Re: [PATCH v2 0/6] ftrace: to kill a daemon (small updates)

[Ingo Molnar <mingo@elte.hu>]

* Steven Rostedt <rostedt@goodmis.org> wrote:

> [
>   Changes since v1:
> 
>     regex fix in x86_64 recordmcount.pl. Now it can handle all
>       mcount+0x... mcount-0x... and mcount, where as the original
>       only handled mcount+0x...
> 
>     Made mcount on start-up to simply return. The current mcount
>       is set up to be replaced with a call to ftrace_record_ip.
>       This is no longer necessary.
> 
>   Note: This patch series is focusing on how calls to mcount in
>     the kernel are converted to nops. It does not address what
>     kind of nop is used. That is a different topic, and should
>     be in a different patch series.
> 
>   Note 2: I have found that the changes here are more stable than
>     the current daemon method, and these patches should be used.
>     It also solves the resume from suspend to ram bug that was
>     reported:
> 
>       http://lkml.org/lkml/2008/8/12/234
>        and
>       http://lkml.org/lkml/2008/8/12/451
> 
>   Note 3: I have already ported this to PowerPC64, but I am waiting
>     for this to be accepted first before submitting those changes.
> ]
> 
> One of the things that bothered me about the latest ftrace code was 
> this annoying daemon that would wake up once a second to see if it had 
> work to do. If it did not, it would go to sleep, otherwise it would do 
> its work and then go to sleep.

i like this concept alot - i've applied the whole lot to 
tip/tracing/ftrace.

Eventually gcc should be extended to provide a separate section for 
instrumentation patch sites, instead of us having to disassemble the 
object code. That would also make the ftrace build faster. Any gcc folks 
interested in that?

	Ingo