Re: [PATCH v20 00/19] tracing: fprobe: function_graph: Multi-function graph and fprobe on fgraph

On Fri, 6 Dec 2024 17:20:03 +0900
Masami Hiramatsu (Google) [off-list ref] wrote:

Hi,

Sorry, I found a problem on arm64 on qemu. Let me recheck it.

[  592.422044]     # test_fprobe_entry: EXPECTATION FAILED at lib/test_fprobe.c:38
[  592.422044]     Expected (preempt_count() == 0 && !({ unsigned long _flags; do { ({ unsigned long __dummy; typeof(_flags) __dummy2; (void)(&__dummy == &__dummy2); 1; }); _flags = arch_local_save_flags(); } while (0); ({ ({ unsigned long __dummy; typeof(_flags) __dummy2; (void)(&__dummy == &__dummy2); 1; }); arch_irqs_disabled_flags(_flags); }); })) to be false, but is true

This checks the handler is called with preempt off.

On x86_64, the ftrace_graph_func calls function_graph_enter_regs() with
ftrace_test_recursion_trylock() as below;

void ftrace_graph_func(unsigned long ip, unsigned long parent_ip,
		       struct ftrace_ops *op, struct ftrace_regs *fregs)
{
	struct pt_regs *regs = &arch_ftrace_regs(fregs)->regs;
	unsigned long *stack = (unsigned long *)kernel_stack_pointer(regs);
	unsigned long return_hooker = (unsigned long)&return_to_handler;
	unsigned long *parent = (unsigned long *)stack;
	int bit;

	if (unlikely(skip_ftrace_return()))
		return;

	bit = ftrace_test_recursion_trylock(ip, *parent);
	if (bit < 0)
		return;

	if (!function_graph_enter_regs(*parent, ip, 0, parent, fregs))
		*parent = return_hooker;

	ftrace_test_recursion_unlock(bit);
}

However, arm64 version does not;

void ftrace_graph_func(unsigned long ip, unsigned long parent_ip,
		       struct ftrace_ops *op, struct ftrace_regs *fregs)
{
	unsigned long return_hooker = (unsigned long)&return_to_handler;
	unsigned long frame_pointer = arch_ftrace_regs(fregs)->fp;
	unsigned long *parent = &arch_ftrace_regs(fregs)->lr;
	unsigned long old;

	if (unlikely(atomic_read(&current->tracing_graph_pause)))
		return;

	old = *parent;

	if (!function_graph_enter_regs(old, ip, frame_pointer,
				       (void *)frame_pointer, fregs)) {
		*parent = return_hooker;
	}
}

Is it a bug or intended?

Thank you,

-- 
Masami Hiramatsu (Google) [off-list ref]