Re: [PATCH 2/2 v2] tracing/syscalls: support for syscalls tracing on x86-64

[Frederic Weisbecker <fweisbec@gmail.com>]

On Sun, Mar 15, 2009 at 06:30:18AM +0100, Ingo Molnar wrote:
> 
> * Andrew Morton <akpm@linux-foundation.org> wrote:
> 
> > On Fri, 13 Mar 2009 15:42:12 +0100 Frederic Weisbecker <fweisbec@gmail.com> wrote:
> > 
> > > +static struct syscall_metadata *find_syscall_meta(unsigned long *syscall)
> > > +{
> > > +	struct syscall_metadata *start;
> > > +	struct syscall_metadata *stop;
> > > +	char str[KSYM_SYMBOL_LEN];
> > > +
> > > +
> > > +	start = (struct syscall_metadata *)__start_syscalls_metadata;
> > > +	stop = (struct syscall_metadata *)__stop_syscalls_metadata;
> > > +	kallsyms_lookup((unsigned long) syscall, NULL, NULL, NULL, str);
> > > +
> > > +	for ( ; start < stop; start++) {
> > > +		if (start->name && !strcmp(start->name, str))
> > > +			return start;
> > > +	}
> > > +	return NULL;
> > > +}
> > 
> > afacit this feature can be enabled when CONFIG_KALLSYMS=n.  
> > Does that make sense?
> 
> It does not make much sense - the function will return NULL and 
> we wont do any tracing. Frederic: FTRACE_SYSCALLS should either 
> select KALLSYMS in kernel/trace/Kconfig, like the STACK_TRACER 
> and FUNCTION_TRACER already does. Or perhaps, for really 
> puritane tracing, we should emit simplified, non-symbolic trace 
> data in the !KALLSYMS case.
> 
> 	Ingo

You're right. But it would be somewhat pointless to trace the syscalls
without their names.
I will make it select KALLSYMS, it seems to me a sane way to fix it.

Thanks.