This is the mail archive of the gcc-help@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: effect of -fPIC on code speed


Ian Lance Taylor <iant@google.com> writes:
> There is a great deal one could say about -fPIC.  It's hard to know what
> should be said in the gcc docs.  To me it seems more appropriate in some
> document about what PIC really is.

Sure, then the gcc doc can say "(see <all-about-pic> for details)"... :)

> What happens here is that when using -fPIC symbol interposition is
> permitted for any globally visible symbol.  That is, the -fPIC code is
> presumed to wind up in a shared library.  An executable may override any
> function in a shared library.  The code in the shared library which
> calls that function is then expected to call the definition in the
> executable instead.  It follows that when using -fPIC the compiler may
> not inline any call to a globally visible function, because that would
> prevent symbol interposition.

Hmm, I guess this makes sense.

It would definitely be nice to have a more precise description in a
document somewhere though.

I'd think it would also be good to have a way of disabling this
particular side-effect -- e.g., for non-shared-library uses of shared
objects, like loadable modules/plugins etc, which presumably don't aim
to provide the same guarantees that shared libraries do.

[Is there more appropriate place to say these things than gcc-help, btw?]

-Miles

-- 
Run away!  Run away!


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]