183 static void __kprobes arch_copy_kprobe(struct kprobe *p)
186 memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE);
187 ripdisp = is_riprel(p->ainsn.insn);
190 * The copied instruction uses the %rip-relative
191 * addressing mode. Adjust the displacement for the
192 * difference between the original location of this
193 * instruction and the location of the copy that will
194 * actually be run. The tricky bit here is making sure
195 * that the sign extension happens correctly in this
196 * calculation, since we need a signed 32-bit result to
197 * be sign-extended to 64 bits when it's added to the
198 * %rip value and yield the same 64-bit result that the
199 * sign-extension of the original signed 32-bit
200 * displacement would have given.
202 s64 disp = (u8 *) p->addr + *ripdisp - (u8 *) p->ainsn.insn;
203 BUG_ON((s64) (s32) disp != disp); /* Sanity check. */
206 p->opcode = *p->addr;
209 void __kprobes arch_arm_kprobe(struct kprobe *p)
211 *p->addr = BREAKPOINT_INSTRUCTION;
212 flush_icache_range((unsigned long) p->addr,
213 (unsigned long) p->addr + sizeof(kprobe_opcode_t));
216 void __kprobes arch_disarm_kprobe(struct kprobe *p)
218 *p->addr = p->opcode;
219 flush_icache_range((unsigned long) p->addr,
220 (unsigned long) p->addr + sizeof(kprobe_opcode_t));
223 void __kprobes arch_remove_kprobe(struct kprobe *p)
225 free_insn_slot(p->ainsn.insn);
228 static inline void save_previous_kprobe(struct kprobe_ctlblk *kcb)
230 kcb->prev_kprobe.kp = kprobe_running();
231 kcb->prev_kprobe.status = kcb->kprobe_status;
232 kcb->prev_kprobe.old_rflags = kcb->kprobe_old_rflags;
233 kcb->prev_kprobe.saved_rflags = kcb->kprobe_saved_rflags;
236 static inline void restore_previous_kprobe(struct kprobe_ctlblk *kcb)
238 __get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
239 kcb->kprobe_status = kcb->prev_kprobe.status;
240 kcb->kprobe_old_rflags = kcb->prev_kprobe.old_rflags;
241 kcb->kprobe_saved_rflags = kcb->prev_kprobe.saved_rflags;
244 static inline void set_current_kprobe(struct kprobe *p, struct pt_regs *regs,
245 struct kprobe_ctlblk *kcb)
247 __get_cpu_var(current_kprobe) = p;
248 kcb->kprobe_saved_rflags = kcb->kprobe_old_rflags
249 = (regs->eflags & (TF_MASK | IF_MASK));
250 if (is_IF_modifier(p->ainsn.insn))
251 kcb->kprobe_saved_rflags &= ~IF_MASK;
254 static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs)
256 regs->eflags |= TF_MASK;
257 regs->eflags &= ~IF_MASK;
258 /*single step inline if the instruction is an int3*/
259 if (p->opcode == BREAKPOINT_INSTRUCTION)
260 regs->rip = (unsigned long)p->addr;
262 regs->rip = (unsigned long)p->ainsn.insn;
265 /* Called with kretprobe_lock held */
266 void __kprobes arch_prepare_kretprobe(struct kretprobe *rp,
267 struct pt_regs *regs)
269 unsigned long *sara = (unsigned long *)regs->rsp;
270 struct kretprobe_instance *ri;
272 if ((ri = get_free_rp_inst(rp)) != NULL) {
275 ri->ret_addr = (kprobe_opcode_t *) *sara;
277 /* Replace the return addr with trampoline addr */
278 *sara = (unsigned long) &kretprobe_trampoline;
286 int __kprobes kprobe_handler(struct pt_regs *regs)
290 kprobe_opcode_t *addr = (kprobe_opcode_t *)(regs->rip - sizeof(kprobe_opcode_t));
291 struct kprobe_ctlblk *kcb;
294 * We don't want to be preempted for the entire
295 * duration of kprobe processing
298 kcb = get_kprobe_ctlblk();
300 /* Check we're not actually recursing */
301 if (kprobe_running()) {
302 p = get_kprobe(addr);
304 if (kcb->kprobe_status == KPROBE_HIT_SS &&
305 *p->ainsn.insn == BREAKPOINT_INSTRUCTION) {
306 regs->eflags &= ~TF_MASK;
307 regs->eflags |= kcb->kprobe_saved_rflags;
309 } else if (kcb->kprobe_status == KPROBE_HIT_SSDONE) {
310 /* TODO: Provide re-entrancy from
311 * post_kprobes_handler() and avoid exception
312 * stack corruption while single-stepping on
313 * the instruction of the new probe.
315 arch_disarm_kprobe(p);
316 regs->rip = (unsigned long)p->addr;
317 reset_current_kprobe();
320 /* We have reentered the kprobe_handler(), since
321 * another probe was hit while within the
322 * handler. We here save the original kprobe
323 * variables and just single step on instruction
324 * of the new probe without calling any user
327 save_previous_kprobe(kcb);
328 set_current_kprobe(p, regs, kcb);
329 kprobes_inc_nmissed_count(p);
330 prepare_singlestep(p, regs);
331 kcb->kprobe_status = KPROBE_REENTER;
335 p = __get_cpu_var(current_kprobe);
336 if (p->break_handler && p->break_handler(p, regs)) {
343 p = get_kprobe(addr);
345 if (*addr != BREAKPOINT_INSTRUCTION) {
347 * The breakpoint instruction was removed right
348 * after we hit it. Another cpu has removed
349 * either a probepoint or a debugger breakpoint
350 * at this address. In either case, no further
351 * handling of this interrupt is appropriate.
352 * Back up over the (now missing) int3 and run
353 * the original instruction.
355 regs->rip = (unsigned long)addr;
358 /* Not one of ours: let kernel handle it */
362 set_current_kprobe(p, regs, kcb);
363 kcb->kprobe_status = KPROBE_HIT_ACTIVE;
365 if (p->pre_handler && p->pre_handler(p, regs))
366 /* handler has already set things up, so skip ss setup */
370 prepare_singlestep(p, regs);
371 kcb->kprobe_status = KPROBE_HIT_SS;
375 preempt_enable_no_resched();
380 * For function-return probes, init_kprobes() establishes a probepoint
381 * here. When a retprobed function returns, this probe is hit and
382 * trampoline_probe_handler() runs, calling the kretprobe's handler.
384 void kretprobe_trampoline_holder(void)
386 asm volatile ( ".global kretprobe_trampoline\n"
387 "kretprobe_trampoline: \n"
392 * Called when we hit the probe point at kretprobe_trampoline
394 int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
396 struct kretprobe_instance *ri = NULL;
397 struct hlist_head *head;
398 struct hlist_node *node, *tmp;
399 unsigned long flags, orig_ret_address = 0;
400 unsigned long trampoline_address =(unsigned long)&kretprobe_trampoline;
402 spin_lock_irqsave(&kretprobe_lock, flags);
403 head = kretprobe_inst_table_head(current);
406 * It is possible to have multiple instances associated with a given
407 * task either because an multiple functions in the call path
408 * have a return probe installed on them, and/or more then one return
409 * return probe was registered for a target function.
411 * We can handle this because:
412 * - instances are always inserted at the head of the list
413 * - when multiple return probes are registered for the same
414 * function, the first instance's ret_addr will point to the
415 * real return address, and all the rest will point to
416 * kretprobe_trampoline
418 hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
419 if (ri->task != current)
420 /* another task is sharing our hash bucket */
423 if (ri->rp && ri->rp->handler)
424 ri->rp->handler(ri, regs);
426 orig_ret_address = (unsigned long)ri->ret_addr;
429 if (orig_ret_address != trampoline_address)
431 * This is the real return address. Any other
432 * instances associated with this task are for
433 * other calls deeper on the call stack
438 BUG_ON(!orig_ret_address || (orig_ret_address == trampoline_address));
439 regs->rip = orig_ret_address;
441 reset_current_kprobe();
442 spin_unlock_irqrestore(&kretprobe_lock, flags);
443 preempt_enable_no_resched();
446 * By returning a non-zero value, we are telling
447 * kprobe_handler() that we don't want the post_handler
448 * to run (and have re-enabled preemption)
454 * Called after single-stepping. p->addr is the address of the
455 * instruction whose first byte has been replaced by the "int 3"
456 * instruction. To avoid the SMP problems that can occur when we
457 * temporarily put back the original opcode to single-step, we
458 * single-stepped a copy of the instruction. The address of this
459 * copy is p->ainsn.insn.
461 * This function prepares to return from the post-single-step
462 * interrupt. We have to fix up the stack as follows:
464 * 0) Except in the case of absolute or indirect jump or call instructions,
465 * the new rip is relative to the copied instruction. We need to make
466 * it relative to the original instruction.
468 * 1) If the single-stepped instruction was pushfl, then the TF and IF
469 * flags are set in the just-pushed eflags, and may need to be cleared.
471 * 2) If the single-stepped instruction was a call, the return address
472 * that is atop the stack is the address following the copied instruction.
473 * We need to make it the address following the original instruction.
475 static void __kprobes resume_execution(struct kprobe *p,
476 struct pt_regs *regs, struct kprobe_ctlblk *kcb)
478 unsigned long *tos = (unsigned long *)regs->rsp;
479 unsigned long next_rip = 0;
480 unsigned long copy_rip = (unsigned long)p->ainsn.insn;
481 unsigned long orig_rip = (unsigned long)p->addr;
482 kprobe_opcode_t *insn = p->ainsn.insn;
484 /*skip the REX prefix*/
485 if (*insn >= 0x40 && *insn <= 0x4f)
489 case 0x9c: /* pushfl */
490 *tos &= ~(TF_MASK | IF_MASK);
491 *tos |= kcb->kprobe_old_rflags;
493 case 0xc3: /* ret/lret */
497 regs->eflags &= ~TF_MASK;
498 /* rip is already adjusted, no more changes required*/
500 case 0xe8: /* call relative - Fix return addr */
501 *tos = orig_rip + (*tos - copy_rip);
504 if ((*insn & 0x30) == 0x10) {
505 /* call absolute, indirect */
506 /* Fix return addr; rip is correct. */
507 next_rip = regs->rip;
508 *tos = orig_rip + (*tos - copy_rip);
509 } else if (((*insn & 0x31) == 0x20) || /* jmp near, absolute indirect */
510 ((*insn & 0x31) == 0x21)) { /* jmp far, absolute indirect */
511 /* rip is correct. */
512 next_rip = regs->rip;
515 case 0xea: /* jmp absolute -- rip is correct */
516 next_rip = regs->rip;
522 regs->eflags &= ~TF_MASK;
524 regs->rip = next_rip;
526 regs->rip = orig_rip + (regs->rip - copy_rip);
530 int __kprobes post_kprobe_handler(struct pt_regs *regs)
532 struct kprobe *cur = kprobe_running();
533 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
538 if ((kcb->kprobe_status != KPROBE_REENTER) && cur->post_handler) {
539 kcb->kprobe_status = KPROBE_HIT_SSDONE;
540 cur->post_handler(cur, regs, 0);
543 resume_execution(cur, regs, kcb);
544 regs->eflags |= kcb->kprobe_saved_rflags;
546 /* Restore the original saved kprobes variables and continue. */
547 if (kcb->kprobe_status == KPROBE_REENTER) {
548 restore_previous_kprobe(kcb);
551 reset_current_kprobe();
553 preempt_enable_no_resched();
556 * if somebody else is singlestepping across a probe point, eflags
557 * will have TF set, in which case, continue the remaining processing
558 * of do_debug, as if this is not a probe hit.
560 if (regs->eflags & TF_MASK)
566 int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
568 struct kprobe *cur = kprobe_running();
569 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
571 if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr))
574 if (kcb->kprobe_status & KPROBE_HIT_SS) {
575 resume_execution(cur, regs, kcb);
576 regs->eflags |= kcb->kprobe_old_rflags;
578 reset_current_kprobe();
579 preempt_enable_no_resched();
585 * Wrapper routine for handling exceptions.
587 int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
588 unsigned long val, void *data)
590 struct die_args *args = (struct die_args *)data;
591 int ret = NOTIFY_DONE;
595 if (kprobe_handler(args->regs))
599 if (post_kprobe_handler(args->regs))
604 /* kprobe_running() needs smp_processor_id() */
606 if (kprobe_running() &&
607 kprobe_fault_handler(args->regs, args->trapnr))
617 int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs)
619 struct jprobe *jp = container_of(p, struct jprobe, kp);
621 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
623 kcb->jprobe_saved_regs = *regs;
624 kcb->jprobe_saved_rsp = (long *) regs->rsp;
625 addr = (unsigned long)(kcb->jprobe_saved_rsp);
627 * As Linus pointed out, gcc assumes that the callee
628 * owns the argument space and could overwrite it, e.g.
629 * tailcall optimization. So, to be absolutely safe
630 * we also save and restore enough stack bytes to cover
633 memcpy(kcb->jprobes_stack, (kprobe_opcode_t *)addr,
634 MIN_STACK_SIZE(addr));
635 regs->eflags &= ~IF_MASK;
636 regs->rip = (unsigned long)(jp->entry);
640 void __kprobes jprobe_return(void)
642 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
644 asm volatile (" xchg %%rbx,%%rsp \n"
646 " .globl jprobe_return_end \n"
647 " jprobe_return_end: \n"
649 (kcb->jprobe_saved_rsp):"memory");
652 int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
654 struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
655 u8 *addr = (u8 *) (regs->rip - 1);
656 unsigned long stack_addr = (unsigned long)(kcb->jprobe_saved_rsp);
657 struct jprobe *jp = container_of(p, struct jprobe, kp);
659 if ((addr > (u8 *) jprobe_return) && (addr < (u8 *) jprobe_return_end)) {
660 if ((long *)regs->rsp != kcb->jprobe_saved_rsp) {
661 struct pt_regs *saved_regs =
662 container_of(kcb->jprobe_saved_rsp,
663 struct pt_regs, rsp);
664 printk("current rsp %p does not match saved rsp %p\n",
665 (long *)regs->rsp, kcb->jprobe_saved_rsp);
666 printk("Saved registers for jprobe %p\n", jp);
667 show_registers(saved_regs);
668 printk("Current registers\n");
669 show_registers(regs);
672 *regs = kcb->jprobe_saved_regs;
673 memcpy((kprobe_opcode_t *) stack_addr, kcb->jprobes_stack,
674 MIN_STACK_SIZE(stack_addr));
675 preempt_enable_no_resched();
681 static struct kprobe trampoline_p = {
682 .addr = (kprobe_opcode_t *) &kretprobe_trampoline,
683 .pre_handler = trampoline_probe_handler
686 int __init arch_init_kprobes(void)
688 return register_kprobe(&trampoline_p);