[Bug tree-optimization/59779] New: [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[Bug tree-optimization/59779] New: [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[danglin at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59779

            Bug ID: 59779
           Summary: [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c
                    scan-tree-dump-times parloops "parallelizing outer
                    loop"
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: danglin at gcc dot gnu.org
              Host: hppa-unknown-linux-gnu
            Target: hppa-unknown-linux-gnu
             Build: hppa-unknown-linux-gnu

Created attachment 31816
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31816&action=edit
Tree dump

Executing on host: /home/dave/gnu/gcc/objdir/gcc/xgcc
-B/home/dave/gnu/gcc/objdi
r/gcc/ /home/dave/gnu/gcc/gcc/gcc/testsuite/gcc.dg/autopar/outer-1.c 
-fno-diagnostics-show-caret -fdiagnostics-color=never   -O2
-ftree-parallelize-loops=4 -fdump-tree-parloops-details -fdump-tree-optimized
-S  -o outer-1.s    (timeout = 300)spawn /home/dave/gnu/gcc/objdir/gcc/xgcc
-B/home/dave/gnu/gcc/objdir/gcc/ /home/
dave/gnu/gcc/gcc/gcc/testsuite/gcc.dg/autopar/outer-1.c
-fno-diagnostics-show-caret -fdiagnostics-color=never -O2
-ftree-parallelize-loops=4 -fdump-tree-parloops-details -fdump-tree-optimized
-S -o outer-1.sPASS: gcc.dg/autopar/outer-1.c (test for excess errors)
FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing
oute
r loop" 1FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times optimized "loopfn"
4

r204343 was ok.  r204829 failed.

[Bug tree-optimization/59779] [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[danglin at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59779

--- Comment #1 from John David Anglin <danglin at gcc dot gnu.org> ---
Created attachment 31817
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31817&action=edit
tree dump

[Bug tree-optimization/59779] [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59779

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.9.0

[Bug tree-optimization/59779] [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59779

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

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

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Can't reproduce this with a cross-compiler to hppa-unknown-linux on current
trunk:
./cc1 -O2 -quiet -ftree-parallelize-loops=4 -fdump-tree-parloops-details
-fdump-tree-optimized outer-1.c -nostdinc; grep 'parallelizing outer loop'
outer-1.c.112t.parloops; grep loopfn outer-1.c.169t.optimized
parallelizing outer loop 9
  __builtin_GOMP_parallel (parloop._loopfn.0, &.paral_data_store.8, 4, 0);
;; Function parloop._loopfn.0 (parloop._loopfn.0, funcdef_no=2, decl_uid=1447,
symbol_order=3)
parloop._loopfn.0 (void * .paral_data_param)

[Bug tree-optimization/59779] [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[dave.anglin at bell dot net]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59779

--- Comment #3 from dave.anglin at bell dot net ---
On 12-Mar-14, at 9:55 AM, jakub at gcc dot gnu.org wrote:

> Can't reproduce this with a cross-compiler to hppa-unknown-linux on  
> current
> trunk:


Could this be a HOST_WIDE_INT issue?  I see this for 32-bit builds on  
both hpux and linux,
but not on 64-bit builds.

Dave
--
John David Anglin    dave.anglin@bell.net

[Bug tree-optimization/59779] [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59779

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-03-13
     Ever confirmed|0                           |1

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Indeed, you're right, seems cddce1 removes the first loop when HOST_WIDE_INT is
32-bit.  So, you really don't need -ftree-parallelize-loop=4 at all, the
testcase would just abort at runtime.  Looking into it.

[Bug tree-optimization/59779] [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59779

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
The following fixes this for me:
--- gcc/tree-dfa.c.jj    2014-01-03 11:40:57.000000000 +0100
+++ gcc/tree-dfa.c    2014-03-13 11:22:20.727848114 +0100
@@ -615,6 +615,15 @@ get_ref_base_and_extent (tree exp, HOST_
     maxsize = -1;

  done:
+  if (bitsize < 0)
+    {
+      *poffset = 0;
+      *psize = -1;
+      *pmax_size = -1;
+
+      return exp;
+    }
+
   if (!bit_offset.fits_shwi ())
     {
       *poffset = 0;
@@ -645,6 +654,9 @@ get_ref_base_and_extent (tree exp, HOST_
     maxsize = tree_to_uhwi (TYPE_SIZE (TREE_TYPE (exp))) - hbit_offset;
     }

+  if (maxsize < 0)
+    maxsize = -1;
+
   /* ???  Due to negative offsets in ARRAY_REF we can end up with
      negative bit_offset here.  We might want to store a zero offset
      in this case.  */

The problem is that the array is 0xbebc2000 bits long, and most of the places
in get_ref_base_and_extent work with maxsize (and bitsize) as uhwi, so we can
easily end up with negative, but not -1, maxsize (or bitsize), but callers
obviously expect sizes to be positive or -1 (special magic value for unknown).

[Bug tree-optimization/59779] [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59779

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

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

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 32344
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32344&action=edit
gcc49-pr59779.patch

Untested fix.

[Bug tree-optimization/59779] [4.9 Regression] FAIL: gcc.dg/autopar/outer-1.c scan-tree-dump-times parloops "parallelizing outer loop"

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59779

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Author: jakub
Date: Thu Mar 13 19:10:05 2014
New Revision: 208554

URL: http://gcc.gnu.org/viewcvs?rev=208554&root=gcc&view=rev
Log:
    PR tree-optimization/59779
    * tree-dfa.c (get_ref_base_and_extent): Use double_int
    type for bitsize and maxsize instead of HOST_WIDE_INT.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/tree-dfa.c