[PATCH v6 3/6] arm64: ftrace: fix a stack tracer's output under function graph tracer

Jungseok Lee jungseoklee85 at gmail.com
Wed Nov 25 03:48:20 PST 2015


On Nov 25, 2015, at 2:29 PM, AKASHI Takahiro wrote:
> On 11/24/2015 10:37 PM, Jungseok Lee wrote:
>> On Nov 18, 2015, at 3:43 PM, AKASHI Takahiro wrote:
>> 
>> Hi Akashi,
>> 
>>> Function graph tracer modifies a return address (LR) in a stack frame
>>> to hook a function return. This will result in many useless entries
>>> (return_to_handler) showing up in
>>> a) a stack tracer's output
>>> b) perf call graph (with perf record -g)
>>> c) dump_backtrace (at panic et al.)
>>> 
>>> For example, in case of a),
>>>  $ echo function_graph > /sys/kernel/debug/tracing/current_tracer
>>>  $ echo 1 > /proc/sys/kernel/stack_trace_enabled
>>>  $ cat /sys/kernel/debug/tracing/stack_trace
>>>        Depth    Size   Location    (54 entries)
>>>        -----    ----   --------
>>>  0)     4504      16   gic_raise_softirq+0x28/0x150
>>>  1)     4488      80   smp_cross_call+0x38/0xb8
>>>  2)     4408      48   return_to_handler+0x0/0x40
>>>  3)     4360      32   return_to_handler+0x0/0x40
>>>  ...
>>> 
>>> In case of b),
>>>  $ echo function_graph > /sys/kernel/debug/tracing/current_tracer
>>>  $ perf record -e mem:XXX:x -ag -- sleep 10
>>>  $ perf report
>>>                  ...
>>>                  |          |          |--0.22%-- 0x550f8
>>>                  |          |          |          0x10888
>>>                  |          |          |          el0_svc_naked
>>>                  |          |          |          sys_openat
>>>                  |          |          |          return_to_handler
>>>                  |          |          |          return_to_handler
>>>                  ...
>>> 
>>> In case of c),
>>>  $ echo function_graph > /sys/kernel/debug/tracing/current_tracer
>>>  $ echo c > /proc/sysrq-trigger
>>>  ...
>>>  Call trace:
>>>  [<ffffffc00044d3ac>] sysrq_handle_crash+0x24/0x30
>>>  [<ffffffc000092250>] return_to_handler+0x0/0x40
>>>  [<ffffffc000092250>] return_to_handler+0x0/0x40
>>>  ...
>>> 
>>> This patch replaces such entries with real addresses preserved in
>>> current->ret_stack[] at unwind_frame(). This way, we can cover all
>>> the cases.
>> 
>> I've observed a strange behavior when playing with case c). Call trace
>> is as follows when function_graph is not used.
>> 
>> Call trace:
>>  [<fffffe00003dc738>] sysrq_handle_crash+0x24/0x30          <- (1)
>>  [<fffffe00003dd2ac>] __handle_sysrq+0x128/0x19c            <- (2)
>>  [<fffffe00003dd730>] write_sysrq_trigger+0x60/0x74
>>  [<fffffe0000249fc4>] proc_reg_write+0x84/0xc0
>>  [<fffffe00001f2638>] __vfs_write+0x44/0x104                <- (3)
>>  [<fffffe00001f2e60>] vfs_write+0x98/0x1a8
>>  [<fffffe00001f3730>] SyS_write+0x50/0xb0
>>  [<fffffe00000939ec>] el0_svc_naked+0x20/0x28               <- (4)
>> 
>> When function_graph is set, some entries, such as do_mem_abort, are added
>> between (1) and (2). In addition, entries from (3) to (4) are not printed
>> out. As tracking down elements of ret_stack[], I realize dump_backtrace()
>> has been terminated before reaching to ret_stack[0]. Have you seen this
>> kind of behavior? I believe push & pop operations work correctly.
>> 
>> Please correct me if I'm wrong.
> 
> Oops, I mis-interpreted the result output.
> You are right.
> This can happen because the original dump_backtrace() wants to trace a stack
> from a function where an exception has taken place, sysrq_handle_crash(), while
> ret_stack[curr_ret_stack] doesn't point to that function, but the very top of
> traced functions in callchains, that is, __do_kernel_fault in your case, probably.

Yes, __do_kernel_fault.

> So it results in replacing entries of return_to_handler to wrong function addresses.
> 
> A fixup! patch attached below (informative only) fixes this issue by once tracing
> all the functions on a stack, but preventing a top few ones from being printed.
> But there is a tricky thing here: we have to use 'regs->pc' instead of frame.pc
> as a trapped function because, as I've already mentioned before, we always miss
> the function when walking through a stack from an exception handler to functions
> in a thread context.
> (Please note that we will introduce a dummy stack frame at interrupt, but not
> at exception.)

Thanks for clear explanation!

Since I'm not the only person to experience the above case, I report one more and
the last observation ;) 

PC and LR information is printed out when function_graph is set.

 PC is at sysrq_handle_crash+0x24/0x30
 LR is at sysrq_handle_crash+0x10/0x30

The logs are as follows when function_graph turns off.

 PC is at sysrq_handle_crash+0x24/0x30
 LR is at __handle_sysrq+0x128/0x19c

I think __show_regs() might have a similar problem when retrieving LR according to
the below informative patch. Thoughts?

>>> Signed-off-by: AKASHI Takahiro <takahiro.akashi at linaro.org>
>>> ---
>>> arch/arm64/include/asm/ftrace.h     |    2 ++
>>> arch/arm64/include/asm/stacktrace.h |    3 +++
>>> arch/arm64/kernel/perf_callchain.c  |    3 +++
>>> arch/arm64/kernel/process.c         |    3 +++
>>> arch/arm64/kernel/return_address.c  |    3 +++
>>> arch/arm64/kernel/stacktrace.c      |   17 +++++++++++++++++
>>> arch/arm64/kernel/time.c            |    3 +++
>>> arch/arm64/kernel/traps.c           |    3 +++
>>> 8 files changed, 37 insertions(+)
>>> 
>>> diff --git a/arch/arm64/include/asm/ftrace.h b/arch/arm64/include/asm/ftrace.h
>>> index c5534fa..3c60f37 100644
>>> --- a/arch/arm64/include/asm/ftrace.h
>>> +++ b/arch/arm64/include/asm/ftrace.h
>>> @@ -28,6 +28,8 @@ struct dyn_arch_ftrace {
>>> 
>>> extern unsigned long ftrace_graph_call;
>>> 
>>> +extern void return_to_handler(void);
>>> +
>>> static inline unsigned long ftrace_call_adjust(unsigned long addr)
>>> {
>>> 	/*
>>> diff --git a/arch/arm64/include/asm/stacktrace.h b/arch/arm64/include/asm/stacktrace.h
>>> index 6fb61c5..801a16db 100644
>>> --- a/arch/arm64/include/asm/stacktrace.h
>>> +++ b/arch/arm64/include/asm/stacktrace.h
>>> @@ -22,6 +22,9 @@ struct stackframe {
>>> 	unsigned long fp;
>>> 	unsigned long sp;
>>> 	unsigned long pc;
>>> +#ifdef CONFIG_FUNCTION_GRAPH_TRACER
>>> +	unsigned int graph;
>>> +#endif
>>> };
>> 
>> How about using int instead of unsigned int to align with cure_ret_stack
>> of struct task_struct?
>> 
>>> extern int unwind_frame(struct task_struct *tsk, struct stackframe *frame);
>>> diff --git a/arch/arm64/kernel/perf_callchain.c b/arch/arm64/kernel/perf_callchain.c
>>> index 797220d..ff46654 100644
>>> --- a/arch/arm64/kernel/perf_callchain.c
>>> +++ b/arch/arm64/kernel/perf_callchain.c
>>> @@ -164,6 +164,9 @@ void perf_callchain_kernel(struct perf_callchain_entry *entry,
>>> 	frame.fp = regs->regs[29];
>>> 	frame.sp = regs->sp;
>>> 	frame.pc = regs->pc;
>>> +#ifdef CONFIG_FUNCTION_GRAPH_TRACER
>>> +	frame.graph = current->curr_ret_stack;
>>> +#endif
>>> 
>>> 	walk_stackframe(current, &frame, callchain_trace, entry);
>>> }
>>> diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
>>> index 98bf546..88d742b 100644
>>> --- a/arch/arm64/kernel/process.c
>>> +++ b/arch/arm64/kernel/process.c
>>> @@ -344,6 +344,9 @@ unsigned long get_wchan(struct task_struct *p)
>>> 	frame.fp = thread_saved_fp(p);
>>> 	frame.sp = thread_saved_sp(p);
>>> 	frame.pc = thread_saved_pc(p);
>>> +#ifdef CONFIG_FUNCTION_GRAPH_TRACER
>>> +	frame.graph = p->curr_ret_stack;
>>> +#endif
>>> 	stack_page = (unsigned long)task_stack_page(p);
>>> 	do {
>>> 		if (frame.sp < stack_page ||
>>> diff --git a/arch/arm64/kernel/return_address.c b/arch/arm64/kernel/return_address.c
>>> index 07b37ac..1718706 100644
>>> --- a/arch/arm64/kernel/return_address.c
>>> +++ b/arch/arm64/kernel/return_address.c
>>> @@ -43,6 +43,9 @@ void *return_address(unsigned int level)
>>> 	frame.fp = (unsigned long)__builtin_frame_address(0);
>>> 	frame.sp = current_stack_pointer;
>>> 	frame.pc = (unsigned long)return_address; /* dummy */
>>> +#ifdef CONFIG_FUNCTION_GRAPH_TRACER
>>> +	frame.graph = current->curr_ret_stack;
>>> +#endif
>>> 
>>> 	walk_stackframe(current, &frame, save_return_addr, &data);
>>> 
>>> diff --git a/arch/arm64/kernel/stacktrace.c b/arch/arm64/kernel/stacktrace.c
>>> index 9c7acf8..0a39049 100644
>>> --- a/arch/arm64/kernel/stacktrace.c
>>> +++ b/arch/arm64/kernel/stacktrace.c
>>> @@ -17,6 +17,7 @@
>>>  */
>>> #include <linux/kernel.h>
>>> #include <linux/export.h>
>>> +#include <linux/ftrace.h>
>>> #include <linux/sched.h>
>>> #include <linux/stacktrace.h>
>>> 
>>> @@ -66,6 +67,19 @@ int notrace unwind_frame(struct task_struct *tsk, struct stackframe *frame)
>>> 	frame->fp = *(unsigned long *)(fp);
>>> 	frame->pc = *(unsigned long *)(fp + 8);
>>> 
>>> +#ifdef CONFIG_FUNCTION_GRAPH_TRACER
>>> +	if (tsk && tsk->ret_stack &&
>>> +			(frame->pc == (unsigned long)return_to_handler)) {
>>> +		/*
>>> +		 * This is a case where function graph tracer has
>>> +		 * modified a return address (LR) in a stack frame
>>> +		 * to hook a function return.
>>> +		 * So replace it to an original value.
>>> +		 */
>>> +		frame->pc = tsk->ret_stack[frame->graph--].ret;
>>> +	}
>>> +#endif /* CONFIG_FUNCTION_GRAPH_TRACER */
>> 
>> There is an index check of ret_stack[] in case of x86 [1]. Even though
>> graph is unsigned int, I think we need to check the value of frame->graph
>> before accessing ret_stack[].
> 
> I'm not sure that the checking is very useful because, if it happens,
> it is a bug. It might make sense to avoid a possible panic though.
> 
>>> +
>>> 	/*
>>> 	 * Check whether we are going to walk through from interrupt stack
>>> 	 * to task stack.
>>> @@ -137,6 +151,9 @@ void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
>>> 		frame.sp = current_stack_pointer;
>>> 		frame.pc = (unsigned long)save_stack_trace_tsk;
>>> 	}
>>> +#ifdef CONFIG_FUNCTION_GRAPH_TRACER
>>> +	frame.graph = tsk->curr_ret_stack;
>>> +#endif
>>> 
>>> 	walk_stackframe(tsk, &frame, save_trace, &data);
>>> 	if (trace->nr_entries < trace->max_entries)
>>> diff --git a/arch/arm64/kernel/time.c b/arch/arm64/kernel/time.c
>>> index 6e5c521..5977969 100644
>>> --- a/arch/arm64/kernel/time.c
>>> +++ b/arch/arm64/kernel/time.c
>>> @@ -52,6 +52,9 @@ unsigned long profile_pc(struct pt_regs *regs)
>>> 	frame.fp = regs->regs[29];
>>> 	frame.sp = regs->sp;
>>> 	frame.pc = regs->pc;
>>> +#ifdef CONFIG_FUNCTION_GRAPH_TRACER
>>> +	frame.graph = -1; /* no task info */
>>> +#endif
>> 
>> graph is unsigned int type. Is this intentional?
> 
> No. This initialization is, I believe, redundant as it is not checked anywhere,
> but I will re-think of it along with the checking above.
> 
> Thanks,
> -Takahiro AKASHI
> 
>> Best Regards
>> Jungseok Lee
>> 
>> [1] arch/x86/kernel/dumpstack.c
>> 
> ----8<----
> From 9ebba7167f7838daf68d8231f04141d2f4d4b7b5 Mon Sep 17 00:00:00 2001
> From: AKASHI Takahiro <takahiro.akashi at linaro.org>
> Date: Wed, 25 Nov 2015 13:29:54 +0900
> Subject: [PATCH] fixup! arm64: ftrace: fix a stack tracer's output under
> function graph tracer
> 
> 
> Signed-off-by: AKASHI Takahiro <takahiro.akashi at linaro.org>
> ---
> arch/arm64/kernel/traps.c |   23 +++++++++++++++++------
> 1 file changed, 17 insertions(+), 6 deletions(-)
> 
> diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c
> index 46053c2..f140029 100644
> --- a/arch/arm64/kernel/traps.c
> +++ b/arch/arm64/kernel/traps.c
> @@ -147,17 +147,14 @@ static void dump_backtrace(struct pt_regs *regs, struct task_struct *tsk)
> {
> 	struct stackframe frame;
> 	unsigned long _irq_stack_ptr = per_cpu(irq_stack_ptr, smp_processor_id());
> +	int skip;
> 
> 	pr_debug("%s(regs = %p tsk = %p)\n", __func__, regs, tsk);
> 
> 	if (!tsk)
> 		tsk = current;
> 
> -	if (regs) {
> -		frame.fp = regs->regs[29];
> -		frame.sp = regs->sp;
> -		frame.pc = regs->pc;
> -	} else if (tsk == current) {
> +	if (tsk == current) {
> 		frame.fp = (unsigned long)__builtin_frame_address(0);
> 		frame.sp = current_stack_pointer;
> 		frame.pc = (unsigned long)dump_backtrace;
> @@ -173,13 +170,27 @@ static void dump_backtrace(struct pt_regs *regs, struct task_struct *tsk)
> 	frame.graph = tsk->curr_ret_stack;
> #endif
> 
> +	skip = (regs ? 1 : 0);
> 	pr_emerg("Call trace:\n");
> 	while (1) {
> 		unsigned long where = frame.pc;
> 		unsigned long stack;
> 		int ret;
> 
> -		dump_backtrace_entry(where);
> +		/* skip until specified stack frame */
> +		if (!skip)
> +			dump_backtrace_entry(where);
> +		else if (frame.fp == regs->regs[29]) {
> +			skip = 0;
> +			/*
> +			 * Mostly, this is the case where this function is
> +			 * called in panic/abort. As exception handler's
> +			 * stack frame does not contain the corresponding pc
> +			 * at which an exception has taken place, use regs->pc
> +			 * instead.
> +			 */
> +			dump_backtrace_entry(regs->pc);
> +		}
> 		ret = unwind_frame(tsk, &frame);
> 		if (ret < 0)
> 			break;
> -- 
> 1.7.9.5

This fixes up the issue I mentioned.

Best Regards
Jungseok Lee


More information about the linux-arm-kernel mailing list