[Bug middle-end/15654] New: ICE with -O2 -fno-tree-dominator-opts

[Bug middle-end/15654] New: ICE with -O2 -fno-tree-dominator-opts

[green at redhat dot com]

I am compiling a large amount of java code with -fno-tree-dominator-opts in
order to work around a bug, when I cam across this problem.  No small test case
yet, but I'll update the offending source file right away.

$ gcj --encoding=UTF-8 -fassume-compiled
--classpath=upstream/src:/home/green/sources/jhbuild-src/rhug/rhino/upstream/toolsrc:/home/green/sources/jhbuild-src/rhug/rhino/../gnu.readline/upstream/src
-O2 -g -O2 -MT upstream/src/org/mozilla/javascript/Interpreter.lo -MD -MP -MF
upstream/src/org/mozilla/javascript/.deps/Interpreter.Tpo -c
/home/green/sources/jhbuild-src/rhug/rhino/upstream/src/org/mozilla/javascript/Interpreter.java
 -fPIC -o upstream/src/org/mozilla/javascript/.libs/Interpreter.o
-fno-tree-dominator-opts
iCodeTop_505 is live-on-entry to BB224  but is not the default def of iCodeTop_13585
iCodeTop_3613 is live-on-entry to BB224  but is not the default def of
iCodeTop_13585
/home/green/sources/jhbuild-src/rhug/rhino/upstream/src/org/mozilla/javascript/Interpreter.java:
In class `org.mozilla.javascript.Interpreter':
/home/green/sources/jhbuild-src/rhug/rhino/upstream/src/org/mozilla/javascript/Interpreter.java:
In method
`org.mozilla.javascript.Interpreter.generateICode(org.mozilla.javascript.Node,int)':
/home/green/sources/jhbuild-src/rhug/rhino/upstream/src/org/mozilla/javascript/Interpreter.java:255:
internal compiler error: in calculate_live_on_entry, at tree-ssa-live.c:713
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.

-- 
           Summary: ICE with -O2 -fno-tree-dominator-opts
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: middle-end
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: green at redhat dot com
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug middle-end/15654] ICE with -O2 -fno-tree-dominator-opts

[green at redhat dot com]

------- Additional Comments From green at redhat dot com  2004-05-25 19:34 -------
Created an attachment (id=6380)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=6380&action=view)
Offending source file

This is the offending source file.  No small test case yet.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-05-25 20:01 -------
This is caused in loop header copy pass.
Zendek, can you look into this?

I do not have a small testcase yet either.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rakdver at gcc dot gnu dot
                   |                            |org
            Summary|[3.5 Regression] ICE with - |[3.5 Regression] ICE in
                   |O2 -fno-tree-dominator-opts |calculate_live_on_entry with
                   |                            |-O2 -fno-tree-dominator-opts


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE with -O2 -fno-tree-dominator-opts

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|middle-end                  |tree-optimization
           Keywords|                            |ice-on-valid-code
            Summary|ICE with -O2 -fno-tree-     |[3.5 Regression] ICE with -
                   |dominator-opts              |O2 -fno-tree-dominator-opts
   Target Milestone|---                         |3.5.0
            Version|unknown                     |3.5.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE with -O2 -fno-tree-dominator-opts

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-05-25 19:53 -------
Here is how to get the source for someone who is going to look into this bug.
cvs -z3 -d:pserver:anoncvs@sources.redhat.com:/cvs/rhug login
> anoncvs
cvs -z3 -d:pserver:anoncvs@sources.redhat.com:/cvs/rhug co rhug/rhino
cd rhug/rhino
setenv GCJFLAGS "-O2 -g -fno-tree-dominator-opts"
./configure
make

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[belyshev at lubercy dot com]

------- Additional Comments From belyshev at lubercy dot com  2004-05-25 23:40 -------
Created an attachment (id=6384)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=6384&action=view)
small testcase (287 bytes)

(gdb) run
Starting program: /usr/local/src/gcc/head/build/gcc/jc1 Interpreter.java -quiet
-O1 -fno-tree-dominator-opts
wiCodeTop_8 is live-on-entry to BB5  and there is no default def.

Breakpoint 1, fancy_abort (file=0x87048ea "../../gcc/gcc/tree-ssa-live.c",
line=713, 
    function=0x8704ac1 "calculate_live_on_entry") at
../../gcc/gcc/diagnostic.c:589
589	  internal_error ("in %s, at %s:%d", function, trim_filename (file),
line);
(gdb) where
#0  fancy_abort (file=0x87048ea "../../gcc/gcc/tree-ssa-live.c", line=713, 
    function=0x8704ac1 "calculate_live_on_entry") at
../../gcc/gcc/diagnostic.c:589
#1  0x081feec6 in calculate_live_on_entry (map=0x884fbf0) at
../../gcc/gcc/tree-ssa-live.c:713
#2  0x081c9117 in coalesce_ssa_name (map=0x884fbf0, flags=8)
    at ../../gcc/gcc/tree-outof-ssa.c:687
#3  0x081cd6e5 in remove_ssa_form (dump=0x0, map=0x884fbf0, flags=8)
    at ../../gcc/gcc/tree-outof-ssa.c:1931
#4  0x081cdfba in rewrite_vars_out_of_ssa (vars=0x884fcf0)
    at ../../gcc/gcc/tree-outof-ssa.c:2092
#5  0x08246bc1 in duplicate_blocks (bbs_to_duplicate=0x8859f30)
    at ../../gcc/gcc/tree-ssa-loop.c:187
#6  0x082472b2 in copy_loop_headers () at ../../gcc/gcc/tree-ssa-loop.c:333
#7  0x081966eb in execute_one_pass (pass=0x878b060) at
../../gcc/gcc/tree-optimize.c:406
#8  0x081967bc in execute_pass_list (pass=0x878b060) at
../../gcc/gcc/tree-optimize.c:435
#9  0x081967e6 in execute_pass_list (pass=0x878aa20) at
../../gcc/gcc/tree-optimize.c:436
#10 0x08196bfa in tree_rest_of_compilation (fndecl=0x404d10e8, nested_p=false)
    at ../../gcc/gcc/tree-optimize.c:530
#11 0x0810b3e4 in java_expand_body (fndecl=0x404d10e8) at
../../gcc/gcc/java/decl.c:1852
#12 0x0866df22 in cgraph_expand_function (node=0x404d62b8) at
../../gcc/gcc/cgraphunit.c:794
#13 0x0866c9cf in cgraph_assemble_pending_functions () at
../../gcc/gcc/cgraphunit.c:295
#14 0x0866cb96 in cgraph_finalize_function (decl=0x404d10e8, nested=false)
    at ../../gcc/gcc/cgraphunit.c:376
#15 0x0810b3c2 in finish_method (fndecl=0x404d10e8) at
../../gcc/gcc/java/decl.c:1844
#16 0x08085009 in source_end_java_method () at parse.y:7536
#17 0x08089384 in java_expand_method_bodies (class=0x404cecb0) at parse.y:8178
#18 0x0809035e in java_expand_classes () at parse.y:9155
#19 0x081448a6 in java_parse_file (set_yydebug=0) at
../../gcc/gcc/java/jcf-parse.c:1111
#20 0x085d38a6 in compile_file () at ../../gcc/gcc/toplev.c:1653
#21 0x085d584a in do_compile () at ../../gcc/gcc/toplev.c:2699
#22 0x085d58af in toplev_main (argc=5, argv=0xbffff804) at
../../gcc/gcc/toplev.c:2731
#23 0x40039b45 in __libc_start_main (main=0xbffffa8f, argc=-1073743199,
ubp_av=0xbffffaba, 
---Type <return> to continue, or q <return> to quit---
    init=0x86e38f0 <__libc_csu_init>, fini=0x86e3950 <__libc_csu_fini>, 
    rtld_fini=0x4000baa0 <_dl_fini>, stack_end=0xbffffb13)
    at ../sysdeps/generic/libc-start.c:209
#24 0x08049c71 in _start () at ../sysdeps/i386/elf/start.S:102
(gdb) 


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-05-25 23:43 -------
Confirmed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2004-05-25 23:43:23
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[rakdver at gcc dot gnu dot org]

------- Additional Comments From rakdver at gcc dot gnu dot org  2004-05-26 23:58 -------
The problem actually is in tailcall optimization, where we remove the call
that defines iCodeTop_8, but not the assigment iCodeTop_9 = iCodeTop_8
that follows it.  I am testing the patch for the problem.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2004-05-25 23:43:23         |2004-05-26 23:58:46
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[green at redhat dot com]

------- Additional Comments From green at redhat dot com  2004-05-27 18:58 -------
(In reply to comment #6)
> The problem actually is in tailcall optimization, where we remove the call
> that defines iCodeTop_8, but not the assigment iCodeTop_9 = iCodeTop_8
> that follows it.  I am testing the patch for the problem.

Thanks to everyone who's helped with this bug!

How's the testing going Zdenek?

AG

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[rakdver at atrey dot karlin dot mff dot cuni dot cz]

------- Additional Comments From rakdver at atrey dot karlin dot mff dot cuni dot cz  2004-05-27 19:01 -------
Subject: Re:  [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

Hello,

> (In reply to comment #6)
> > The problem actually is in tailcall optimization, where we remove the call
> > that defines iCodeTop_8, but not the assigment iCodeTop_9 = iCodeTop_8
> > that follows it.  I am testing the patch for the problem.
> 
> Thanks to everyone who's helped with this bug!
> 
> How's the testing going Zdenek?

the machine I tested it on crashed, so I am rerunning the tests.  Here
is the patch, if someone wants to try it.

Zdenek

Index: tree-tailcall.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/tree-tailcall.c,v
retrieving revision 2.3
diff -c -3 -p -r2.3 tree-tailcall.c
*** tree-tailcall.c	15 May 2004 09:39:29 -0000	2.3
--- tree-tailcall.c	26 May 2004 23:55:40 -0000
*************** eliminate_tail_call (struct tailcall *t)
*** 646,651 ****
--- 646,652 ----
    stmt_ann_t ann;
    vdef_optype vdefs;
    unsigned i;
+   block_stmt_iterator bsi;
  
    stmt = bsi_stmt (t->call_bsi);
    get_stmt_operands (stmt);
*************** eliminate_tail_call (struct tailcall *t)
*** 665,670 ****
--- 666,686 ----
  
    first = ENTRY_BLOCK_PTR->succ->dest;
  
+   /* Remove the code after call_bsi that will become unreachable.  The
+      possibly unreachable code in other blocks is removed later in
+      cfg cleanup.  */
+   bsi = t->call_bsi;
+   bsi_next (&bsi);
+   while (!bsi_end_p (bsi))
+     {
+       /* Do not remove the return statement, so that redirect_edge_and_branch
+ 	 sees how the block ends.  */
+       if (TREE_CODE (bsi_stmt (bsi)) == RETURN_EXPR)
+ 	break;
+ 
+       bsi_remove (&bsi);
+     }
+ 
    /* Replace the call by a jump to the start of function.  */
    e = redirect_edge_and_branch (t->call_block->succ, first);
    if (!e)
*************** eliminate_tail_call (struct tailcall *t)
*** 682,688 ****
         param = TREE_CHAIN (param),
         args = TREE_CHAIN (args))
      {
-       
        for (phi = phi_nodes (first); phi; phi = TREE_CHAIN (phi))
  	if (param == SSA_NAME_VAR (PHI_RESULT (phi)))
  	  break;
--- 698,703 ----


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[green at redhat dot com]

------- Additional Comments From green at redhat dot com  2004-05-27 20:57 -------
(In reply to comment #8) 
> the machine I tested it on crashed, so I am rerunning the tests.  Here
> is the patch, if someone wants to try it.

Thanks Zdenek.   This patch appears to get me past the problem!

(now I have a new problem, and will file a new case)

AG

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-06-09 00:15 -------
Zdenek did you retest the patch? If so can you submit it as I think you can reproduce the ICE when 
bootstrapping with -O2 -fno-tree-dominator-opts.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[rakdver at gcc dot gnu dot org]

------- Additional Comments From rakdver at gcc dot gnu dot org  2004-06-09 00:20 -------
The patch is

http://gcc.gnu.org/ml/gcc-patches/2004-05/msg01837.html

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-07-11 19:57 -------
Subject: Bug 15654

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	rakdver@gcc.gnu.org	2004-07-11 19:57:47

Modified files:
	gcc            : ChangeLog tree-tailcall.c 

Log message:
	PR tree-optimization/15654
	* tree-tailcall.c (eliminate_tail_call): Remove unreachable code.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.4474&r2=2.4475
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-tailcall.c.diff?cvsroot=gcc&r1=2.14&r2=2.15



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654

[Bug tree-optimization/15654] [3.5 Regression] ICE in calculate_live_on_entry with -O2 -fno-tree-dominator-opts

[rakdver at gcc dot gnu dot org]

------- Additional Comments From rakdver at gcc dot gnu dot org  2004-07-11 19:58 -------
Fixed.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15654