On Fri, Jun 10, 2011 at 12:51 AM, Xinliang David Li <davi...@google.com> wrote: > Patch is temporally rolled back. > > Richard, looks like deeper pass manager cleanup is needed -- I would > like to delay that. For now, this leaves us two choices -- 1) do cfunc > push/pop, or 2) do pass dump while executing. None of them is ideal, > but safe enough.
Well. It seems we should take a step back and look at the whole picture and try to figure out how it should look like in the end (maybe do that in London). For now I prefer 1) over 2). Thanks, Richard. > Thanks, > > David > > On Thu, Jun 9, 2011 at 3:32 PM, Xinliang David Li <davi...@google.com> wrote: >> Though I can not reproduce it, it might be related to what Richard >> mentioned in the review -- The TODO's are executed though the pass is >> not. This opened up a can of worm -- I will revert the patches for >> now. >> >> David >> >> >> On Thu, Jun 9, 2011 at 3:05 PM, H.J. Lu <hjl.to...@gmail.com> wrote: >>> On Tue, Jun 7, 2011 at 11:54 AM, Xinliang David Li <davi...@google.com> >>> wrote: >>>> Please review the attached two patches. >>>> >>>> In the first patch, gate functions are cleaned up. All the per >>>> function legality checks are moved into the executor and the >>>> optimization heuristic checks (optimize for size) remain in the >>>> gators. These allow the the following overriding order: >>>> >>>> common flags (O2, -ftree-vrp, -fgcse etc) <--- compiler >>>> heuristic (optimize for size/speed) <--- -fdisable/enable forcing pass >>>> options <--- legality check >>>> >>>> Testing under going. Ok for trunk? >>>> >>> >>> This caused: >>> >>> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49350 >>> >>> -- >>> H.J. >>> >> >
