Re: AccessController speedup

["Scott Gilbertson" <scottg at mantatest dot com>]

> Scott Gilbertson wrote:
> >
> > I'm getting a segmentation fault...
> > Do you suppose "--disable-tls" or "--enable-sjlj-exceptions" has
anything to
> > do with it?
>
> As Andrew has noted, the Linux targets assume that DWARF-2
> unwinding information is available and no one seems to test
> the SJLJ case. If you can help it, do not use SJLJ-based EH
> ...
> Ranjit.

Thanks, Andrew and Ranjit, for the clear and complete explanations.  I'll
stop using SJLJ.

Andrew:
>Your post suggested that there was infinite recursion in the
>backtrace.  You could have a look to see what's causing that.
SJLJ, apparently -- NullPointerException in _Unwind_GetIPInfo
(unwind-sjlj.c).

I've re-built without --enable-sjlj-exceptions (but still
with --disable-tls), and statically built programs (including some using awt
and xlib) are running again.  I had originally gone to SJLJ because it
solved a different startup crash using the dwarf2 unwinder with my specific
build system (others weren't having that problem), but I guess that's been
fixed.  I'm smiling again.

The only problem I've seen so far is that exception stack dumps don't work
(static or dynamic).  I'm guessing that's because it needs TLS, which I
think is problematic on my old glibc.  Does that sound right?