riscv: Add KPROBES_ON_FTRACE supported

This patch adds support for kprobes on ftrace call sites to avoids
much of the overhead with regular kprobes. Try it with simple
steps:

 echo 'p:myprobe sys_clone a0=%a0 a1=%a1 stack_val=+4($stack)' > /sys/kernel/de
bug/tracing/kprobe_events
 echo 1 > /sys/kernel/debug/tracing/events/kprobes/enable
 cat /sys/kernel/debug/tracing/trace
 tracer: nop

 entries-in-buffer/entries-written: 1/1   #P:1

                                _-----=> irqs-off
                               / _----=> need-resched
                              | / _---=> hardirq/softirq
                              || / _--=> preempt-depth
                              ||| /     delay
           TASK-PID     CPU#  ||||   TIMESTAMP  FUNCTION
              | |         |   ||||      |         |
              sh-92      [000] ....   369.899962: myprobe: (sys_clone+0x0/0x28) a0=0x1200011 a1=0x0 stack_val=0x201c20ffffffe0
 cat /sys/kernel/debug/kprobes/list
ffffffe00020b584  k  sys_clone+0x0    [FTRACE]
                                       ^^^^^^

Signed-off-by: Guo Ren <guoren@linux.alibaba.com>
Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>

arch/riscv/Kconfig

@@ -72,6 +72,7 @@ config RISCV
 	select HAVE_GENERIC_VDSO if MMU && 64BIT
 	select HAVE_IRQ_TIME_ACCOUNTING
 	select HAVE_KPROBES
+	select HAVE_KPROBES_ON_FTRACE
 	select HAVE_KRETPROBES
 	select HAVE_PCI
 	select HAVE_PERF_EVENTS

arch/riscv/kernel/probes/Makefile

@@ -1,4 +1,5 @@
 # SPDX-License-Identifier: GPL-2.0
 obj-$(CONFIG_KPROBES)		+= kprobes.o decode-insn.o simulate-insn.o
 obj-$(CONFIG_KPROBES)		+= kprobes_trampoline.o
+obj-$(CONFIG_KPROBES_ON_FTRACE)	+= ftrace.o
 CFLAGS_REMOVE_simulate-insn.o = $(CC_FLAGS_FTRACE)

arch/riscv/kernel/probes/ftrace.c

@@ -0,0 +1,53 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/kprobes.h>
+
+/* Ftrace callback handler for kprobes -- called under preepmt disabed */
+void kprobe_ftrace_handler(unsigned long ip, unsigned long parent_ip,
+			   struct ftrace_ops *ops, struct ftrace_regs *regs)
+{
+	struct kprobe *p;
+	struct kprobe_ctlblk *kcb;
+
+	p = get_kprobe((kprobe_opcode_t *)ip);
+	if (unlikely(!p) || kprobe_disabled(p))
+		return;
+
+	kcb = get_kprobe_ctlblk();
+	if (kprobe_running()) {
+		kprobes_inc_nmissed_count(p);
+	} else {
+		unsigned long orig_ip = instruction_pointer(&(regs->regs));
+
+		instruction_pointer_set(&(regs->regs), ip);
+
+		__this_cpu_write(current_kprobe, p);
+		kcb->kprobe_status = KPROBE_HIT_ACTIVE;
+		if (!p->pre_handler || !p->pre_handler(p, &(regs->regs))) {
+			/*
+			 * Emulate singlestep (and also recover regs->pc)
+			 * as if there is a nop
+			 */
+			instruction_pointer_set(&(regs->regs),
+				(unsigned long)p->addr + MCOUNT_INSN_SIZE);
+			if (unlikely(p->post_handler)) {
+				kcb->kprobe_status = KPROBE_HIT_SSDONE;
+				p->post_handler(p, &(regs->regs), 0);
+			}
+			instruction_pointer_set(&(regs->regs), orig_ip);
+		}
+
+		/*
+		 * If pre_handler returns !0, it changes regs->pc. We have to
+		 * skip emulating post_handler.
+		 */
+		__this_cpu_write(current_kprobe, NULL);
+	}
+}
+NOKPROBE_SYMBOL(kprobe_ftrace_handler);
+
+int arch_prepare_kprobe_ftrace(struct kprobe *p)
+{
+	p->ainsn.api.insn = NULL;
+	return 0;
+}