Re: [PATCH 3/3] tracing: add latency format to function_graph tracer
From: Török Edwin
Date: Fri Sep 11 2009 - 12:08:12 EST
On 2009年09月11日 18:50, Steven Rostedt wrote:
>
On Fri, 2009年09月11日 at 18:38 +0300, Török Edwin wrote:
>
>
> On 2009年09月11日 18:18, Frederic Weisbecker wrote:
>
>
>
>> On Fri, Sep 11, 2009 at 11:11:56AM -0400, Steven Rostedt wrote:
>
>>
>
>
>
>
>
>>> Ug! I just noticed that tgid was added to struct trace_entry, with the
>
>>> only user as the user stack entry. This should be in the user stack
>
>>> field not something that goes into every event!
>
>>>
>
>>> I guess I need to fix that too.
>
>>>
>
>>>
>
> Indeed tgid should be part of struct userstack_entry, and set in
>
> ftrace_trace_userstack.
>
> Do you want to me to write up a patch for that, or have you already
>
> fixed it?
>
>
>
>
I've just fixed it, thanks!
Ok.
BTW any plans on integrating an in-kernel unwinder like systemtap has?
Even if I build libc with framepointers, a userspace stacktrace on
x86-64 can't go beyond most pthreads routines, like __read_nocancel
(they're written in asm maybe?).
Best regards,
--Edwin
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
http://vger.kernel.org/majordomo-info.html
Please read the FAQ at
http://www.tux.org/lkml/