GDBJIT: emit .eh_frame section on x64.

GDBJIT: emit .eh_frame section on x64.

This enables proper stack unwinding on x64.

Currently this requires V8 to be compiled without snapshot and --gdbjit-full to be enabled.

Original patch by Sanjoy Das (http://codereview.chromium.org/6371011/)

Committed: http://code.google.com/p/v8/source/detail?r=6575

[Erik Corry, 2011-02-02 13:10:24]

lgtm

But no tests makes me sad :-(

http://codereview.chromium.org/6250104/diff/1/src/gdb-jit.cc
File src/gdb-jit.cc (right):

http://codereview.chromium.org/6250104/diff/1/src/gdb-jit.cc#newcode659
src/gdb-jit.cc:659: code_(code),
Indentation

http://codereview.chromium.org/6250104/diff/1/src/gdb-jit.cc#newcode677
src/gdb-jit.cc:677: uintptr_t CodeStart() const {
Why can't these return Address?

http://codereview.chromium.org/6250104/diff/1/src/gdb-jit.cc#newcode770
src/gdb-jit.cc:770: w->Write<intptr_t>(desc_->CodeStart());
w->Write<Address>(...

http://codereview.chromium.org/6250104/diff/1/src/gdb-jit.cc#newcode969
src/gdb-jit.cc:969: DW_CFA_ADVANCE_LOC = 0X40,
We use a small x everywhere in V8, so we should do it here too for consistency.

http://codereview.chromium.org/6250104/diff/1/src/gdb-jit.cc#newcode1022
src/gdb-jit.cc:1022: for (uint32_t i = 0; i < (kPointerSize - align); i++)
Missing {} on multiline for statement.

http://codereview.chromium.org/6250104/diff/1/src/gdb-jit.cc#newcode1345
src/gdb-jit.cc:1345: // function prologue and epilogue are the same for all
we rely on ... are the same -> we rely on ... being the same