[Bug tree-optimization/37031] New: ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[Bug tree-optimization/37031] New: ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[janis at gcc dot gnu dot org]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 3037 bytes --]

Current mainline fails to compile h264ref from SPEC CPU2006 with some sets of
options.  It's somewhat intermittent but complaints from valgrind appear to be
consistent.  Here's the failure for -O3 -ftree-loop-linear using cc1 from
within a script; using the same command directly didn't result in the failure:

bug.c: In function ‘foo’:
bug.c:13: internal compiler error: vector VEC(lambda_vector,base) index domain
error, in gather_interchange_stats at tree-loop-linear.c:150
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

valgrind reports:

==13989== Invalid read of size 4
==13989==    at 0x106DBC34: gather_interchange_stats (lambda.h:31)
==13989==    by 0x106DC5B8: linear_transform_loops (tree-loop-linear.c:211)
==13989==    by 0x10432CC8: tree_linear_transform (tree-ssa-loop.c:259)
==13989==    by 0x1029D19C: execute_one_pass (passes.c:1282)
==13989==    by 0x1029D480: execute_pass_list (passes.c:1330)
==13989==    by 0x1029D498: execute_pass_list (passes.c:1331)
==13989==    by 0x1029D498: execute_pass_list (passes.c:1331)
==13989==    by 0x103999D0: tree_rest_of_compilation (tree-optimize.c:418)
==13989==    by 0x10547DBC: cgraph_expand_function (cgraphunit.c:1038)
==13989==    by 0x1054A2D0: cgraph_optimize (cgraphunit.c:1100)
==13989==    by 0x10017584: c_write_global_declarations (c-decl.c:8038)
==13989==    by 0x10338A2C: toplev_main (toplev.c:967)
==13989==  Address 0x76A1BD8 is 0 bytes inside a block of size 20 free'd
==13989==    at 0xFFBAE98: realloc (in
/usr/lib/valgrind/ppc32-linux/vgpreload_memcheck.so)
==13989==    by 0x10720DD8: xrealloc (xmalloc.c:179)
==13989==    by 0x105152DC: vec_heap_o_reserve_1 (vec.c:176)
==13989==    by 0x1069FC98: lambda_compute_access_matrices (lambda.h:32)
==13989==    by 0x106DC458: linear_transform_loops (tree-loop-linear.c:353)
==13989==    by 0x10432CC8: tree_linear_transform (tree-ssa-loop.c:259)
==13989==    by 0x1029D19C: execute_one_pass (passes.c:1282)
==13989==    by 0x1029D480: execute_pass_list (passes.c:1330)
==13989==    by 0x1029D498: execute_pass_list (passes.c:1331)
==13989==    by 0x1029D498: execute_pass_list (passes.c:1331)
==13989==    by 0x103999D0: tree_rest_of_compilation (tree-optimize.c:418)
==13989==    by 0x10547DBC: cgraph_expand_function (cgraphunit.c:1038)

I'll attach the testcase, the valgrind output, and the script I used with delta
to initially cut down the testcase.  I've seen the failure with GCC back to
20080429 but haven't yet tried a reghunt.


-- 
           Summary: ICE for h264ref in gather_interchange_stats with -ftree-
                    loop-linear
           Product: gcc
           Version: 4.4.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: janis at gcc dot gnu dot org
GCC target triplet: powerpc-unknown-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[janis at gcc dot gnu dot org]

------- Comment #1 from janis at gcc dot gnu dot org  2008-08-05 17:41 -------
Created an attachment (id=16026)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=16026&action=view)
minimized testcase


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[janis at gcc dot gnu dot org]

------- Comment #2 from janis at gcc dot gnu dot org  2008-08-05 17:43 -------
Created an attachment (id=16027)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=16027&action=view)
output of valgrind and GCC for minimized testcase


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[janis at gcc dot gnu dot org]

------- Comment #3 from janis at gcc dot gnu dot org  2008-08-05 17:44 -------
Created an attachment (id=16028)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=16028&action=view)
script to compile testcase under valgrind


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[janis at gcc dot gnu dot org]

------- Comment #4 from janis at gcc dot gnu dot org  2008-08-05 21:24 -------
A regression showed that the problem shown by valgrind was introduced by this
patch:

    http://gcc.gnu.org/viewcvs?view=rev&rev=135672

    r135672 | spop | 2008-05-20 19:11:56 +0000 (Tue, 20 May 2008)


-- 

janis at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
            Summary|ICE for h264ref in          |[4.4 Regression] ICE for
                   |gather_interchange_stats    |h264ref in
                   |with -ftree-loop-linear     |gather_interchange_stats
                   |                            |with -ftree-loop-linear
   Target Milestone|---                         |4.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[jakub at gcc dot gnu dot org]

------- Comment #5 from jakub at gcc dot gnu dot org  2008-09-15 13:34 -------
Can you still reproduce it?  With x86_64-linux -> powerpc64-linux cross I see
no errors from:
valgrind ./cc1 -O3 -ftree-loop-linear /tmp/bug.c  -m32 -maltivec
(just a few leaks).


-- 

jakub at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[janis at gcc dot gnu dot org]

------- Comment #6 from janis at gcc dot gnu dot org  2008-09-16 17:21 -------
Valgrind no longer complains for this testcase, but the I still get the
reported ICE compiling q_matrix.c with "-O3 -ftree-loop-linear".


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[jakub at gcc dot gnu dot org]

------- Comment #7 from jakub at gcc dot gnu dot org  2009-01-08 11:04 -------
Created an attachment (id=17055)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=17055&action=view)
pr37031.i

Slightly less minimized testcase which reproduces (under valgrind only) on both
powerpc64-linux native and x86_64-linux -> powerpc64-linux cross, with
-O3 -ftree-loop-linear -m{32,64}.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[jakub at gcc dot gnu dot org]

------- Comment #8 from jakub at gcc dot gnu dot org  2009-01-08 13:39 -------
Testing a patch.


-- 

jakub at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |jakub at gcc dot gnu dot org
                   |dot org                     |
             Status|WAITING                     |ASSIGNED
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2009-01-08 13:39:09
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[jakub at gcc dot gnu dot org]

------- Comment #9 from jakub at gcc dot gnu dot org  2009-01-08 13:51 -------
Created an attachment (id=17056)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=17056&action=view)
gcc44-pr37031.patch

Fix I'm going to bootstrap/regtest.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[spop at gcc dot gnu dot org]

------- Comment #10 from spop at gcc dot gnu dot org  2009-01-08 14:18 -------
Subject: Re:  [4.4 Regression] ICE for h264ref in gather_interchange_stats with
-ftree-loop-linear

Patch looks good.  Thanks for this fix.

Sebastian


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[jakub at gcc dot gnu dot org]

------- Comment #11 from jakub at gcc dot gnu dot org  2009-01-08 16:01 -------
Subject: Bug 37031

Author: jakub
Date: Thu Jan  8 16:01:42 2009
New Revision: 143188

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=143188
Log:
        PR tree-optimization/37031
        * lambda-code.c (lambda_collect_parameters): Call pointer_set_destroy
        on parameter_set.
        (build_access_matrix): Reserve correct size for AM_MATRIX vector,
        allocate it using gc instead of heap, use VEC_quick_push instead of
        VEC_safe_push.
        * graphite.c (build_access_matrix): Allocate AM_MATRIX vector using gc
        instead of heap, use VEC_quick_push instead of VEC_safe_push.
        * tree-data-ref.h (struct access_matrix): Change matrix to gc
        allocated vector from heap allocated.
        * lambda.h: Add DEF_VEC_ALLOC_P for gc allocated lambda_vector.
        * tree-loop-linear.c (linear_transform_loops): Allocate nest
        vector only after perfect_loop_nest_depth call.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/graphite.c
    trunk/gcc/lambda-code.c
    trunk/gcc/lambda.h
    trunk/gcc/tree-data-ref.h
    trunk/gcc/tree-loop-linear.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031

[Bug tree-optimization/37031] [4.4 Regression] ICE for h264ref in gather_interchange_stats with -ftree-loop-linear

[jakub at gcc dot gnu dot org]

------- Comment #12 from jakub at gcc dot gnu dot org  2009-01-08 16:03 -------
Fixed.


-- 

jakub at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37031