[PATCH][cxxabi] ARM EHABI zero-cost exception handling for libc++abi

[Jonathan Roelofs]

On 5/4/14, 1:03 PM, Renato Golin wrote:
> On 3 May 2014 17:29, Logan Chien <tzuhsiang.chien at gmail.com> wrote:
>> IMO, maybe we can focus on my patch first?  Although, it still relies on the
>> unwinding facilities from libgcc, I feel that we can replace them
>> step-by-step.
>
> Hi Logan,
>
> This sounds like a good plan. I had a skim through your patch and it
> seems good, but I haven't tested any of it.
>
> I modified Clang to include --rtlib=compiler-rt on Linux, but that
> still includes both gcc_s and gcc_eh. The idea was to use libc++abi,
> or libunwinder as it would be called when moved to compiler-rt bundle.
>
> On Clang, I can only see we replacing each of { gcc_s, gcc_eh }
> entirely. It seems to me that doing the unwind move first would be
> ideal, especially if we can make it to compiler-rt by that time.
>
>
>> Once the language-independent unwinding library is complete, we
>> can ifdef (or remove) the function call to libgcc.  Does this plan sound
>> reasonable?
This plan sounds OK to me, but I am a little hesitant. I want to be careful 
about how we use the reserved bits of the unwind caches, because some of it is 
implementation details of libgcc_s, and I'd rather not have a bunch of "this is 
unspecified in the spec, but we're doing it this way because libgcc's unwinder 
does it this way" in there permanently.  Can you add notes where such 
assumptions are made in the libc++abi part so that we can fix them later?

Again, sorry I've been so slow at this.

Jon
>
> Yes. Thanks for working on this!
>
> cheers,
> --renato
>

-- 
Jon Roelofs
jonathan at codesourcery.com
CodeSourcery / Mentor Embedded