[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From pinskia at gcc dot gnu dot org 2005-01-04 15:43 --- Fixed. -- What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED Target Milestone|--- |4.0.0 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From cvs-commit at gcc dot gnu dot org 2005-01-04 14:37 --- Subject: Bug 18308 CVSROOT:/cvs/gcc Module name:gcc Changes by: [EMAIL PROTECTED] 2005-01-04 14:37:24 Modified files: gcc: ChangeLog tree-if-conv.c dojump.c gcc/testsuite : ChangeLog Added files: gcc/testsuite/gcc.dg/vect: pr18308.c Log message: gcc: 2005-01-04 Paolo Bonzini <[EMAIL PROTECTED]> Devang Patel <[EMAIL PROTECTED]> PR tree-optimization/18308 * tree-if-conv.c (add_to_dst_predicate_list): Gimplify the operands before creating a new expression. * dojump.c (do_jump): Make drop_through_label available for all cases. Add expansion of COND_EXPR. gcc/testsuite: 2005-01-04 Paolo Bonzini <[EMAIL PROTECTED]> * gcc.dg/vect/pr18308.c: New testcase. Patches: http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.7019&r2=2.7020 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-if-conv.c.diff?cvsroot=gcc&r1=2.24&r2=2.25 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/dojump.c.diff?cvsroot=gcc&r1=1.35&r2=1.36 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.4844&r2=1.4845 http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.dg/vect/pr18308.c.diff?cvsroot=gcc&r1=NONE&r2=1.1 -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From bonzini at gcc dot gnu dot org 2004-12-28 18:49 --- http://gcc.gnu.org/ml/gcc-patches/2004-12/msg02012.html -- What|Removed |Added Keywords||patch http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
-- What|Removed |Added AssignedTo|unassigned at gcc dot gnu |bonzini at gcc dot gnu dot |dot org |org Status|NEW |ASSIGNED Last reconfirmed|2004-11-05 13:51:36 |2004-12-28 17:45:12 date|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From bonzini at gcc dot gnu dot org 2004-12-28 17:44 --- Looking at it. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From dpatel at apple dot com 2004-11-18 19:09 --- Subject: Re: ICE in do_jump, at dojump.c:274 On Nov 18, 2004, at 12:03 AM, paolo dot bonzini at lu dot unisi dot ch wrote: > While I can work on a fix, those COND_EXPR were *not* valid GIMPLE at > the time the patch was written. I agree with you. While cleanup this stuff, you even cc'ed us and left some stuff around. May be we can update rewrite_outof_ssa()? See example I included above. - Devang -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From paolo dot bonzini at lu dot unisi dot ch 2004-11-18 08:03 --- Subject: Re: ICE in do_jump, at dojump.c:274 > And that would mean it was caused by: > * dojump.c (do_jump) TRUTH_ANDIF_EXPR, TRUTH_ORIF_EXPR, COMPOUND_EXPR>: > Abort on gimplified cases. While I can work on a fix, those COND_EXPR were *not* valid GIMPLE at the time the patch was written. Paolo -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From dpatel at apple dot com 2004-11-18 02:33
---
Subject: Re: ICE in do_jump, at dojump.c:274
Andrew,
You can try following to fix tree level if-conversion. I have not
tested it completely yet.
-
Devang
Index: tree-if-conv.c
===
RCS file: /cvs/gcc/gcc/gcc/tree-if-conv.c,v
retrieving revision 2.19
diff -Idpatel.pbxuser -c -3 -p -r2.19 tree-if-conv.c
*** tree-if-conv.c 16 Nov 2004 20:02:48 - 2.19
--- tree-if-conv.c 18 Nov 2004 02:32:03 -
*** add_to_dst_predicate_list (struct loop *
*** 639,645
new_cond = unshare_expr (cond);
else
{
! tree tmp_stmt;
/* new_cond == prev_cond AND cond */
tree tmp = build (TRUTH_AND_EXPR, boolean_type_node,
unshare_expr (prev_cond), cond);
--- 639,655
new_cond = unshare_expr (cond);
else
{
! tree tmp_stmt = NULL_TREE;
! tree tmp_stmts1 = NULL_TREE;
! tree tmp_stmts2 = NULL_TREE;
! prev_cond = force_gimple_operand (unshare_expr (prev_cond),
&tmp_stmts1, true, NULL);
! if (tmp_stmts1)
! bsi_insert_before (bsi, tmp_stmts1, BSI_SAME_STMT);
!
! cond = force_gimple_operand (unshare_expr (cond), &tmp_stmts2,
true, NULL);
! if (tmp_stmts2)
! bsi_insert_before (bsi, tmp_stmts2, BSI_SAME_STMT);
!
/* new_cond == prev_cond AND cond */
tree tmp = build (TRUTH_AND_EXPR, boolean_type_node,
unshare_expr (prev_cond), cond);
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From pinskia at gcc dot gnu dot org 2004-11-18 02:31 --- And that would mean it was caused by: * dojump.c (do_jump) : Abort on gimplified cases. -- What|Removed |Added CC||bonzini at gnu dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From pinskia at gcc dot gnu dot org 2004-11-18 02:29 --- So someone removed too much from do_jump in the first place. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From dpatel at apple dot com 2004-11-18 02:26
---
After I update tree-level if-conversion to force gimple operands appropriately,
rewrite_out_of_ssa() is
converting following ...
bar()
{
_Bool _ifc_.3;
_Bool _ifc_.2;
_Bool D.1339;
_Bool D.1336;
_Bool D.1337;
_Bool D.1338;
_Bool _ifc_.1;
unsigned int ivtmp.0;
int k;
int j;
int i;
# BLOCK 0
# PRED: ENTRY [100.0%] (fallthru,exec)
k_21 = j_6 != 0 ? 2 : 0;
k_5 = j_6 == 0 ? k_21 : 2;
if (k_5 != 0) goto ; else goto ;
# SUCC: 1 [46.5%] (true,exec) 2 [53.5%] (false,exec)
# BLOCK 1
# PRED: 0 [46.5%] (true,exec)
:;
# .GLOBAL_VAR_10 = V_MAY_DEF <.GLOBAL_VAR_9>;
foo () [tail call];
# SUCC: 2 [100.0%] (fallthru,exec)
# BLOCK 2
# PRED: 0 [53.5%] (false,exec) 1 [100.0%] (fallthru,exec)
:;
return;
# SUCC: EXIT [100.0%]
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From dpatel at apple dot com 2004-11-18 01:28 --- I meant, Bigger issue is : if-convert COND_EXPR is not vectorized and do_jump() does not handle it. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From dpatel at apple dot com 2004-11-18 01:27 --- Bigger issue is : if-convert COND_EXPR is not vectorized and do_jump() does not aborts(). -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From pinskia at gcc dot gnu dot org 2004-11-05 13:51 --- ifcvt produces non-gimple: _ifc_.2_15 = (j_6 != 0 || !_ifc_.1_16) && ivtmp.0_4 != 0; Confirmed. -- What|Removed |Added CC||dpatel at apple dot com Status|UNCONFIRMED |NEW Ever Confirmed||1 Last reconfirmed|-00-00 00:00:00 |2004-11-05 13:51:36 date|| http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
[Bug tree-optimization/18308] ICE in do_jump, at dojump.c:274
--- Additional Comments From reichelt at gcc dot gnu dot org 2004-11-05 13:25 --- Zdenek, the problem appeared with your patch http://gcc.gnu.org/ml/gcc-cvs/2004-10/msg00030.html -- What|Removed |Added CC||rakdver at gcc dot gnu dot ||org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18308
