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]

Re: Local update flow info assumes valid situations are illegal


On Jan  4, 2001, Richard Henderson <rth@redhat.com> wrote:

> I'm somewhat surprised that EPILOGUE_USES plus the global
> life recalculation at the end of optimize_mode_switching
> is not sufficient.

It has been sufficient for SH4, but EPILOGUE_USES doesn't set fpscr
live on SH3E or sh4-single-only, which is where the bug showed up.
Matt had posted a patch to fix it, I approved it, but then decided
there might be a better fix (along the lines of the patch I posted
today).  It turned out I was wrong, so, Matt, please go ahead and
install your patch.

> Are references to the fpscr emitted before optimize_mode_switching?
> If so, why, and if not, why are things going wrong with the split
> in sched1?  There's nothing in between the o_m_s life update...

I'd have to study the test-case (that I've just got from Matt) to
answer these questions.  Please let me know if their answer is still
relevant (maybe the clarification above is enough).

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

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