Segfault calling dlopen()'ed executable built with -pie and using TLS

[Cary Coutant]

>> Maybe I misunderstood your original question. Yes, you're calling the
>> payload's main() from your loader, and that's OK because your loader
>> has already started up through the crt startup code. But I thought
>> you
>> were trying to link the payload binary so that it could also serve as
>> an executable that could be launched by itself rather than via the
>> loader. To do that, you need the startup code in the crt*.o files,
>> and
>> that code will call your payload's main().
>
> Yes, I understand.
> But what does *not* work is exactly "calling the payload's main() from
> the loader". On the other hand, starting the payload directly *does*
> work.

Your original question was why you couldn't dlopen the PIE binary, and
I explained that the TLS model was wrong. We were then talking about
building your payload as a real shared library (i.e., with -shared
instead of -pie), so that you could dlopen it from your loader. But in
order to also be able to launch that as a standalone executable, you
need startup code, and that's what I was talking about
(theoretically). When you use your launcher, it will just call main,
but when you try to run it standalone, it needs to begin execution at
_start in order to initialize everything before calling main.

Case (a): Build as -pie.
  - Can launch as standalone.
  - Cannot dlopen from separate launcher.

Case (b): Build as -shared.
  - Can dlopen from separate launcher.
  - Need to develop PIC startup code in order to launch as standalone
(might work in theory, but no current support).

-cary