On Wed, Jun 25, 2014 at 3:35 PM, Kai Tietz <ktiet...@googlemail.com> wrote: > Hello, > > so there seems to be a fallout caused by moving peephole2 pass. See PR/61608. > So we need indeed 2 peephole2 passes. > > ChangeLog > > 2014-06-25 Kai Tietz <kti...@redhat.com> > > PR rtl-optimization/61608 > * passes.def (peephole2): Readd peephole2 pass > before if-after-reload pass. > > Tested for arm*-none-*, i686-w64-cygwin, x86_64-unknown-linux-gnu. Ok > for apply?
Uh, every time I try to improve compile-time by removing passes somebody else adds other passes back. Please try to avoid this. Thanks, Richard. > Regards, > Kai > > Index: passes.def > =================================================================== > --- passes.def (Revision 211971) > +++ passes.def (Arbeitskopie) > @@ -396,6 +396,7 @@ along with GCC; see the file COPYING3. If not see > NEXT_PASS (pass_rtl_dse2); > NEXT_PASS (pass_stack_adjustments); > NEXT_PASS (pass_jump2); > + NEXT_PASS (pass_peephole2); > NEXT_PASS (pass_if_after_reload); > NEXT_PASS (pass_regrename); > NEXT_PASS (pass_cprop_hardreg); > @@ -407,8 +408,7 @@ along with GCC; see the file COPYING3. If not see > We have a single indirect branch in the entire function > before duplicate-compute-gotos pass. This vastly reduces > the size of the CFG. > - For preventing to run peephole2 pass twice, its run after > - the jump2 got removed. */ > + We need to run peephole2 pass twice. See PR/61608. */ > NEXT_PASS (pass_peephole2); > NEXT_PASS (pass_branch_target_load_optimize2); > NEXT_PASS (pass_leaf_regs);
