[PATCH] x86_64 specific function return probes

The following patch adds the x86_64 architecture specific implementation
for function return probes.

Function return probes is a mechanism built on top of kprobes that allows
a caller to register a handler to be called when a given function exits.
For example, to instrument the return path of sys_mkdir:

static int sys_mkdir_exit(struct kretprobe_instance *i, struct pt_regs *regs)
{
	printk("sys_mkdir exited\n");
	return 0;
}
static struct kretprobe return_probe = {
	.handler = sys_mkdir_exit,
};

<inside setup function>

return_probe.kp.addr = (kprobe_opcode_t *) kallsyms_lookup_name("sys_mkdir");
if (register_kretprobe(&return_probe)) {
	printk(KERN_DEBUG "Unable to register return probe!\n");
	/* do error path */
}

<inside cleanup function>
unregister_kretprobe(&return_probe);

The way this works is that:

* At system initialization time, kernel/kprobes.c installs a kprobe
  on a function called kretprobe_trampoline() that is implemented in
  the arch/x86_64/kernel/kprobes.c  (More on this later)

* When a return probe is registered using register_kretprobe(),
  kernel/kprobes.c will install a kprobe on the first instruction of the
  targeted function with the pre handler set to arch_prepare_kretprobe()
  which is implemented in arch/x86_64/kernel/kprobes.c.

* arch_prepare_kretprobe() will prepare a kretprobe instance that stores:
  - nodes for hanging this instance in an empty or free list
  - a pointer to the return probe
  - the original return address
  - a pointer to the stack address

  With all this stowed away, arch_prepare_kretprobe() then sets the return
  address for the targeted function to a special trampoline function called
  kretprobe_trampoline() implemented in arch/x86_64/kernel/kprobes.c

* The kprobe completes as normal, with control passing back to the target
  function that executes as normal, and eventually returns to our trampoline
  function.

* Since a kprobe was installed on kretprobe_trampoline() during system
  initialization, control passes back to kprobes via the architecture
  specific function trampoline_probe_handler() which will lookup the
  instance in an hlist maintained by kernel/kprobes.c, and then call
  the handler function.

* When trampoline_probe_handler() is done, the kprobes infrastructure
  single steps the original instruction (in this case just a top), and
  then calls trampoline_post_handler().  trampoline_post_handler() then
  looks up the instance again, puts the instance back on the free list,
  and then makes a long jump back to the original return instruction.

So to recap, to instrument the exit path of a function this implementation
will cause four interruptions:

  - A breakpoint at the very beginning of the function allowing us to
    switch out the return address
  - A single step interruption to execute the original instruction that
    we replaced with the break instruction (normal kprobe flow)
  - A breakpoint in the trampoline function where our instrumented function
    returned to
  - A single step interruption to execute the original instruction that
    we replaced with the break instruction (normal kprobe flow)

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

2 files changed, 113 insertions, 1 deletions

diff --git a/arch/x86_64/kernel/kprobes.c b/arch/x86_64/kernel/kprobes.c
index f77f8a0..203672c 100644
--- a/arch/x86_64/kernel/kprobes.c
+++ b/arch/x86_64/kernel/kprobes.c
@@ -27,6 +27,8 @@
  *		<prasanna@in.ibm.com> adapted for x86_64
  * 2005-Mar	Roland McGrath <roland@redhat.com>
  *		Fixed to handle %rip-relative addressing mode correctly.
+ * 2005-May     Rusty Lynch <rusty.lynch@intel.com>
+ *              Added function return probes functionality
  */
 
 #include <linux/config.h>
@@ -240,6 +242,50 @@ static void prepare_singlestep(struct kprobe *p, struct pt_regs *regs)
 		regs->rip = (unsigned long)p->ainsn.insn;
 }
 
+struct task_struct  *arch_get_kprobe_task(void *ptr)
+{
+	return ((struct thread_info *) (((unsigned long) ptr) &
+					(~(THREAD_SIZE -1))))->task;
+}
+
+void arch_prepare_kretprobe(struct kretprobe *rp, struct pt_regs *regs)
+{
+	unsigned long *sara = (unsigned long *)regs->rsp;
+	struct kretprobe_instance *ri;
+	static void *orig_ret_addr;
+
+	/*
+	 * Save the return address when the return probe hits
+	 * the first time, and use it to populate the (krprobe
+	 * instance)->ret_addr for subsequent return probes at
+	 * the same addrress since stack address would have
+	 * the kretprobe_trampoline by then.
+	 */
+	if (((void*) *sara) != kretprobe_trampoline)
+		orig_ret_addr = (void*) *sara;
+
+	if ((ri = get_free_rp_inst(rp)) != NULL) {
+		ri->rp = rp;
+		ri->stack_addr = sara;
+		ri->ret_addr = orig_ret_addr;
+		add_rp_inst(ri);
+		/* Replace the return addr with trampoline addr */
+		*sara = (unsigned long) &kretprobe_trampoline;
+	} else {
+		rp->nmissed++;
+	}
+}
+
+void arch_kprobe_flush_task(struct task_struct *tk)
+{
+	struct kretprobe_instance *ri;
+	while ((ri = get_rp_inst_tsk(tk)) != NULL) {
+		*((unsigned long *)(ri->stack_addr)) =
+					(unsigned long) ri->ret_addr;
+		recycle_rp_inst(ri);
+	}
+}
+
 /*
  * Interrupts are disabled on entry as trap3 is an interrupt gate and they
  * remain disabled thorough out this function.
@@ -317,6 +363,55 @@ no_kprobe:
 }
 
 /*
+ * For function-return probes, init_kprobes() establishes a probepoint
+ * here. When a retprobed function returns, this probe is hit and
+ * trampoline_probe_handler() runs, calling the kretprobe's handler.
+ */
+ void kretprobe_trampoline_holder(void)
+ {
+ 	asm volatile (  ".global kretprobe_trampoline\n"
+ 			"kretprobe_trampoline: \n"
+ 			"nop\n");
+ }
+
+/*
+ * Called when we hit the probe point at kretprobe_trampoline
+ */
+int trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
+{
+	struct task_struct *tsk;
+	struct kretprobe_instance *ri;
+	struct hlist_head *head;
+	struct hlist_node *node;
+	unsigned long *sara = (unsigned long *)regs->rsp - 1;
+
+	tsk = arch_get_kprobe_task(sara);
+	head = kretprobe_inst_table_head(tsk);
+
+	hlist_for_each_entry(ri, node, head, hlist) {
+		if (ri->stack_addr == sara && ri->rp) {
+			if (ri->rp->handler)
+				ri->rp->handler(ri, regs);
+		}
+	}
+	return 0;
+}
+
+void trampoline_post_handler(struct kprobe *p, struct pt_regs *regs,
+						unsigned long flags)
+{
+	struct kretprobe_instance *ri;
+	/* RA already popped */
+	unsigned long *sara = ((unsigned long *)regs->rsp) - 1;
+
+	while ((ri = get_rp_inst(sara))) {
+		regs->rip = (unsigned long)ri->ret_addr;
+		recycle_rp_inst(ri);
+	}
+	regs->eflags &= ~TF_MASK;
+}
+
+/*
  * Called after single-stepping.  p->addr is the address of the
  * instruction whose first byte has been replaced by the "int 3"
  * instruction.  To avoid the SMP problems that can occur when we
@@ -404,7 +499,8 @@ int post_kprobe_handler(struct pt_regs *regs)
 	if (current_kprobe->post_handler)
 		current_kprobe->post_handler(current_kprobe, regs, 0);
 
-	resume_execution(current_kprobe, regs);
+	if (current_kprobe->post_handler != trampoline_post_handler)
+		resume_execution(current_kprobe, regs);
 	regs->eflags |= kprobe_saved_rflags;
 
 	unlock_kprobes();
diff --git a/arch/x86_64/kernel/process.c b/arch/x86_64/kernel/process.c
index dce8bab..e59d1f9 100644
--- a/arch/x86_64/kernel/process.c
+++ b/arch/x86_64/kernel/process.c
@@ -34,6 +34,7 @@
 #include <linux/ptrace.h>
 #include <linux/utsname.h>
 #include <linux/random.h>
+#include <linux/kprobes.h>
 
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
@@ -293,6 +294,14 @@ void exit_thread(void)
 {
 	struct task_struct *me = current;
 	struct thread_struct *t = &me->thread;
+
+	/*
+	 * Remove function-return probe instances associated with this task
+	 * and put them back on the free list. Do not insert an exit probe for
+	 * this function, it will be disabled by kprobe_flush_task if you do.
+	 */
+	kprobe_flush_task(me);
+
 	if (me->thread.io_bitmap_ptr) { 
 		struct tss_struct *tss = &per_cpu(init_tss, get_cpu());
 
@@ -312,6 +321,13 @@ void flush_thread(void)
 	struct task_struct *tsk = current;
 	struct thread_info *t = current_thread_info();
 
+	/*
+	 * Remove function-return probe instances associated with this task
+	 * and put them back on the free list. Do not insert an exit probe for
+	 * this function, it will be disabled by kprobe_flush_task if you do.
+	 */
+	kprobe_flush_task(tsk);
+
 	if (t->flags & _TIF_ABI_PENDING)
 		t->flags ^= (_TIF_ABI_PENDING | _TIF_IA32);