public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Patrick O'Neill <patrick@rivosinc.com>
To: 钟居哲 <juzhe.zhong@rivai.ai>, 丁乐华 <lehua.ding@rivai.ai>
Cc: "kito.cheng" <kito.cheng@gmail.com>,
	"rdapp.gcc" <rdapp.gcc@gmail.com>, palmer <palmer@rivosinc.com>,
	Jeff Law <jeffreyalaw@gmail.com>,
	gcc-patches <gcc-patches@gcc.gnu.org>
Subject: Re: [PATCH V3 00/11] Refactor and cleanup vsetvl pass
Date: Mon, 23 Oct 2023 16:42:38 -0700	[thread overview]
Message-ID: <ac46c375-a4df-7dc4-cf98-b93dbe3f24d0@rivosinc.com> (raw)
In-Reply-To: <A5160E5D1259456F+2023102406505459464692@rivai.ai>

[-- Attachment #1: Type: text/plain, Size: 13631 bytes --]

When configuring, pass in --enable-checking=rtl
If you're using riscv-gnu-toolchain, pass in --enable-gcc-checking=rtl

The -freport-bug output attached to the bug report has the full 
configure command used:
/scratch/tc-testing/tc-trunk/build-rtl-checking/../gcc/configure 
--target=riscv64-unknown-linux-gnu 
--prefix=/scratch/tc-testing/tc-trunk/build-rtl-checking 
--with-sysroot=/scratch/tc-testing/tc-trunk/build-rtl-checking/sysroot 
--with-newlib --without-headers --disable-shared --disable-threads 
--with-system-zlib --enable-tls --enable-languages=c --disable-libatomic 
--disable-libmudflap --disable-libssp --disable-libquadmath 
--disable-libgomp --disable-nls --disable-bootstrap --src=../../gcc 
--enable-checking=rtl --disable-multilib --with-abi=lp64d 
--with-arch=rv64gcv --with-tune=rocket --with-isa-spec=20191213 
'CFLAGS_FOR_TARGET=-O2    -mcmodel=medlow' 'CXXFLAGS_FOR_TARGET=-O2    
-mcmodel=medlow'

On 10/23/23 15:50, 钟居哲 wrote:
> I didn't reproduce it. How to enable RTL checking ?
>
> ------------------------------------------------------------------------
> juzhe.zhong@rivai.ai
>
>     *From:* Patrick O'Neill <mailto:patrick@rivosinc.com>
>     *Date:* 2023-10-24 06:46
>     *To:* 钟居哲 <mailto:juzhe.zhong@rivai.ai>; 丁乐华
>     <mailto:lehua.ding@rivai.ai>
>     *CC:* kito.cheng <mailto:kito.cheng@gmail.com>; rdapp.gcc
>     <mailto:rdapp.gcc@gmail.com>; palmer <mailto:palmer@rivosinc.com>;
>     Jeff Law <mailto:jeffreyalaw@gmail.com>; gcc-patches
>     <mailto:gcc-patches@gcc.gnu.org>
>     *Subject:* Re: [PATCH V3 00/11] Refactor and cleanup vsetvl pass
>
>     You're on top of it - thanks for fixing this! I'll send the testcase.
>
>     Unrelated to this failure, I'm seeing a build failure on glibc
>     rv32/64gcv when RTL checking is enabled.
>     https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111947
>
>     Thanks,
>     Patrick
>
>     On 10/23/23 14:41, 钟居哲 wrote:
>>     I have fixed it:
>>     https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=0c4bd1321a6def5eb44c530e83b01a415633b660
>>
>>     Plz verify it and send a patch with testcase pr111941.c if you
>>     confirm it has been fixed on the trunk.
>>
>>     Thanks.
>>     ------------------------------------------------------------------------
>>     juzhe.zhong@rivai.ai
>>
>>         *From:* Patrick O'Neill <mailto:patrick@rivosinc.com>
>>         *Date:* 2023-10-24 02:30
>>         *To:* Lehua Ding <mailto:lehua.ding@rivai.ai>
>>         *CC:* kito.cheng <mailto:kito.cheng@gmail.com>; rdapp.gcc
>>         <mailto:rdapp.gcc@gmail.com>; palmer
>>         <mailto:palmer@rivosinc.com>; Jeff Law
>>         <mailto:jeffreyalaw@gmail.com>; gcc-patches
>>         <mailto:gcc-patches@gcc.gnu.org>; 钟居哲
>>         <mailto:juzhe.zhong@rivai.ai>
>>         *Subject:* Re: [PATCH V3 00/11] Refactor and cleanup vsetvl pass
>>         Hi Lehua,
>>         This patch causes a build failure with newlib 4.1.0 with
>>         -march=rv64gv_zbb.
>>         I've creduced the failure here:
>>         https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111941
>>         Thanks,
>>         Patrick
>>         On 10/19/23 20:58, Lehua Ding wrote:
>>         > Committed, thanks Patrick and Juzhe.
>>         >
>>         > On 2023/10/20 2:04, Patrick O'Neill wrote:
>>         >> I tested it this morning on my machine and it passed!
>>         >>
>>         >> Tested against:
>>         >> 04d6c74564b7eb51660a00b35353aeab706b5a50
>>         >>
>>         >> Using targets:
>>         >> glibc rv32gcv qemu
>>         >> glibc rv64gcv qemu
>>         >>
>>         >> This patch series does not introduce any new failures.
>>         >>
>>         >> Here's a list of *resolved* failures by this patch series:
>>         >> rv64gcv:
>>         >> FAIL: gfortran.dg/host_assoc_function_7.f90   -O3
>>         >> -fomit-frame-pointer -funroll-loops -fpeel-loops -ftracer
>>         >> -finline-functions  execution test
>>         >> FAIL: gfortran.dg/host_assoc_function_7.f90   -O3 -g
>>         execution test
>>         >>
>>         >> rv32gcv:
>>         >> FAIL: gcc.target/riscv/rvv/autovec/binop/narrow_run-1.c
>>         execution test
>>         >> FAIL: gfortran.dg/host_assoc_function_7.f90   -O3
>>         >> -fomit-frame-pointer -funroll-loops -fpeel-loops -ftracer
>>         >> -finline-functions  execution test
>>         >> FAIL: gfortran.dg/host_assoc_function_7.f90   -O3 -g
>>         execution test
>>         >>
>>         >> Thanks for the quick revision Lehua!
>>         >>
>>         >> Tested-by: Patrick O'Neill <patrick@rivosinc.com>
>>         >>
>>         >> Patrick
>>         >>
>>         >> On 10/19/23 01:50, 钟居哲 wrote:
>>         >>> LGTM now. But wait for Patrick CI testing.
>>         >>>
>>         >>> Hi, @Patrick. Could you apply this patch and trigger CI
>>         in your
>>         >>> github  so that we can see the full running result.
>>         >>>
>>         >>> Issues · patrick-rivos/riscv-gnu-toolchain · GitHub
>>         >>> <https://github.com/patrick-rivos/riscv-gnu-toolchain/issues>
>>         >>>
>>         >>>
>>         ------------------------------------------------------------------------
>>
>>         >>>
>>         >>> juzhe.zhong@rivai.ai
>>         >>>
>>         >>>     *From:* Lehua Ding <mailto:lehua.ding@rivai.ai>
>>         >>>     *Date:* 2023-10-19 16:33
>>         >>>     *To:* gcc-patches <mailto:gcc-patches@gcc.gnu.org>
>>         >>>     *CC:* juzhe.zhong <mailto:juzhe.zhong@rivai.ai>;
>>         kito.cheng
>>         >>> <mailto:kito.cheng@gmail.com>; rdapp.gcc
>>         >>> <mailto:rdapp.gcc@gmail.com>; palmer
>>         <mailto:palmer@rivosinc.com>;
>>         >>>     jeffreyalaw <mailto:jeffreyalaw@gmail.com>; lehua.ding
>>         >>> <mailto:lehua.ding@rivai.ai>
>>         >>>     *Subject:* [PATCH V3 00/11] Refactor and cleanup
>>         vsetvl pass
>>         >>>     This patch refactors and cleanups the vsetvl pass in
>>         order to make
>>         >>>     the code
>>         >>>     easier to modify and understand. This patch does
>>         several things:
>>         >>>     1. Introducing a virtual CFG for vsetvl infos and
>>         Phase 1, 2 and 3
>>         >>>     only maintain
>>         >>>        and modify this virtual CFG. Phase 4 performs
>>         insertion,
>>         >>>     modification and
>>         >>>        deletion of vsetvl insns based on the virtual CFG.
>>         The Basic
>>         >>>     block in the
>>         >>>        virtual CFG is called vsetvl_block_info and the vsetvl
>>         >>>     information inside
>>         >>>        is called vsetvl_info.
>>         >>>     2. Combine Phase 1 and 2 into a single Phase 1 and
>>         unified the
>>         >>>     demand system,
>>         >>>        this Phase only fuse local vsetvl info in forward
>>         direction.
>>         >>>     3. Refactor Phase 3, change the logic for determining
>>         whether to
>>         >>>     uplift vsetvl
>>         >>>        info to a pred basic block to a more unified
>>         method that there
>>         >>>     is a vsetvl
>>         >>>        info in the vsetvl defintion reaching in
>>         compatible with it.
>>         >>>     4. Place all modification operations to the RTL in
>>         Phase 4 and
>>         >>>     Phase 5.
>>         >>>        Phase 4 is responsible for inserting, modifying
>>         and deleting
>>         >>> vsetvl
>>         >>>        instructions based on fully optimized vsetvl
>>         infos. Phase 5
>>         >>>     removes the avl
>>         >>>        operand from the RVV instruction and removes the
>>         unused dest
>>         >>>     operand
>>         >>>        register from the vsetvl insns.
>>         >>>     These modifications resulted in some testcases
>>         needing to be
>>         >>>     updated. The reasons
>>         >>>     for updating are summarized below:
>>         >>>     1. more optimized
>>         >>> vlmax_back_prop-25.c/vlmax_back_prop-26.c/vlmax_conflict-3.c/
>>         >>> vlmax_conflict-12.c/vsetvl-13.c/vsetvl-23.c/
>>         >>> avl_single-23.c/avl_single-89.c/avl_single-95.c/pr109773-1.c
>>         >>>     2. less unnecessary fusion
>>         >>> avl_single-46.c/imm_bb_prop-1.c/pr109743-2.c/vsetvl-18.c
>>         >>>     3. local fuse direction (backward -> forward)
>>         >>>        scalar_move-1.c/
>>         >>>     4. add some bugfix testcases.
>>         >>>        pr111037-3.c/pr111037-4.c
>>         >>>        avl_single-89.c
>>         >>>     PR target/111037
>>         >>>     PR target/111234
>>         >>>     PR target/111725
>>         >>>     Lehua Ding (11):
>>         >>>       RISC-V: P1: Refactor
>>         >>> avl_info/vl_vtype_info/vector_insn_info/vector_block_info
>>         >>>       RISC-V: P2: Refactor and cleanup demand system
>>         >>>       RISC-V: P3: Refactor vector_infos_manager
>>         >>>       RISC-V: P4: move method from pass_vsetvl to pre_vsetvl
>>         >>>       RISC-V: P5: combine phase 1 and 2
>>         >>>       RISC-V: P6: Add computing reaching definition data flow
>>         >>>       RISC-V: P7: Move earliest fuse and lcm code to
>>         pre_vsetvl class
>>         >>>       RISC-V: P8: Refactor emit-vsetvl phase and delete post
>>         >>> optimization
>>         >>>       RISC-V: P9: Cleanup and reorganize helper functions
>>         >>>       RISC-V: P10: Delete riscv-vsetvl.h and adjust
>>         riscv-vsetvl.def
>>         >>>       RISC-V: P11: Adjust and add testcases
>>         >>> gcc/config/riscv/riscv-vsetvl.cc              | 6502
>>         >>> +++++++----------
>>         >>> gcc/config/riscv/riscv-vsetvl.def             |  641 +-
>>         >>> gcc/config/riscv/riscv-vsetvl.h               |  488 --
>>         >>> gcc/config/riscv/t-riscv                      |    2 +-
>>         >>> .../gcc.target/riscv/rvv/base/scalar_move-1.c |    2 +-
>>         >>> .../riscv/rvv/vsetvl/avl_single-104.c         |   35 +
>>         >>> .../riscv/rvv/vsetvl/avl_single-105.c         |   23 +
>>         >>> .../riscv/rvv/vsetvl/avl_single-106.c         |   34 +
>>         >>> .../riscv/rvv/vsetvl/avl_single-107.c         |   41 +
>>         >>> .../riscv/rvv/vsetvl/avl_single-108.c         |   41 +
>>         >>> .../riscv/rvv/vsetvl/avl_single-109.c         |   45 +
>>         >>> .../riscv/rvv/vsetvl/avl_single-23.c          |    7 +-
>>         >>> .../riscv/rvv/vsetvl/avl_single-46.c          |    3 +-
>>         >>> .../riscv/rvv/vsetvl/avl_single-84.c          |    5 +-
>>         >>> .../riscv/rvv/vsetvl/avl_single-89.c          |    8 +-
>>         >>> .../riscv/rvv/vsetvl/avl_single-95.c          |    2 +-
>>         >>> .../riscv/rvv/vsetvl/imm_bb_prop-1.c          |    7 +-
>>         >>> .../gcc.target/riscv/rvv/vsetvl/pr109743-2.c  |    2 +-
>>         >>> .../gcc.target/riscv/rvv/vsetvl/pr109773-1.c  |    2 +-
>>         >>>     .../riscv/rvv/{base => vsetvl}/pr111037-1.c   |    0
>>         >>>     .../riscv/rvv/{base => vsetvl}/pr111037-2.c   |    0
>>         >>> .../gcc.target/riscv/rvv/vsetvl/pr111037-3.c  |   16 +
>>         >>> .../gcc.target/riscv/rvv/vsetvl/pr111037-4.c  |   16 +
>>         >>> .../riscv/rvv/vsetvl/vlmax_back_prop-25.c     |   10 +-
>>         >>> .../riscv/rvv/vsetvl/vlmax_back_prop-26.c     |   10 +-
>>         >>> .../riscv/rvv/vsetvl/vlmax_conflict-12.c      |    1 -
>>         >>> .../riscv/rvv/vsetvl/vlmax_conflict-3.c       |    2 +-
>>         >>> .../gcc.target/riscv/rvv/vsetvl/vsetvl-13.c   |    4 +-
>>         >>> .../gcc.target/riscv/rvv/vsetvl/vsetvl-18.c   |    4 +-
>>         >>> .../gcc.target/riscv/rvv/vsetvl/vsetvl-23.c   |    2 +-
>>         >>>     30 files changed, 3263 insertions(+), 4692 deletions(-)
>>         >>>     delete mode 100644 gcc/config/riscv/riscv-vsetvl.h
>>         >>>     create mode 100644
>>         >>> gcc/testsuite/gcc.target/riscv/rvv/vsetvl/avl_single-104.c
>>         >>>     create mode 100644
>>         >>> gcc/testsuite/gcc.target/riscv/rvv/vsetvl/avl_single-105.c
>>         >>>     create mode 100644
>>         >>> gcc/testsuite/gcc.target/riscv/rvv/vsetvl/avl_single-106.c
>>         >>>     create mode 100644
>>         >>> gcc/testsuite/gcc.target/riscv/rvv/vsetvl/avl_single-107.c
>>         >>>     create mode 100644
>>         >>> gcc/testsuite/gcc.target/riscv/rvv/vsetvl/avl_single-108.c
>>         >>>     create mode 100644
>>         >>> gcc/testsuite/gcc.target/riscv/rvv/vsetvl/avl_single-109.c
>>         >>>     rename gcc/testsuite/gcc.target/riscv/rvv/{base =>
>>         >>>     vsetvl}/pr111037-1.c (100%)
>>         >>>     rename gcc/testsuite/gcc.target/riscv/rvv/{base =>
>>         >>>     vsetvl}/pr111037-2.c (100%)
>>         >>>     create mode 100644
>>         >>> gcc/testsuite/gcc.target/riscv/rvv/vsetvl/pr111037-3.c
>>         >>>     create mode 100644
>>         >>> gcc/testsuite/gcc.target/riscv/rvv/vsetvl/pr111037-4.c
>>         >>>     --     2.36.3
>>         >>>
>>         >
>>

  reply	other threads:[~2023-10-23 23:42 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-10-19  8:33 Lehua Ding
2023-10-19  8:33 ` [PATCH V3 01/11] RISC-V: P1: Refactor avl_info/vl_vtype_info/vector_insn_info/vector_block_info Lehua Ding
2023-10-19  8:33 ` [PATCH V3 02/11] RISC-V: P2: Refactor and cleanup demand system Lehua Ding
2023-10-19  8:33 ` [PATCH V3 03/11] RISC-V: P3: Refactor vector_infos_manager Lehua Ding
2023-10-19  8:33 ` [PATCH V3 04/11] RISC-V: P4: move method from pass_vsetvl to pre_vsetvl Lehua Ding
2023-10-19  8:33 ` [PATCH V3 05/11] RISC-V: P5: Combine phase 1 and 2 Lehua Ding
2023-10-19  8:33 ` [PATCH V3 06/11] RISC-V: P6: Add computing reaching definition data flow Lehua Ding
2023-10-19  8:33 ` [PATCH V3 07/11] RISC-V: P7: Move earliest fuse and lcm code to pre_vsetvl class Lehua Ding
2023-10-19  8:33 ` [PATCH V3 08/11] RISC-V: P8: Refactor emit-vsetvl phase and delete post optimization Lehua Ding
2023-10-19  8:33 ` [PATCH V3 09/11] RISC-V: P9: Cleanup and reorganize helper functions Lehua Ding
2023-10-19  8:33 ` [PATCH V3 10/11] RISC-V: P10: Delete riscv-vsetvl.h and adjust riscv-vsetvl.def Lehua Ding
2023-10-19  8:33 ` [PATCH V3 11/11] RISC-V: P11: Adjust and add testcases Lehua Ding
2023-10-19  8:38 ` [PATCH V3 00/11] Refactor and cleanup vsetvl pass Robin Dapp
2023-10-19  8:43   ` Lehua Ding
2023-10-19  8:50 ` 钟居哲
2023-10-19 18:04   ` Patrick O'Neill
2023-10-20  2:20     ` Lehua Ding
2023-10-20  3:58     ` Lehua Ding
2023-10-23 18:30       ` Patrick O'Neill
2023-10-23 21:41         ` 钟居哲
2023-10-23 22:46           ` Patrick O'Neill
2023-10-23 22:50             ` 钟居哲
2023-10-23 23:42               ` Patrick O'Neill [this message]
2023-10-24  0:51                 ` juzhe.zhong
2023-10-24  1:01                   ` Patrick O'Neill
2023-10-24  2:27                     ` juzhe.zhong

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=ac46c375-a4df-7dc4-cf98-b93dbe3f24d0@rivosinc.com \
    --to=patrick@rivosinc.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=jeffreyalaw@gmail.com \
    --cc=juzhe.zhong@rivai.ai \
    --cc=kito.cheng@gmail.com \
    --cc=lehua.ding@rivai.ai \
    --cc=palmer@rivosinc.com \
    --cc=rdapp.gcc@gmail.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).