[Bug sanitizer/64888] New: ubsan doesn't work with openmp

[Bug sanitizer/64888] New: ubsan doesn't work with openmp

[mikulas at artax dot karlin.mff.cuni.cz]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="UTF-8", Size: 4558 bytes --]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

            Bug ID: 64888
           Summary: ubsan doesn't work with openmp
           Product: gcc
           Version: 4.9.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: sanitizer
          Assignee: unassigned at gcc dot gnu.org
          Reporter: mikulas at artax dot karlin.mff.cuni.cz
                CC: dodji at gcc dot gnu.org, dvyukov at gcc dot gnu.org,
                    jakub at gcc dot gnu.org, kcc at gcc dot gnu.org

Undefined behavior sanitizer and openmp don't work together. Ubsan generates
some internal variables and openmp complains that shareability of these
variables haven't been defined. Try to compile this program with
gcc -fsanitize=undefined -fopenmp openmp-ubsan.c

#include <omp.h>

int main(void)
{
        unsigned a = 1;
        unsigned b = 2;
        unsigned c;
#pragma omp parallel default(none) shared(a,b) private(c)
        {
                c = a / b;
        }
}

openmp-ubsan.c: In function ‘main’:
openmp-ubsan.c:10:9: error: ‘*.Lubsan_data0’ not specified in enclosing
parallel
   c = a / b;
         ^
openmp-ubsan.c:8:9: error: enclosing parallel
 #pragma omp parallel default(none) shared(a,b) private(c)
         ^
>From gcc-bugs-return-475644-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Sat Jan 31 17:23:44 2015
Return-Path: <gcc-bugs-return-475644-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 30289 invoked by alias); 31 Jan 2015 17:23:43 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 30259 invoked by uid 48); 31 Jan 2015 17:23:40 -0000
From: "ysato at users dot sourceforge.jp" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug target/64889] New: [h8300] ICE maybe_record_trace_start, at dwarf2cfi.c:2318
Date: Sat, 31 Jan 2015 17:23:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: new
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: target
X-Bugzilla-Version: 5.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: ysato at users dot sourceforge.jp
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: bug_id short_desc product version bug_status bug_severity priority component assigned_to reporter attachments.created
Message-ID: <bug-64889-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2015-01/txt/msg03638.txt.bz2
Content-length: 1401

https://gcc.gnu.org/bugzilla/show_bug.cgi?idd889

            Bug ID: 64889
           Summary: [h8300] ICE maybe_record_trace_start, at
                    dwarf2cfi.c:2318
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: ysato at users dot sourceforge.jp

Created attachment 34639
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id4639&actioníit
reproduce on ICE

reporoduce code attached.

$ h8300-elf-gcc -mh -mint32 -Os -g -fomit-frame-pointer -c namei.c
namei.c: In function 'SYSC_mknodat':
namei.c:38:6: warning: implicit declaration of function 'new_decode_dev'
[-Wimplicit-function-declaration]
      new_decode_dev(dev));
      ^
../../linux/namei.c:42:1: internal compiler error: in maybe_record_trace_start,
at dwarf2cfi.c:2318
 }
 ^
0x6d1d87 maybe_record_trace_start
    ../../gcc/dwarf2cfi.c:2318
0x6d2640 scan_trace
    ../../gcc/dwarf2cfi.c:2496
0x6d2e4a create_cfi_notes
    ../../gcc/dwarf2cfi.c:2650
0x6d2e4a execute_dwarf2_frame
    ../../gcc/dwarf2cfi.c:3006
0x6d2e4a execute
    ../../gcc/dwarf2cfi.c:3486
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.

[Bug sanitizer/64888] ubsan doesn't work with openmp

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

Marek Polacek <mpolacek at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2015-01-31
                 CC|                            |mpolacek at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #1 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
Yikes.  Confirmed.

[Bug sanitizer/64888] ubsan doesn't work with openmp

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

--- Comment #2 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
Started looking into this.  Seems that we should handle specially those
artificial ubsan decls such as *.Lubsan_data in omp_default_clause, i.e. treat
them as shared?  But I hardly know what I'm talking about; inviting Jakub to
comment.

[Bug sanitizer/64888] ubsan doesn't work with openmp

[mikulas at artax dot karlin.mff.cuni.cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

--- Comment #3 from mikulas at artax dot karlin.mff.cuni.cz ---
I think you should treat these artifical variables as private, not shared. If
you treated them as shared, threads would race with each other when accessing
them.

[Bug sanitizer/64888] ubsan doesn't work with openmp

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

--- Comment #4 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
But these *.Lubsan_data0 are read-only I think, so I don't think there's a
problem if multiple threads read such a variable.

[Bug sanitizer/64888] ubsan doesn't work with openmp

[claudiokozicky at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

Claudio Kozický <claudiokozicky at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |claudiokozicky at gmail dot com

--- Comment #5 from Claudio Kozický <claudiokozicky at gmail dot com> ---
Created attachment 51627
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51627&action=edit
A minimal testcase to reproduce the bug.

[Bug sanitizer/64888] ubsan doesn't work with openmp

[claudiokozicky at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

--- Comment #6 from Claudio Kozický <claudiokozicky at gmail dot com> ---
Created attachment 51628
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51628&action=edit
GCC version

This is the version of GCC with which I have reproduced the bug with.

[Bug sanitizer/64888] ubsan doesn't work with openmp

[claudiokozicky at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

Claudio Kozický <claudiokozicky at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #51628|0                           |1
        is obsolete|                            |

--- Comment #7 from Claudio Kozický <claudiokozicky at gmail dot com> ---
Created attachment 51629
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51629&action=edit
GCC 11.1.0 compilation log on Arch Linux

executed command: gcc -v -save-temps -fsanitize=undefined -fopenmp
minimal_example.c

[Bug sanitizer/64888] ubsan doesn't work with openmp

[claudiokozicky at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

--- Comment #8 from Claudio Kozický <claudiokozicky at gmail dot com> ---
Created attachment 51630
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51630&action=edit
GCC 11.1.0 preprocessed source

executed command: gcc -v -save-temps -fsanitize=undefined -fopenmp
minimal_example.c

[Bug middle-end/64888] ubsan doesn't work with openmp

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |jakub at gcc dot gnu.org

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 51636
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51636&action=edit
gcc12-pr64888.patch

Untested fix.

[Bug middle-end/64888] ubsan doesn't work with openmp

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

--- Comment #10 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:40dd9d839e52f679d8eabc1c5ca0ca17a5ccfd14

commit r12-4599-g40dd9d839e52f679d8eabc1c5ca0ca17a5ccfd14
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Thu Oct 21 10:27:44 2021 +0200

    openmp: For default(none) ignore variables created by ubsan_create_data
[PR64888]

    We weren't ignoring the ubsan variables created by c-ubsan.c before
gimplification
    (others are added later).  One way to fix this would be to introduce
further
    UBSAN_ internal functions and lower it later (sanopt pass) like other ifns,
    this patch instead recognizes those magic vars by name/name of type and
DECL_ARTIFICIAL
    and TYPE_ARTIFICIAL.

    2021-10-21  Jakub Jelinek  <jakub@redhat.com>

            PR middle-end/64888
    gcc/c-family/
            * c-omp.c (c_omp_predefined_variable): Return true also for
            ubsan_create_data created artificial variables.
    gcc/testsuite/
            * c-c++-common/ubsan/pr64888.c: New test.

[Bug middle-end/64888] ubsan doesn't work with openmp

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

--- Comment #11 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:930ffe933186b66ee33479fe384d4aa08e9f7459

commit r11-9327-g930ffe933186b66ee33479fe384d4aa08e9f7459
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Thu Oct 21 10:27:44 2021 +0200

    openmp: For default(none) ignore variables created by ubsan_create_data
[PR64888]

    We weren't ignoring the ubsan variables created by c-ubsan.c before
gimplification
    (others are added later).  One way to fix this would be to introduce
further
    UBSAN_ internal functions and lower it later (sanopt pass) like other ifns,
    this patch instead recognizes those magic vars by name/name of type and
DECL_ARTIFICIAL
    and TYPE_ARTIFICIAL.

    2021-10-21  Jakub Jelinek  <jakub@redhat.com>

            PR middle-end/64888
    gcc/c-family/
            * c-omp.c (c_omp_predefined_variable): Return true also for
            ubsan_create_data created artificial variables.
    gcc/testsuite/
            * c-c++-common/ubsan/pr64888.c: New test.

    (cherry picked from commit 40dd9d839e52f679d8eabc1c5ca0ca17a5ccfd14)

[Bug middle-end/64888] ubsan doesn't work with openmp

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

--- Comment #12 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:cf3a927af0c040478280d90559e3e3088256c6fc

commit r10-10651-gcf3a927af0c040478280d90559e3e3088256c6fc
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Thu Oct 21 10:27:44 2021 +0200

    openmp: For default(none) ignore variables created by ubsan_create_data
[PR64888]

    We weren't ignoring the ubsan variables created by c-ubsan.c before
gimplification
    (others are added later).  One way to fix this would be to introduce
further
    UBSAN_ internal functions and lower it later (sanopt pass) like other ifns,
    this patch instead recognizes those magic vars by name/name of type and
DECL_ARTIFICIAL
    and TYPE_ARTIFICIAL.

    2021-10-21  Jakub Jelinek  <jakub@redhat.com>

            PR middle-end/64888
    gcc/c-family/
            * c-omp.c (c_omp_predefined_variable): Return true also for
            ubsan_create_data created artificial variables.
    gcc/testsuite/
            * c-c++-common/ubsan/pr64888.c: New test.

    (cherry picked from commit 40dd9d839e52f679d8eabc1c5ca0ca17a5ccfd14)

[Bug middle-end/64888] ubsan doesn't work with openmp

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

--- Comment #13 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-9 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:a76e866211798da138c51abea390bcd605b6da02

commit r9-10107-ga76e866211798da138c51abea390bcd605b6da02
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Thu Oct 21 10:27:44 2021 +0200

    openmp: For default(none) ignore variables created by ubsan_create_data
[PR64888]

    We weren't ignoring the ubsan variables created by c-ubsan.c before
gimplification
    (others are added later).  One way to fix this would be to introduce
further
    UBSAN_ internal functions and lower it later (sanopt pass) like other ifns,
    this patch instead recognizes those magic vars by name/name of type and
DECL_ARTIFICIAL
    and TYPE_ARTIFICIAL.

    2021-10-21  Jakub Jelinek  <jakub@redhat.com>

            PR middle-end/64888
    gcc/c-family/
            * c-omp.c (c_omp_predefined_variable): Return true also for
            ubsan_create_data created artificial variables.
    gcc/testsuite/
            * c-c++-common/ubsan/pr64888.c: New test.

    (cherry picked from commit 40dd9d839e52f679d8eabc1c5ca0ca17a5ccfd14)

[Bug middle-end/64888] ubsan doesn't work with openmp

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64888

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|ASSIGNED                    |RESOLVED

--- Comment #14 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed.