This is the mail archive of the
mailing list for the binutils project.
Re: [PATCH 1/2] [WebAssembly] GAS support
- From: Pip Cet <pipcet at gmail dot com>
- To: Nick Clifton <nickc at redhat dot com>
- Cc: binutils at sourceware dot org
- Date: Thu, 30 Mar 2017 11:32:20 +0000
- Subject: Re: [PATCH 1/2] [WebAssembly] GAS support
- Authentication-results: sourceware.org; auth=none
- References: <CAOqdjBeNadWo_P=ApawgM5QUo6PinMokF0QHKRHyot8_Yaot-A@mail.gmail.com> <firstname.lastname@example.org>
On Thu, Mar 30, 2017 at 10:05 AM, Nick Clifton <email@example.com> wrote:
> Hi Pip,
>> This patch adds support for assembling WebAssembly opcodes into ELF
>> binaries (which can then be converted into WebAssembly modules). In
>> order to pass tests, it also adds the first wasm32-specific relocation
>> (other than R_WASM32_NONE), a simple absolute 32-bit relocation.
> Thanks for doing this. I have applied the patch basically as is. The
> only changes I made were to add entries in the gas/NEWS and binutils/NEWS
> files, mentioning this new support. Oh - and I have tweaked your gas
> documentation, so that it is now built as part of the general assembler
> documentation. I did have to rearrange the c-wasm32.texi file a bit in
> order to make the texinfo conversion work, so you may want to check that.
> (Oh, c-wasm32.texi is a renamed version of your wasm32.texi file. The
> convention is that target specific texinfo files in the gas/doc directory
> are named c-<target>.texi).
Thanks! I'd actually noticed the c-wasm32.texi thing yesterday, but
you were faster than I was in making the correction. I'd just like to
point out I really appreciate those; I am looking over them.
> One thing that I did note is that you are defining these new relocation names
> in bfd/reloc.c but not handling them in elf32-wasm.c. Are they going to be
> subject to a future patch ?
Yes, that was my intention. Disassembler support is probably up next
(and is independent of any ABI). After that, it's pretty much the
linker changes, which still need some work, particularly in order to
document their limitations and figure out a way to leave the generic