Re: patch to reduce register pressure from strength reduction

[Toon Moene <toon at moene dot indiv dot nluug dot nl>]

Joern Rennecke wrote:
> 
> > Program received signal SIGSEGV, Segmentation fault.
> > copy_loop_body (copy_start=0x82e40d4, copy_end=0x82e4144,
> > map=0xbfff423c, exit_label=0x83304a8, last_iteration=0,
> > unroll_type=UNROLL_MODULO,
> >     start_label=0x82e410c, loop_end=0x82e4e80, insert_before=0x82e4190,
> > copy_notes_from=0x82e4190) at ../../egcs/gcc/unroll.c:1733
> >
> > Here's the code:
> 
> I have made a patch that allows me to compile your testcase.
> I haven't analyzed the generated code, so please check it out.

For some reason I can't apply your patch:

[toon@moene compilers]$ patch -p0 -V t -d snapshots/egcs/gcc < biv2.diff
patching file `loop.h'
Hunk #1 FAILED at 101.
Hunk #2 FAILED at 113.
2 out of 3 hunks FAILED -- saving rejects to loop.h.rej
patching file `loop.c'
Hunk #2 FAILED at 4176.
Hunk #3 FAILED at 4612.
Hunk #4 succeeded at 5267 (offset 4 lines).
Hunk #6 FAILED at 7271.
Hunk #7 FAILED at 7312.
4 out of 7 hunks FAILED -- saving rejects to loop.c.rej
patching file `unroll.c'
Hunk #1 FAILED at 1801.
Hunk #2 FAILED at 2840.
Hunk #3 FAILED at 2896.
Hunk #4 FAILED at 2959.
Hunk #5 FAILED at 3038.
Hunk #6 FAILED at 3079.
6 out of 6 hunks FAILED -- saving rejects to unroll.c.rej

(this was also true of your previous patch, but I was able to derive the
changes "by hand" - this one, however, is rather large).

Are you sure your patch is against a recent CVS'd tree ?

[ I hope I do not sound too negative - I _do_ appreciate your work in
  improving {b,g}iv-combining; that's the whole reason I immediately
  jumped in to try it :-) ]

TIA,

-- 
Toon Moene (toon@moene.indiv.nluug.nl)
Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
Phone: +31 346 214290; Fax: +31 346 214286
g77 Support: fortran@gnu.org; egcs: egcs-bugs@cygnus.com