tile: support kprobes on tilegx

This change includes support for Kprobes, Jprobes and Return Probes.

Reviewed-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Signed-off-by: Tony Lu <zlu@tilera.com>
Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>

14 files changed, 718 insertions, 5 deletions

diff --git a/arch/tile/Kconfig b/arch/tile/Kconfig
index d2277c4f949..616e2344ec6 100644
--- a/arch/tile/Kconfig
+++ b/arch/tile/Kconfig
@@ -125,6 +125,8 @@ config TILEGX
 	select HAVE_FUNCTION_GRAPH_TRACER
 	select HAVE_DYNAMIC_FTRACE
 	select HAVE_FTRACE_MCOUNT_RECORD
+	select HAVE_KPROBES
+	select HAVE_KRETPROBES
 
 config TILEPRO
 	def_bool !TILEGX
diff --git a/arch/tile/include/asm/Kbuild b/arch/tile/include/asm/Kbuild
index b17b9b8e53c..4c0b3c24c39 100644
--- a/arch/tile/include/asm/Kbuild
+++ b/arch/tile/include/asm/Kbuild
@@ -15,7 +15,6 @@ generic-y += ioctl.h
 generic-y += ioctls.h
 generic-y += ipcbuf.h
 generic-y += irq_regs.h
-generic-y += kdebug.h
 generic-y += local.h
 generic-y += msgbuf.h
 generic-y += mutex.h
diff --git a/arch/tile/include/asm/kdebug.h b/arch/tile/include/asm/kdebug.h
new file mode 100644
index 00000000000..5bbbfa904c2
--- /dev/null
+++ b/arch/tile/include/asm/kdebug.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2012 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   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, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ */
+
+#ifndef _ASM_TILE_KDEBUG_H
+#define _ASM_TILE_KDEBUG_H
+
+#include <linux/notifier.h>
+
+enum die_val {
+	DIE_OOPS = 1,
+	DIE_BREAK,
+	DIE_SSTEPBP,
+	DIE_PAGE_FAULT,
+	DIE_COMPILED_BPT
+};
+
+#endif /* _ASM_TILE_KDEBUG_H */
diff --git a/arch/tile/include/asm/kprobes.h b/arch/tile/include/asm/kprobes.h
new file mode 100644
index 00000000000..d8f9a83943b
--- /dev/null
+++ b/arch/tile/include/asm/kprobes.h
@@ -0,0 +1,79 @@
+/*
+ * arch/tile/include/asm/kprobes.h
+ *
+ * Copyright 2012 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   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, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ */
+
+#ifndef _ASM_TILE_KPROBES_H
+#define _ASM_TILE_KPROBES_H
+
+#include <linux/types.h>
+#include <linux/ptrace.h>
+#include <linux/percpu.h>
+
+#include <arch/opcode.h>
+
+#define __ARCH_WANT_KPROBES_INSN_SLOT
+#define MAX_INSN_SIZE			2
+
+#define kretprobe_blacklist_size 0
+
+typedef tile_bundle_bits kprobe_opcode_t;
+
+#define flush_insn_slot(p)						\
+	flush_icache_range((unsigned long)p->addr,			\
+			   (unsigned long)p->addr +			\
+			   (MAX_INSN_SIZE * sizeof(kprobe_opcode_t)))
+
+struct kprobe;
+
+/* Architecture specific copy of original instruction. */
+struct arch_specific_insn {
+	kprobe_opcode_t *insn;
+};
+
+struct prev_kprobe {
+	struct kprobe *kp;
+	unsigned long status;
+	unsigned long saved_pc;
+};
+
+#define MAX_JPROBES_STACK_SIZE 128
+#define MAX_JPROBES_STACK_ADDR \
+	(((unsigned long)current_thread_info()) + THREAD_SIZE - 32 \
+		- sizeof(struct pt_regs))
+
+#define MIN_JPROBES_STACK_SIZE(ADDR)					\
+	((((ADDR) + MAX_JPROBES_STACK_SIZE) > MAX_JPROBES_STACK_ADDR)	\
+		? MAX_JPROBES_STACK_ADDR - (ADDR)			\
+		: MAX_JPROBES_STACK_SIZE)
+
+/* per-cpu kprobe control block. */
+struct kprobe_ctlblk {
+	unsigned long kprobe_status;
+	unsigned long kprobe_saved_pc;
+	unsigned long jprobe_saved_sp;
+	struct prev_kprobe prev_kprobe;
+	struct pt_regs jprobe_saved_regs;
+	char jprobes_stack[MAX_JPROBES_STACK_SIZE];
+};
+
+extern tile_bundle_bits breakpoint2_insn;
+extern tile_bundle_bits breakpoint_insn;
+
+void arch_remove_kprobe(struct kprobe *);
+
+extern int kprobe_exceptions_notify(struct notifier_block *self,
+			     unsigned long val, void *data);
+
+#endif /* _ASM_TILE_KPROBES_H */
diff --git a/arch/tile/include/asm/ptrace.h b/arch/tile/include/asm/ptrace.h
index 73b681b566f..0d25c21bcd6 100644
--- a/arch/tile/include/asm/ptrace.h
+++ b/arch/tile/include/asm/ptrace.h
@@ -33,6 +33,7 @@ typedef unsigned long pt_reg_t;
 
 #ifndef __ASSEMBLY__
 
+#define regs_return_value(regs) ((regs)->regs[0])
 #define instruction_pointer(regs) ((regs)->pc)
 #define profile_pc(regs) instruction_pointer(regs)
 #define user_stack_pointer(regs) ((regs)->sp)
diff --git a/arch/tile/include/uapi/arch/opcode_tilegx.h b/arch/tile/include/uapi/arch/opcode_tilegx.h
index c14d02c8160..d76ff2db745 100644
--- a/arch/tile/include/uapi/arch/opcode_tilegx.h
+++ b/arch/tile/include/uapi/arch/opcode_tilegx.h
@@ -61,6 +61,7 @@ typedef tilegx_bundle_bits tile_bundle_bits;
 #define TILE_BUNDLE_ALIGNMENT_IN_BYTES TILEGX_BUNDLE_ALIGNMENT_IN_BYTES
 #define TILE_LOG2_BUNDLE_ALIGNMENT_IN_BYTES \
   TILEGX_LOG2_BUNDLE_ALIGNMENT_IN_BYTES
+#define TILE_BPT_BUNDLE TILEGX_BPT_BUNDLE
 
 /* 64-bit pattern for a { bpt ; nop } bundle. */
 #define TILEGX_BPT_BUNDLE 0x286a44ae51485000ULL
diff --git a/arch/tile/include/uapi/arch/opcode_tilepro.h b/arch/tile/include/uapi/arch/opcode_tilepro.h
index 71b763b8ce8..4451cff1a86 100644
--- a/arch/tile/include/uapi/arch/opcode_tilepro.h
+++ b/arch/tile/include/uapi/arch/opcode_tilepro.h
@@ -71,6 +71,7 @@ typedef tilepro_bundle_bits tile_bundle_bits;
 #define TILE_BUNDLE_ALIGNMENT_IN_BYTES TILEPRO_BUNDLE_ALIGNMENT_IN_BYTES
 #define TILE_LOG2_BUNDLE_ALIGNMENT_IN_BYTES \
   TILEPRO_LOG2_BUNDLE_ALIGNMENT_IN_BYTES
+#define TILE_BPT_BUNDLE TILEPRO_BPT_BUNDLE
 
 /* 64-bit pattern for a { bpt ; nop } bundle. */
 #define TILEPRO_BPT_BUNDLE 0x400b3cae70166000ULL
diff --git a/arch/tile/kernel/Makefile b/arch/tile/kernel/Makefile
index 2e6eaa11b60..b7c8b5e19d5 100644
--- a/arch/tile/kernel/Makefile
+++ b/arch/tile/kernel/Makefile
@@ -28,5 +28,6 @@ endif
 obj-$(CONFIG_TILE_USB)		+= usb.o
 obj-$(CONFIG_TILE_HVGLUE_TRACE)	+= hvglue_trace.o
 obj-$(CONFIG_FUNCTION_TRACER)	+= ftrace.o mcount_64.o
+obj-$(CONFIG_KPROBES)		+= kprobes.o
 
 obj-y				+= vdso/
diff --git a/arch/tile/kernel/kprobes.c b/arch/tile/kernel/kprobes.c
new file mode 100644
index 00000000000..27cdcacbe81
--- /dev/null
+++ b/arch/tile/kernel/kprobes.c
@@ -0,0 +1,528 @@
+/*
+ * arch/tile/kernel/kprobes.c
+ * Kprobes on TILE-Gx
+ *
+ * Some portions copied from the MIPS version.
+ *
+ * Copyright (C) IBM Corporation, 2002, 2004
+ * Copyright 2006 Sony Corp.
+ * Copyright 2010 Cavium Networks
+ *
+ * Copyright 2012 Tilera Corporation. All Rights Reserved.
+ *
+ *   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, version 2.
+ *
+ *   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, GOOD TITLE or
+ *   NON INFRINGEMENT.  See the GNU General Public License for
+ *   more details.
+ */
+
+#include <linux/kprobes.h>
+#include <linux/kdebug.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include <asm/cacheflush.h>
+
+#include <arch/opcode.h>
+
+DEFINE_PER_CPU(struct kprobe *, current_kprobe) = NULL;
+DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
+
+tile_bundle_bits breakpoint_insn = TILEGX_BPT_BUNDLE;
+tile_bundle_bits breakpoint2_insn = TILEGX_BPT_BUNDLE | DIE_SSTEPBP;
+
+/*
+ * Check whether instruction is branch or jump, or if executing it
+ * has different results depending on where it is executed (e.g. lnk).
+ */
+static int __kprobes insn_has_control(kprobe_opcode_t insn)
+{
+	if (get_Mode(insn) != 0) {   /* Y-format bundle */
+		if (get_Opcode_Y1(insn) != RRR_1_OPCODE_Y1 ||
+		    get_RRROpcodeExtension_Y1(insn) != UNARY_RRR_1_OPCODE_Y1)
+			return 0;
+
+		switch (get_UnaryOpcodeExtension_Y1(insn)) {
+		case JALRP_UNARY_OPCODE_Y1:
+		case JALR_UNARY_OPCODE_Y1:
+		case JRP_UNARY_OPCODE_Y1:
+		case JR_UNARY_OPCODE_Y1:
+		case LNK_UNARY_OPCODE_Y1:
+			return 1;
+		default:
+			return 0;
+		}
+	}
+
+	switch (get_Opcode_X1(insn)) {
+	case BRANCH_OPCODE_X1:	/* branch instructions */
+	case JUMP_OPCODE_X1:	/* jump instructions: j and jal */
+		return 1;
+
+	case RRR_0_OPCODE_X1:   /* other jump instructions */
+		if (get_RRROpcodeExtension_X1(insn) != UNARY_RRR_0_OPCODE_X1)
+			return 0;
+		switch (get_UnaryOpcodeExtension_X1(insn)) {
+		case JALRP_UNARY_OPCODE_X1:
+		case JALR_UNARY_OPCODE_X1:
+		case JRP_UNARY_OPCODE_X1:
+		case JR_UNARY_OPCODE_X1:
+		case LNK_UNARY_OPCODE_X1:
+			return 1;
+		default:
+			return 0;
+		}
+	default:
+		return 0;
+	}
+}
+
+int __kprobes arch_prepare_kprobe(struct kprobe *p)
+{
+	unsigned long addr = (unsigned long)p->addr;
+
+	if (addr & (sizeof(kprobe_opcode_t) - 1))
+		return -EINVAL;
+
+	if (insn_has_control(*p->addr)) {
+		pr_notice("Kprobes for control instructions are not "
+			  "supported\n");
+		return -EINVAL;
+	}
+
+	/* insn: must be on special executable page on tile. */
+	p->ainsn.insn = get_insn_slot();
+	if (!p->ainsn.insn)
+		return -ENOMEM;
+
+	/*
+	 * In the kprobe->ainsn.insn[] array we store the original
+	 * instruction at index zero and a break trap instruction at
+	 * index one.
+	 */
+	memcpy(&p->ainsn.insn[0], p->addr, sizeof(kprobe_opcode_t));
+	p->ainsn.insn[1] = breakpoint2_insn;
+	p->opcode = *p->addr;
+
+	return 0;
+}
+
+void __kprobes arch_arm_kprobe(struct kprobe *p)
+{
+	unsigned long addr_wr;
+
+	/* Operate on writable kernel text mapping. */
+	addr_wr = (unsigned long)p->addr - MEM_SV_START + PAGE_OFFSET;
+
+	if (probe_kernel_write((void *)addr_wr, &breakpoint_insn,
+		sizeof(breakpoint_insn)))
+		pr_err("%s: failed to enable kprobe\n", __func__);
+
+	smp_wmb();
+	flush_insn_slot(p);
+}
+
+void __kprobes arch_disarm_kprobe(struct kprobe *kp)
+{
+	unsigned long addr_wr;
+
+	/* Operate on writable kernel text mapping. */
+	addr_wr = (unsigned long)kp->addr - MEM_SV_START + PAGE_OFFSET;
+
+	if (probe_kernel_write((void *)addr_wr, &kp->opcode,
+		sizeof(kp->opcode)))
+		pr_err("%s: failed to enable kprobe\n", __func__);
+
+	smp_wmb();
+	flush_insn_slot(kp);
+}
+
+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 save_previous_kprobe(struct kprobe_ctlblk *kcb)
+{
+	kcb->prev_kprobe.kp = kprobe_running();
+	kcb->prev_kprobe.status = kcb->kprobe_status;
+	kcb->prev_kprobe.saved_pc = kcb->kprobe_saved_pc;
+}
+
+static void __kprobes restore_previous_kprobe(struct kprobe_ctlblk *kcb)
+{
+	__this_cpu_write(current_kprobe, kcb->prev_kprobe.kp);
+	kcb->kprobe_status = kcb->prev_kprobe.status;
+	kcb->kprobe_saved_pc = kcb->prev_kprobe.saved_pc;
+}
+
+static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs,
+			struct kprobe_ctlblk *kcb)
+{
+	__this_cpu_write(current_kprobe, p);
+	kcb->kprobe_saved_pc = regs->pc;
+}
+
+static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs)
+{
+	/* Single step inline if the instruction is a break. */
+	if (p->opcode == breakpoint_insn ||
+	    p->opcode == breakpoint2_insn)
+		regs->pc = (unsigned long)p->addr;
+	else
+		regs->pc = (unsigned long)&p->ainsn.insn[0];
+}
+
+static int __kprobes kprobe_handler(struct pt_regs *regs)
+{
+	struct kprobe *p;
+	int ret = 0;
+	kprobe_opcode_t *addr;
+	struct kprobe_ctlblk *kcb;
+
+	addr = (kprobe_opcode_t *)regs->pc;
+
+	/*
+	 * 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[0] == breakpoint_insn) {
+				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 {
+			if (*addr != breakpoint_insn) {
+				/*
+				 * The breakpoint instruction was removed by
+				 * another cpu right after we hit, no further
+				 * handling of this interrupt is appropriate.
+				 */
+				ret = 1;
+				goto no_kprobe;
+			}
+			p = __this_cpu_read(current_kprobe);
+			if (p->break_handler && p->break_handler(p, regs))
+				goto ss_probe;
+		}
+		goto no_kprobe;
+	}
+
+	p = get_kprobe(addr);
+	if (!p) {
+		if (*addr != breakpoint_insn) {
+			/*
+			 * The breakpoint instruction was removed right
+			 * after we hit it.  Another cpu has removed
+			 * either a probepoint or a debugger breakpoint
+			 * at this address.  In either case, no further
+			 * handling of this interrupt is appropriate.
+			 */
+			ret = 1;
+		}
+		/* Not one of ours: let kernel handle it. */
+		goto no_kprobe;
+	}
+
+	set_current_kprobe(p, regs, kcb);
+	kcb->kprobe_status = KPROBE_HIT_ACTIVE;
+
+	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;
+}
+
+/*
+ * Called after single-stepping.  p->addr is the address of the
+ * instruction that has been replaced by the breakpoint. 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.
+ *
+ * This function prepares to return from the post-single-step
+ * breakpoint trap.
+ */
+static void __kprobes resume_execution(struct kprobe *p,
+				       struct pt_regs *regs,
+				       struct kprobe_ctlblk *kcb)
+{
+	unsigned long orig_pc = kcb->kprobe_saved_pc;
+	regs->pc = orig_pc + 8;
+}
+
+static inline int 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, kcb);
+
+	/* 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();
+
+	return 1;
+}
+
+static inline int kprobe_fault_handler(struct pt_regs *regs, int trapnr)
+{
+	struct kprobe *cur = kprobe_running();
+	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+
+	if (cur->fault_handler && cur->fault_handler(cur, regs, trapnr))
+		return 1;
+
+	if (kcb->kprobe_status & KPROBE_HIT_SS) {
+		/*
+		 * We are here because the instruction being single
+		 * stepped caused a page fault. We reset the current
+		 * kprobe and the ip points back to the probe address
+		 * and allow the page fault handler to continue as a
+		 * normal page fault.
+		 */
+		resume_execution(cur, regs, kcb);
+		reset_current_kprobe();
+		preempt_enable_no_resched();
+	}
+	return 0;
+}
+
+/*
+ * Wrapper routine 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_BREAK:
+		if (kprobe_handler(args->regs))
+			ret = NOTIFY_STOP;
+		break;
+	case DIE_SSTEPBP:
+		if (post_kprobe_handler(args->regs))
+			ret = NOTIFY_STOP;
+		break;
+	case DIE_PAGE_FAULT:
+		/* 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);
+	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+
+	kcb->jprobe_saved_regs = *regs;
+	kcb->jprobe_saved_sp = regs->sp;
+
+	memcpy(kcb->jprobes_stack, (void *)kcb->jprobe_saved_sp,
+	       MIN_JPROBES_STACK_SIZE(kcb->jprobe_saved_sp));
+
+	regs->pc = (unsigned long)(jp->entry);
+
+	return 1;
+}
+
+/* Defined in the inline asm below. */
+void jprobe_return_end(void);
+
+void __kprobes jprobe_return(void)
+{
+	asm volatile(
+		"bpt\n\t"
+		".globl jprobe_return_end\n"
+		"jprobe_return_end:\n");
+}
+
+int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs)
+{
+	struct kprobe_ctlblk *kcb = get_kprobe_ctlblk();
+
+	if (regs->pc >= (unsigned long)jprobe_return &&
+	    regs->pc <= (unsigned long)jprobe_return_end) {
+		*regs = kcb->jprobe_saved_regs;
+		memcpy((void *)kcb->jprobe_saved_sp, kcb->jprobes_stack,
+		       MIN_JPROBES_STACK_SIZE(kcb->jprobe_saved_sp));
+		preempt_enable_no_resched();
+
+		return 1;
+	}
+	return 0;
+}
+
+/*
+ * 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(
+		"nop\n\t"
+		".global kretprobe_trampoline\n"
+		"kretprobe_trampoline:\n\t"
+		"nop\n\t"
+		: : : "memory");
+}
+
+void kretprobe_trampoline(void);
+
+void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri,
+				      struct pt_regs *regs)
+{
+	ri->ret_addr = (kprobe_opcode_t *) regs->lr;
+
+	/* Replace the return addr with trampoline addr */
+	regs->lr = (unsigned long)kretprobe_trampoline;
+}
+
+/*
+ * 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 *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 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, 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);
+	instruction_pointer(regs) = orig_ret_address;
+
+	reset_current_kprobe();
+	kretprobe_hash_unlock(current, &flags);
+	preempt_enable_no_resched();
+
+	hlist_for_each_entry_safe(ri, 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;
+}
+
+int __kprobes arch_trampoline_kprobe(struct kprobe *p)
+{
+	if (p->addr == (kprobe_opcode_t *)kretprobe_trampoline)
+		return 1;
+
+	return 0;
+}
+
+static struct kprobe trampoline_p = {
+	.addr = (kprobe_opcode_t *)kretprobe_trampoline,
+	.pre_handler = trampoline_probe_handler
+};
+
+int __init arch_init_kprobes(void)
+{
+	register_kprobe(&trampoline_p);
+	return 0;
+}
diff --git a/arch/tile/kernel/smp.c b/arch/tile/kernel/smp.c
index 6cc520d71d2..0ae1c594d88 100644
--- a/arch/tile/kernel/smp.c
+++ b/arch/tile/kernel/smp.c
@@ -20,6 +20,7 @@
 #include <linux/irq.h>
 #include <linux/module.h>
 #include <asm/cacheflush.h>
+#include <asm/homecache.h>
 
 HV_Topology smp_topology __write_once;
 EXPORT_SYMBOL(smp_topology);
@@ -167,9 +168,16 @@ static void ipi_flush_icache_range(void *info)
 void flush_icache_range(unsigned long start, unsigned long end)
 {
 	struct ipi_flush flush = { start, end };
-	preempt_disable();
-	on_each_cpu(ipi_flush_icache_range, &flush, 1);
-	preempt_enable();
+
+	/* If invoked with irqs disabled, we can not issue IPIs. */
+	if (irqs_disabled())
+		flush_remote(0, HV_FLUSH_EVICT_L1I, NULL, 0, 0, 0,
+			NULL, NULL, 0);
+	else {
+		preempt_disable();
+		on_each_cpu(ipi_flush_icache_range, &flush, 1);
+		preempt_enable();
+	}
 }
 
 
diff --git a/arch/tile/kernel/traps.c b/arch/tile/kernel/traps.c
index a1bbc5de4d0..cfff6f958d5 100644
--- a/arch/tile/kernel/traps.c
+++ b/arch/tile/kernel/traps.c
@@ -15,6 +15,7 @@
 #include <linux/sched.h>
 #include <linux/kernel.h>
 #include <linux/kprobes.h>
+#include <linux/kdebug.h>
 #include <linux/module.h>
 #include <linux/reboot.h>
 #include <linux/uaccess.h>
@@ -214,6 +215,43 @@ static const char *const int_name[] = {
 #endif
 };
 
+static int do_bpt(struct pt_regs *regs)
+{
+	unsigned long bundle, bcode, bpt;
+
+	bundle = *(unsigned long *)instruction_pointer(regs);
+
+	/*
+	 * bpt shoule be { bpt; nop }, which is 0x286a44ae51485000ULL.
+	 * we encode the unused least significant bits for other purpose.
+	 */
+	bpt = bundle & ~((1ULL << 12) - 1);
+	if (bpt != TILE_BPT_BUNDLE)
+		return 0;
+
+	bcode = bundle & ((1ULL << 12) - 1);
+	/*
+	 * notify the kprobe handlers, if instruction is likely to
+	 * pertain to them.
+	 */
+	switch (bcode) {
+	/* breakpoint_insn */
+	case 0:
+		notify_die(DIE_BREAK, "debug", regs, bundle,
+			INT_ILL, SIGTRAP);
+		break;
+	/* breakpoint2_insn */
+	case DIE_SSTEPBP:
+		notify_die(DIE_SSTEPBP, "single_step", regs, bundle,
+			INT_ILL, SIGTRAP);
+		break;
+	default:
+		return 0;
+	}
+
+	return 1;
+}
+
 void __kprobes do_trap(struct pt_regs *regs, int fault_num,
 		       unsigned long reason)
 {
@@ -221,6 +259,11 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num,
 	int signo, code;
 	unsigned long address = 0;
 	bundle_bits instr;
+	int is_kernel = !user_mode(regs);
+
+	/* Handle breakpoints, etc. */
+	if (is_kernel && fault_num == INT_ILL && do_bpt(regs))
+		return;
 
 	/* Re-enable interrupts, if they were previously enabled. */
 	if (!(regs->flags & PT_FLAGS_DISABLE_IRQ))
@@ -230,7 +273,7 @@ void __kprobes do_trap(struct pt_regs *regs, int fault_num,
 	 * If it hits in kernel mode and we can't fix it up, just exit the
 	 * current process and hope for the best.
 	 */
-	if (!user_mode(regs)) {
+	if (is_kernel) {
 		const char *name;
 		char buf[100];
 		if (fixup_exception(regs))  /* ILL_TRANS or UNALIGN_DATA */
diff --git a/arch/tile/kernel/vmlinux.lds.S b/arch/tile/kernel/vmlinux.lds.S
index 673d00afa83..aab995578dc 100644
--- a/arch/tile/kernel/vmlinux.lds.S
+++ b/arch/tile/kernel/vmlinux.lds.S
@@ -43,6 +43,7 @@ SECTIONS
     HEAD_TEXT
     SCHED_TEXT
     LOCK_TEXT
+    KPROBES_TEXT
     IRQENTRY_TEXT
     __fix_text_end = .;   /* tile-cpack won't rearrange before this */
     TEXT_TEXT
diff --git a/arch/tile/mm/fault.c b/arch/tile/mm/fault.c
index 502664ae1e6..64eec3f9584 100644
--- a/arch/tile/mm/fault.c
+++ b/arch/tile/mm/fault.c
@@ -34,6 +34,7 @@
 #include <linux/hugetlb.h>
 #include <linux/syscalls.h>
 #include <linux/uaccess.h>
+#include <linux/kdebug.h>
 
 #include <asm/pgalloc.h>
 #include <asm/sections.h>
@@ -721,6 +722,17 @@ void do_page_fault(struct pt_regs *regs, int fault_num,
 {
 	int is_page_fault;
 
+#ifdef CONFIG_KPROBES
+	/*
+	 * This is to notify the fault handler of the kprobes.  The
+	 * exception code is redundant as it is also carried in REGS,
+	 * but we pass it anyhow.
+	 */
+	if (notify_die(DIE_PAGE_FAULT, "page fault", regs, -1,
+		       regs->faultnum, SIGSEGV) == NOTIFY_STOP)
+		return;
+#endif
+
 #ifdef __tilegx__
 	/*
 	 * We don't need early do_page_fault_ics() support, since unlike
diff --git a/samples/kprobes/kprobe_example.c b/samples/kprobes/kprobe_example.c
index ebf5e0c368e..366db1a9fb6 100644
--- a/samples/kprobes/kprobe_example.c
+++ b/samples/kprobes/kprobe_example.c
@@ -37,6 +37,11 @@ static int handler_pre(struct kprobe *p, struct pt_regs *regs)
 			" status = 0x%lx\n",
 		p->addr, regs->cp0_epc, regs->cp0_status);
 #endif
+#ifdef CONFIG_TILEGX
+	printk(KERN_INFO "pre_handler: p->addr = 0x%p, pc = 0x%lx,"
+			" ex1 = 0x%lx\n",
+		p->addr, regs->pc, regs->ex1);
+#endif
 
 	/* A dump_stack() here will give a stack backtrace */
 	return 0;
@@ -58,6 +63,10 @@ static void handler_post(struct kprobe *p, struct pt_regs *regs,
 	printk(KERN_INFO "post_handler: p->addr = 0x%p, status = 0x%lx\n",
 		p->addr, regs->cp0_status);
 #endif
+#ifdef CONFIG_TILEGX
+	printk(KERN_INFO "post_handler: p->addr = 0x%p, ex1 = 0x%lx\n",
+		p->addr, regs->ex1);
+#endif
 }
 
 /*