[Bug libgomp/59534] New: [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[Bug libgomp/59534] New: [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

            Bug ID: 59534
           Summary: [4.9 Regression] FAIL: libgomp.fortran/retval1.f90
                    execution test due to denormals
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libgomp
          Assignee: unassigned at gcc dot gnu.org
          Reporter: ubizjak at gmail dot com
                CC: jakub at gcc dot gnu.org

This testcase failure is exposed on alpha, which by default generates exception
when denormals are generated:

spawn [open ...]^M

Program received signal SIGFPE: Floating-point exception - erroneous arithmetic
operation.

Program received signal SIGFPE: Floating-point exception - erroneous arithmetic
operation.

Backtrace for this error:

Backtrace for this error:
#0  0x2000004F1D7
#0  0x2000004F1D7
WARNING: program timed out.
FAIL: libgomp.fortran/retval1.f90  -O2  execution test


Running the test under gdb shows:

Program received signal SIGFPE, Arithmetic exception.
[Switching to Thread 0x20000c591a0 (LWP 24795)]
0x0000000120000d6c in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) list
90        end if
91        l = .false.
92      !$omp parallel firstprivate (f5, e5) shared (is_f5) num_threads (2) &
93      !$omp reduction (.or.:l)
94        l = .not. is_f5 .and. e5 .ne. 8
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
96        if (omp_get_thread_num () .eq. 0) e5 = 8
97        if (omp_get_thread_num () .eq. 1) e5 = 14
98        f5 = e5 - 4.5
99      !$omp barrier

   0x0000000120000d50 <+368>:   ldah    t0,-2(gp)
   0x0000000120000d54 <+372>:   lda     s0,14
   0x0000000120000d58 <+376>:   lds     $f2,30828(t0)
   0x0000000120000d5c <+380>:   br      0x120000c6c
<master.3.f5_._omp_fn.0+140>
   0x0000000120000d60 <+384>:   ldah    t0,-2(gp)
   0x0000000120000d64 <+388>:   lds     $f11,30832(t0)
=> 0x0000000120000d68 <+392>:   cmpteq  $f10,$f11,$f11
   0x0000000120000d6c <+396>:   ftoit   $f11,s2
   0x0000000120000d70 <+400>:   cmpeq   s2,0,s2

We can put breakpoint at the faulting insn, to obtain following session:

[New Thread 0x20000c591a0 (LWP 24957)]
[Switching to Thread 0x20000c591a0 (LWP 24957)]

Breakpoint 3, 0x0000000120000d68 in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) i r f10 f11
f10            6.5      (raw 0x401a000000000000)
f11            6.5      (raw 0x401a000000000000)
(gdb) c
Continuing.
[Switching to Thread 0x20000458000 (LWP 24956)]

Breakpoint 3, 0x0000000120000d68 in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) i r f10 f11
f10            6.5      (raw 0x401a000000000000)
f11            6.5      (raw 0x401a000000000000)
(gdb) c
Continuing.
[Switching to Thread 0x20000c591a0 (LWP 24957)]

Breakpoint 3, 0x0000000120000d68 in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) i r f10 f11
f10            2.1219957909652723e-314  (raw 0x0000000100000000)
f11            6.5      (raw 0x401a000000000000)
(gdb) si

Program received signal SIGFPE, Arithmetic exception.
0x0000000120000d6c in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)

This problem started somewhere between 15. October [1] and 3. November [2].

[1] http://gcc.gnu.org/ml/gcc-testresults/2013-10/msg01200.html
[2] http://gcc.gnu.org/ml/gcc-testresults/2013-11/msg00175.html

[Bug libgomp/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

Uroš Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|unknown                     |4.9.0

--- Comment #1 from Uroš Bizjak <ubizjak at gmail dot com> ---
The same problem can be observed on x86_64-linux-gnu, but this target doesn't
trigger an exception on denormals by default.

The breakpoint should be set in retval1.f90:95, that corresponds to
master.3.f5_._omp_fn.0. The exact location of the failing insn is:

   0x0000000000400a40 <+240>:   xor    %edx,%edx
   0x0000000000400a42 <+242>:   mov    $0x1,%r12d
=> 0x0000000000400a48 <+248>:   ucomiss 0x7f9(%rip),%xmm0        # 0x401248
   0x0000000000400a4f <+255>:   setp   %dl
   0x0000000000400a52 <+258>:   cmove  %edx,%r12d
   0x0000000000400a56 <+262>:   and    %eax,%r12d
   0x0000000000400a59 <+265>:   jmpq   0x40097e <master.3.f5_._omp_fn.0+46>

[New Thread 0x7ffff7866700 (LWP 29056)]

Breakpoint 4, 0x0000000000400a48 in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) p/f $xmm0
$13 = ( (6.5, 0, 0, 0), (5.3723379766381528e-315, 0), (0, 0, -48, 64, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0), (0, 16592, 0, 0, 0, 0, 0, 0), (6.5, 0, 0, 0),
(5.3723379766381528e-315, 0), 0 )
(gdb) c
Continuing.
[Switching to Thread 0x7ffff7866700 (LWP 29056)]

Following session can be obtained:

(gdb) r
Breakpoint 4, 0x0000000000400a48 in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) p/f $xmm0
$14 = ( (6.5, 0, 0, 0), (5.3723379766381528e-315, 0), (0, 0, -48, 64, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0), (0, 16592, 0, 0, 0, 0, 0, 0), (6.5, 0, 0, 0),
(5.3723379766381528e-315, 0), 0 )
(gdb) c
Continuing.

Breakpoint 4, 0x0000000000400a48 in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) p/f $xmm0
$15 = ( (1.12103877e-44, 0, 0, 0), (3.9525251667299724e-323, 0), (8, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0), (8, 0, 0, 0, 0, 0, 0, 0), (1.12103877e-44,
0, 0, 0), (3.9525251667299724e-323, 0), 0 )

Checking mxcsr and stepping a single insn shows that DE flag is set after:

(gdb) i r mxcsr
mxcsr          0x1f80   [ IM DM ZM OM UM PM ]
(gdb) si
0x0000000000400a4f      95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) i r mxcsr
mxcsr          0x1f82   [ DE IM DM ZM OM UM PM ]

The position of the PC:

   0x0000000000400a48 <+248>:   ucomiss 0x7f9(%rip),%xmm0        # 0x401248
=> 0x0000000000400a4f <+255>:   setp   %dl
>From gcc-bugs-return-437795-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Dec 16 22:26:18 2013
Return-Path: <gcc-bugs-return-437795-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 27643 invoked by alias); 16 Dec 2013 22:26:18 -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 27612 invoked by uid 48); 16 Dec 2013 22:26:14 -0000
From: "ubizjak at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug libgomp/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90  execution test due to denormals
Date: Mon, 16 Dec 2013 22:26:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: libgomp
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-59534-4-yz2CYP6Wkt@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59534-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59534-4@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: 2013-12/txt/msg01450.txt.bz2
Content-length: 2047

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

--- Comment #2 from Uroš Bizjak <ubizjak at gmail dot com> ---
Sorry, Firefox's editor has some issues, this part should be at the top of
session, so:

Following session can be obtained:

(gdb) r
The program being debugged has been started already.
Start it from the beginning? (y or n) y
Starting program: /home/uros/test/./retval1.exe 
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
[New Thread 0x7ffff7866700 (LWP 29056)]

Breakpoint 4, 0x0000000000400a48 in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) p/f $xmm0
$13 = ( (6.5, 0, 0, 0), (5.3723379766381528e-315, 0), (0, 0, -48, 64, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0), (0, 16592, 0, 0, 0, 0, 0, 0), (6.5, 0, 0, 0),
(5.3723379766381528e-315, 0), 0 )
(gdb) c
Continuing.
[Switching to Thread 0x7ffff7866700 (LWP 29056)]

Breakpoint 4, 0x0000000000400a48 in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) p/f $xmm0
$14 = ( (6.5, 0, 0, 0), (5.3723379766381528e-315, 0), (0, 0, -48, 64, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0), (0, 16592, 0, 0, 0, 0, 0, 0), (6.5, 0, 0, 0),
(5.3723379766381528e-315, 0), 0 )
(gdb) c
Continuing.

Breakpoint 4, 0x0000000000400a48 in master.3.f5_._omp_fn.0 () at
/home/uros/gcc-svn/trunk/libgomp/testsuite/libgomp.fortran/retval1.f90:95
95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) p/f $xmm0
$15 = ( (1.12103877e-44, 0, 0, 0), (3.9525251667299724e-323, 0), (8, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0), (8, 0, 0, 0, 0, 0, 0, 0), (1.12103877e-44,
0, 0, 0), (3.9525251667299724e-323, 0), 0 )
(gdb) i r mxcsr
mxcsr          0x1f80   [ IM DM ZM OM UM PM ]
(gdb) si
0x0000000000400a4f      95        l = l .or. (is_f5 .and. f5 .ne. 6.5)
(gdb) i r mxcsr
mxcsr          0x1f82   [ DE IM DM ZM OM UM PM ]
>From gcc-bugs-return-437796-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Dec 16 22:35:30 2013
Return-Path: <gcc-bugs-return-437796-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 4410 invoked by alias); 16 Dec 2013 22:35:30 -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 4383 invoked by uid 48); 16 Dec 2013 22:35:26 -0000
From: "d.g.gorbachev at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug lto/59468] [4.9 Regression] ICE on invalid C++ code with LTO in gimple_get_virt_method_for_binfo, at gimple-fold.c:3224
Date: Mon, 16 Dec 2013 22:35:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: lto
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords: ice-on-invalid-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: d.g.gorbachev at gmail dot com
X-Bugzilla-Status: ASSIGNED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: hubicka at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.9.0
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-59468-4-lJsPuTX8eK@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59468-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59468-4@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: 2013-12/txt/msg01451.txt.bz2
Content-length: 222

http://gcc.gnu.org/bugzilla/show_bug.cgi?idY468

--- Comment #5 from Dmitry Gorbachev <d.g.gorbachev at gmail dot com> ---
I also saw something similar with normal, non-invalid code. This TC is from
Delta-reduced code.

[Bug libgomp/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[ubizjak at gmail dot com]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="UTF-8", Size: 2963 bytes --]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

--- Comment #3 from Uroš Bizjak <ubizjak at gmail dot com> ---
The denormal in f5.0_16 is directly loaded from

  f5.0_16 = .omp_data_i_14(D)->f5.0;
  _17 = .omp_data_i_14(D)->is_f5;
  _18 = *_17;
  _43 = e5.1_15 != 8;
  if (_18 < _43)
    goto <bb 3>;
  else
    goto <bb 13>;

  <bb 3>:
...
  <bb 13>:
  _40 = f5.0_16 != 6.5e+0;
  l_36 = _40 & _18;
  goto <bb 3>;

That corresponds to:
    ...
    movss    8(%rdi), %xmm0
    movl    (%rax), %eax
    cmpl    $8, %ebp
    setne    %dl
    cmpl    %edx, %eax
    jae    .L2
    movl    $1, %r12d
.L13:
    ...
.L2:
    xorl    %edx, %edx
    movl    $1, %r12d
    ucomiss    .LC1(%rip), %xmm0
    setp    %dl
    cmove    %edx, %r12d
    andl    %eax, %r12d
    jmp    .L13
>From gcc-bugs-return-437798-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon Dec 16 22:37:07 2013
Return-Path: <gcc-bugs-return-437798-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 6130 invoked by alias); 16 Dec 2013 22:37:07 -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 6071 invoked by uid 48); 16 Dec 2013 22:37:04 -0000
From: "d.g.gorbachev at gmail dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug lto/59468] [4.9 Regression] ICE on invalid C++ code with LTO in gimple_get_virt_method_for_binfo, at gimple-fold.c:3224
Date: Mon, 16 Dec 2013 22:37:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: lto
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords: ice-on-invalid-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: d.g.gorbachev at gmail dot com
X-Bugzilla-Status: ASSIGNED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: hubicka at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.9.0
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: attachments.created
Message-ID: <bug-59468-4-2FT9pgRRuh@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59468-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59468-4@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: 2013-12/txt/msg01453.txt.bz2
Content-length: 279

http://gcc.gnu.org/bugzilla/show_bug.cgi?idY468

--- Comment #6 from Dmitry Gorbachev <d.g.gorbachev at gmail dot com> ---
Created attachment 31453
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id1453&actioníit
TC: ICE in function_and_variable_visibility, at ipa.c:997

[Bug libgomp/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

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

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

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
This is broken by r204194.
I'm afraid it is generally unsafe to do it if the inner comparison (the second
one) is floating point comparison (regardless of -ffast-math or not), because
one can have:
if (some_condition)
  {
    if (float_val != 6.5)
      ...
  }
else
  {
    if (float_val2 != 7.2)
      ...
  }
and float_val could be uninitialized/random if !some_condition, similarly
float_val2 uninitialized/random if some_condition.  In that case even with
-ffast-math it could fail, the uninitialized value could be say sNaN (or
denormal as in this case).

[Bug libgomp/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note in this fortran testcase: l .or. (is_f5 .and. f5 .ne. 6.5) is already not
short-circuit by the definition of the language.  Though I don't know if the
reduction part for OMP can be done by not short-circuiting though.

[Bug libgomp/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
We are doing the correct thing for this testcase (the testcase should change
also) though most likely you can generate a testcase which short-circuits the
check.

And then we should add a check for trapping of the comparisons.

[Bug libgomp/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #6)
> We are doing the correct thing for this testcase 

See http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57160 for the reasons why I say
this testcase is being handled correctly.

[Bug libgomp/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

--- Comment #8 from Uroš Bizjak <ubizjak at gmail dot com> ---
(In reply to Andrew Pinski from comment #6)
> We are doing the correct thing for this testcase (the testcase should change
> also) though most likely you can generate a testcase which short-circuits
> the check.

Let's ask a nice fortraner to fix the testcase then...

> And then we should add a check for trapping of the comparisons.

IEEE exception handling is not yet implemented, but alpha traps on denormals by
default (that is how the problem was discovered). OTOH, if I add -mieee to the
compile options, the exception on alpha goes away (as expected).
>From gcc-bugs-return-437811-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Tue Dec 17 07:15:43 2013
Return-Path: <gcc-bugs-return-437811-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 20086 invoked by alias); 17 Dec 2013 07:15:43 -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 19809 invoked by uid 48); 17 Dec 2013 07:15:38 -0000
From: "samsonov at google dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug sanitizer/59136] [4.9 Regression] llvm-symbolizer shouldn't be started always
Date: Tue, 17 Dec 2013 07:15:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: sanitizer
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: samsonov at google dot com
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P1
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.9.0
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-59136-4-KT4gl2Z3fx@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59136-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59136-4@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: 2013-12/txt/msg01466.txt.bz2
Content-length: 268

http://gcc.gnu.org/bugzilla/show_bug.cgi?idY136

--- Comment #13 from Alexey Samsonov <samsonov at google dot com> ---
I don't think fork() issue will be relevant here, at least for the minimalistic
TSan test cases. Let's wait till we have libbacktrace symbolizer.

[Bug testsuite/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

Uroš Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-12-17
                 CC|                            |burnus at gcc dot gnu.org
          Component|libgomp                     |testsuite
     Ever confirmed|0                           |1

--- Comment #9 from Uroš Bizjak <ubizjak at gmail dot com> ---
Recategorized as (fortran) testsuite bug.

Adding CC of fortran expert.
>From gcc-bugs-return-437845-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Tue Dec 17 14:43:16 2013
Return-Path: <gcc-bugs-return-437845-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 3336 invoked by alias); 17 Dec 2013 14:43:15 -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 3235 invoked by uid 55); 17 Dec 2013 14:43:10 -0000
From: "rguenther at suse dot de" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/52272] [4.7/4.8/4.9 regression] Performance regression of 410.bwaves on x86.
Date: Tue, 17 Dec 2013 14:43:00 -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: 4.7.0
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
X-Bugzilla-Who: rguenther at suse dot de
X-Bugzilla-Status: ASSIGNED
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: rguenth at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.8.3
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-52272-4-7rMZh2SKLo@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-52272-4@http.gcc.gnu.org/bugzilla/>
References: <bug-52272-4@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: 2013-12/txt/msg01500.txt.bz2
Content-length: 3129

http://gcc.gnu.org/bugzilla/show_bug.cgi?idR272

--- Comment #22 from rguenther at suse dot de <rguenther at suse dot de> ---
On 12/17/13 9:29 AM, amker.cheng at gmail dot com wrote:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?idR272
>
> bin.cheng <amker.cheng at gmail dot com> changed:
>
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                  CC|                            |amker.cheng at gmail dot com
>
> --- Comment #21 from bin.cheng <amker.cheng at gmail dot com> ---
> Hi Richard,
> I looked into PR50955 for which the mentioned commit causing this PR is
> applied:
>
> Commit
> 2012-02-06  Richard Guenther  <rguenther@suse.de>
>
>         PR tree-optimization/50955
>         * tree-ssa-loop-ivopts.c (get_computation_cost_at): Artificially
>         raise cost of expressions that replace an address with an
>         expression based on a different pointer.
>
> I noticed that the offending non-linear use in PR50955 is actually from memory
> reference.  If I understand the issue correct, the whole alias issue is
> introduced by rewriting iv use with one base_object through candidate with
> another incompatible base_object, and it is related to memory reference.  An
> genuine non-linear iv use (the pointer never de-referenced, like in this PR)
> won't have this issue.
>
> So I come up this idea to relax the condition:
>
> -  if (address_p)
> +  if (address_p
> +      || (use->iv->base_object
> +         && cand->iv->base_object
> +         && POINTER_TYPE_P (TREE_TYPE (use->iv->base_object))
> +         && POINTER_TYPE_P (TREE_TYPE (cand->iv->base_object))))
>      {
>        /* Do not try to express address of an object with computation based
>          on address of a different object.  This may cause problems in rtl
>
> to non-linear uses which truly occurred in memory reference, something like:
>
> -  if (address_p)
> +  if (address_p
> +      || (use->in_mem_ref_p
> +         && use->iv->base_object
> +         && cand->iv->base_object
> +         && POINTER_TYPE_P (TREE_TYPE (use->iv->base_object))
> +         && POINTER_TYPE_P (TREE_TYPE (cand->iv->base_object))))
>      {
>        /* Do not try to express address of an object with computation based
>          on address of a different object.  This may cause problems in rtl
>
> The flag in_mem_ref_p can be set for appropriate uses when finding interesting
> address uses.
>
> With this change, this PR should be resolved while not violating PR50955.
>
> I am not very much into 50955, so how does this sound? I can send a patch for
> review if the idea is in right direction.

I'm not 100% sure.

> BTW, I cannot reproduce 50955 with the reported revision of GCC.  The store
> isn't deleted by pass_cd_dce, though it is re-written just as the PR reported.
> So maybe I just misunderstood something.

It's been too long to remember ;)  The issue boils down to a bogus
TMR_SYMBOL. Later passes may forward a !use->in_mem_ref into a mem-ref,
exposing the issue as well.

Richard.

> Any words?
>
> Thanks,
> bin
>

[Bug testsuite/59534] [4.9 Regression] FAIL: libgomp.fortran/retval1.f90 execution test due to denormals

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59534

--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Author: jakub
Date: Tue Dec 17 15:17:00 2013
New Revision: 206051

URL: http://gcc.gnu.org/viewcvs?rev=206051&root=gcc&view=rev
Log:
    PR testsuite/59534
    * testsuite/libgomp.fortran/retval1.f90 (e5): Avoid non-shortcircuited
    comparisons.

Modified:
    trunk/libgomp/ChangeLog
    trunk/libgomp/testsuite/libgomp.fortran/retval1.f90