Re: [PATCH, middle-end]: Fix PR tree-optimization/{31966, 32533}

[Uros Bizjak <ubizjak at gmail dot com>]

Richard Guenther wrote:
> > > > > 2007-07-02  Uros Bizjak  <ubizjak@gmail.com>
> > > > >
> > > > >         PR tree-optimization/31966
> > > > >         PR tree-optimization/32533
> >
> > > On the 4.2 branch the testcase for PR32533 ICEs:
> > >
> > > 
> > /abuild/rguenther/gcc-4_2-branch/gcc/testsuite/gfortran.dg/pr32533.f90:
> > > In function 't':^M
> > > 
> > /abuild/rguenther/gcc-4_2-branch/gcc/testsuite/gfortran.dg/pr32533.f90:6: 
> >
> > > error: incorrect sharing of tree nodes^M
> > > D.1370_44 = M.2D.1253_16 > 1;^M
> > > ^M
> > > M.2D.1253_16 > 1;^M
> > > ^M
> > > 
> > /abuild/rguenther/gcc-4_2-branch/gcc/testsuite/gfortran.dg/pr32533.f90:6: 
> >
> > > internal compiler error: verify_stmts failed^M
> > > Please submit a full bug report,^M
> > > with preprocessed source if appropriate.^M
> > > See <URL:http://gcc.gnu.org/bugs.html> for instructions.^M
>
> This seems to fix it.
>
> Index: tree-if-conv.c
> ===================================================================
> --- tree-if-conv.c      (revision 126670)
> +++ tree-if-conv.c      (working copy)
> @@ -764,7 +764,8 @@ find_phi_replacement_condition (struct l
>      value as condition. Various targets use different means to 
> communicate
>      condition in vector compare operation. Using gimple value allows 
> compiler
>      to emit vector compare and select RTL without exposing compare's
> result.  */
> -  *cond = force_gimple_operand (*cond, &new_stmts, false, NULL_TREE);
> +  *cond = force_gimple_operand (unshare_expr (*cond),
> +                               &new_stmts, false, NULL_TREE);
>   if (new_stmts)
>     bsi_insert_before (bsi, new_stmts, BSI_SAME_STMT);
>   if (!is_gimple_reg (*cond) && !is_gimple_condexpr (*cond))
It looks that by touching tree-if-conv.c I somehow inherited this source ;)

Anyway, Richard's fix was bootstrapped and regression tested on 
i686-pc-linux-gnu 4.2 branch for all default languages, bootstrapped and 
regression tested mainline on x86_64-pc-linux-gnu  for all default 
languages, bootstrap on gcc-4.1 branch is currently running.

Having hit the tree node sharing violation, the proposed patch is kind 
of obvious (but only exposed by --enable-checking config).
OK for mainline and branches?

2007-07-16  Richard Guenther  <rguenther@suse.de>
           Uros Bizjak  <ubizjak@gmail.com>

       * tree-if-conv.c (find_phi_replacement_condition): Unshare "*cond"
       before forcing it to gimple operand.

Uros.