[Bug target/59685] New: ICE: in output_2305, at config/i386/sse.md:8919 with -march=bdver1 -mavx512f

[Bug target/59685] New: ICE: in output_2305, at config/i386/sse.md:8919 with -march=bdver1 -mavx512f

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59685

            Bug ID: 59685
           Summary: ICE: in output_2305, at config/i386/sse.md:8919 with
                    -march=bdver1 -mavx512f
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: zsojka at seznam dot cz

Created attachment 31573
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31573&action=edit
reduced testcase

Compiler output:
$ gcc -march=bdver1 -mavx512f testcase.c            
testcase.c: In function 'foo':
testcase.c:7:1: internal compiler error: in output_2305, at
config/i386/sse.md:8919
 }
 ^
0xfffaf3 output_2305
        /mnt/svn/gcc-trunk/gcc/config/i386/sse.md:8919
0x8702f5 final_scan_insn(rtx_def*, _IO_FILE*, int, int, int*)
        /mnt/svn/gcc-trunk/gcc/final.c:2929
0x8719e5 final(rtx_def*, _IO_FILE*, int)
        /mnt/svn/gcc-trunk/gcc/final.c:2024
0x871f0e rest_of_handle_final
        /mnt/svn/gcc-trunk/gcc/final.c:4438
0x871f0e execute
        /mnt/svn/gcc-trunk/gcc/final.c:4513
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.


$ gcc -v                                
Using built-in specs.
COLLECT_GCC=/mnt/svn/gcc-trunk/binary-latest/bin/gcc
COLLECT_LTO_WRAPPER=/mnt/svn/gcc-trunk/binary-206332-lto-fortran-checking-yes-rtl-df/libexec/gcc/x86_64-unknown-linux-gnu/4.9.0/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: /mnt/svn/gcc-trunk//configure --enable-checking=yes,rtl,df
--enable-languages=c,c++,lto,fortran
--prefix=/mnt/svn/gcc-trunk/binary-206332-lto-fortran-checking-yes-rtl-df/
--without-cloog --without-ppl
Thread model: posix
gcc version 4.9.0 20140104 (experimental) (GCC)

[Bug target/59685] ICE: in output_2305, at config/i386/sse.md:8919 with -march=bdver1 -mavx512f

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59685

Uroš Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2014-01-20
           Assignee|unassigned at gcc dot gnu.org      |ubizjak at gmail dot com
   Target Milestone|---                         |4.9.0
     Ever confirmed|0                           |1

--- Comment #1 from Uroš Bizjak <ubizjak at gmail dot com> ---
Just missing V16SF mode in *andnot<mode>3<mask_name>.

Patch in testing:

--cut here--
Index: sse.md
===================================================================
--- sse.md      (revision 206844)
+++ sse.md      (working copy)
@@ -8906,6 +8906,8 @@
       tmp = "pandn";
       break;

+   case MODE_V16SF:
+      gcc_assert (TARGET_AVX512F);
    case MODE_V8SF:
       gcc_assert (TARGET_AVX);
    case MODE_V4SF:
--cut here--
>From gcc-bugs-return-441051-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Jan 20 19:24:27 2014
Return-Path: <gcc-bugs-return-441051-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 3468 invoked by alias); 20 Jan 2014 19:24:26 -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 3434 invoked by uid 48); 20 Jan 2014 19:24:22 -0000
From: "fragabr at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug libstdc++/56779] [4.8/4.9 Regression] libstdc++.so: undefined reference to `libintl_textdomain'
Date: Mon, 20 Jan 2014 19:24:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: libstdc++
X-Bugzilla-Version: 4.8.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: major
X-Bugzilla-Who: fragabr at gmail dot com
X-Bugzilla-Status: WAITING
X-Bugzilla-Priority: P3
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-56779-4-gLoszCcMYX@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-56779-4@http.gcc.gnu.org/bugzilla/>
References: <bug-56779-4@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: 2014-01/txt/msg02193.txt.bz2
Content-length: 1362

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56779

--- Comment #14 from Dâniel Fraga <fragabr at gmail dot com> ---
(In reply to Jonathan Wakely from comment #12)
> So I need to know which GNU/Linux distro you are using, and why it does not
> provide textdomain in glibc.

Hi Jonathan:

1) I'm using a customized installation, more like Linux from scratch (remember
that everything worked fine before gcc 4.8.0)

2) Regarding libintl.h, I have from both glibc and gettext:

/usr/include/libintl.h (glibc)
/usr/local/include/libintl.h (gettext)

3) g++ -E -x c++ - <<< '#include <libintl.h>' | head

# 1 "<stdin>"
# 1 "<command-line>"
# 1 "/usr/include/stdc-predef.h" 1 3 4
# 1 "<command-line>" 2
# 1 "<stdin>"
# 1 "/usr/local/include/libintl.h" 1 3
# 20 "/usr/local/include/libintl.h" 3
# 1 "/usr/include/locale.h" 1 3 4
# 25 "/usr/include/locale.h" 3 4
# 1 "/usr/include/features.h" 1 3 4

4) which msgfmt

/usr/local/bin/msgfmt

5) I tested the patch and it didn't work. Same problem:

  /usr/local/lib64/libstdc++.so: undefined reference to `libintl_gettext'

  /usr/local/lib64/libstdc++.so: undefined reference to `libintl_textdomain'

  /usr/local/lib64/libstdc++.so: undefined reference to
  `libintl_bindtextdomain'

  collect2: error: ld returned 1 exit status

***********

Any other hints? Just ask and I'll test anything you want. Thanks.
>From gcc-bugs-return-441052-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Jan 20 19:29:04 2014
Return-Path: <gcc-bugs-return-441052-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 6406 invoked by alias); 20 Jan 2014 19:29:03 -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 6369 invoked by uid 48); 20 Jan 2014 19:29:00 -0000
From: "redi at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug libstdc++/56779] [4.8/4.9 Regression] libstdc++.so: undefined reference to `libintl_textdomain'
Date: Mon, 20 Jan 2014 19:29:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: libstdc++
X-Bugzilla-Version: 4.8.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: major
X-Bugzilla-Who: redi at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.8.3
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: bug_status
Message-ID: <bug-56779-4-UzSes3ynoX@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-56779-4@http.gcc.gnu.org/bugzilla/>
References: <bug-56779-4@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: 2014-01/txt/msg02194.txt.bz2
Content-length: 1198

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56779

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW

--- Comment #15 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Dâniel Fraga from comment #14)
> 
> 1) I'm using a customized installation, more like Linux from scratch
> (remember that everything worked fine before gcc 4.8.0)

I don't understand how it worked previously, I think it should always have
failed!

> 2) Regarding libintl.h, I have from both glibc and gettext:
> 
> /usr/include/libintl.h (glibc)
> /usr/local/include/libintl.h (gettext)

Thanks for the quick response, that explains the problem.

G++ finds the version in /usr/local/include first, so the library gets a
dependency on libintl_textdomain, but the configure checks find textdomain in
glibc, so thinks there is no need to link to libintl.


> Any other hints? Just ask and I'll test anything you want. Thanks.

If you remove /usr/local/include/libintl.h it should work, but that's not a
clean solution.
>From gcc-bugs-return-441053-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Jan 20 19:50:09 2014
Return-Path: <gcc-bugs-return-441053-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 13434 invoked by alias); 20 Jan 2014 19:50:08 -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 13392 invoked by uid 48); 20 Jan 2014 19:50:05 -0000
From: "glisse at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug libstdc++/56779] [4.8/4.9 Regression] libstdc++.so: undefined reference to `libintl_textdomain'
Date: Mon, 20 Jan 2014 19:50:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: libstdc++
X-Bugzilla-Version: 4.8.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: major
X-Bugzilla-Who: glisse at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P3
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-56779-4-eXl1awTt2N@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-56779-4@http.gcc.gnu.org/bugzilla/>
References: <bug-56779-4@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: 2014-01/txt/msg02195.txt.bz2
Content-length: 561

http://gcc.gnu.org/bugzilla/show_bug.cgi?idV779

--- Comment #16 from Marc Glisse <glisse at gcc dot gnu.org> ---
>    AC_SEARCH_LIBS(gettext, intl, [], USE_NLS=no)

seems a bit crude, as it results in this link test:

#ifdef __cplusplus
extern "C"
#endif
char gettext ();
int
main ()
{
return gettext ();
  ;
  return 0;
}

when here we would need something closer to:

#include <libintl.h>
int main (int argc, char**argv)
{
  gettext(argv[0]);
  return 0;
}

so that if libintl.h redirects gettext to libintl_gettext the link test is for
the right symbol.

[Bug target/59685] ICE: in output_2305, at config/i386/sse.md:8919 with -march=bdver1 -mavx512f

[uros at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59685

--- Comment #2 from uros at gcc dot gnu.org ---
Author: uros
Date: Mon Jan 20 20:45:35 2014
New Revision: 206856

URL: http://gcc.gnu.org/viewcvs?rev=206856&root=gcc&view=rev
Log:
    PR target/59685
    * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
    mode attribute in insn output.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/i386/sse.md

[Bug target/59685] ICE: in output_2305, at config/i386/sse.md:8919 with -march=bdver1 -mavx512f

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59685

Uroš Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
                URL|                            |http://gcc.gnu.org/ml/gcc-p
                   |                            |atches/2014-01/msg01285.htm
                   |                            |l
         Resolution|---                         |FIXED

--- Comment #3 from Uroš Bizjak <ubizjak at gmail dot com> ---
Fixed.
>From gcc-bugs-return-441063-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Jan 20 20:55:30 2014
Return-Path: <gcc-bugs-return-441063-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 26834 invoked by alias); 20 Jan 2014 20:55:30 -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 26797 invoked by uid 48); 20 Jan 2014 20:55:27 -0000
From: "glisse at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug other/59893] New: Use LTO for libgcc.a, libstdc++.a, etc
Date: Mon, 20 Jan 2014 20:55:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: new
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: other
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords: build, lto
X-Bugzilla-Severity: enhancement
X-Bugzilla-Who: glisse 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: bug_id short_desc product version bug_status keywords bug_severity priority component assigned_to reporter
Message-ID: <bug-59893-4@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: 2014-01/txt/msg02205.txt.bz2
Content-length: 1454

http://gcc.gnu.org/bugzilla/show_bug.cgi?idY893

            Bug ID: 59893
           Summary: Use LTO for libgcc.a, libstdc++.a, etc
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Keywords: build, lto
          Severity: enhancement
          Priority: P3
         Component: other
          Assignee: unassigned at gcc dot gnu.org
          Reporter: glisse at gcc dot gnu.org

Hello,

LTO is not really a brand new, experimental and exotic option anymore. I
believe that by default, on systems that support it, we should build the static
target libraries that are part of gcc with -flto (and obviously
-ffat-lto-objects). This should have no impact on people not using LTO (well,
slightly longer bootstrap and a little bit of hard drive wasted), and people
using LTO actually expect it to apply to libstdc++ when building statically.
This should "solve" PR 59048 for instance where a simple std::string function
is hidden in libstdc++ and we thus miss an optimization. Assuming it works, it
could also help to have new/delete inlined from libsupc++, so the middle-end
optimizations on malloc/free have a chance to apply.

It may be as easy as adding the flags to C(XX)FLAGS_FOR_TARGET or it may be
much harder (I probably should have tried it before filing this PR), but it
seems we'll have to get there eventually.

This is quite different from bootstrap-lto as it applies to the target, not the
host.