[for-next][PATCH 05/27] function_graph: Handle tail calls for stack unwinding

[Steven Rostedt]

From: "Masami Hiramatsu (Google)" <mhiramat@xxxxxxxxxx>

For the tail-call, there would be 2 or more ftrace_ret_stacks on the
ret_stack, which records "return_to_handler" as the return address except
for the last one.  But on the real stack, there should be 1 entry because
tail-call reuses the return address on the stack and jump to the next
function.

In ftrace_graph_ret_addr() that is used for stack unwinding, skip tail
calls as a real stack unwinder would do.

Link: https://lore.kernel.org/linux-trace-kernel/171509096221.162236.8806372072523195752.stgit@devnote2
Link: https://lore.kernel.org/linux-trace-kernel/20240603190821.717065217@xxxxxxxxxxx

Cc: Mark Rutland <mark.rutland@xxxxxxx>
Cc: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Cc: Alexei Starovoitov <alexei.starovoitov@xxxxxxxxx>
Cc: Florent Revest <revest@xxxxxxxxxxxx>
Cc: Martin KaFai Lau <martin.lau@xxxxxxxxx>
Cc: bpf <bpf@xxxxxxxxxxxxxxx>
Cc: Sven Schnelle <svens@xxxxxxxxxxxxx>
Cc: Alexei Starovoitov <ast@xxxxxxxxxx>
Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
Cc: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
Cc: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
Cc: Alan Maguire <alan.maguire@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Guo Ren <guoren@xxxxxxxxxx>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@xxxxxxxxxx>
Signed-off-by: Steven Rostedt (Google) <rostedt@xxxxxxxxxxx>
---
 kernel/trace/fgraph.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)

diff --git a/kernel/trace/fgraph.c b/kernel/trace/fgraph.c
index aae51f746828..8de2a2662281 100644
--- a/kernel/trace/fgraph.c
+++ b/kernel/trace/fgraph.c
@@ -594,16 +594,26 @@ unsigned long ftrace_graph_ret_addr(struct task_struct *task, int *idx,
 				    unsigned long ret, unsigned long *retp)
 {
 	struct ftrace_ret_stack *ret_stack;
+	unsigned long return_handler = (unsigned long)dereference_kernel_function_descriptor(return_to_handler);
 	int i = task->curr_ret_stack;
 
-	if (ret != (unsigned long)dereference_kernel_function_descriptor(return_to_handler))
+	if (ret != return_handler)
 		return ret;
 
 	while (i > 0) {
 		ret_stack = get_ret_stack(current, i, &i);
 		if (!ret_stack)
 			break;
-		if (ret_stack->retp == retp)
+		/*
+		 * For the tail-call, there would be 2 or more ftrace_ret_stacks on
+		 * the ret_stack, which records "return_to_handler" as the return
+		 * address except for the last one.
+		 * But on the real stack, there should be 1 entry because tail-call
+		 * reuses the return address on the stack and jump to the next function.
+		 * Thus we will continue to find real return address.
+		 */
+		if (ret_stack->retp == retp &&
+		    ret_stack->ret != return_handler)
 			return ret_stack->ret;
 	}
 
@@ -614,10 +624,11 @@ unsigned long ftrace_graph_ret_addr(struct task_struct *task, int *idx,
 				    unsigned long ret, unsigned long *retp)
 {
 	struct ftrace_ret_stack *ret_stack;
+	unsigned long return_handler = (unsigned long)dereference_kernel_function_descriptor(return_to_handler);
 	int offset = task->curr_ret_stack;
 	int i;
 
-	if (ret != (unsigned long)dereference_kernel_function_descriptor(return_to_handler))
+	if (ret != return_handler)
 		return ret;
 
 	if (!idx)
@@ -626,6 +637,8 @@ unsigned long ftrace_graph_ret_addr(struct task_struct *task, int *idx,
 	i = *idx;
 	do {
 		ret_stack = get_ret_stack(task, offset, &offset);
+		if (ret_stack && ret_stack->ret == return_handler)
+			continue;
 		i--;
 	} while (i >= 0 && ret_stack);
 
-- 
2.43.0