This is the mail archive of the gcc@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]
Other format: [Raw text]

Re: PR 6394


>  > Jeff, should we go ahead with a temporary fix for the ICE in the machine
>  > definition or is there a chance for a better fix?  I am thinking of the
>  > patch that I previously sent, modified to handle the situation in both
>  > 32bit and 64bit code.  This would be just for the branch.
> Let's see how fixing the 'T' constraint handling pans out.  While in theory
> we need to handle address loads into FP registers, in practice we shouldn't.

Trying to fix the 'T' constraint was in fact the approach I tried first.
Hopefully, I didn't do it correctly.  In my original patch, the 'A'
contraint was placed before the 'T' constraint because it accepted address
loads into FP registers.  If your 'T' fixes works and it proves necessary
to handle address loads into FP registers, possibly '!' could be used
to severely disparage the 'A' FP alternative.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)


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