[Bug ada/65519] New: Unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE if ZCX_By_Default is set False

[Bug ada/65519] New: Unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE if ZCX_By_Default is set False

[syq at debian dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

            Bug ID: 65519
           Summary: Unable to coalesce ssa_names 2 and 87 which are marked
                    as MUST COALESCE if ZCX_By_Default is set False
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: ada
          Assignee: unassigned at gcc dot gnu.org
          Reporter: syq at debian dot org

Created attachment 35101
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35101&action=edit
the required files

/home/syq/gcctest/gcc-5-5-20150321/build/./gcc/xgcc
-B/home/syq/gcctest/gcc-5-5-20150321/build/./gcc/ -B/usr/x86_64-linux-gnu/bin/
-B/usr/x86_64-linux-gnu/lib/ -isystem /usr/x86_64-linux-gnu/include -isystem
/usr/x86_64-linux-gnu/sys-include -isystem
/home/syq/gcctest/gcc-5-5-20150321/build/sys-include    -c -g -O2   -W -Wall
-gnatpg -gnata -nostdinc   g-sptavs.ads -o g-sptavs.o

Unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE.
I.3_2(ab) and  I.3_87(ab)
+===========================GNAT BUG DETECTED==============================+
| 5.0.0 20150321 (experimental) [trunk revision 221551] (x86_64-linux-gnu) GCC
error:|
| SSA corruption                                                           |
| Error detected around g-spitbo.ads:373:12                                |
| Please submit a bug report; see http://gcc.gnu.org/bugs.html.            |
| Use a subject line meaningful to you and us to track the bug.            |
| Include the entire contents of this bug box in the report.               |
| Include the exact command that you entered.                              |
| Also include sources listed below.                                       |
+==========================================================================+

Please include these source files with error report
Note that list may not be accurate in some cases,
so please double check that the problem can still
be reproduced with the set of files listed.
Consider also -gnatd.n switch (see debug.adb).

system.ads
g-sptavs.ads
g-spitbo.ads
gnat.ads
ada.ads
a-finali.ads
s-finroo.ads
a-string.ads
a-strunb.ads
a-strmap.ads
a-charac.ads
a-chlat1.ads
s-atocou.ads
interfac.ads
a-tags.ads
s-stoele.ads
s-stalib.ads
a-unccon.ads
s-exctab.ads
a-stream.ads
s-soflin.ads
a-except.ads
s-parame.ads
s-traent.ads
s-stache.ads
s-unstyp.ads
s-stratt.ads
s-secsta.ads
s-stposu.ads
s-stopoo.ads
s-finmas.ads
s-pooglo.ads
s-ststop.ads
s-string.ads
a-uncdea.ads
g-spitbo.adb
a-stunau.ads
g-debuti.ads
g-io.ads
s-strhas.ads
s-valint.ads
s-strhas.adb
s-stalib.adb
s-memory.ads
s-stoele.adb


raised TYPES.UNRECOVERABLE_ERROR : comperr.adb:423
../gcc-interface/Makefile:311: recipe for target 'g-sptavs.o' failed


I can reproduce it on mips64el and amd64.

[Bug ada/65519] Unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE if ZCX_By_Default is set False

[syq at debian dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #1 from YunQiang Su <syq at debian dot org> ---
It only fails when using -O2 +, while
build successfully when using -O1.

[Bug ada/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2015-03-23
                 CC|                            |ebotcazou at gcc dot gnu.org
   Target Milestone|---                         |5.0
            Summary|Unable to coalesce          |[5 regression] unable to
                   |ssa_names 2 and 87 which    |coalesce ssa_names 2 and 87
                   |are marked as MUST COALESCE |which are marked as MUST
                   |if ZCX_By_Default is set    |COALESCE
                   |False                       |
     Ever confirmed|0                           |1

--- Comment #2 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
I can reproduce.

[Bug ada/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
                 CC|ebotcazou at gcc dot gnu.org       |
           Assignee|unassigned at gcc dot gnu.org      |ebotcazou at gcc dot gnu.org

--- Comment #3 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Investigating.

[Bug ada/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #4 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
This boils down to:

with Ada.Strings.Unbounded; use Ada.Strings.Unbounded;
with Interfaces;            use Interfaces;

package P is

   type Arr is array (Unsigned_32 range <>) of Unbounded_String;

end P;

compiled at -O2 with SJLJ exceptions.

[Bug ada/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #5 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
It's forwprop4 propagating an abnormal SSA name:

  <bb 5>:
  [...]
  I.3_30(ab) = I.3_1 + 1;
  _31 = (interfaces__unsigned_32) I.3_30(ab);

[...]

  <bb 13>:
  _43 = v.P_ARRAY;
  _45 = (sizetype) _31;

is changed into:

  <bb 13>:
  _43 = v.P_ARRAY;
  _71 = I.3_30(ab) & 4294967295;

but I.3_87(ab) is live in BB 13.

[Bug ada/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #6 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
The problematic statement is created by the gimple-match stuff:

Applying pattern match.pd:761, gimple-match.c:1727
Applying pattern match.pd:625, gimple-match.c:1525
gimple_simplified to _71 = I.3_30(ab) & 4294967295;
_45 = _71;

and falls through the cracks of replace_stmt_with_simplification:

  /* Play safe and do not allow abnormals to be mentioned in
     newly created statements.  See also maybe_push_res_to_seq.  */
  if ((TREE_CODE (ops[0]) == SSA_NAME
       && SSA_NAME_OCCURS_IN_ABNORMAL_PHI (ops[0]))
      || (ops[1]
      && TREE_CODE (ops[1]) == SSA_NAME
      && SSA_NAME_OCCURS_IN_ABNORMAL_PHI (ops[1]))
      || (ops[2]
      && TREE_CODE (ops[2]) == SSA_NAME
      && SSA_NAME_OCCURS_IN_ABNORMAL_PHI (ops[2])))
    return false;

because ops[0] is already the new SSA_NAME (_71), which was created by:

tree
gimple_build (gimple_seq *seq, location_t loc,
          enum tree_code code, tree type, tree op0, tree op1,
          tree (*valueize)(tree))
{
  tree res = gimple_simplify (code, type, op0, op1, seq, valueize);
  if (!res)
    {
      if (gimple_in_ssa_p (cfun))
    res = make_ssa_name (type);
      else
    res = create_tmp_reg (type);
      gimple stmt = gimple_build_assign (res, code, op0, op1);
      gimple_set_location (stmt, loc);
      gimple_seq_add_stmt_without_update (seq, stmt);
    }
  return res;
}

I don't think that gimple_build can fail, so maybe the way out is to test
  stmt_references_abnormal_ssa_name (SSA_NAME_DEF_STMT (ops[x]))
instead of just
  SSA_NAME_OCCURS_IN_ABNORMAL_PHI (ops [x]))
in replace_stmt_with_simplification.

[Bug ada/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

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

--- Comment #7 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> I don't think that gimple_build can fail, so maybe the way out is to test
>   stmt_references_abnormal_ssa_name (SSA_NAME_DEF_STMT (ops[x]))
> instead of just
>   SSA_NAME_OCCURS_IN_ABNORMAL_PHI (ops [x]))
> in replace_stmt_with_simplification.

Richard, does gimple-fold.c only fold the outermost expression like
fold-const.c or does it work recursively?  In the former case, the above change
would be OK but, in the latter case, some changes in genmatch.c itself might be
needed.

[Bug tree-optimization/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #8 from Richard Biener <rguenth at gcc dot gnu.org> ---
Created attachment 35121
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35121&action=edit
patch

gimple-fold works like fold - it tries to re-simplify what you feed it, but
only
the outermost level (so it doesn't really recurse).  Still I think this bug
needs changing code-generation like with the attached patch which I am testing
right now.

I always forget how to force SJLJ EH on x86_64-linux for ada so I didn't manage
to reproduce the issue or check if the patch fixes it ...

[Bug tree-optimization/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #35101|0                           |1
        is obsolete|                            |

--- Comment #10 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Created attachment 35122
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35122&action=edit
Reduced testcase

[Bug tree-optimization/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #11 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> I always forget how to force SJLJ EH on x86_64-linux for ada so I didn't
> manage to reproduce the issue or check if the patch fixes it ...

Yes, it does.

[Bug tree-optimization/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #12 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Eric Botcazou from comment #11)
> > I always forget how to force SJLJ EH on x86_64-linux for ada so I didn't
> > manage to reproduce the issue or check if the patch fixes it ...
> 
> Yes, it does.

Verified.  I'll add

Index: gcc/testsuite/gnat.dg/opt48.adb
===================================================================
--- gcc/testsuite/gnat.dg/opt48.adb     (revision 0)
+++ gcc/testsuite/gnat.dg/opt48.adb     (working copy)
@@ -0,0 +1,11 @@
+-- { dg-do compile }
+-- { dg-options "-O2" }
+
+with Ada.Strings.Unbounded; use Ada.Strings.Unbounded;
+with Interfaces;            use Interfaces;
+
+package Opt48 is
+
+  type Arr is array (Unsigned_32 range <>) of Unbounded_String;
+
+end P;

given that passes in my bootstrap/test run.

[Bug tree-optimization/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #13 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> I'll add
> 
> Index: gcc/testsuite/gnat.dg/opt48.adb
> ===================================================================
> --- gcc/testsuite/gnat.dg/opt48.adb     (revision 0)
> +++ gcc/testsuite/gnat.dg/opt48.adb     (working copy)
> @@ -0,0 +1,11 @@
> +-- { dg-do compile }
> +-- { dg-options "-O2" }
> +
> +with Ada.Strings.Unbounded; use Ada.Strings.Unbounded;
> +with Interfaces;            use Interfaces;
> +
> +package Opt48 is
> +
> +  type Arr is array (Unsigned_32 range <>) of Unbounded_String;
> +
> +end P;
> 
> given that passes in my bootstrap/test run.

Careful, it's a package spec only so it needs to have .ads extension and placed
in gnat.dg/specs.

[Bug tree-optimization/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #14 from rguenther at suse dot de <rguenther at suse dot de> ---
On Tue, 24 Mar 2015, ebotcazou at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519
> 
> --- Comment #13 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> > I'll add
> > 
> > Index: gcc/testsuite/gnat.dg/opt48.adb
> > ===================================================================
> > --- gcc/testsuite/gnat.dg/opt48.adb     (revision 0)
> > +++ gcc/testsuite/gnat.dg/opt48.adb     (working copy)
> > @@ -0,0 +1,11 @@
> > +-- { dg-do compile }
> > +-- { dg-options "-O2" }
> > +
> > +with Ada.Strings.Unbounded; use Ada.Strings.Unbounded;
> > +with Interfaces;            use Interfaces;
> > +
> > +package Opt48 is
> > +
> > +  type Arr is array (Unsigned_32 range <>) of Unbounded_String;
> > +
> > +end P;
> > 
> > given that passes in my bootstrap/test run.
> 
> Careful, it's a package spec only so it needs to have .ads extension and placed
> in gnat.dg/specs.

Ok, moved.

[Bug middle-end/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1
             Status|ASSIGNED                    |RESOLVED
          Component|tree-optimization           |middle-end
         Resolution|---                         |FIXED

--- Comment #17 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.

[Bug middle-end/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #18 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Wed Mar 25 08:46:20 2015
New Revision: 221654

URL: https://gcc.gnu.org/viewcvs?rev=221654&root=gcc&view=rev
Log:
2015-03-25  Richard Biener  <rguenther@suse.de>

    PR middle-end/65519
    * genmatch.c (expr::gen_transform): Re-write to avoid
    using gimple_build.

    * gnat.dg/specs/opt2.ads: New testcase.

Added:
    trunk/gcc/testsuite/gnat.dg/specs/opt2.ads
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/genmatch.c
    trunk/gcc/testsuite/ChangeLog

[Bug middle-end/65519] [5 regression] unable to coalesce ssa_names 2 and 87 which are marked as MUST COALESCE

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65519

--- Comment #19 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Thu Mar 26 09:26:26 2015
New Revision: 221690

URL: https://gcc.gnu.org/viewcvs?rev=221690&root=gcc&view=rev
Log:
2015-03-26  Richard Biener  <rguenther@suse.de>

    PR middle-end/65519
    * gnat.dg/specs/opt2.ads: Fix testcase.

Modified:
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gnat.dg/specs/opt2.ads