Implement support for CFI unwinding information

[fitzgen]

trafficstars

Section 6.4 of the DWARF 4 standard.

---

TODO:

• [ ] https://github.com/gimli-rs/gimli/issues/112: Be robust against overflow / casting outside of target type's range
• [x] https://github.com/gimli-rs/gimli/issues/113: Support .eh_frame augmentation
• [x] https://github.com/gimli-rs/gimli/issues/114: Benches
• [x] https://github.com/gimli-rs/gimli/issues/115: Self-parse integration test (requires the .eh_frame augmentation)
• [ ] https://github.com/gimli-rs/gimli/issues/116: Output this info in dwarfdump.rs
• [x] https://github.com/gimli-rs/gimli/issues/117: A single, orchestrating DebugFrame::unwind_info_for_address(&self, ctx: &mut UninitializedUnwindContext, address: u64) -> UnwindTableRow method (for folks who don't have any kind of CIE/FDE caching or anything like that)

[fitzgen]

Going to work on this, now that it seems like @tromey is pretty close to having expressions landable.

[fitzgen]

Also, http://refspecs.linuxfoundation.org/LSB_4.1.0/LSB-Core-generic/LSB-Core-generic/ehframechpt.html (I think that is the most up to date version of the .eh_frame spec)

Is it worth supporting .debug_frame in addition to .eh_frame? It seems like everything uses the latter these days, AFAICT...

@tromey perhaps you might know?

[fitzgen]

Is it worth supporting .debug_frame in addition to .eh_frame? It seems like everything uses the latter these days, AFAICT...

I think we want both. Can't rely on either one always existing.

Probably want to also support .ARM.exidx and .ARM.extab for arm/android... Although maybe that should be a different crate?

[fitzgen]

Probably want to also support .ARM.exidx and .ARM.extab for arm/android... Although maybe that should be a different crate?

http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038b/IHI0038B_ehabi.pdf