[Bug fortran/26444] New: gfortran does not compile cp2k

[Bug fortran/26444] New: gfortran does not compile cp2k

[jv244 at cam dot ac dot uk]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1369 bytes --]

gfortran currently yields ICEs trying to compile the cp2k code. To make it
easier to address these, I'll attach a 'preprocessed source'. 

cp2k.f90: In function ‘timeset_full’:
cp2k.f90:10751: internal compiler error: Segmentation fault

I've decided to make it the full program, so that it can be compiled into an
executable. As soon as gfortran does generate a correct executable, I'll setup
a nightly or weekly regression tester (frontend bugs / compile time /
performance).

The code is quite standard conforming and portable, but as is usual there are a
few exceptions (best documented by the NAG f95 command line options needed
-dusty -dcfuns -mismatch_all -kind=byte and that it is missing some extensions
such as iargc or etime) , these are very common and ifort,g95 just compile with
the defaults. On the other hand the thing as a whole seems quite challening,
two version of xlf90 run into an ICE, and pgf90 is generating .mod files for
about an hour now (g95 compiles in about 10min).


-- 
           Summary: gfortran does not compile cp2k
           Product: gcc
           Version: 4.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: jv244 at cam dot ac dot uk


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[jv244 at cam dot ac dot uk]

------- Comment #1 from jv244 at cam dot ac dot uk  2006-02-23 15:55 -------
Ok, bugzilla kindly doesn't allow to attach anything more than 1Mb. If any of
the bugmasters can provide permanent space for a ~5Mb file (compressed) they
can write  me.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[jb at gcc dot gnu dot org]

------- Comment #2 from jb at gcc dot gnu dot org  2006-02-23 18:37 -------
I have the current CVS of cp2k, it fails with

gfortran  -c -O3 -g -ffast-math -fomit-frame-pointer message_passing.f90
...
message_passing.f90: In function 'mp_perf_env_create':
message_passing.f90:58: internal compiler error: in add_virtual_operand, at
tree-ssa-operands.c:1867

Confirmed.

And yes, it seems cp2k is a good testsuite for modern Fortran features.


-- 

jb at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2006-02-23 18:37:53
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[jv244 at cam dot ac dot uk]

------- Comment #3 from jv244 at cam dot ac dot uk  2006-02-23 18:56 -------
which is in fact different from the bug I see. This is one reason I would
prefer to have a single file (the version in CVS will change rapidly, and
requires some additional setup which is avoided if we talk about a specific
file), so that we know what we're talking about.

I've added Andrew, I guess he might have an easy way to make the file
permanently available. (I can email it, but don't have permanent webspace).


-- 

jv244 at cam dot ac dot uk changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

Re: [Bug fortran/26444] gfortran does not compile cp2k

[Daniel Berlin]

On Thu, 2006-02-23 at 18:37 +0000, jb at gcc dot gnu dot org wrote:
> 
> ------- Comment #2 from jb at gcc dot gnu dot org  2006-02-23 18:37 -------
> I have the current CVS of cp2k, it fails with
> 
> gfortran  -c -O3 -g -ffast-math -fomit-frame-pointer message_passing.f90
> ...
> message_passing.f90: In function 'mp_perf_env_create':
> message_passing.f90:58: internal compiler error: in add_virtual_operand, at
> tree-ssa-operands.c:1867
> 
> Confirmed.
> 
> And yes, it seems cp2k is a good testsuite for modern Fortran features.
> 
> 

This assert means some pass changed TMT usage without the right update
flags.
Andrew, can you try to figure out what pass did this (it should be
relatively simple to see what the last pass touching the statement in
question is).

[Bug fortran/26444] gfortran does not compile cp2k

[dberlin at dberlin dot org]

------- Comment #4 from dberlin at gcc dot gnu dot org  2006-02-23 19:25 -------
Subject: Re:  gfortran does not compile cp2k

On Thu, 2006-02-23 at 18:37 +0000, jb at gcc dot gnu dot org wrote:
> 
> ------- Comment #2 from jb at gcc dot gnu dot org  2006-02-23 18:37 -------
> I have the current CVS of cp2k, it fails with
> 
> gfortran  -c -O3 -g -ffast-math -fomit-frame-pointer message_passing.f90
> ...
> message_passing.f90: In function 'mp_perf_env_create':
> message_passing.f90:58: internal compiler error: in add_virtual_operand, at
> tree-ssa-operands.c:1867
> 
> Confirmed.
> 
> And yes, it seems cp2k is a good testsuite for modern Fortran features.
> 
> 

This assert means some pass changed TMT usage without the right update
flags.
Andrew, can you try to figure out what pass did this (it should be
relatively simple to see what the last pass touching the statement in
question is).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[dberlin at dberlin dot org]

------- Comment #5 from dberlin at gcc dot gnu dot org  2006-02-23 19:25 -------
Subject: Re:  gfortran does not compile cp2k

On Thu, 2006-02-23 at 18:37 +0000, jb at gcc dot gnu dot org wrote:
> 
> ------- Comment #2 from jb at gcc dot gnu dot org  2006-02-23 18:37 -------
> I have the current CVS of cp2k, it fails with
> 
> gfortran  -c -O3 -g -ffast-math -fomit-frame-pointer message_passing.f90
> ...
> message_passing.f90: In function 'mp_perf_env_create':
> message_passing.f90:58: internal compiler error: in add_virtual_operand, at
> tree-ssa-operands.c:1867
> 
> Confirmed.
> 
> And yes, it seems cp2k is a good testsuite for modern Fortran features.
> 
> 

This assert means some pass changed TMT usage without the right update
flags.
Andrew, can you try to figure out what pass did this (it should be
relatively simple to see what the last pass touching the statement in
question is).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[jb at gcc dot gnu dot org]

------- Comment #6 from jb at gcc dot gnu dot org  2006-02-23 20:18 -------
Created an attachment (id=10904)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10904&action=view)
Causes ICE when compiled with -O3

This is a self-contained example that show the tree bug. It doesn't appear with
-O2, you need -O3.

message_passing.f90: In function 'mp_perf_env_create':
message_passing.f90:462: internal compiler error: in add_virtual_operand, at
tree-ssa-operands.c:1867


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[pinskia at gcc dot gnu dot org]

------- Comment #7 from pinskia at gcc dot gnu dot org  2006-02-24 17:42 -------
I cannot reproduce this with "4.2.0 20060218".


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[pinskia at gcc dot gnu dot org]

------- Comment #8 from pinskia at gcc dot gnu dot org  2006-02-24 17:47 -------
I can reproduce this on the 20th though.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[pinskia at gcc dot gnu dot org]

------- Comment #10 from pinskia at gcc dot gnu dot org  2006-02-24 17:53 -------
Note -O3 enables complete unrolling at the tree level, I think there might be
another related bug to this one.  I bet it is failing in VRP2 also, just like
the other one.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[pinskia at gcc dot gnu dot org]

------- Comment #9 from pinskia at gcc dot gnu dot org  2006-02-24 17:48 -------
So what this means I could not have caused this as the only patch after my 18th
GCC and my 20th GCC is not a patch which I committed that would have caused
this.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug fortran/26444] gfortran does not compile cp2k

[pinskia at gcc dot gnu dot org]

------- Comment #11 from pinskia at gcc dot gnu dot org  2006-02-24 17:54 -------
(In reply to comment #10)
> Note -O3 enables complete unrolling at the tree level, I think there might be
> another related bug to this one.  I bet it is failing in VRP2 also, just like
> the other one.

That would be PR 26443 by the way.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug tree-optimization/26444] [4.2 Regression] gfortran does not compile cp2k

[pinskia at gcc dot gnu dot org]

------- Comment #12 from pinskia at gcc dot gnu dot org  2006-02-24 18:14 -------
(In reply to comment #11)
> (In reply to comment #10)
> > Note -O3 enables complete unrolling at the tree level, I think there might be
> > another related bug to this one.  I bet it is failing in VRP2 also, just like
> > the other one.
> 
> That would be PR 26443 by the way.

And it fails in the same way PR 26443 fails as that it is ICE after VRP2.
Here is the "reduced" testcase (there are no complex types in there):
MODULE message_passing
  TYPE mp_perf_type
    CHARACTER ( LEN = 20 ) :: name
    REAL :: time
  END TYPE mp_perf_type
  INTEGER, PARAMETER :: MAX_PERF = 12
  TYPE mp_perf_env_type
     INTEGER :: ref_count, id_nr
     TYPE ( mp_perf_type ), DIMENSION ( MAX_PERF ) :: mp_perfs
  END TYPE mp_perf_env_type
  CHARACTER ( LEN = 20 ), PARAMETER :: sname ( MAX_PERF ) =  &
   (/"MP_Group            ", "MP_Bcast            ", "MP_Allreduce        ", &
     "MP_Gather           ", "MP_Sync             ", "MP_Alltoall         ", &
     "MP_SendRecv         ", "MP_ISendRecv        ", "MP_Wait             ", &
     "MP_comm_split       ", "MP_ISend            ", "MP_IRecv            "/)
CONTAINS
SUBROUTINE mp_perf_env_create(perf_env)
    TYPE(mp_perf_env_type), POINTER :: perf_env
    perf_env%id_nr=last_mp_perf_env_id
    DO i = 1, MAX_PERF
       perf_env%mp_perfs(i) % name = sname(i)
       perf_env%mp_perfs(i) % time = 0.0
    END DO
END SUBROUTINE mp_perf_env_create
END MODULE message_passing


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  BugsThisDependsOn|                            |26443
          Component|fortran                     |tree-optimization
           Keywords|                            |ice-on-valid-code
            Summary|gfortran does not compile   |[4.2 Regression] gfortran
                   |cp2k                        |does not compile cp2k
   Target Milestone|---                         |4.2.0
            Version|4.1.0                       |4.2.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug tree-optimization/26444] [4.2 Regression] ICE in add_virtual_operand with -O3

[martin at mpa-garching dot mpg dot de]

------- Comment #13 from martin at mpa-garching dot mpg dot de  2006-03-01 12:39 -------
It appears that this bug also prevents FFTW 3.1 from compiling.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444

[Bug tree-optimization/26444] [4.2 Regression] ICE in add_virtual_operand with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #14 from pinskia at gcc dot gnu dot org  2006-03-01 18:32 -------
Confirmed, this is fixed with the patch for PR 26443.

*** This bug has been marked as a duplicate of 26443 ***


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26444