[Bug tree-optimization/59617] New: [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[Bug tree-optimization/59617] New: [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[kirill.yukhin at intel dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

            Bug ID: 59617
           Summary: [vectorizer] ICE in vectorizable_mask_load_store with
                    AVX-512F's gathers enabled.
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: kirill.yukhin at intel dot com

Created attachment 31529
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31529&action=edit
Reproducer

Hello, I am going to check in a patch, which will
enable AVX-512F new gathers instructions.

New gathers are use mask registers (%kN), while
vectorizable_mask_load_store asserts mask type to be compatible
with operand type.

This fails 416.gamess to build with -mavx512f -Ofast.

Reproducer attached.

Reproduce:
$ gfortran -S -Ofast -mavx512f hss2a.fppized.f

Back trace:
0xbcfc27 vectorizable_mask_load_store
        /export/users/kyukhin/gcc/git/gcc/gcc/tree-vect-stmts.c:1901
0xbddf6c vectorizable_call
        /export/users/kyukhin/gcc/git/gcc/gcc/tree-vect-stmts.c:2172
0xbe1021 vect_transform_stmt(gimple_statement_base*, gimple_stmt_iterator*,
bool*, _slp_tree*, _slp_instance*)
        /export/users/kyukhin/gcc/git/gcc/gcc/tree-vect-stmts.c:7017
0xbe4731 vect_transform_loop(_loop_vec_info*)
        /export/users/kyukhin/gcc/git/gcc/gcc/tree-vect-loop.c:6046
0xc00838 vectorize_loops()
        /export/users/kyukhin/gcc/git/gcc/gcc/tree-vectorizer.c:476

I'll check my patch in with disabled gathers, so to enable it -
remove `#if 0' in gcc/config/i386/i386.c

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Author: jakub
Date: Mon Jan 13 19:16:30 2014
New Revision: 206585

URL: http://gcc.gnu.org/viewcvs?rev=206585&root=gcc&view=rev
Log:
    PR target/59617
    * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
    AVX512F gather builtins.
    * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
    on gather decls with INTEGER_TYPE masktype.
    (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
    directly into the builtin rather than hoisting it before loop.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/i386/i386.c
    trunk/gcc/tree-vect-stmts.c

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[kyukhin at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #2 from Kirill Yukhin <kyukhin at gcc dot gnu.org> ---
Author: kyukhin
Date: Wed Jan 29 17:35:19 2014
New Revision: 207274

URL: http://gcc.gnu.org/viewcvs?rev=207274&root=gcc&view=rev
Log:
gcc/testsuite/

        PR target/59617
    * gcc.target/i386/avx512f-gather-2.c: Remove XPASS.
    * gcc.target/i386/avx512f-gather-5.c: Ditto.



Modified:
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.target/i386/avx512f-gather-2.c
    trunk/gcc/testsuite/gcc.target/i386/avx512f-gather-5.c

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
The test gcc.target/i386/avx512f-gather-5.c fails on darwin

FAIL: gcc.target/i386/avx512f-gather-5.c scan-assembler gather[^\\n]*zmm

There is no 'gather' in the avx512f-gather-5.s file.

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #4 from Uroš Bizjak <ubizjak at gmail dot com> ---
(In reply to Dominique d'Humieres from comment #3)
> The test gcc.target/i386/avx512f-gather-5.c fails on darwin
> 
> FAIL: gcc.target/i386/avx512f-gather-5.c scan-assembler gather[^\\n]*zmm
> 
> There is no 'gather' in the avx512f-gather-5.s file.

I have compiled the preprocessed file with a crosscompiler from
x86_64-pc-linux-gnu to x86_64-apple-darwin10.8.0 (gcc was configured with
--target=x86_64-apple-darwin10.8.0) without problems.

(preprocessed file was obtained from the native x86_64-pc-linux-gnu
compilation).

$ ~/gcc-build-xxx/gcc/cc1 -O3 -mavx512f -quiet avx512f-gather-5.i
$ grep gather avx512f-gather-5.s
        vgatherdps      (%rbx,%zmm0,4), %zmm2{%k2}
        vgatherdps      (%rdx,%zmm0,4), %zmm1{%k3}
>From gcc-bugs-return-448095-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Wed Apr 02 07:53:14 2014
Return-Path: <gcc-bugs-return-448095-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 9427 invoked by alias); 2 Apr 2014 07:53:14 -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 9392 invoked by uid 55); 2 Apr 2014 07:53:10 -0000
From: "rguenth at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug bootstrap/60719] With --program-prefix=$target_alias --program-suffix=-$version install-driver breaks
Date: Wed, 02 Apr 2014 07:53:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: bootstrap
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords: build
X-Bugzilla-Severity: normal
X-Bugzilla-Who: rguenth at gcc dot gnu.org
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:
Message-ID: <bug-60719-4-PcwCmGw3E8@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-60719-4@http.gcc.gnu.org/bugzilla/>
References: <bug-60719-4@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: 2014-04/txt/msg00115.txt.bz2
Content-length: 478

http://gcc.gnu.org/bugzilla/show_bug.cgi?id`719

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Wed Apr  2 07:52:36 2014
New Revision: 209010

URL: http://gcc.gnu.org/viewcvs?rev 9010&root=gcc&view=rev
Log:
2014-04-02  Richard Biener  <rguenther@suse.de>

    PR bootstrap/60719
    * Makefile.in (install-driver): Guard extra installs with special
    names properly.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/Makefile.in

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #5 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
Created attachment 32530
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32530&action=edit
Preprocessed file for avx512f-gather-5.c on x86_64-apple-darwin13

(In reply to UroÅ Bizjak from comment #4)
> I have compiled the preprocessed file with a crosscompiler from
> x86_64-pc-linux-gnu to x86_64-apple-darwin10.8.0 (gcc was configured with
> --target=x86_64-apple-darwin10.8.0) without problems.
>
> (preprocessed file was obtained from the native x86_64-pc-linux-gnu
> compilation).
>
> $ ~/gcc-build-xxx/gcc/cc1 -O3 -mavx512f -quiet avx512f-gather-5.i
> $ grep gather avx512f-gather-5.s
>        vgatherdps      (%rbx,%zmm0,4), %zmm2{%k2}
>        vgatherdps      (%rdx,%zmm0,4), %zmm1{%k3}

Could you compare your preprocessed file obtained from the native
x86_64-pc-linux-gnu to the attached one obtained from the native
x86_64-apple-darwin13?
>From gcc-bugs-return-448171-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Thu Apr 03 08:59:17 2014
Return-Path: <gcc-bugs-return-448171-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 30118 invoked by alias); 3 Apr 2014 08:59:17 -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 30057 invoked by uid 48); 3 Apr 2014 08:59:07 -0000
From: "ubizjak at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.
Date: Thu, 03 Apr 2014 08:59:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: ubizjak at gmail dot com
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: attachments.created
Message-ID: <bug-59617-4-Xv9HdrK43Q@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59617-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59617-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-04/txt/msg00191.txt.bz2
Content-length: 320

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #6 from Uroš Bizjak <ubizjak at gmail dot com> ---
Created attachment 32531
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32531&action=edit
Preprocessed file for avx512f-gather-5.c on x86_64-pc-linux-gnu

Preprocessed file from Comment #4.
>From gcc-bugs-return-448172-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Thu Apr 03 09:04:44 2014
Return-Path: <gcc-bugs-return-448172-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 32634 invoked by alias); 3 Apr 2014 09:04:44 -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 32594 invoked by uid 48); 3 Apr 2014 09:04:39 -0000
From: "ubizjak at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.
Date: Thu, 03 Apr 2014 09:04:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: ubizjak at gmail dot com
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:
Message-ID: <bug-59617-4-Uk5LoKA2H0@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59617-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59617-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-04/txt/msg00192.txt.bz2
Content-length: 1423

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #7 from Uroš Bizjak <ubizjak at gmail dot com> ---
(In reply to Dominique d'Humieres from comment #5)
> Created attachment 32530 [details]
> Preprocessed file for avx512f-gather-5.c on x86_64-apple-darwin13
> 
> (In reply to UroÅ Bizjak from comment #4)
> > I have compiled the preprocessed file with a crosscompiler from
> > x86_64-pc-linux-gnu to x86_64-apple-darwin10.8.0 (gcc was configured with
> > --target=x86_64-apple-darwin10.8.0) without problems.
> >
> > (preprocessed file was obtained from the native x86_64-pc-linux-gnu
> > compilation).
> >
> > $ ~/gcc-build-xxx/gcc/cc1 -O3 -mavx512f -quiet avx512f-gather-5.i
> > $ grep gather avx512f-gather-5.s
> >        vgatherdps      (%rbx,%zmm0,4), %zmm2{%k2}
> >        vgatherdps      (%rdx,%zmm0,4), %zmm1{%k3}
> 
> Could you compare your preprocessed file obtained from the native
> x86_64-pc-linux-gnu to the attached one obtained from the native
> x86_64-apple-darwin13?

linux preprocessed file has a couple of *__restrict modifiers in:

-# 6 "avx512f-gather-4.c" 2
+# 6 "/opt/gcc/_clean/gcc/testsuite/gcc.target/i386/avx512f-gather-4.c" 2


 int a[1024], b[1024], c[1024], d[1024];

 __attribute__((noinline, noclone)) void
-foo (float *__restrict p, float *__restrict q, float *__restrict r,
+foo (float * p, float * q, float * r,
      int s1, int s2, int s3)
 {
   int i;
>From gcc-bugs-return-448173-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Thu Apr 03 09:11:01 2014
Return-Path: <gcc-bugs-return-448173-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 2567 invoked by alias); 3 Apr 2014 09:11:00 -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 2529 invoked by uid 48); 3 Apr 2014 09:10:55 -0000
From: "ubizjak at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.
Date: Thu, 03 Apr 2014 09:11:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: ubizjak at gmail dot com
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:
Message-ID: <bug-59617-4-zDtDILpNsw@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59617-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59617-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-04/txt/msg00193.txt.bz2
Content-length: 483

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #8 from Uroš Bizjak <ubizjak at gmail dot com> ---
(In reply to Uroš Bizjak from comment #7)
> 
> linux preprocessed file has a couple of *__restrict modifiers in:

Indeed, there is no vgather in the resulting assembly when darwin version of
the preprocessed file is compiled. OTOH, when __restrict modifiers are added to
foo, darwin file compiles to the assembly that includes expected vgather insns.
>From gcc-bugs-return-448174-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Thu Apr 03 09:30:11 2014
Return-Path: <gcc-bugs-return-448174-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 16115 invoked by alias); 3 Apr 2014 09:30:10 -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 16049 invoked by uid 48); 3 Apr 2014 09:30:06 -0000
From: "pault at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug fortran/60717] Wrong code with recursive procedure with unlimited polymorphic dummy argument
Date: Thu, 03 Apr 2014 09:30:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: fortran
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: pault at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: pault at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-60717-4-HdDqqdYHxN@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-60717-4@http.gcc.gnu.org/bugzilla/>
References: <bug-60717-4@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: 2014-04/txt/msg00194.txt.bz2
Content-length: 612

http://gcc.gnu.org/bugzilla/show_bug.cgi?id`717

--- Comment #5 from Paul Thomas <pault at gcc dot gnu.org> ---
(In reply to Dominique d'Humieres from comment #4)
> > Could there be a connection http://gcc.gnu.org/bugzilla/show_bug.cgi?idX085 ?
>
> May be, but pr58085 is not fixed by the patch in comment 2 (submitted at
> http://gcc.gnu.org/ml/fortran/2014-04/msg00005.html).

Thanks for bringing this PR to my attention. I would point a finger at
trans-stmt.c(trans_associate_var), which, ignoring the error in comment 2, was
one of the areas that I suspected would benefit from this fix.

Cheers

Paul

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #9 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
If I compile the following test with -S -O3 -mavx512f -save-temps

// #include "avx512f-check.h"

#define N 1024
int a[N], b[N], c[N], d[N];

__attribute__((noinline, noclone)) void
foo (float *__restrict p, float *__restrict q, float *__restrict r,
     int s1, int s2, int s3)
{
  int i;
  for (i = 0; i < N; i++)
    p[i] = q[a[i] * s1 + b[i] * s2 + s3] * r[c[i] * s1 + d[i] * s2 + s3];
}

grepping the preprocess file for restrict I see

foo (float *__restrict p, float *__restrict q, float *__restrict r,

If I uncomment the include (adding
-I/opt/gcc/_clean/gcc/testsuite/gcc.target/i386/),
the restrict qualifiers disappear.

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #10 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
I have found the origin of the problem. It is due to

#include <stdlib.h>

which after some steps include

sys/cdefs.h

which contains the following lines

/*
 * We use `__restrict' as a way to define the `restrict' type qualifier
 * without disturbing older software that is unaware of C99 keywords.
 */
#if __STDC_VERSION__ < 199901
#define __restrict
#else
#define __restrict      restrict
#endif

Note that using something such as -D__STDC_VERSION__=201401 gives a lot of
errors of the kind

/usr/include/stdlib.h:154:51: error: conflicting types for 'restrict'
 size_t  mbstowcs(wchar_t * __restrict , const char * __restrict, size_t);

I don't have any idea about the best way to fix it.

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[kirill.yukhin at intel dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #11 from Yukhin Kirill <kirill.yukhin at intel dot com> ---
Maybe simply do:
#ifdef __restrict
#undef __restrict

In some common header (say, avx512f-check.h)?

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #12 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> Maybe simply do:
> #ifdef __restrict
> #undef __restrict
>
> In some common header (say, avx512f-check.h)?

It works (added in avx512f-check.h, with a #endif!-). It would probably require
a comment referring to darwin.

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #13 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
Another option would be to add -std=c99 for darwin.

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #15 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> I would suggest that this is the Right Thing to do.

I agree. Patch tested

--- ../_clean/gcc/testsuite/gcc.target/i386/avx512f-gather-5.c    2014-01-29
19:02:01.000000000 +0100
+++ gcc/testsuite/gcc.target/i386/avx512f-gather-5.c    2014-04-03
15:17:05.000000000 +0200
@@ -1,5 +1,6 @@
 /* { dg-do compile } */
 /* { dg-options "-O3 -mavx512f" } */
+/* { dg-additional-options "-std=c99" { target *-*-darwin* } } */

 #include "avx512f-gather-4.c"

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[kirill.yukhin at intel dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #16 from Yukhin Kirill <kirill.yukhin at intel dot com> ---
(In reply to Dominique d'Humieres from comment #15)
> 19:02:01.000000000 +0100
> +++ gcc/testsuite/gcc.target/i386/avx512f-gather-5.c	2014-04-03
> 15:17:05.000000000 +0200
> @@ -1,5 +1,6 @@
>  /* { dg-do compile } */
>  /* { dg-options "-O3 -mavx512f" } */
> +/* { dg-additional-options "-std=c99" { target *-*-darwin* } } */
>  
>  #include "avx512f-gather-4.c"

Then I think we need to add such a line in all files with __restrict keyword.
>From gcc-bugs-return-448197-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Thu Apr 03 13:25:08 2014
Return-Path: <gcc-bugs-return-448197-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 22561 invoked by alias); 3 Apr 2014 13:25:08 -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 22484 invoked by uid 48); 3 Apr 2014 13:25:04 -0000
From: "ubizjak at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.
Date: Thu, 03 Apr 2014 13:25:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: ubizjak at gmail dot com
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:
Message-ID: <bug-59617-4-9HdbTQddE8@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59617-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59617-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-04/txt/msg00217.txt.bz2
Content-length: 496

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #17 from Uroš Bizjak <ubizjak at gmail dot com> ---
(In reply to Iain Sandoe from comment #14)
> (In reply to Dominique d'Humieres from comment #13)
> > Another option would be to add -std=c99 for darwin.
> 
> I would suggest that this is the Right Thing to do.

There is quite some usage of __restrict in gcc.target/i386/ directory. Why
doesn't this problem trigger with e.g. avx2-gather-4.c, which also uses
__restrict?
>From gcc-bugs-return-448198-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Thu Apr 03 13:29:25 2014
Return-Path: <gcc-bugs-return-448198-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 26368 invoked by alias); 3 Apr 2014 13:29:25 -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 26320 invoked by uid 48); 3 Apr 2014 13:29:21 -0000
From: "charles.baylis at linaro dot org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug target/60609] [4.8/4.9 Regression] Error: value of 256 too large for field of 1 bytes at 68242
Date: Thu, 03 Apr 2014 13:29:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: target
X-Bugzilla-Version: 4.8.2
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: charles.baylis at linaro dot org
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.8.3
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-60609-4-a4g6TyrbFk@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-60609-4@http.gcc.gnu.org/bugzilla/>
References: <bug-60609-4@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: 2014-04/txt/msg00218.txt.bz2
Content-length: 216

http://gcc.gnu.org/bugzilla/show_bug.cgi?id`609

--- Comment #5 from Charles Baylis <charles.baylis at linaro dot org> ---

I have proposed a fix for this:

http://gcc.gnu.org/ml/gcc-patches/2014-04/msg00122.html

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #18 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> There is quite some usage of __restrict in gcc.target/i386/ directory.
> Why doesn't this problem trigger with e.g. avx2-gather-4.c, which also uses
>  __restrict?

AFAICT there is no scan in avx2-gather-4.c. Also the fact that all the other
supported tests pass may be because the __restrict is not needed for the tested
optimization (wild guess!-).

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[iains at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

--- Comment #19 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Uroš Bizjak from comment #17)
> (In reply to Iain Sandoe from comment #14)
> > (In reply to Dominique d'Humieres from comment #13)
> > > Another option would be to add -std=c99 for darwin.
> > 
> > I would suggest that this is the Right Thing to do.
> 
> There is quite some usage of __restrict in gcc.target/i386/ directory. Why
> doesn't this problem trigger with e.g. avx2-gather-4.c, which also uses
> __restrict?

presumably, other tests don't include the system header that is the source of
the problem (which can be checked).

My understanding is that restrict is c99 (and, the addition of  -std= was only
being proposed for darwin, so other targets would be unaffected)

however, that said, I am of course open to other solutions.
>From gcc-bugs-return-448201-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Thu Apr 03 13:41:31 2014
Return-Path: <gcc-bugs-return-448201-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 1425 invoked by alias); 3 Apr 2014 13:41:30 -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 1375 invoked by uid 48); 3 Apr 2014 13:41:26 -0000
From: "tim.moloney at mrsl dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/60731] [4.7/4.8/4.9 Regression] dynamic library not getting reinitialized on multiple calls to dlopen()
Date: Thu, 03 Apr 2014 13:41:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: c++
X-Bugzilla-Version: unknown
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: tim.moloney at mrsl dot com
X-Bugzilla-Status: RESOLVED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: jason at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.7.4
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-60731-4-tErJrReMzh@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-60731-4@http.gcc.gnu.org/bugzilla/>
References: <bug-60731-4@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: 2014-04/txt/msg00221.txt.bz2
Content-length: 202

http://gcc.gnu.org/bugzilla/show_bug.cgi?id`731

--- Comment #6 from Tim Moloney <tim.moloney at mrsl dot com> ---
I created glibc bug #16805
(https://sourceware.org/bugzilla/show_bug.cgi?id\x16805).

[Bug tree-optimization/59617] [vectorizer] ICE in vectorizable_mask_load_store with AVX-512F's gathers enabled.

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59617

Richard Biener <rguenth at gcc dot gnu.org> changed:

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

--- Comment #20 from Richard Biener <rguenth at gcc dot gnu.org> ---
I'm sure this is fixed.