Re: [PATCH] trace: Fix build breakage without CONFIG_PERF_EVENTS

[Steven Rostedt <rostedt@goodmis.org>]

On Tue, 2012-02-28 at 15:43 +0100, Jiri Olsa wrote:
> On Tue, Feb 28, 2012 at 02:03:35PM +0000, Mark Brown wrote:
> > On Tue, Feb 28, 2012 at 09:02:09AM -0500, Steven Rostedt wrote:
> > > On Tue, 2012-02-28 at 13:37 +0000, Mark Brown wrote:
> > 
> > > > ARM.  If it was not possible to switch off PERF_EVENTS it shouldn't be a
> > > > config option in the code as this is just creating complexity and
> > > > potential breakage.
> > 
> > > It is possible to switch off PERF_EVENTS, just not on x86 :-/
> > 
> > Well, if you insist on using these restricted architectures :P
> 
> reproduced ;) looking for fix..
> 
> jirka
> 
> 
> [jolsa@m tip]$ make   ARCH=arm CROSS_COMPILE=armv5tel-redhat-linux-gnueabi- uImage 
>   CHK     include/linux/version.h
>   CHK     include/generated/utsrelease.h
> make[1]: `include/generated/mach-types.h' is up to date.
>   CALL    scripts/checksyscalls.sh
>   CHK     include/generated/compile.h
>   CC      kernel/trace/trace_export.o
> In file included from kernel/trace/trace_export.c:197:
> kernel/trace/trace_entries.h:58: error: ‘perf_ftrace_event_register’
> undeclared here (not in a function)
> make[2]: *** [kernel/trace/trace_export.o] Error 1
> make[1]: *** [kernel/trace] Error 2
> make: *** [kernel] Error 2

Hmm, I'm surprised my cross build test didn't catch this.

Crap, I think I only ran the defconfig tests, and not the test that also
enables tracing.

I'll update my config to do both automatically.

Thanks!

-- Steve