[deliverable/linux.git] / arch / s390 / kernel / kprobes.c

/*
 *  Kernel Probes (KProbes)
 *
 * 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, 2006
 *
 * s390 port, used ppc64 as template. Mike Grundy <grundym@us.ibm.com>
 */

#include <linux/kprobes.h>
#include <linux/ptrace.h>
#include <linux/preempt.h>
#include <linux/stop_machine.h>
#include <linux/kdebug.h>
#include <linux/uaccess.h>
#include <asm/cacheflush.h>
#include <asm/sections.h>
#include <linux/module.h>

DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);

struct kretprobe_blackpoint kretprobe_blacklist[] = {{NULL, NULL}};

int __kprobes arch_prepare_kprobe(struct kprobe *p)
{
	/* Make sure the probe isn't going on a difficult instruction */
	if (is_prohibited_opcode((kprobe_opcode_t *) p->addr))
		return -EINVAL;

	if ((unsigned long)p->addr & 0x01)
		return -EINVAL;

	/* Use the get_insn_slot() facility for correctness */
	if (!(p->ainsn.insn = get_insn_slot()))
		return -ENOMEM;

	memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));

	get_instruction_type(&p->ainsn);
	p->opcode = *p->addr;
	return 0;
}

int __kprobes is_prohibited_opcode(kprobe_opcode_t *instruction)
{
	switch (*(__u8 *) instruction) {
	case 0x0c:	/* bassm */
	case 0x0b:	/* bsm	 */
	case 0x83:	/* diag  */
	case 0x44:	/* ex	 */
		return -EINVAL;
	}
	switch (*(__u16 *) instruction) {
	case 0x0101:	/* pr	 */
	case 0xb25a:	/* bsa	 */
	case 0xb240:	/* bakr  */
	case 0xb258:	/* bsg	 */
	case 0xb218:	/* pc	 */
	case 0xb228:	/* pt	 */
		return -EINVAL;
	}
	return 0;
}

void __kprobes get_instruction_type(struct arch_specific_insn *ainsn)
{
	/* default fixup method */
	ainsn->fixup = FIXUP_PSW_NORMAL;

	/* save r1 operand */
	ainsn->reg = (*ainsn->insn & 0xf0) >> 4;

	/* save the instruction length (pop 5-5) in bytes */
	switch (*(__u8 *) (ainsn->insn) >> 6) {
	case 0:
		ainsn->ilen = 2;
		break;
	case 1:
	case 2:
		ainsn->ilen = 4;
		break;
	case 3:
		ainsn->ilen = 6;
		break;
	}

	switch (*(__u8 *) ainsn->insn) {
	case 0x05:	/* balr	*/
	case 0x0d:	/* basr */
		ainsn->fixup = FIXUP_RETURN_REGISTER;
		/* if r2 = 0, no branch will be taken */
		if ((*ainsn->insn & 0x0f) == 0)
			ainsn->fixup |= FIXUP_BRANCH_NOT_TAKEN;
		break;
	case 0x06:	/* bctr	*/
	case 0x07:	/* bcr	*/
		ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
		break;
	case 0x45:	/* bal	*/
	case 0x4d:	/* bas	*/
		ainsn->fixup = FIXUP_RETURN_REGISTER;
		break;
	case 0x47:	/* bc	*/
	case 0x46:	/* bct	*/
	case 0x86:	/* bxh	*/
	case 0x87:	/* bxle	*/
		ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
		break;
	case 0x82:	/* lpsw	*/
		ainsn->fixup = FIXUP_NOT_REQUIRED;
		break;
	case 0xb2:	/* lpswe */
		if (*(((__u8 *) ainsn->insn) + 1) == 0xb2) {
			ainsn->fixup = FIXUP_NOT_REQUIRED;
		}
		break;
	case 0xa7:	/* bras	*/
		if ((*ainsn->insn & 0x0f) == 0x05) {
			ainsn->fixup |= FIXUP_RETURN_REGISTER;
		}
		break;
	case 0xc0:
		if ((*ainsn->insn & 0x0f) == 0x00  /* larl  */
			|| (*ainsn->insn & 0x0f) == 0x05) /* brasl */
		ainsn->fixup |= FIXUP_RETURN_REGISTER;
		break;
	case 0xeb:
		if (*(((__u8 *) ainsn->insn) + 5 ) == 0x44 ||	/* bxhg  */
			*(((__u8 *) ainsn->insn) + 5) == 0x45) {/* bxleg */
			ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
		}
		break;
	case 0xe3:	/* bctg	*/
		if (*(((__u8 *) ainsn->insn) + 5) == 0x46) {
			ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
		}
		break;
	}
}

static int __kprobes swap_instruction(void *aref)
{
	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
	unsigned long status = kcb->kprobe_status;
	struct ins_replace_args *args = aref;
	int rc;

	kcb->kprobe_status = KPROBE_SWAP_INST;
	rc = probe_kernel_write(args->ptr, &args->new, sizeof(args->new));
	kcb->kprobe_status = status;
	return rc;
}

void __kprobes arch_arm_kprobe(struct kprobe *p)
{
	struct ins_replace_args args;

	args.ptr = p->addr;
	args.old = p->opcode;
	args.new = BREAKPOINT_INSTRUCTION;
	stop_machine(swap_instruction, &args, NULL);
}

void __kprobes arch_disarm_kprobe(struct kprobe *p)
{
	struct ins_replace_args args;

	args.ptr = p->addr;
	args.old = BREAKPOINT_INSTRUCTION;
	args.new = p->opcode;
	stop_machine(swap_instruction, &args, NULL);
}

void __kprobes arch_remove_kprobe(struct kprobe *p)
{
	if (p->ainsn.insn) {
		free_insn_slot(p->ainsn.insn, 0);
		p->ainsn.insn = NULL;
	}
}

static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs)
{
	per_cr_bits kprobe_per_regs[1];

	memset(kprobe_per_regs, 0, sizeof(per_cr_bits));
	regs->psw.addr = (unsigned long)p->ainsn.insn | PSW_ADDR_AMODE;

	/* Set up the per control reg info, will pass to lctl */
	kprobe_per_regs[0].em_instruction_fetch = 1;
	kprobe_per_regs[0].starting_addr = (unsigned long)p->ainsn.insn;
	kprobe_per_regs[0].ending_addr = (unsigned long)p->ainsn.insn + 1;

	/* Set the PER control regs, turns on single step for this address */
	__ctl_load(kprobe_per_regs, 9, 11);
	regs->psw.mask |= PSW_MASK_PER;
	regs->psw.mask &= ~(PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK);
}

static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb)
{
	kcb->prev_kprobe.kp = kprobe_running();
	kcb->prev_kprobe.status = kcb->kprobe_status;
	kcb->prev_kprobe.kprobe_saved_imask = kcb->kprobe_saved_imask;
	memcpy(kcb->prev_kprobe.kprobe_saved_ctl, kcb->kprobe_saved_ctl,
					sizeof(kcb->kprobe_saved_ctl));
}

static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb)
{
	__get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
	kcb->kprobe_status = kcb->prev_kprobe.status;
	kcb->kprobe_saved_imask = kcb->prev_kprobe.kprobe_saved_imask;
	memcpy(kcb->kprobe_saved_ctl, kcb->prev_kprobe.kprobe_saved_ctl,
					sizeof(kcb->kprobe_saved_ctl));
}

static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs,
						struct kprobe_ctlblk *kcb)
{
	__get_cpu_var(current_kprobe) = p;
	/* Save the interrupt and per flags */
	kcb->kprobe_saved_imask = regs->psw.mask &
	    (PSW_MASK_PER | PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK);
	/* Save the control regs that govern PER */
	__ctl_store(kcb->kprobe_saved_ctl, 9, 11);
}

void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
					struct pt_regs *regs)
{
	ri->ret_addr = (kprobe_opcode_t *) regs->gprs[14];

	/* Replace the return addr with trampoline addr */
	regs->gprs[14] = (unsigned long)&kretprobe_trampoline;
}

static int __kprobes kprobe_handler(struct pt_regs *regs)
{
	struct kprobe *p;
	int ret = 0;
	unsigned long *addr = (unsigned long *)
		((regs->psw.addr & PSW_ADDR_INSN) - 2);
	struct kprobe_ctlblk *kcb;

	/*
	 * We don't want to be preempted for the entire
	 * duration of kprobe processing
	 */
	preempt_disable();
	kcb = get_kprobe_ctlblk();

	/* Check we're not actually recursing */
	if (kprobe_running()) {
		p = get_kprobe(addr);
		if (p) {
			if (kcb->kprobe_status == KPROBE_HIT_SS &&
			    *p->ainsn.insn == BREAKPOINT_INSTRUCTION) {
				regs->psw.mask &= ~PSW_MASK_PER;
				regs->psw.mask |= kcb->kprobe_saved_imask;
				goto no_kprobe;
			}
			/* 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 {
			p = __get_cpu_var(current_kprobe);
			if (p->break_handler && p->break_handler(p, regs)) {
				goto ss_probe;
			}
		}
		goto no_kprobe;
	}

	p = get_kprobe(addr);
	if (!p)
		/*
		 * No kprobe at this address. The fault has not been
		 * caused by a kprobe breakpoint. The race of breakpoint
		 * vs. kprobe remove does not exist because on s390 we
		 * use stop_machine to arm/disarm the breakpoints.
		 */
		goto no_kprobe;

	kcb->kprobe_status = KPROBE_HIT_ACTIVE;
	set_current_kprobe(p, regs, kcb);
	if (p->pre_handler && p->pre_handler(p, regs))
		/* handler has already set things up, so skip ss setup */
		return 1;

ss_probe:
	prepare_singlestep(p, regs);
	kcb->kprobe_status = KPROBE_HIT_SS;
	return 1;

no_kprobe:
	preempt_enable_no_resched();
	return ret;
}

/*
 * Function return probe trampoline:
 *	- init_kprobes() establishes a probepoint here
 *	- When the probed function returns, this probe
 *		causes the handlers to fire
 */
static void __used kretprobe_trampoline_holder(void)
{
	asm volatile(".global kretprobe_trampoline\n"
		     "kretprobe_trampoline: bcr 0,0\n");
}

/*
 * Called when the probe at kretprobe trampoline is hit
 */
static int __kprobes trampoline_probe_handler(struct kprobe *p,
					      struct pt_regs *regs)
{
	struct kretprobe_instance *ri = NULL;
	struct hlist_head *head, empty_rp;
	struct hlist_node *node, *tmp;
	unsigned long flags, orig_ret_address = 0;
	unsigned long trampoline_address = (unsigned long)&kretprobe_trampoline;

	INIT_HLIST_HEAD(&empty_rp);
	kretprobe_hash_lock(current, &head, &flags);

	/*
	 * It is possible to have multiple instances associated with a given
	 * task either because an multiple functions in the call path
	 * have a return probe installed on them, and/or more than one return
	 * return probe was registered for a target function.
	 *
	 * We can handle this because:
	 *     - instances are always inserted at the head of the list
	 *     - when multiple return probes are registered for the same
	 *	 function, the first instance's ret_addr will point to the
	 *	 real return address, and all the rest will point to
	 *	 kretprobe_trampoline
	 */
	hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
		if (ri->task != current)
			/* another task is sharing our hash bucket */
			continue;

		if (ri->rp && ri->rp->handler)
			ri->rp->handler(ri, regs);

		orig_ret_address = (unsigned long)ri->ret_addr;
		recycle_rp_inst(ri, &empty_rp);

		if (orig_ret_address != trampoline_address) {
			/*
			 * This is the real return address. Any other
			 * instances associated with this task are for
			 * other calls deeper on the call stack
			 */
			break;
		}
	}
	kretprobe_assert(ri, orig_ret_address, trampoline_address);
	regs->psw.addr = orig_ret_address | PSW_ADDR_AMODE;

	reset_current_kprobe();
	kretprobe_hash_unlock(current, &flags);
	preempt_enable_no_resched();

	hlist_for_each_entry_safe(ri, node, tmp, &empty_rp, hlist) {
		hlist_del(&ri->hlist);
		kfree(ri);
	}
	/*
	 * By returning a non-zero value, we are telling
	 * kprobe_handler() that we don't want the post_handler
	 * to run (and have re-enabled preemption)
	 */
	return 1;
}

/*
 * Called after single-stepping.  p->addr is the address of the
 * instruction whose first byte has been replaced by the "breakpoint"
 * instruction.  To avoid the SMP problems that can occur when we
 * temporarily put back the original opcode to single-step, we
 * single-stepped a copy of the instruction.  The address of this
 * copy is p->ainsn.insn.
 */
static void __kprobes resume_execution(struct kprobe *p, struct pt_regs *regs)
{
	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();

	regs->psw.addr &= PSW_ADDR_INSN;

	if (p->ainsn.fixup & FIXUP_PSW_NORMAL)
		regs->psw.addr = (unsigned long)p->addr +
				((unsigned long)regs->psw.addr -
				 (unsigned long)p->ainsn.insn);

	if (p->ainsn.fixup & FIXUP_BRANCH_NOT_TAKEN)
		if ((unsigned long)regs->psw.addr -
		    (unsigned long)p->ainsn.insn == p->ainsn.ilen)
			regs->psw.addr = (unsigned long)p->addr + p->ainsn.ilen;

	if (p->ainsn.fixup & FIXUP_RETURN_REGISTER)
		regs->gprs[p->ainsn.reg] = ((unsigned long)p->addr +
						(regs->gprs[p->ainsn.reg] -
						(unsigned long)p->ainsn.insn))
						| PSW_ADDR_AMODE;

	regs->psw.addr |= PSW_ADDR_AMODE;
	/* turn off PER mode */
	regs->psw.mask &= ~PSW_MASK_PER;
	/* Restore the original per control regs */
	__ctl_load(kcb->kprobe_saved_ctl, 9, 11);
	regs->psw.mask |= kcb->kprobe_saved_imask;
}

static int __kprobes post_kprobe_handler(struct pt_regs *regs)
{
	struct kprobe *cur = kprobe_running();
	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();

	if (!cur)
		return 0;

	if ((kcb->kprobe_status != KPROBE_REENTER) && cur->post_handler) {
		kcb->kprobe_status = KPROBE_HIT_SSDONE;
		cur->post_handler(cur, regs, 0);
	}

	resume_execution(cur, regs);

	/*Restore back the original saved kprobes variables and continue. */
	if (kcb->kprobe_status == KPROBE_REENTER) {
		restore_previous_kprobe(kcb);
		goto out;
	}
	reset_current_kprobe();
out:
	preempt_enable_no_resched();

	/*
	 * if somebody else is singlestepping across a probe point, psw mask
	 * will have PER set, in which case, continue the remaining processing
	 * of do_single_step, as if this is not a probe hit.
	 */
	if (regs->psw.mask & PSW_MASK_PER) {
		return 0;
	}

	return 1;
}

int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
{
	struct kprobe *cur = kprobe_running();
	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
	const struct exception_table_entry *entry;

	switch(kcb->kprobe_status) {
	case KPROBE_SWAP_INST:
		/* We are here because the instruction replacement failed */
		return 0;
	case KPROBE_HIT_SS:
	case KPROBE_REENTER:
		/*
		 * We are here because the instruction being single
		 * stepped caused a page fault. We reset the current
		 * kprobe and the nip points back to the probe address
		 * and allow the page fault handler to continue as a
		 * normal page fault.
		 */
		regs->psw.addr = (unsigned long)cur->addr | PSW_ADDR_AMODE;
		regs->psw.mask &= ~PSW_MASK_PER;
		regs->psw.mask |= kcb->kprobe_saved_imask;
		if (kcb->kprobe_status == KPROBE_REENTER)
			restore_previous_kprobe(kcb);
		else
			reset_current_kprobe();
		preempt_enable_no_resched();
		break;
	case KPROBE_HIT_ACTIVE:
	case KPROBE_HIT_SSDONE:
		/*
		 * We increment the nmissed count for accounting,
		 * we can also use npre/npostfault count for accouting
		 * these specific fault cases.
		 */
		kprobes_inc_nmissed_count(cur);

		/*
		 * We come here because instructions in the pre/post
		 * handler caused the page_fault, this could happen
		 * if handler tries to access user space by
		 * copy_from_user(), get_user() etc. Let the
		 * user-specified handler try to fix it first.
		 */
		if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr))
			return 1;

		/*
		 * In case the user-specified fault handler returned
		 * zero, try to fix up.
		 */
		entry = search_exception_tables(regs->psw.addr & PSW_ADDR_INSN);
		if (entry) {
			regs->psw.addr = entry->fixup | PSW_ADDR_AMODE;
			return 1;
		}

		/*
		 * fixup_exception() could not handle it,
		 * Let do_page_fault() fix it.
		 */
		break;
	default:
		break;
	}
	return 0;
}

/*
 * Wrapper routine to for handling exceptions.
 */
int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
				       unsigned long val, void *data)
{
	struct die_args *args = (struct die_args *)data;
	int ret = NOTIFY_DONE;

	switch (val) {
	case DIE_BPT:
		if (kprobe_handler(args->regs))
			ret = NOTIFY_STOP;
		break;
	case DIE_SSTEP:
		if (post_kprobe_handler(args->regs))
			ret = NOTIFY_STOP;
		break;
	case DIE_TRAP:
		/* kprobe_running() needs smp_processor_id() */
		preempt_disable();
		if (kprobe_running() &&
		    kprobe_fault_handler(args->regs, args->trapnr))
			ret = NOTIFY_STOP;
		preempt_enable();
		break;
	default:
		break;
	}
	return ret;
}

int __kprobes setjmp_pre_handler(struct kprobe *p, struct pt_regs *regs)
{
	struct jprobe *jp = container_of(p, struct jprobe, kp);
	unsigned long addr;
	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();

	memcpy(&kcb->jprobe_saved_regs, regs, sizeof(struct pt_regs));

	/* setup return addr to the jprobe handler routine */
	regs->psw.addr = (unsigned long)(jp->entry) | PSW_ADDR_AMODE;

	/* r14 is the function return address */
	kcb->jprobe_saved_r14 = (unsigned long)regs->gprs[14];
	/* r15 is the stack pointer */
	kcb->jprobe_saved_r15 = (unsigned long)regs->gprs[15];
	addr = (unsigned long)kcb->jprobe_saved_r15;

	memcpy(kcb->jprobes_stack, (kprobe_opcode_t *) addr,
	       MIN_STACK_SIZE(addr));
	return 1;
}

void __kprobes jprobe_return(void)
{
	asm volatile(".word 0x0002");
}

void __kprobes jprobe_return_end(void)
{
	asm volatile("bcr 0,0");
}

int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
{
	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
	unsigned long stack_addr = (unsigned long)(kcb->jprobe_saved_r15);

	/* Put the regs back */
	memcpy(regs, &kcb->jprobe_saved_regs, sizeof(struct pt_regs));
	/* put the stack back */
	memcpy((kprobe_opcode_t *) stack_addr, kcb->jprobes_stack,
	       MIN_STACK_SIZE(stack_addr));
	preempt_enable_no_resched();
	return 1;
}

static struct kprobe trampoline_p = {
	.addr = (kprobe_opcode_t *) & kretprobe_trampoline,
	.pre_handler = trampoline_probe_handler
};

int __init arch_init_kprobes(void)
{
	return register_kprobe(&trampoline_p);
}

int __kprobes arch_trampoline_kprobe(struct kprobe *p)
{
	if (p->addr == (kprobe_opcode_t *) & kretprobe_trampoline)
		return 1;
	return 0;
}
Commit	Line	Data
4ba069b8 MG	1	/*
	2	* Kernel Probes (KProbes)
	3	*
	4	* This program is free software; you can redistribute it and/or modify
	5	* it under the terms of the GNU General Public License as published by
	6	* the Free Software Foundation; either version 2 of the License, or
	7	* (at your option) any later version.
	8	*
	9	* This program is distributed in the hope that it will be useful,
	10	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	11	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	12	* GNU General Public License for more details.
	13	*
	14	* You should have received a copy of the GNU General Public License
	15	* along with this program; if not, write to the Free Software
	16	* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
	17	*
	18	* Copyright (C) IBM Corporation, 2002, 2006
	19	*
	20	* s390 port, used ppc64 as template. Mike Grundy <grundym@us.ibm.com>
	21	*/
	22
4ba069b8 MG	23	#include <linux/kprobes.h>
	24	#include <linux/ptrace.h>
	25	#include <linux/preempt.h>
	26	#include <linux/stop_machine.h>
1eeb66a1	27	#include <linux/kdebug.h>
a2b53673	28	#include <linux/uaccess.h>
4ba069b8	29	#include <asm/cacheflush.h>
4ba069b8	30	#include <asm/sections.h>
4ba069b8 MG	31	#include <linux/module.h>
	32
	33	DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
	34	DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
	35
f438d914 MH	36	struct kretprobe_blackpoint kretprobe_blacklist[] = {{NULL, NULL}};
f438d914 MH	37
4ba069b8 MG	38	int __kprobes arch_prepare_kprobe(struct kprobe *p)
	39	{
	40	/* Make sure the probe isn't going on a difficult instruction */
	41	if (is_prohibited_opcode((kprobe_opcode_t *) p->addr))
	42	return -EINVAL;
	43
5532bd0f	44	if ((unsigned long)p->addr & 0x01)
4ba069b8	45	return -EINVAL;
4ba069b8 MG	46
	47	/* Use the get_insn_slot() facility for correctness */
	48	if (!(p->ainsn.insn = get_insn_slot()))
	49	return -ENOMEM;
	50
	51	memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t));
	52
	53	get_instruction_type(&p->ainsn);
	54	p->opcode = *p->addr;
	55	return 0;
	56	}
	57
	58	int __kprobes is_prohibited_opcode(kprobe_opcode_t *instruction)
	59	{
	60	switch ((__u8 ) instruction) {
	61	case 0x0c: /* bassm */
	62	case 0x0b: /* bsm */
	63	case 0x83: /* diag */
	64	case 0x44: /* ex */
	65	return -EINVAL;
	66	}
	67	switch ((__u16 ) instruction) {
	68	case 0x0101: /* pr */
	69	case 0xb25a: /* bsa */
	70	case 0xb240: /* bakr */
	71	case 0xb258: /* bsg */
	72	case 0xb218: /* pc */
	73	case 0xb228: /* pt */
	74	return -EINVAL;
	75	}
	76	return 0;
	77	}
	78
	79	void __kprobes get_instruction_type(struct arch_specific_insn *ainsn)
	80	{
	81	/* default fixup method */
	82	ainsn->fixup = FIXUP_PSW_NORMAL;
	83
	84	/* save r1 operand */
	85	ainsn->reg = (*ainsn->insn & 0xf0) >> 4;
	86
	87	/* save the instruction length (pop 5-5) in bytes */
9c5f225f	88	switch ((__u8 ) (ainsn->insn) >> 6) {
4ba069b8 MG	89	case 0:
	90	ainsn->ilen = 2;
	91	break;
	92	case 1:
	93	case 2:
	94	ainsn->ilen = 4;
	95	break;
	96	case 3:
	97	ainsn->ilen = 6;
	98	break;
	99	}
	100
	101	switch ((__u8 ) ainsn->insn) {
	102	case 0x05: /* balr */
	103	case 0x0d: /* basr */
	104	ainsn->fixup = FIXUP_RETURN_REGISTER;
	105	/* if r2 = 0, no branch will be taken */
	106	if ((*ainsn->insn & 0x0f) == 0)
	107	ainsn->fixup \|= FIXUP_BRANCH_NOT_TAKEN;
	108	break;
	109	case 0x06: /* bctr */
	110	case 0x07: /* bcr */
	111	ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
	112	break;
	113	case 0x45: /* bal */
	114	case 0x4d: /* bas */
	115	ainsn->fixup = FIXUP_RETURN_REGISTER;
	116	break;
	117	case 0x47: /* bc */
	118	case 0x46: /* bct */
	119	case 0x86: /* bxh */
	120	case 0x87: /* bxle */
	121	ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
	122	break;
	123	case 0x82: /* lpsw */
	124	ainsn->fixup = FIXUP_NOT_REQUIRED;
	125	break;
	126	case 0xb2: /* lpswe */
	127	if ((((__u8 ) ainsn->insn) + 1) == 0xb2) {
	128	ainsn->fixup = FIXUP_NOT_REQUIRED;
	129	}
	130	break;
	131	case 0xa7: /* bras */
	132	if ((*ainsn->insn & 0x0f) == 0x05) {
	133	ainsn->fixup \|= FIXUP_RETURN_REGISTER;
	134	}
	135	break;
	136	case 0xc0:
	137	if ((ainsn->insn & 0x0f) == 0x00 / larl */
	138	\|\| (ainsn->insn & 0x0f) == 0x05) / brasl */
	139	ainsn->fixup \|= FIXUP_RETURN_REGISTER;
	140	break;
	141	case 0xeb:
	142	if ((((__u8 ) ainsn->insn) + 5 ) == 0x44 \|\| /* bxhg */
	143	(((__u8 ) ainsn->insn) + 5) == 0x45) {/* bxleg */
	144	ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
	145	}
	146	break;
	147	case 0xe3: /* bctg */
	148	if ((((__u8 ) ainsn->insn) + 5) == 0x46) {
	149	ainsn->fixup = FIXUP_BRANCH_NOT_TAKEN;
	150	}
	151	break;
	152	}
153	}
154
155	static int __kprobes swap_instruction(void *aref)
156	{
acf01800 HC	157	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
acf01800 HC	158	unsigned long status = kcb->kprobe_status;
4ba069b8	159	struct ins_replace_args *args = aref;
acf01800	160	int rc;
a2b53673	161
acf01800 HC	162	kcb->kprobe_status = KPROBE_SWAP_INST;
	163	rc = probe_kernel_write(args->ptr, &args->new, sizeof(args->new));
	164	kcb->kprobe_status = status;
	165	return rc;
4ba069b8 MG	166	}
	167
	168	void __kprobes arch_arm_kprobe(struct kprobe *p)
	169	{
4ba069b8 MG	170	struct ins_replace_args args;
	171
	172	args.ptr = p->addr;
	173	args.old = p->opcode;
	174	args.new = BREAKPOINT_INSTRUCTION;
9b1a4d38	175	stop_machine(swap_instruction, &args, NULL);
4ba069b8 MG	176	}
	177
	178	void __kprobes arch_disarm_kprobe(struct kprobe *p)
	179	{
4ba069b8 MG	180	struct ins_replace_args args;
	181
	182	args.ptr = p->addr;
	183	args.old = BREAKPOINT_INSTRUCTION;
	184	args.new = p->opcode;
9b1a4d38	185	stop_machine(swap_instruction, &args, NULL);
4ba069b8 MG	186	}
	187
	188	void __kprobes arch_remove_kprobe(struct kprobe *p)
	189	{
12941560 MH	190	if (p->ainsn.insn) {
	191	free_insn_slot(p->ainsn.insn, 0);
	192	p->ainsn.insn = NULL;
	193	}
4ba069b8 MG	194	}
	195
	196	static void __kprobes prepare_singlestep(struct kprobe p, struct pt_regs regs)
	197	{
	198	per_cr_bits kprobe_per_regs[1];
	199
	200	memset(kprobe_per_regs, 0, sizeof(per_cr_bits));
	201	regs->psw.addr = (unsigned long)p->ainsn.insn \| PSW_ADDR_AMODE;
	202
	203	/* Set up the per control reg info, will pass to lctl */
	204	kprobe_per_regs[0].em_instruction_fetch = 1;
	205	kprobe_per_regs[0].starting_addr = (unsigned long)p->ainsn.insn;
	206	kprobe_per_regs[0].ending_addr = (unsigned long)p->ainsn.insn + 1;
	207
	208	/* Set the PER control regs, turns on single step for this address */
	209	__ctl_load(kprobe_per_regs, 9, 11);
	210	regs->psw.mask \|= PSW_MASK_PER;
	211	regs->psw.mask &= ~(PSW_MASK_IO \| PSW_MASK_EXT \| PSW_MASK_MCHECK);
	212	}
	213
	214	static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb)
	215	{
	216	kcb->prev_kprobe.kp = kprobe_running();
	217	kcb->prev_kprobe.status = kcb->kprobe_status;
	218	kcb->prev_kprobe.kprobe_saved_imask = kcb->kprobe_saved_imask;
	219	memcpy(kcb->prev_kprobe.kprobe_saved_ctl, kcb->kprobe_saved_ctl,
	220	sizeof(kcb->kprobe_saved_ctl));
	221	}
	222
	223	static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb)
	224	{
	225	__get_cpu_var(current_kprobe) = kcb->prev_kprobe.kp;
	226	kcb->kprobe_status = kcb->prev_kprobe.status;
	227	kcb->kprobe_saved_imask = kcb->prev_kprobe.kprobe_saved_imask;
	228	memcpy(kcb->kprobe_saved_ctl, kcb->prev_kprobe.kprobe_saved_ctl,
	229	sizeof(kcb->kprobe_saved_ctl));
	230	}
	231
	232	static void __kprobes set_current_kprobe(struct kprobe p, struct pt_regs regs,
	233	struct kprobe_ctlblk *kcb)
	234	{
	235	__get_cpu_var(current_kprobe) = p;
	236	/* Save the interrupt and per flags */
	237	kcb->kprobe_saved_imask = regs->psw.mask &
	238	(PSW_MASK_PER \| PSW_MASK_IO \| PSW_MASK_EXT \| PSW_MASK_MCHECK);
	239	/* Save the control regs that govern PER */
	240	__ctl_store(kcb->kprobe_saved_ctl, 9, 11);
	241	}
	242
4c4308cb	243	void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
4ba069b8 MG	244	struct pt_regs *regs)
4ba069b8 MG	245	{
4c4308cb	246	ri->ret_addr = (kprobe_opcode_t *) regs->gprs[14];
4ba069b8	247
4c4308cb CH	248	/* Replace the return addr with trampoline addr */
4c4308cb CH	249	regs->gprs[14] = (unsigned long)&kretprobe_trampoline;
4ba069b8 MG	250	}
	251
	252	static int __kprobes kprobe_handler(struct pt_regs *regs)
	253	{
	254	struct kprobe *p;
	255	int ret = 0;
	256	unsigned long addr = (unsigned long )
	257	((regs->psw.addr & PSW_ADDR_INSN) - 2);
	258	struct kprobe_ctlblk *kcb;
	259
	260	/*
	261	* We don't want to be preempted for the entire
	262	* duration of kprobe processing
	263	*/
	264	preempt_disable();
	265	kcb = get_kprobe_ctlblk();
	266
	267	/* Check we're not actually recursing */
	268	if (kprobe_running()) {
	269	p = get_kprobe(addr);
	270	if (p) {
	271	if (kcb->kprobe_status == KPROBE_HIT_SS &&
	272	*p->ainsn.insn == BREAKPOINT_INSTRUCTION) {
	273	regs->psw.mask &= ~PSW_MASK_PER;
	274	regs->psw.mask \|= kcb->kprobe_saved_imask;
	275	goto no_kprobe;
	276	}
	277	/* We have reentered the kprobe_handler(), since
	278	* another probe was hit while within the handler.
	279	* We here save the original kprobes variables and
	280	* just single step on the instruction of the new probe
	281	* without calling any user handlers.
	282	*/
	283	save_previous_kprobe(kcb);
	284	set_current_kprobe(p, regs, kcb);
	285	kprobes_inc_nmissed_count(p);
	286	prepare_singlestep(p, regs);
	287	kcb->kprobe_status = KPROBE_REENTER;
	288	return 1;
	289	} else {
	290	p = __get_cpu_var(current_kprobe);
	291	if (p->break_handler && p->break_handler(p, regs)) {
	292	goto ss_probe;
	293	}
	294	}
	295	goto no_kprobe;
	296	}
	297
	298	p = get_kprobe(addr);
f794c827 MS	299	if (!p)
	300	/*
	301	* No kprobe at this address. The fault has not been
	302	* caused by a kprobe breakpoint. The race of breakpoint
	303	* vs. kprobe remove does not exist because on s390 we
9b1a4d38	304	* use stop_machine to arm/disarm the breakpoints.
f794c827	305	*/
4ba069b8	306	goto no_kprobe;
4ba069b8 MG	307
	308	kcb->kprobe_status = KPROBE_HIT_ACTIVE;
	309	set_current_kprobe(p, regs, kcb);
	310	if (p->pre_handler && p->pre_handler(p, regs))
	311	/* handler has already set things up, so skip ss setup */
	312	return 1;
	313
	314	ss_probe:
	315	prepare_singlestep(p, regs);
	316	kcb->kprobe_status = KPROBE_HIT_SS;
	317	return 1;
	318
	319	no_kprobe:
	320	preempt_enable_no_resched();
	321	return ret;
	322	}
	323
	324	/*
	325	* Function return probe trampoline:
	326	* - init_kprobes() establishes a probepoint here
	327	* - When the probed function returns, this probe
	328	* causes the handlers to fire
	329	*/
a806170e	330	static void __used kretprobe_trampoline_holder(void)
4ba069b8 MG	331	{
	332	asm volatile(".global kretprobe_trampoline\n"
	333	"kretprobe_trampoline: bcr 0,0\n");
	334	}
	335
	336	/*
	337	* Called when the probe at kretprobe trampoline is hit
	338	*/
2b67fc46 HC	339	static int __kprobes trampoline_probe_handler(struct kprobe *p,
2b67fc46 HC	340	struct pt_regs *regs)
4ba069b8 MG	341	{
4ba069b8 MG	342	struct kretprobe_instance *ri = NULL;
99219a3f	343	struct hlist_head *head, empty_rp;
4ba069b8 MG	344	struct hlist_node node, tmp;
	345	unsigned long flags, orig_ret_address = 0;
	346	unsigned long trampoline_address = (unsigned long)&kretprobe_trampoline;
	347
99219a3f	348	INIT_HLIST_HEAD(&empty_rp);
ef53d9c5	349	kretprobe_hash_lock(current, &head, &flags);
4ba069b8 MG	350
	351	/*
	352	* It is possible to have multiple instances associated with a given
	353	* task either because an multiple functions in the call path
025dfdaf	354	* have a return probe installed on them, and/or more than one return
4ba069b8 MG	355	* return probe was registered for a target function.
	356	*
	357	* We can handle this because:
	358	* - instances are always inserted at the head of the list
	359	* - when multiple return probes are registered for the same
	360	* function, the first instance's ret_addr will point to the
	361	* real return address, and all the rest will point to
	362	* kretprobe_trampoline
	363	*/
	364	hlist_for_each_entry_safe(ri, node, tmp, head, hlist) {
	365	if (ri->task != current)
	366	/* another task is sharing our hash bucket */
	367	continue;
	368
	369	if (ri->rp && ri->rp->handler)
	370	ri->rp->handler(ri, regs);
	371
	372	orig_ret_address = (unsigned long)ri->ret_addr;
99219a3f	373	recycle_rp_inst(ri, &empty_rp);
4ba069b8 MG	374
	375	if (orig_ret_address != trampoline_address) {
	376	/*
	377	* This is the real return address. Any other
	378	* instances associated with this task are for
	379	* other calls deeper on the call stack
	380	*/
	381	break;
	382	}
	383	}
a5a60a2b	384	kretprobe_assert(ri, orig_ret_address, trampoline_address);
4ba069b8 MG	385	regs->psw.addr = orig_ret_address \| PSW_ADDR_AMODE;
	386
	387	reset_current_kprobe();
ef53d9c5	388	kretprobe_hash_unlock(current, &flags);
4ba069b8 MG	389	preempt_enable_no_resched();
4ba069b8 MG	390
99219a3f	391	hlist_for_each_entry_safe(ri, node, tmp, &empty_rp, hlist) {
	392	hlist_del(&ri->hlist);
	393	kfree(ri);
	394	}
4ba069b8 MG	395	/*
	396	* By returning a non-zero value, we are telling
	397	* kprobe_handler() that we don't want the post_handler
	398	* to run (and have re-enabled preemption)
	399	*/
	400	return 1;
	401	}
	402
	403	/*
	404	* Called after single-stepping. p->addr is the address of the
	405	* instruction whose first byte has been replaced by the "breakpoint"
	406	* instruction. To avoid the SMP problems that can occur when we
	407	* temporarily put back the original opcode to single-step, we
	408	* single-stepped a copy of the instruction. The address of this
	409	* copy is p->ainsn.insn.
	410	*/
	411	static void __kprobes resume_execution(struct kprobe p, struct pt_regs regs)
	412	{
	413	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
	414
	415	regs->psw.addr &= PSW_ADDR_INSN;
	416
	417	if (p->ainsn.fixup & FIXUP_PSW_NORMAL)
	418	regs->psw.addr = (unsigned long)p->addr +
	419	((unsigned long)regs->psw.addr -
	420	(unsigned long)p->ainsn.insn);
	421
	422	if (p->ainsn.fixup & FIXUP_BRANCH_NOT_TAKEN)
	423	if ((unsigned long)regs->psw.addr -
	424	(unsigned long)p->ainsn.insn == p->ainsn.ilen)
	425	regs->psw.addr = (unsigned long)p->addr + p->ainsn.ilen;
	426
	427	if (p->ainsn.fixup & FIXUP_RETURN_REGISTER)
	428	regs->gprs[p->ainsn.reg] = ((unsigned long)p->addr +
	429	(regs->gprs[p->ainsn.reg] -
	430	(unsigned long)p->ainsn.insn))
	431	\| PSW_ADDR_AMODE;
	432
	433	regs->psw.addr \|= PSW_ADDR_AMODE;
	434	/* turn off PER mode */
	435	regs->psw.mask &= ~PSW_MASK_PER;
	436	/* Restore the original per control regs */
	437	__ctl_load(kcb->kprobe_saved_ctl, 9, 11);
	438	regs->psw.mask \|= kcb->kprobe_saved_imask;
	439	}
	440
	441	static int __kprobes post_kprobe_handler(struct pt_regs *regs)
	442	{
	443	struct kprobe *cur = kprobe_running();
	444	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
	445
	446	if (!cur)
	447	return 0;
	448
	449	if ((kcb->kprobe_status != KPROBE_REENTER) && cur->post_handler) {
	450	kcb->kprobe_status = KPROBE_HIT_SSDONE;
	451	cur->post_handler(cur, regs, 0);
	452	}
	453
	454	resume_execution(cur, regs);
	455
	456	/Restore back the original saved kprobes variables and continue. /
	457	if (kcb->kprobe_status == KPROBE_REENTER) {
	458	restore_previous_kprobe(kcb);
459	goto out;
460	}
461	reset_current_kprobe();
462	out:
463	preempt_enable_no_resched();
464
465	/*
466	* if somebody else is singlestepping across a probe point, psw mask
467	* will have PER set, in which case, continue the remaining processing
468	* of do_single_step, as if this is not a probe hit.
469	*/
470	if (regs->psw.mask & PSW_MASK_PER) {
471	return 0;
472	}
473
474	return 1;
475	}
476
33464e3b	477	int __kprobes kprobe_fault_handler(struct pt_regs *regs, int trapnr)
4ba069b8 MG	478	{
	479	struct kprobe *cur = kprobe_running();
	480	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
	481	const struct exception_table_entry *entry;
	482
	483	switch(kcb->kprobe_status) {
	484	case KPROBE_SWAP_INST:
	485	/* We are here because the instruction replacement failed */
	486	return 0;
	487	case KPROBE_HIT_SS:
	488	case KPROBE_REENTER:
	489	/*
	490	* We are here because the instruction being single
	491	* stepped caused a page fault. We reset the current
	492	* kprobe and the nip points back to the probe address
	493	* and allow the page fault handler to continue as a
	494	* normal page fault.
	495	*/
	496	regs->psw.addr = (unsigned long)cur->addr \| PSW_ADDR_AMODE;
	497	regs->psw.mask &= ~PSW_MASK_PER;
	498	regs->psw.mask \|= kcb->kprobe_saved_imask;
	499	if (kcb->kprobe_status == KPROBE_REENTER)
	500	restore_previous_kprobe(kcb);
	501	else
	502	reset_current_kprobe();
	503	preempt_enable_no_resched();
	504	break;
	505	case KPROBE_HIT_ACTIVE:
	506	case KPROBE_HIT_SSDONE:
	507	/*
	508	* We increment the nmissed count for accounting,
	509	* we can also use npre/npostfault count for accouting
	510	* these specific fault cases.
	511	*/
	512	kprobes_inc_nmissed_count(cur);
	513
	514	/*
	515	* We come here because instructions in the pre/post
	516	* handler caused the page_fault, this could happen
	517	* if handler tries to access user space by
	518	* copy_from_user(), get_user() etc. Let the
	519	* user-specified handler try to fix it first.
	520	*/
	521	if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr))
	522	return 1;
	523
	524	/*
	525	* In case the user-specified fault handler returned
	526	* zero, try to fix up.
	527	*/
	528	entry = search_exception_tables(regs->psw.addr & PSW_ADDR_INSN);
	529	if (entry) {
	530	regs->psw.addr = entry->fixup \| PSW_ADDR_AMODE;
	531	return 1;
	532	}
	533
	534	/*
	535	* fixup_exception() could not handle it,
	536	* Let do_page_fault() fix it.
	537	*/
	538	break;
	539	default:
	540	break;
	541	}
542	return 0;
543	}
544
545	/*
546	* Wrapper routine to for handling exceptions.
547	*/
548	int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
549	unsigned long val, void *data)
550	{
551	struct die_args args = (struct die_args )data;
552	int ret = NOTIFY_DONE;
553
554	switch (val) {
555	case DIE_BPT:
556	if (kprobe_handler(args->regs))
557	ret = NOTIFY_STOP;
558	break;
559	case DIE_SSTEP:
560	if (post_kprobe_handler(args->regs))
561	ret = NOTIFY_STOP;
562	break;
563	case DIE_TRAP:
4ba069b8 MG	564	/* kprobe_running() needs smp_processor_id() */
	565	preempt_disable();
	566	if (kprobe_running() &&
	567	kprobe_fault_handler(args->regs, args->trapnr))
	568	ret = NOTIFY_STOP;
	569	preempt_enable();
	570	break;
	571	default:
	572	break;
	573	}
	574	return ret;
	575	}
	576
	577	int __kprobes setjmp_pre_handler(struct kprobe p, struct pt_regs regs)
	578	{
	579	struct jprobe *jp = container_of(p, struct jprobe, kp);
	580	unsigned long addr;
	581	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
	582
	583	memcpy(&kcb->jprobe_saved_regs, regs, sizeof(struct pt_regs));
	584
	585	/* setup return addr to the jprobe handler routine */
	586	regs->psw.addr = (unsigned long)(jp->entry) \| PSW_ADDR_AMODE;
	587
	588	/* r14 is the function return address */
	589	kcb->jprobe_saved_r14 = (unsigned long)regs->gprs[14];
	590	/* r15 is the stack pointer */
	591	kcb->jprobe_saved_r15 = (unsigned long)regs->gprs[15];
	592	addr = (unsigned long)kcb->jprobe_saved_r15;
	593
	594	memcpy(kcb->jprobes_stack, (kprobe_opcode_t *) addr,
	595	MIN_STACK_SIZE(addr));
	596	return 1;
	597	}
	598
	599	void __kprobes jprobe_return(void)
	600	{
	601	asm volatile(".word 0x0002");
	602	}
	603
	604	void __kprobes jprobe_return_end(void)
	605	{
	606	asm volatile("bcr 0,0");
	607	}
	608
	609	int __kprobes longjmp_break_handler(struct kprobe p, struct pt_regs regs)
	610	{
	611	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
	612	unsigned long stack_addr = (unsigned long)(kcb->jprobe_saved_r15);
	613
	614	/* Put the regs back */
	615	memcpy(regs, &kcb->jprobe_saved_regs, sizeof(struct pt_regs));
	616	/* put the stack back */
	617	memcpy((kprobe_opcode_t *) stack_addr, kcb->jprobes_stack,
	618	MIN_STACK_SIZE(stack_addr));
	619	preempt_enable_no_resched();
	620	return 1;
	621	}
	622
	623	static struct kprobe trampoline_p = {
	624	.addr = (kprobe_opcode_t *) & kretprobe_trampoline,
	625	.pre_handler = trampoline_probe_handler
	626	};
	627
628	int __init arch_init_kprobes(void)
629	{
630	return register_kprobe(&trampoline_p);
631	}
bf8f6e5b AM	632
	633	int __kprobes arch_trampoline_kprobe(struct kprobe *p)
	634	{
	635	if (p->addr == (kprobe_opcode_t *) & kretprobe_trampoline)
	636	return 1;
	637	return 0;
	638	}