[Bug tree-optimization/21155] New: ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[Bug tree-optimization/21155] New: ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[janis at gcc dot gnu dot org]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1725 bytes --]

GCC mainline gets an ICE compiling crafty from SPEC CPU2000 on
powerpc64-linux:
                                                                                
elm3b149% /home/janis/tools/gcc-mline-20050421/bin/gcc -m64 -O2 -ftree-vectorize
-maltivec -c crafty_bug.c
crafty_bug.c: In function ‘main’:
crafty_bug.c:31: internal compiler error: tree check: expected ssa_name, have
var_decl in ssa_mark_def_sites, at tree-into-ssa.c:3055
                                                                                
The test case, extracted from crafty's main.c, is still pretty big,
but taking out anything else made it work.  I tried it on another
system where it got a segfault instead, and some small changes caused
segfaults with the compiler which I was reducing it.
                                                                                
I first see the ICE with this patch from law but it might merely be
a latent bug, particularly since the test case is so touchy:
                                                                                
   http://gcc.gnu.org/ml/gcc-cvs/2005-04/msg00808.html

-- 
           Summary: ICE in ssa tree check compiling crafty with -ftree-
                    vectorize -maltivec
           Product: gcc
           Version: 4.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: janis at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: powerpc64-linux
  GCC host triplet: powerpc64-linux
GCC target triplet: powerpc64-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[janis at gcc dot gnu dot org]

------- Additional Comments From janis at gcc dot gnu dot org  2005-04-21 23:09 -------
Created an attachment (id=8705)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=8705&action=view)
minimized testcase


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
            Summary|ICE in ssa tree check       |[4.1 Regression] ICE in ssa
                   |compiling crafty with -     |tree check compiling crafty
                   |ftree-vectorize -maltivec   |with -ftree-vectorize -
                   |                            |maltivec
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[janis at gcc dot gnu dot org]

------- Additional Comments From janis at gcc dot gnu dot org  2005-04-29 22:27 -------
Today I'm seeing the same ICE building mesa and sometimes building 176.gcc.
Tests apsi. fma3d, gap, and gcc all cause GCC to segfault with "-O2
-ftree-vectorize -maltivec", some with -m32, some with -m64, and some with
both; some ICE in multiple source files for the same test (I'm using -j8).
Using a .i file sometimes makes the problem disappear, as do small changes
to the source code or using gdb.

I realize this isn't particularly helpful, but I wanted to record the failures
I'm seeing so that whoever looks into this doesn't give up if the problem
seems to have gone away.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[janis at gcc dot gnu dot org]

------- Additional Comments From janis at gcc dot gnu dot org  2005-04-29 22:39 -------
That last sentence didn't make much sense, sorry; I meant that if you can't
initially reproduce the problem, please ask me if I've got more recent
failures to try out.  Meanwhile, I'll post any short testcases I can come
up with.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[janis at gcc dot gnu dot org]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 846 bytes --]


------- Additional Comments From janis at gcc dot gnu dot org  2005-05-16 20:41 -------
The failure still exists with mainline GCC built last night:

elm3b145% /opt/gcc-nightly/mline-20050516/bin/gcc -O2 -ftree-vectorize -maltivec
-m64 crafty_bug.c
crafty_bug.c: In function ‘main’:
crafty_bug.c:31: internal compiler error: tree check: expected ssa_name, have
var_decl in is_old_name, at tree-into-ssa.c:467

SPEC CPU2000 tests gcc, crafty, and mesa all get this ICE.  Tests gap, facerec,
and apsi get segfaults in the compiler and resist attempts at mininizing test
cases.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-05-26 18:49 -------
Confirmed, this is most likely the same bug as PR 21653.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
  BugsThisDependsOn|                            |21653
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-05-26 18:49:07
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[dorit at il dot ibm dot com]

------- Additional Comments From dorit at il dot ibm dot com  2005-05-30 14:47 -------
> crafty_bug.c:31: internal compiler error: tree check: expected ssa_name, have
> var_decl in is_old_name, at tree-into-ssa.c:467
> SPEC CPU2000 tests gcc, crafty, and mesa all get this ICE.  

I've seen this error on powerpc-apple-darwin in gcc, fma3d and mesa. I don't 
see this error anymore anywhere in SPEC after applying this patch: 
http://gcc.gnu.org/ml/gcc-patches/2005-05/msg02478.html (wasn't approved yet)

> Tests gap, facerec,
> and apsi get segfaults in the compiler and resist attempts at mininizing test
> cases.

I don't see these anymore. Maybe this related to PR21639 that was recently 
solved? (after this patch:
http://gcc.gnu.org/ml/gcc-patches/2005-05/msg02477.html)



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[pinskia at gcc dot gnu dot org]

-- 
Bug 21155 depends on bug 21653, which changed state.

Bug 21653 Summary: [4.1 Regression] gcc.dg/vect failures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21653

           What    |Old Value                   |New Value
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[janis at gcc dot gnu dot org]

------- Additional Comments From janis at gcc dot gnu dot org  2005-05-31 22:06 -------
With sources from an hour ago I get the same ICE compiling 176.gcc and 191.fma3d
on powerpc64-linux using "-m64 -O2 -ftree-vectorize -maltivec"; all other SPEC
CPU2000 tests compile and run successfully using those options with the same
compiler.  These sources include both of the patches mentioned in comment #6.

Dorit, can you reproduce these using a powerpc64-linux cross compiler?  To test
an ICE it's only necessary to build cc1; I can send you the script I use.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[dorit at il dot ibm dot com]

------- Additional Comments From dorit at il dot ibm dot com  2005-06-01 11:24 -------
I think Keith is looking into these SPEC ICEs on powerpc-linux.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2005-06-11 14:22 -------
Janis, since Keith's patch went in, could you please retry
with the current sources?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[janis at gcc dot gnu dot org]

------- Additional Comments From janis at gcc dot gnu dot org  2005-06-13 17:45 -------
With mainline built last night I get no failures building and running SPEC CPU2000
on powerpc64-linux with "-ftree-vectorize -maltivec -mabi=altivec" for -m32 and
-m64; it looks fixed to me!

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155

[Bug tree-optimization/21155] [4.1 Regression] ICE in ssa tree check compiling crafty with -ftree-vectorize -maltivec

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-06-13 17:49 -------
Fixed.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21155