public inbox for binutils@sourceware.org
 help / color / mirror / Atom feed
From: Palmer Dabbelt <palmer@dabbelt.com>
To: binutils@sourceware.org
Cc: research_trasio@irq.a4lg.com, binutils@sourceware.org
Subject: Re: [PATCH 0/5] RISC-V: Add Ratified Cache Management Operation ISA Extensions (with paren)
Date: Thu, 31 Mar 2022 20:33:03 -0700 (PDT)	[thread overview]
Message-ID: <mhng-7b3bc2e8-1d48-46a0-bba5-d92b1d60ae6a@palmer-mbp2014> (raw)
In-Reply-To: <cover.1644373764.git.research_trasio@irq.a4lg.com>

On Tue, 08 Feb 2022 18:29:27 PST (-0800), binutils@sourceware.org wrote:
> This patchset is a modified version of:
> <https://sourceware.org/pipermail/binutils/2021-December/118909.html>
>
> GitHub:
> <https://github.com/a4lg/binutils-gdb/tree/riscv-CMOs> (original; 2021-12-16, rebased)
> <https://github.com/a4lg/binutils-gdb/tree/riscv-CMOs-with-paren> (new: 2022-02-09)
>
>
> There seems a discussion to change operand of cbo.* instructions on:
> <https://github.com/riscv/riscv-CMOs/issues/47>.

Looks like that still hasn't been resolved?  I don't really care either 
way, just that whatever we do is in agreement with the spec as it'd be 
silly to differ from LLVM here.

> I have a little preference for old one but it's not bad to have both
> options (as complete patchsets).
>
> Note that however, the original patchset is *NOT OUTDATED* yet.
> It implements Zicbom/Zicboz instruction like:
>
>     cbo.zero    a0
>
> This new patchset implements Zicbom/Zicboz instructions like:
>
>     cbo.zero    (a0)
>     cbo.zero    0(a0)    # only zero is allowed

There's some precedent for accepting both flavors of instructions like 
that, so if there's already an assembler out there that does things one 
way and the spec does the other I guess we could accept both?  I'd still 
prefer to just have everyone agree on the specification side, though.

> Actually, my first attempt to implement cbo.* instructions used the same
> format as *new* patchset (I modified before submitting the first
> patchset for compliance to the specification document).
>
> This version requires extra checking (0(a0) actually allows
> EXPRESSION_RESULTS_IN_ZERO(a0)) so I added a few extra tests.
> It will also reject operand format in older patchset.
>
> Old patchset and new patchset are mutually exclusive and should follow
> the conclusion of the discussion above.
>
> Thanks,
> Tsukasa
>
>
>
>
> Tsukasa OI (5):
>   RISC-V: Add mininal support for Zicbo[mpz]
>   RISC-V: Cache management instructions
>   RISC-V: Cache management instruction testcases
>   RISC-V: Prefetch hint instructions and operand set
>   RISC-V: Prefetch hint instruction testcases
>
>  bfd/elfxx-riscv.c                     |  9 +++++++++
>  gas/config/tc-riscv.c                 | 18 ++++++++++++++++++
>  gas/testsuite/gas/riscv/zicbom-fail.d |  3 +++
>  gas/testsuite/gas/riscv/zicbom-fail.l |  7 +++++++
>  gas/testsuite/gas/riscv/zicbom-fail.s |  7 +++++++
>  gas/testsuite/gas/riscv/zicbom.d      | 15 +++++++++++++++
>  gas/testsuite/gas/riscv/zicbom.s      |  7 +++++++
>  gas/testsuite/gas/riscv/zicbop-fail.d |  3 +++
>  gas/testsuite/gas/riscv/zicbop-fail.l |  4 ++++
>  gas/testsuite/gas/riscv/zicbop-fail.s |  4 ++++
>  gas/testsuite/gas/riscv/zicbop.d      | 12 ++++++++++++
>  gas/testsuite/gas/riscv/zicbop.s      |  4 ++++
>  gas/testsuite/gas/riscv/zicboz-fail.d |  3 +++
>  gas/testsuite/gas/riscv/zicboz-fail.l |  3 +++
>  gas/testsuite/gas/riscv/zicboz-fail.s |  3 +++
>  gas/testsuite/gas/riscv/zicboz.d      | 12 ++++++++++++
>  gas/testsuite/gas/riscv/zicboz.s      |  4 ++++
>  include/opcode/riscv-opc.h            | 16 ++++++++++++++++
>  include/opcode/riscv.h                |  3 +++
>  opcodes/riscv-dis.c                   |  4 ++++
>  opcodes/riscv-opc.c                   |  9 +++++++++
>  21 files changed, 150 insertions(+)
>  create mode 100644 gas/testsuite/gas/riscv/zicbom-fail.d
>  create mode 100644 gas/testsuite/gas/riscv/zicbom-fail.l
>  create mode 100644 gas/testsuite/gas/riscv/zicbom-fail.s
>  create mode 100644 gas/testsuite/gas/riscv/zicbom.d
>  create mode 100644 gas/testsuite/gas/riscv/zicbom.s
>  create mode 100644 gas/testsuite/gas/riscv/zicbop-fail.d
>  create mode 100644 gas/testsuite/gas/riscv/zicbop-fail.l
>  create mode 100644 gas/testsuite/gas/riscv/zicbop-fail.s
>  create mode 100644 gas/testsuite/gas/riscv/zicbop.d
>  create mode 100644 gas/testsuite/gas/riscv/zicbop.s
>  create mode 100644 gas/testsuite/gas/riscv/zicboz-fail.d
>  create mode 100644 gas/testsuite/gas/riscv/zicboz-fail.l
>  create mode 100644 gas/testsuite/gas/riscv/zicboz-fail.s
>  create mode 100644 gas/testsuite/gas/riscv/zicboz.d
>  create mode 100644 gas/testsuite/gas/riscv/zicboz.s
>
>
> base-commit: a37603c43f8da7983ed53b567ea30ce66066daa2

  parent reply	other threads:[~2022-04-01  3:33 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-02-09  2:29 Tsukasa OI
2022-02-09  2:29 ` [PATCH 1/5] RISC-V: Add mininal support for Zicbo[mpz] Tsukasa OI
2022-02-09  2:29 ` [PATCH 2/5] RISC-V: Cache management instructions Tsukasa OI
2022-02-09  2:29 ` [PATCH 3/5] RISC-V: Cache management instruction testcases Tsukasa OI
2022-02-09  2:29 ` [PATCH 4/5] RISC-V: Prefetch hint instructions and operand set Tsukasa OI
2022-02-25  2:50   ` Nelson Chu
2022-02-25  7:07     ` Tsukasa OI
2022-03-18  7:50       ` Nelson Chu
2022-02-09  2:29 ` [PATCH 5/5] RISC-V: Prefetch hint instruction testcases Tsukasa OI
2022-04-01  3:33 ` Palmer Dabbelt [this message]
2022-04-01  4:04   ` [PATCH 0/5] RISC-V: Add Ratified Cache Management Operation ISA Extensions (with paren) Kito Cheng
2022-04-01  4:15     ` Palmer Dabbelt

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=mhng-7b3bc2e8-1d48-46a0-bba5-d92b1d60ae6a@palmer-mbp2014 \
    --to=palmer@dabbelt.com \
    --cc=binutils@sourceware.org \
    --cc=research_trasio@irq.a4lg.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).