This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [tree-ssa] New regressions as of 2003-11-04


> On Tue, 2003-11-11 at 03:11, Zdenek Dvorak wrote:
> 
> > +       for (bsi = bsi_start (bb); !bsi_end_p (bsi); bsi_next (&bsi))
> > + 	{
> > + 	  stmt = bsi_stmt (bsi);
> > + 	  if (TREE_CODE (stmt) != LABEL_EXPR)
> > + 	    break;
> > + 
> Shouldn't this be 'continue'?  Or are we back to one label per basic
> block?
No, but the labels are consetuctivly at the beggining of basic block.
We can not have labels in the middle of basic blocks after real
instructions.

(BTW this check and others is also in the verify_flow_info patch I sent
earlier, perhaps we can take a look on that one and I can apply those
bits that does not fail with current tree, so we don't end up re-doing
everything)

Honza
> 
> > + 	  if (label_to_block (LABEL_EXPR_LABEL (stmt)) != bb)
> > + 	    {
> > + 	      fprintf (stderr, "Label_to_block error in bb %d\n",
> > + 		       bb->index);
> >
> I'd say something like 'Label %s does not belong in block %d".
> 
> 
> Diego.
> 


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]