On Sat, 2019-12-14 at 15:33 +0100, John Paul Adrian Glaubitz wrote: > Hi Stefan! > > > The problems are in the gcc implementation > > > > - the lra implementation is buggy > > - the compare elimination can't handle parallel's containing a compare > > - df-core considers parallel's containing a compare also as a USE > > - some optimizations/mechanisms do only work if HAVE_CC0 is defined > > - way more ... > > Are you talking about GCC in general or about the m68k backend? > > In any case, I would appreciate it if you could file bug reports where you > are seeing issues and put me in the CC of these bug reports. Absolutely true for both cases.
> > We are planning to have another Bountysource campaign for the m68k > backend to fund the LRA transition and - if necessary - fix other bugs > in the backend, so constructive feedback is necessary and absolutely > welcome. That may not be necessary. I wouldn't expect the conversion to LRA for m68k to be terribly hard. I would expect some slight problems with operand predicates/constraints as LRA is a bit more strict, but they're usually not terribly hard to track down. What's also considerably easier is you can flip LRA on and iterate on issues easily -- that wasn't really possible with the cc0 transition. > I can live with the fact that my patch was refuted since I simply use > > my *working* fork, where I fixed the issues mentioned above. > > I would prefer getting all bugs in GCC reported to the GCC Bugzilla so that > they can be resolved in GCC itself. We don't gain anything if important bug > fixes are found in forks only. Exactly. jeff