Re: [libunwind] parse_dynamic under DWARF

[David Mosberger]

Max,

I think I'm done as far as the x86-64-support merging is concerned.  I
left out the parse_dynamic() changes though (just lack of time to
review them at the moment, not because I'm against them).  You may
want to review my changes carefully (particularly
http://unwind.bkbits.net:8080/libunwind/address@hidden) to make sure I
didn't break anything inadvertently.

On the testdrive.hp.com machine, which runs RH enterprise server 3),
"make check" now reports 8 of 21 tests as failed.  Among the failed
ones:

        {L,G}test-exc:  these tend to be fairly fragile, so it may
                        not mean much
        {L,G}test-dyn1: expected as long as dynamic unwind info isn't
                        supported
        test-setjmp:    also expected (lack of setjmp implementation)
        test-ptrace:    failes as expected due to lack of ability
                        to locate dynamic unwind info list

Also, tests/bt crashes when attempting to unwind across a signal
delivered on an alternate signal-stack.  Not sure what that is about.

On the postive side, tests/bt now prints the correct stack-pointer
(RSP) values, so there is some progress. ;-)

I'll next work on some other stuff, so the x86-64 lock is yours
again.

Thanks,

        --david