[KJ] value of "__attribute__((no_instrument_function))"??

[KJ] value of "__attribute__((no_instrument_function))"??

[Robert P. J. Day]

  i just noticed that there are a *very* small number of routines in
the source tree tagged with the attribute of "no_instrument_function".

$ grep -rw no_instrument_function .
./net/rxrpc/main.c:__attribute__((no_instrument_function));
./net/rxrpc/main.c:__attribute__((no_instrument_function));
./fs/afs/main.c:__attribute__((no_instrument_function));
./fs/afs/main.c:__attribute__((no_instrument_function));

  from the gcc manual:

"If -finstrument-functions is given, profiling function calls will be
generated at entry and exit of most user-compiled functions. Functions
with this attribute will not be so instrumented."

  that *seems* like a useful attribute so the obvious question is --
why is it not used more widely?  that seems kind of odd.  (in fact,
the functions with these attributes are all commented out within an
"#if 0" so they make no difference, anyway.)

  so ... is there any application to that attribute *at all* within
the kernel?  thanks.

rday

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
https://lists.osdl.org/mailman/listinfo/kernel-janitors

Re: [KJ] value of "__attribute__((no_instrument_function))"??

[Nishanth Aravamudan]

On 19.01.2007 [03:09:07 -0500], Robert P. J. Day wrote:
> 
>   i just noticed that there are a *very* small number of routines in
> the source tree tagged with the attribute of "no_instrument_function".
> 
> $ grep -rw no_instrument_function .
> ./net/rxrpc/main.c:__attribute__((no_instrument_function));
> ./net/rxrpc/main.c:__attribute__((no_instrument_function));
> ./fs/afs/main.c:__attribute__((no_instrument_function));
> ./fs/afs/main.c:__attribute__((no_instrument_function));
> 
>   from the gcc manual:
> 
> "If -finstrument-functions is given, profiling function calls will be
> generated at entry and exit of most user-compiled functions. Functions
> with this attribute will not be so instrumented."
> 
>   that *seems* like a useful attribute so the obvious question is --
> why is it not used more widely?  that seems kind of odd.  (in fact,
> the functions with these attributes are all commented out within an
> "#if 0" so they make no difference, anyway.)

It seems like a useful attribute to *disable* instrumentation in the
kernel? Why?

Thanks,
Nish

-- 
Nishanth Aravamudan <nacc@us.ibm.com>
IBM Linux Technology Center
_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
https://lists.osdl.org/mailman/listinfo/kernel-janitors

Re: [KJ] value of "__attribute__((no_instrument_function))"??

[Robert P. J. Day]

On Fri, 19 Jan 2007, Nishanth Aravamudan wrote:

> On 19.01.2007 [03:09:07 -0500], Robert P. J. Day wrote:
> >
> >   i just noticed that there are a *very* small number of routines in
> > the source tree tagged with the attribute of "no_instrument_function".
> >
> > $ grep -rw no_instrument_function .
> > ./net/rxrpc/main.c:__attribute__((no_instrument_function));
> > ./net/rxrpc/main.c:__attribute__((no_instrument_function));
> > ./fs/afs/main.c:__attribute__((no_instrument_function));
> > ./fs/afs/main.c:__attribute__((no_instrument_function));
> >
> >   from the gcc manual:
> >
> > "If -finstrument-functions is given, profiling function calls will be
> > generated at entry and exit of most user-compiled functions. Functions
> > with this attribute will not be so instrumented."
> >
> >   that *seems* like a useful attribute so the obvious question is --
> > why is it not used more widely?  that seems kind of odd.  (in fact,
> > the functions with these attributes are all commented out within an
> > "#if 0" so they make no difference, anyway.)
>
> It seems like a useful attribute to *disable* instrumentation in the
> kernel? Why?

yes, and the point it -- are there places where people *should* be
using it in the kernel?  as i mentioned, there are only four instances
of it, and all four of those are commented out.  i was just curious.

rday
_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
https://lists.osdl.org/mailman/listinfo/kernel-janitors