Philippe Gerum kirjoitti:
Heikki Lindholm wrote:
Philippe Gerum kirjoitti:
Heikki Lindholm wrote:
libpthread_rt won't produce usable exec apps on ppc64 if it uses
--wrap.
There seems to be no reason to use --wrap for building the lib itself.
Good catch for the lib flags. However, substituting __real calls with
the genuine ones seems wrong to me. You'd just end up causing the
same issue.
I don't think so. Although, I'm not sure how ld is supposed to handle
--wrap case for libraries, this works in the dynamic case, at least,
for the ppc32 and ppc64 (tested on debian sarge and gentoo ppc64),
whereas leaving the __real*'s in dynamic library will fail with
unrelocatable symbols.
Ok, that's another issue. We still need to call __real entry points, but
we also have to care for the specific issue of dynamic linking which
does not substitute the calls:
Excerpt from ld's info page, talking about wrapping malloc:
You may wish to provide a `__real_malloc' function as well, so that
links without the `--wrap' option will succeed. If you do this,
you should not put the definition of `__real_malloc' in the same
file as `__wrap_malloc'; if you do, the assembler may resolve the
call before the linker has a chance to wrap it to `malloc'.
So we need a separate .c file glued to the lib, which basically does:
__real_foo ()
{
foo();
}
For all __real symbols referenced.
Right. Indeed, it seems to be the case that --wrap does not work the
same way for dynamic linking as static. Fortunately, there aren't too
many __real's used in there.
-- Heikki
