Re: [Jack-Devel] [Fwd: Re: [Guitarix-developer] jack session crash]

On Thu, Nov 3, 2011 at 12:28 PM, Andreas Degert <[hidden]> wrote:

> Thanks for solving the problem (especially Nedko). Would be nice if it
> could be documented more prominently, since fPIC (and also fPIE) produce
> less efficient code and one would not normally use those options (and
> if not used it fails silently). I didn't find any references to it, and
> also not in code I looked at (e.g. building jackd tools or qtractor).

i think it may be a bug in gcc, actually. i did quite a lot of reading
on the issue, and it appears that ELF actually allows for two
different kinds of weak-ness, one of which is guaranteed to work in
the way it is described everywhere, and one which is more complex. i
will likely file a bug on this issue, because i agree that -fPIC
and/or -fPIE should not be necessary, but first i have to get some
guidance on where the bug would be (gcc or ld or ld.so)

i also tried testing on OS X but the test design i used won't work
there because of the way the OS X linker embeds explict paths into
library dependencies in mach-o files. however, i have tested with the
jack1/jack2 "switch" on that platform and found that weak linkage
works fine there even if the library is missing a symbol.

1320339784.19943_0.ltw:2,a <CAFa_cKkiSHAkKJyuJyXmkKcumoeBwRUMp0KXXf4X5gBTnSunQg at mail dot gmail dot com>