[Bug fortran/67758] New: [GCC 6 regression] ICE on invalid code

[Bug fortran/67758] New: [GCC 6 regression] ICE on invalid code

[ienkovich at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

            Bug ID: 67758
           Summary: [GCC 6 regression] ICE on invalid code
           Product: gcc
           Version: 6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: ienkovich at gcc dot gnu.org
  Target Milestone: ---

>cat test.f
      COMMON /FMCOM / X(80 000 000)
            CALL T(XX(A))
      COMMON /FMCOM / XX(80 000 000)
>builds/trunk/64/20150927/bin/gfortran test.f
f951: internal compiler error: Segmentation fault
0xb1878f crash_signal
        /export/users/gnutester/stability/svn/trunk/gcc/toplev.c:353
0x3b21c3567f ???
       
/home/glibctest/rpmbuild/BUILD/glibc-2.17-c758a686/signal/../sysdeps/unix/sysv/linux/x86_64/sigaction.c:0
0x66eaf3 gfc_restore_last_undo_checkpoint()
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/symbol.c:3205
0x630bff reject_statement
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/parse.c:2296
0x630d6c match_word_omp_simd
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/parse.c:99
0x634621 match_word
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/parse.c:442
0x634621 decode_statement
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/parse.c:442
0x6351c4 next_free
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/parse.c:1059
0x6351c4 next_statement
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/parse.c:1293
0x638b61 parse_executable
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/parse.c:4706
0x639f7e parse_progunit
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/parse.c:5125
0x63b163 gfc_parse_file()
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/parse.c:5665
0x67cca2 gfc_be_parse_file
        /export/users/gnutester/stability/svn/trunk/gcc/fortran/f95-lang.c:209
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.


Regression started 20150808.

[Bug fortran/67758] [GCC 6 regression] ICE on invalid use of COMMON

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
   Last reconfirmed|                            |2015-09-29
            Summary|[GCC 6 regression] ICE on   |[GCC 6 regression] ICE on
                   |invalid code                |invalid use of COMMON
     Ever confirmed|0                           |1

--- Comment #1 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
WORKSFORME!

pr67758.f90:1:26:

       COMMON /FMCOM / X(80 000 000)
                          1
Error: Expected another dimension in array declaration at (1)
pr67758.f90:3:24:

       COMMON /FMCOM / XX(80 000 000)
                        1
Error: PROCEDURE attribute conflicts with COMMON attribute in 'xx' at (1)

with r227016 (2015-08-19) and all the revisions I have tested (I have added an
END statement). Could you please post the output of 'gfortran -v'?

[Bug fortran/67758] [GCC 6 regression] ICE on invalid use of COMMON

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

kargl at gcc dot gnu.org changed:

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

--- Comment #2 from kargl at gcc dot gnu.org ---
(In reply to Dominique d'Humieres from comment #1)
> WORKSFORME!
> 
> pr67758.f90:1:26:
> 
>        COMMON /FMCOM / X(80 000 000)
>                           1
> Error: Expected another dimension in array declaration at (1)
> pr67758.f90:3:24:
> 
>        COMMON /FMCOM / XX(80 000 000)
>                         1
> Error: PROCEDURE attribute conflicts with COMMON attribute in 'xx' at (1)
> 
> with r227016 (2015-08-19) and all the revisions I have tested (I have added
> an END statement). Could you please post the output of 'gfortran -v'?

troutmask:sgk[204] gfc6 -c r5.f
f951: internal compiler error: Segmentation fault
no stack trace because unwind library not available
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
troutmask:sgk[205] gfc5 -c r5.f
r5.f:3:24:

       COMMON /FMCOM / XX(80 000 000)
                        1
Error: PROCEDURE attribute conflicts with COMMON attribute in 'xx' at (1)
f951: Error: Unexpected end of file in 'r5.f'
troutmask:sgk[206] gfc49 -c r5.f
r5.f:3.24:

      COMMON /FMCOM / XX(80 000 000)                                    
                        1
Error: PROCEDURE attribute conflicts with COMMON attribute in 'xx' at (1)
Error: Unexpected end of file in 'r5.f'

Looks like a regression in symbol.c.

(gdb) p p->common_block->head->common_next
$9 = (gfc_symbol *) 0x0
(gdb) list
3205              else
3206                {
3207                  gfc_symbol *cparent, *csym;
3208
3209                  cparent = p->common_block->head;
3210                  csym = cparent->common_next;
3211
3212                  while (csym != p)
3213                    {
3214                      cparent = csym;
3215                      csym = csym->common_next;
3216                    }
3217
3218                  gcc_assert(cparent->common_next == p);
3219                  cparent->common_next = csym->common_next;
3220                }
3221            }

[Bug fortran/67758] [GCC 6 regression] ICE on invalid use of COMMON

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

--- Comment #3 from Steve Kargl <sgk at troutmask dot apl.washington.edu> ---
On Tue, Sep 29, 2015 at 07:56:16PM +0000, kargl at gcc dot gnu.org wrote:
> 
> troutmask:sgk[204] gfc6 -c r5.f
> f951: internal compiler error: Segmentation fault
> no stack trace because unwind library not available
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <http://gcc.gnu.org/bugs.html> for instructions.

This fixes the regression.

Index: symbol.c
===================================================================
--- symbol.c    (revision 228264)
+++ symbol.c    (working copy)
@@ -3211,6 +3211,8 @@ gfc_restore_last_undo_checkpoint (void)

              while (csym != p)
                {
+                 if (!csym)
+                   goto syntax;
                  cparent = csym;
                  csym = csym->common_next;
                }
@@ -3242,6 +3244,11 @@ gfc_restore_last_undo_checkpoint (void)

   if (!single_undo_checkpoint_p ())
     pop_undo_change_set (latest_undo_chgset);
+
+  return;
+
+syntax:
+   gfc_error ("syntax or semantics issue at %C");
 }

But, the real issue is that the second COMMON is not
allowed to appear in the execution block of a program.
gfortran should probably error out earlier.

[Bug fortran/67758] [GCC 6 regression] ICE on invalid use of COMMON

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

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

--- Comment #4 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
OK. I can reproduce the ICE in free form (*.f90) by replacing '80 000 000' with
'80000000'. The change occurred between r226476 (2015-08-02, error) and r227016
(2015-08-19, ICE), likely r226732 (pr59746).

> But, the real issue is that the second COMMON is not
> allowed to appear in the execution block of a program.
> gfortran should probably error out earlier.

Agreed, although the original error

Error: PROCEDURE attribute conflicts with COMMON attribute in 'xx' at (1)

may not be obvious to understand.

[Bug fortran/67758] [GCC 6 regression] ICE on invalid use of COMMON

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

--- Comment #5 from Mikael Morin <mikael at gcc dot gnu.org> ---
Author: mikael
Date: Sun Oct  4 12:30:16 2015
New Revision: 228457

URL: https://gcc.gnu.org/viewcvs?rev=228457&root=gcc&view=rev
Log:
Fix fortran common-related error recovery ICE.

Upon reverting a symbol in a common block (after throwing an error),
the compiler was ICEing because the symbol's common_block field was set,
but the symbol was not in the common block's list of symbols.

Fixed by both adding the symbol to the common block list and setting
the symbol's common_block field at the same time.
Furthermore, the gfc_add_in_common call is delayed and its result is
ignored, so that its error messages are ignored and the compiler has
the opportunity to give a better error message.
Another gfc_add_in_common call is added later during resolution
to emit again the missing errors.

        PR fortran/67758
gcc/fortran/
        * match.c (gfc_match_common): Delay the common_block pointer
        assignment after error checking.
        Delay the call to gfc_add_in_common attribute after the handling
        of array specs.
        * resolve.c (resolve_common_vars): Call gfc_add_in_common again.
gcc/testsuite/
        * gfortran.dg/common_24.f: New.


Added:
    trunk/gcc/testsuite/gfortran.dg/common_24.f
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/match.c
    trunk/gcc/fortran/resolve.c
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/67758] [GCC 6 regression] ICE on invalid use of COMMON

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

Mikael Morin <mikael at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|kargl at gcc dot gnu.org           |mikael at gcc dot gnu.org

--- Comment #7 from Mikael Morin <mikael at gcc dot gnu.org> ---
(In reply to Dominique d'Humieres from comment #6)
> If I compile the test as free form before r228458, I get the following ICE
> 
You're right, the fix needs a fix.

[Bug fortran/67758] [6 Regression] ICE on invalid use of COMMON

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-invalid-code
           Priority|P3                          |P4
   Target Milestone|---                         |6.0
            Summary|[GCC 6 regression] ICE on   |[6 Regression] ICE on
                   |invalid use of COMMON       |invalid use of COMMON

[Bug fortran/67758] [6 Regression] ICE on invalid use of COMMON

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

--- Comment #8 from Mikael Morin <mikael at gcc dot gnu.org> ---
Author: mikael
Date: Sun Oct 18 17:17:21 2015
New Revision: 228947

URL: https://gcc.gnu.org/viewcvs?rev=228947&root=gcc&view=rev
Log:
Fix common-related error recovery ICE.

Fix an inconsistent state, between the in_common attribute
and the common_block pointer.

 - adding a symbol to a common block list in gfc_match_common is delayed
   after the call to gfc_add_in_common.
 - gfc_restore_latest_undo_checkpoint is changed to check the common_block
   pointer directly instead of the in_common attribute.
 - gfc_restore_old_symbol is changed to also restore
   the common-related pointers.  This is done using a new function created
   to factor the related memory management.
 - In gfc_restore_last_undo_checkpoint, when a symbol has been removed
   from the common block linked list, its common_next pointer is cleared.

        PR fortran/67758
gcc/fortran/
        * gfortran.h (gfc_symbol): Expand comment.
        * match.c (gfc_match_common): Delay adding the symbol to
        the common_block after the gfc_add_in_common call.
        * symbol.c (gfc_free_symbol): Move common block memory handling...
        (gfc_set_symbol_common_block): ... here as a new function.
        (restore_old_symbol): Restore common block fields.
        (gfc_restore_last_undo_checkpoint):
        Check the common_block pointer instead of the in_common attribute.
        When a symbol has been removed from the common block linked list,
        clear its common_next pointer.
gcc/testsuite/
        * gfortran.dg/common_25.f90: New file.


Added:
    trunk/gcc/testsuite/gfortran.dg/common_25.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/gfortran.h
    trunk/gcc/fortran/match.c
    trunk/gcc/fortran/symbol.c
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/67758] [6 Regression] ICE on invalid use of COMMON

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67758

Mikael Morin <mikael at gcc dot gnu.org> changed:

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

--- Comment #9 from Mikael Morin <mikael at gcc dot gnu.org> ---
This should be fixed now.