This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH] RISC-V: Add %got_pcrel_hi operator
- From: James Clarke <jrtc27 at jrtc27 dot com>
- To: Palmer Dabbelt <palmer at sifive dot com>
- Cc: Alex Bradbury <asb at asbradbury dot org>, Andrew Waterman <andrew at sifive dot com>, Jim Wilson <jimw at sifive dot com>, binutils at sourceware dot org
- Date: Thu, 8 Nov 2018 11:35:52 +0000
- Subject: Re: [PATCH] RISC-V: Add %got_pcrel_hi operator
- References: <mhng-5e1b9ff2-4e61-4d79-bde1-40658663ba21@palmer-si-x1c4>
On 8 Nov 2018, at 02:10, Palmer Dabbelt <palmer@sifive.com> wrote:
> On Wed, 07 Nov 2018 17:22:18 PST (-0800), jrtc27@jrtc27.com wrote:
>> On 8 Nov 2018, at 00:40, Jim Wilson <jimw@sifive.com> wrote:
>>> On 11/7/18 10:27 AM, James Clarke wrote:
>>>> This operator allows compilers to expand the "la" pseduo-instruction and
>>>> potentially split up the two instructions.
>>>> gas/
>>>> * config/tc-riscv.c (percent_op_utype): Add entry for a new
>>>> %got_pcrel_hi operator.
>>>> * testsuite/gas/riscv/no-relax-reloc.d: Update for new
>>>> %got_pcrel_hi operator test.
>>>> * testsuite/gas/riscv/no-relax-reloc.s: Add a %got_pcrel_hi
>>>> operator test.
>>>> * testsuite/gas/riscv/relax-reloc.d: Update for new
>>>> %got_pcrel_hi operator test.
>>>> * testsuite/gas/riscv/relax-reloc.s: Add a %got_pcrel_hi
>>>> operator test.
>>> We have a list of assembler operators in the psABI document
>>> https://github.com/riscv/riscv-elf-psabi-doc
>>> We also have a list of assembler operators in the assembler manual
>>> https://github.com/riscv/riscv-asm-manual
>>> This looks a little redundant. It isn't clear why assembler syntax is
>>> mentioned in the psABI. It doesn't seem relevant to that. Anyways, these
>>> should be updated.
>
> It's a bit off topic, but IIRC we didn't have an assembly manual when the psABI doc was started so it ended up in there. It doesn't make a whole lot of sense, I'd be OK dropping the assembler syntax from the psABI doc.
That's a discussion for later. I've opened [1] against riscv-asm-manual, so
let's take this discussion there.
James