[Bug gcov-profile/59003] New: [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback

[Bug gcov-profile/59003] New: [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback

[octoploid at yandex dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

            Bug ID: 59003
           Summary: [4.9 Regression] profiledbootstrap miscompiles gcc
                    during stagefeedback
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: gcov-profile
          Assignee: unassigned at gcc dot gnu.org
          Reporter: octoploid at yandex dot com

% ../gcc/configure --enable-checking=release --disable-werror
--disable-multilib --enable-languages=c,c++
 % make -j4 BOOT_CFLAGS="-march=native -O3 -pipe" STAGE1_CFLAGS="-march=native
-O3 -pipe" CFLAGS_FOR_TARGET="-march=native -O3 -pipe"
CXXFLAGS_FOR_TARGET="-march=native -O3 -pipe" profiledbootstrap
...
/var/tmp/gcc_build_dir/./gcc/xgcc -B/var/tmp/gcc_build_dir/./gcc/
-B/usr/local/x86_64-unknown-linux-gnu/bin/
-B/usr/local/x86_64-unknown-linux-gnu/lib/ -isystem
/usr/local/x86_64-unknown-linux-gnu/include -isystem
/usr/local/x86_64-unknown-linux-gnu/sys-include    -march=native -O3 -pipe -O2 
-march=native -O3 -pipe -DIN_GCC    -W -Wall -Wno-narrowing -Wwrite-strings
-Wcast-qual -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition 
-isystem ./include   -fpic -mlong-double-80 -g -DIN_LIBGCC2 -fbuilding-libgcc
-fno-stack-protector   -fpic -mlong-double-80 -I. -I. -I../.././gcc
-I../../../gcc/libgcc -I../../../gcc/libgcc/. -I../../../gcc/libgcc/../gcc
-I../../../gcc/libgcc/../include -I../../../gcc/libgcc/config/libbid
-DENABLE_DECIMAL_BID_FORMAT -DHAVE_CC_TLS  -DUSE_TLS -o bid128_noncomp.o -MT
bid128_noncomp.o -MD -MP -MF bid128_noncomp.dep -c
../../../gcc/libgcc/config/libbid/bid128_noncomp.c
/var/tmp/gcc_build_dir/./gcc/xgcc -B/var/tmp/gcc_build_dir/./gcc/
-B/usr/local/x86_64-unknown-linux-gnu/bin/
-B/usr/local/x86_64-unknown-linux-gnu/lib/ -isystem
/usr/local/x86_64-unknown-linux-gnu/include -isystem
/usr/local/x86_64-unknown-linux-gnu/sys-include    -march=native -O3 -pipe -O2 
-march=native -O3 -pipe -DIN_GCC    -W -Wall -Wno-narrowing -Wwrite-strings
-Wcast-qual -Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition 
-isystem ./include   -fpic -mlong-double-80 -g -DIN_LIBGCC2 -fbuilding-libgcc
-fno-stack-protector   -fpic -mlong-double-80 -I. -I. -I../.././gcc
-I../../../gcc/libgcc -I../../../gcc/libgcc/. -I../../../gcc/libgcc/../gcc
-I../../../gcc/libgcc/../include -I../../../gcc/libgcc/config/libbid
-DENABLE_DECIMAL_BID_FORMAT -DHAVE_CC_TLS  -DUSE_TLS -o bid128_fma.o -MT
bid128_fma.o -MD -MP -MF bid128_fma.dep -c
../../../gcc/libgcc/config/libbid/bid128_fma.c
In file included from ../../../gcc/libgcc/config/libbid/bid_internal.h:27:0,
                 from ../../../gcc/libgcc/config/libbid/bid64_noncomp.c:24:
../../../gcc/libgcc/config/libbid/bid64_noncomp.c: In function ‘__bid64_isInf’:
../../../gcc/libgcc/config/libbid/bid_conf.h:258:21: internal compiler error:
Segmentation fault
 #define bid64_isInf __bid64_isInf
                     ^
../../../gcc/libgcc/config/libbid/bid64_noncomp.c:220:1: note: in expansion of
macro ‘bid64_isInf’
 bid64_isInf (UINT64 x _EXC_MASKS_PARAM _EXC_INFO_PARAM) {
 ^
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
make[3]: *** [bid64_noncomp.o] Error 1

Looks like gcc got miscompiled during stagefeedback.
-march=native is amdfam10 on my machine.
Without --enable-checking=release gcc builds fine.
>From gcc-bugs-return-433554-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Tue Nov 05 13:11:41 2013
Return-Path: <gcc-bugs-return-433554-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 14235 invoked by alias); 5 Nov 2013 13:11:40 -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 11441 invoked by uid 48); 5 Nov 2013 13:09:38 -0000
From: "burnus at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug fortran/57616] Support -fsanitize=undefined in Fortran
Date: Tue, 05 Nov 2013 13:11:00 -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: diagnostic
X-Bugzilla-Severity: normal
X-Bugzilla-Who: burnus 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: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-57616-4-S38AkQTUQt@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-57616-4@http.gcc.gnu.org/bugzilla/>
References: <bug-57616-4@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: 2013-11/txt/msg00331.txt.bz2
Content-length: 328

http://gcc.gnu.org/bugzilla/show_bug.cgi?idW616

--- Comment #1 from Tobias Burnus <burnus at gcc dot gnu.org> ---
Additionally, check whether -fsanitize=vla-bound (are there more usans?) should
be handled.

Check whether MPX bounds checking should be implemented, cf.
http://gcc.gnu.org/ml/gcc-patches/2013-10/msg02660.html

[Bug gcov-profile/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[octoploid at yandex dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

Markus Trippelsdorf <octoploid at yandex dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[4.9 Regression]            |[4.9 Regression]
                   |profiledbootstrap           |profiledbootstrap
                   |miscompiles gcc during      |miscompiles gcc during
                   |stagefeedback               |stagefeedback
                   |                            |--with-tune=amdfam10

--- Comment #1 from Markus Trippelsdorf <octoploid at yandex dot com> ---
The following is enough to reproduce the issue:

 % ../gcc/configure --with-tune=amdfam10 --enable-checking=release
--disable-werror --disable-multilib --enable-languages=c,c++
--with-build-config="bootstrap-lto bootstrap-O3"
 % make profiledbootstrap

Maybe it is another target problem?

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[octoploid at yandex dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

Markus Trippelsdorf <octoploid at yandex dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
             Target|                            |x86_64-*-*
                 CC|                            |hubicka at ucw dot cz
          Component|gcov-profile                |target

--- Comment #2 from Markus Trippelsdorf <octoploid at yandex dot com> ---
Started with r203937.

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[octoploid at yandex dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

--- Comment #3 from Markus Trippelsdorf <octoploid at yandex dot com> ---
Switching off misaligned_move_string_pro_epilogues for AMD "fixes" 
the issue:

diff --git a/gcc/config/i386/x86-tune.def b/gcc/config/i386/x86-tune.def
index 4c13c3a0ec69..af58b3e1f77e 100644
--- a/gcc/config/i386/x86-tune.def
+++ b/gcc/config/i386/x86-tune.def
@@ -264,7 +264,7 @@ DEF_TUNE (X86_TUNE_SINGLE_STRINGOP, "single_stringop",
m_386 | m_P4_NOCONA)
    FIXME: This may actualy be a win on more targets than listed here.  */
 DEF_TUNE (X86_TUNE_MISALIGNED_MOVE_STRING_PRO_EPILOGUES,
          "misaligned_move_string_pro_epilogues",
-         m_386 | m_486 | m_CORE_ALL | m_AMD_MULTIPLE | m_GENERIC)
+         m_386 | m_486 | m_CORE_ALL | m_GENERIC)

 /* X86_TUNE_USE_SAHF: Controls use of SAHF.  */
 DEF_TUNE (X86_TUNE_USE_SAHF, "use_sa

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[hubicka at ucw dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

--- Comment #4 from Jan Hubicka <hubicka at ucw dot cz> ---
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003
> 
> --- Comment #3 from Markus Trippelsdorf <octoploid at yandex dot com> ---
> Switching off misaligned_move_string_pro_epilogues for AMD "fixes" 
> the issue:

We need to figure out why it causes problem for AMD and not for generic/core. 
I will
try to reproduce the ICE...

Honza

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[octoploid at yandex dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

--- Comment #5 from Markus Trippelsdorf <octoploid at yandex dot com> ---
Looks like gcc/tree-ssa-pre.c gets miscompiled:

Program received signal SIGSEGV, Segmentation fault.
[Switching to process 8327]
0x0000000000d156d3 in get_or_alloc_expr_for_name(tree_node*) [clone
.lto_priv.4065] ()
(gdb) bt
#0  0x0000000000d156d3 in get_or_alloc_expr_for_name(tree_node*) [clone
.lto_priv.4065] ()
#1  0x0000000000f47468 in compute_avail() ()
#2  0x00000000011068cf in (anonymous namespace)::pass_pre::execute() [clone
.lto_priv.5361] ()
#3  0x0000000000c8f3a2 in execute_one_pass(opt_pass*) ()
#4  0x0000000000c9065b in execute_pass_list(opt_pass*) ()
#5  0x0000000000e405f5 in expand_function(cgraph_node*) [clone .lto_priv.5292]
()
#6  0x00000000010ed32b in compile() ()
#7  0x00000000010ed8da in finalize_compilation_unit() ()
#8  0x0000000000ffe208 in c_write_global_declarations() ()
#9  0x00000000010b0c5f in compile_file() ()
#10 0x00000000010b2d14 in toplev_main(int, char**) ()
#11 0x00007ffff7756f90 in __libc_start_main () from /lib/libc.so.6
#12 0x000000000104d236 in _start ()

Is it possible to somehow disassemble slim object files?

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[octoploid at yandex dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

--- Comment #6 from Markus Trippelsdorf <octoploid at yandex dot com> ---
Created attachment 31377
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31377&action=edit
disassemled get_or_alloc_expr_for_name(tree_node*) good

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[octoploid at yandex dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

--- Comment #7 from Markus Trippelsdorf <octoploid at yandex dot com> ---
Created attachment 31378
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31378&action=edit
disassemled get_or_alloc_expr_for_name(tree_node*) bad

I've attached the disassembly of get_or_alloc_expr_for_name()
Good: misaligned_move_string_pro_epilogues off
Bad: misaligned_move_string_pro_epilogues on

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 31907
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31907&action=edit
gcc49-pr59003-test.patch

Reduced testcase (with s/amdfam10/sandybridge/ it works just fine).

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I believe the bug is in expand_small_movmem_or_setmem, if
GET_MODE_SIZE (mode) < size, then we do size / GET_MODE_SIZE (mode) first
stores
starting at destmem's address (correct), but then do just a single store
starting at destmem's address + count - GET_MODE_SIZE (mode), which is not
correct.  We should have done again size / GET_MODE_SIZE (mode) stores starting
at destmem's address + count - size.

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #31907|0                           |1
        is obsolete|                            |
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2014-01-21
           Assignee|unassigned at gcc dot gnu.org      |jakub at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 31908
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31908&action=edit
gcc49-pr59003.patch

Untested fix.

[Bug target/59003] [4.9 Regression] profiledbootstrap miscompiles gcc during stagefeedback --with-tune=amdfam10

[trippels at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59003

Markus Trippelsdorf <trippels at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
                 CC|                            |trippels at gcc dot gnu.org
         Resolution|---                         |FIXED

--- Comment #12 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
Fixed. Thanks.