[Bug tree-optimization/54570] New: [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[Bug tree-optimization/54570] New: [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

             Bug #: 54570
           Summary: [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c
                    execution test
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: rguenth@gcc.gnu.org


I broke it by making early FRE more reliably figure out address equivalences.

extern void abort (void);
union A
{
  int a1;
  char a2[3];
};
int main()
{
  union A *a = __builtin_malloc (212);
  if (__builtin_object_size (&a->a1, 1) != 4
      || __builtin_object_size (&a->a2, 1) != 212)
    abort ();
  return 0;
}

Would break in similar ways with struct { int a[]; } a; and &a vs. &a.a.

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org
   Target Milestone|---                         |4.8.0

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-09-13 15:15:31 UTC ---
We talked about disabling SCCVNs offset-based field VN here:

        case COMPONENT_REF:
          /* The field decl is enough to unambiguously specify the field,
             a matching type is not necessary and a mismatching type
             is always a spurious difference.  */
          temp.type = NULL_TREE;
          temp.op0 = TREE_OPERAND (ref, 1);
          temp.op1 = TREE_OPERAND (ref, 2);
          {
            tree this_offset = component_ref_field_offset (ref);
            if (this_offset
...

thus keep temp.off == -1 or adjust the equality function.

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org
   Target Milestone|---                         |4.8.0

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-09-13 15:15:31 UTC ---
We talked about disabling SCCVNs offset-based field VN here:

        case COMPONENT_REF:
          /* The field decl is enough to unambiguously specify the field,
             a matching type is not necessary and a mismatching type
             is always a spurious difference.  */
          temp.type = NULL_TREE;
          temp.op0 = TREE_OPERAND (ref, 1);
          temp.op1 = TREE_OPERAND (ref, 2);
          {
            tree this_offset = component_ref_field_offset (ref);
            if (this_offset
...

thus keep temp.off == -1 or adjust the equality function.

--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-09-13 15:17:10 UTC ---
And introducing a global flag whether any __builtin_object_size calls where
in the code and not pessimize opts if not.

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

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

--- Comment #3 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-09-13 23:55:35 UTC ---
*** Bug 54573 has been marked as a duplicate of this bug. ***

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-09-13
                 CC|                            |ebotcazou at gcc dot
                   |                            |gnu.org
     Ever Confirmed|0                           |1

--- Comment #4 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-09-13 23:58:36 UTC ---
By the duplicate.

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

--- Comment #5 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-09-14 01:03:09 UTC ---
(In reply to comment #3)
> *** Bug 54573 has been marked as a duplicate of this bug. ***

Sorry, but bugzilla just didn't find this for me when doing due diligence.
Putting "builtin-object-size-8.c" in the Comment / default choice ("contains
all of the strings") or choosing "contains the string" which used to be
foolproof, doesn't find anything.  I don't know what to do about that.  Other
similar search strings still find PR:s, like "attr-weakref-1.c" and
"PT_GNU_EH_FRAME".

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

--- Comment #6 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-09-14 01:30:57 UTC ---
(In reply to comment #5)
> Sorry, but bugzilla just didn't find this for me when doing due diligence.

Update: now it does.  Weird.  I guess the database wasn't updated or something.
Whatever.

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ubizjak at gmail dot com

--- Comment #7 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2012-09-19 10:24:57 UTC ---
*** Bug 54624 has been marked as a duplicate of this bug. ***

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

--- Comment #8 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-09-19 12:08:23 UTC ---
Recent added -Og also FAILs some of the -chk testcases at runtime.

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

--- Comment #9 from Richard Biener <rguenth at gcc dot gnu.org> 2012-10-26 10:32:53 UTC ---
Note that attributes on __builtin_object_size are wrong - it is marked as pure
which means that

 struct A { char a[4]; int i; } x;

 __builtin_object_size (&x, 1) == __builtin_object_size (&x.a[0], 1)

this only does not "break" as we fold the individual object size calls
to (different) constants before checking whether they are equivalent calls.
Same effect if it were const instead of pure.

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

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

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

--- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> 2012-12-11 12:54:43 UTC ---
An alternative suggestion was to allow arbitrary complex addresses (well,
"arbitrary" == gimplified ADDR_EXPRs) in call arguments (either in general
or just for specially marked builtins).  That way they escape SSA based CSE.

Yet another variant would be to have an optional 2nd operand for ADDR_EXPRs
for the Frontend to fill in, specifying the "size" of the object at that
address.  Preserving that across propagation/substitution would be required
of course (details on what the 'size' of &a[2] with int a[4]; would be is
still to be determined).

I'm leaning towards trying to have explicit information tracked from their
origin rather than trying to re-discover them after optimizations obfuscated
them.

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

--- Comment #11 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-12-11 14:35:06 UTC ---
Author: jakub
Date: Tue Dec 11 14:34:57 2012
New Revision: 194401

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=194401
Log:
    PR tree-optimization/54570
    * gcc.dg/builtin-object-size-8.c: Xfail.
    * gcc.dg/builtin-object-size-13.c: New test.

Added:
    trunk/gcc/testsuite/gcc.dg/builtin-object-size-13.c
Modified:
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.dg/builtin-object-size-8.c

[Bug tree-optimization/54570] [4.8/4.9 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.8.0                       |4.9.0
            Summary|[4.8 Regression] FAIL:      |[4.8/4.9 Regression] FAIL:
                   |gcc.dg/builtin-object-size- |gcc.dg/builtin-object-size-
                   |8.c execution test          |8.c execution test

--- Comment #12 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-12-11 14:45:27 UTC ---
Deferring for 4.9.0, this isn't safely fixable for 4.8.0 and hopefully doesn't
cause problems in too many real-world programs.

[Bug tree-optimization/54570] [4.8/4.9 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P1                          |P2

[Bug tree-optimization/54570] [4.8/4.9 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

--- Comment #13 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Mon Nov 18 15:25:05 2013
New Revision: 204966

URL: http://gcc.gnu.org/viewcvs?rev=204966&root=gcc&view=rev
Log:
2013-11-18  Richard Biener  <rguenther@suse.de>

    PR tree-optimization/59125
    PR tree-optimization/54570
    * tree-ssa-sccvn.c (copy_reference_ops_from_ref): When inlining
    is not complete do not treat component-references with offset zero
    but different fields as equal.
    * tree-object-size.c: Include tree-phinodes.h and ssa-iterators.h.
    (compute_object_sizes): Apply TLC.  Propagate the constant
    results into all uses and fold their stmts.
    * passes.def (pass_all_optimizations): Move pass_object_sizes
    after the first pass_forwprop and before pass_fre.

    * gcc.dg/builtin-object-size-8.c: Un-xfail.
    * gcc.dg/builtin-object-size-14.c: New testcase.
    * gcc.dg/strlenopt-14gf.c: Adjust.
    * gcc.dg/strlenopt-1f.c: Likewise.
    * gcc.dg/strlenopt-4gf.c: Likewise.

Added:
    trunk/gcc/testsuite/gcc.dg/builtin-object-size-14.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/passes.def
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.dg/builtin-object-size-8.c
    trunk/gcc/testsuite/gcc.dg/strlenopt-14gf.c
    trunk/gcc/testsuite/gcc.dg/strlenopt-1f.c
    trunk/gcc/testsuite/gcc.dg/strlenopt-4gf.c
    trunk/gcc/tree-object-size.c
    trunk/gcc/tree-ssa-sccvn.c

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.9.0
            Summary|[4.8/4.9 Regression] FAIL:  |[4.8 Regression] FAIL:
                   |gcc.dg/builtin-object-size- |gcc.dg/builtin-object-size-
                   |8.c execution test          |8.c execution test
      Known to fail|                            |4.8.2

--- Comment #14 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed on trunk sofar.

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

Bug 54570 depends on bug 59362, which changed state.

Bug 59362 Summary: [4.9 Regression] Abort in fini_object_sizes
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59362

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

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

--- Comment #16 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Thu Jan  9 15:25:34 2014
New Revision: 206467

URL: http://gcc.gnu.org/viewcvs?rev=206467&root=gcc&view=rev
Log:
2014-01-09  Richard Biener  <rguenther@suse.de>

    Backport from mainline
    2013-11-18  Richard Biener  <rguenther@suse.de>

    PR tree-optimization/59125
    PR tree-optimization/54570
    * tree-ssa-sccvn.c (copy_reference_ops_from_ref): When inlining
    is not complete do not treat component-references with offset zero
    but different fields as equal.
    * tree-object-size.c: Include tree-phinodes.h and ssa-iterators.h.
    (compute_object_sizes): Apply TLC.  Propagate the constant
    results into all uses and fold their stmts.
    * passes.def (pass_all_optimizations): Move pass_object_sizes
    after the first pass_forwprop and before pass_fre.

    * gcc.dg/builtin-object-size-8.c: Un-xfail.
    * gcc.dg/builtin-object-size-14.c: New testcase.
    * gcc.dg/strlenopt-14gf.c: Adjust.
    * gcc.dg/strlenopt-1f.c: Likewise.
    * gcc.dg/strlenopt-4gf.c: Likewise.

    2013-12-03  Jakub Jelinek  <jakub@redhat.com>

    PR tree-optimization/59362
    * tree-object-size.c (object_sizes): Change into array of
    vec<unsigned HOST_WIDE_INT>.
    (compute_builtin_object_size): Check computed bitmap for
    non-NULL instead of object_sizes.  Call safe_grow on object_sizes
    vector if new SSA_NAMEs appeared.
    (init_object_sizes): Check computed bitmap for non-NULL.
    Call safe_grow on object_sizes elements instead of initializing
    it with XNEWVEC.
    (fini_object_sizes): Call release on object_sizes elements, don't
    set it to NULL.

    * gcc.c-torture/compile/pr59362.c: New test.

Added:
    branches/gcc-4_8-branch/gcc/testsuite/gcc.c-torture/compile/pr59362.c
    branches/gcc-4_8-branch/gcc/testsuite/gcc.dg/builtin-object-size-14.c
Modified:
    branches/gcc-4_8-branch/gcc/ChangeLog
    branches/gcc-4_8-branch/gcc/passes.c
    branches/gcc-4_8-branch/gcc/testsuite/ChangeLog
    branches/gcc-4_8-branch/gcc/testsuite/gcc.dg/builtin-object-size-8.c
    branches/gcc-4_8-branch/gcc/testsuite/gcc.dg/strlenopt-14gf.c
    branches/gcc-4_8-branch/gcc/testsuite/gcc.dg/strlenopt-1f.c
    branches/gcc-4_8-branch/gcc/testsuite/gcc.dg/strlenopt-4gf.c
    branches/gcc-4_8-branch/gcc/tree-object-size.c
    branches/gcc-4_8-branch/gcc/tree-ssa-sccvn.c

[Bug tree-optimization/54570] [4.8 Regression] FAIL: gcc.dg/builtin-object-size-8.c execution test

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54570

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
      Known to work|                            |4.8.3
         Resolution|---                         |FIXED

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