public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "acoplan at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug target/98177] New: [11 Regression] SVE: ICE in expand_direct_optab_fn, at internal-fn.c:3368
Date: Mon, 07 Dec 2020 15:21:19 +0000	[thread overview]
Message-ID: <bug-98177-4@http.gcc.gnu.org/bugzilla/> (raw)

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98177

            Bug ID: 98177
           Summary: [11 Regression] SVE: ICE in expand_direct_optab_fn, at
                    internal-fn.c:3368
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: acoplan at gcc dot gnu.org
  Target Milestone: ---

For the following C++ testcase:

int a, b;
short c;
void d(long e) {
  for (int f = 0; f < b; f += 1)
    for (short g = 0; g < c; g += 5)
      a = (short)e;
}

we ICE with -Ofast -march=armv8.2-a+sve -msve-vector-bits=128 since
r11-4912-g46c705e70e078f6a1920d92e49042125d5e18495.

To reproduce:

$ aarch64-elf-gcc -c -S -Ofast -march=armv8.2-a+sve -msve-vector-bits=128
test.cc
during RTL pass: expand
test.cc: In function 'void d(long int)':
test.cc:3:6: internal compiler error: in expand_direct_optab_fn, at
internal-fn.c:3368
    3 | void d(long e) {
      |      ^
0xd63e0f expand_direct_optab_fn
        /home/alecop01/toolchain/src/gcc/gcc/internal-fn.c:3368
0xd64978 expand_REDUC_MAX
        /home/alecop01/toolchain/src/gcc/gcc/internal-fn.def:209
0xd6631e expand_internal_call(internal_fn, gcall*)
        /home/alecop01/toolchain/src/gcc/gcc/internal-fn.c:4089
0xd66338 expand_internal_call(gcall*)
        /home/alecop01/toolchain/src/gcc/gcc/internal-fn.c:4097
0xae74ad expand_call_stmt
        /home/alecop01/toolchain/src/gcc/gcc/cfgexpand.c:2740
0xae74ad expand_gimple_stmt_1
        /home/alecop01/toolchain/src/gcc/gcc/cfgexpand.c:3835
0xae74ad expand_gimple_stmt
        /home/alecop01/toolchain/src/gcc/gcc/cfgexpand.c:3999
0xaf1827 expand_gimple_basic_block
        /home/alecop01/toolchain/src/gcc/gcc/cfgexpand.c:6043
0xaf2ede execute
        /home/alecop01/toolchain/src/gcc/gcc/cfgexpand.c:6727
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.

Curiously, I can only reproduce the ICE when the testcase is treated as C++. It
seems we fail to vectorize the testcase if we instead treat it as C.

             reply	other threads:[~2020-12-07 15:21 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-12-07 15:21 acoplan at gcc dot gnu.org [this message]
2020-12-07 15:21 ` [Bug target/98177] " acoplan at gcc dot gnu.org
2020-12-07 17:22 ` ktkachov at gcc dot gnu.org
2020-12-08  7:10 ` rguenth at gcc dot gnu.org
2020-12-18 18:21 ` cvs-commit at gcc dot gnu.org
2021-01-04  9:53 ` wirkus at gcc dot gnu.org

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=bug-98177-4@http.gcc.gnu.org/bugzilla/ \
    --to=gcc-bugzilla@gcc.gnu.org \
    --cc=gcc-bugs@gcc.gnu.org \
    /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).