Re: Boot-time dumping of ftrace fuctiongraph buffer

[Ahmad Fatoum <a.fatoum@pengutronix.de>]

Hello Masami-san,

On 02.02.24 07:04, Masami Hiramatsu (Google) wrote:
> On Thu, 1 Feb 2024 13:21:37 +0100
> Ahmad Fatoum <a.fatoum@pengutronix.de> wrote:
>> For drivers that don't call dev_err_probe, I find myself sometimes doing printf
>> debugging inside the probe function.

[snip]

>> I would like to replace this with the function graph tracer:
>>
>>   - record the probe function, configured over kernel command line
>>     (The device indefinitely deferring probe is printed to the console,
>>      so I know what I am looking for on the next boot)
>>
>>   - Dump the function graph trace
>>
>>   - See if the last call before (non-devm) cleanup is getting a clock, a GPIO,
>>     a regulator or w/e.
> 
> What kind of information you prints by the printk()?

Just littering around some:

  pr_notice("%s:%d\n", __func__, __LINE__);

to find where the probe function exits.

> If the target (suspicious driver probe function) is obvious, you can use kprobe
> event and tp_printk. Or, even if you don't know, if you are sure which function
> is the starting/ending point, you can use bootconfig to record the specific part
> of execution in the ring buffer, and dump it as Steve said.
> 
> In Documentation/trace/boottime-trace.rst, there is an example.
> -----
> With the trigger action and kprobes, you can trace function-graph while
> a function is called. For example, this will trace all function calls in
> the pci_proc_init()::
> 
>   ftrace {
>         tracing_on = 0
>         tracer = function_graph
>         event.kprobes {
>                 start_event {
>                         probes = "pci_proc_init"
>                         actions = "traceon"
>                 }
>                 end_event {
>                         probes = "pci_proc_init%return"
>                         actions = "traceoff"
>                 }
>         }
>   }

I get comparable tracing with the kernel command line, but lacked a way to dump
it at boot-time. Sorry should have been clearer.

Grepping the kernel source for `ftrace_func_command' lets me think that
what I am after might be:

  end_event {
    actions = "traceoff", "dump"
  }

I need to try it out. Do you happen to know if the equivalent is possible
without bootconfig?

Thank you!
Ahmad


> -----
> 
> Thank you,
> 
>>
>> For this to be maximally useful, I need to configure this not only at boot-time,
>> but also dump the ftrace buffer at boot time. Probe deferral can hinder the kernel from
>> calling init and providing a shell, where I could read /sys/kernel/tracing/trace.
>>
>> I found following two mechanisms that looked relevant, but seem not to
>> do exactly what I want:
>>
>>   - tp_printk: seems to be related to trace points only and not usable
>>     for the function graph output
>>
>>   - dump_on_oops: I don't get an Oops if probe deferral times out, but maybe
>>     one could patch the kernel to check a oops_on_probe_deferral or dump_on_probe_deferral
>>     kernel command line parameter in deferred_probe_timeout_work_func()?
>>
>>
>> Is there existing support that I am missing? Any input on whether this
>> would be a welcome feature to have?
>>
>> Thanks!
>>
>> Cheers,
>> Ahmad
>>     
>> -- 
>> Pengutronix e.K.                           |                             |
>> Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
>> 31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
>> Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
>>  
> 
> 

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |