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


In message <200204302327.g3UNR2X26790@desire.geoffk.org>, Geoff Keating writes:
 > > cc: dje@watson.ibm.com, geoffk@geoffk.org, gcc@gcc.gnu.org
 > > Reply-To: law@redhat.com
 > > From: law@redhat.com
 > > Date: Tue, 30 Apr 2002 17:13:28 -0600
 > > X-OriginalArrivalTime: 30 Apr 2002 23:10:46.0384 (UTC) FILETIME=[42787700:
 > 01C1F09C]
 > > 
 > > In message <200204302128.g3ULSB8S000435@hiauly1.hia.nrc.ca>, "John David A
 > nglin
 > > " writes:
 > >  > >  In message <200204302007.g3UK7DeK000134@hiauly1.hia.nrc.ca>, "John D
 > avid 
 > >  > > Anglin" writes:
 > >  > >  > DImode is also allowed in FPRs on the PA.  The class used by globa
 > l_all
 > >  > oc
 > >  > >  > for the pseudo was GENERAL_OR_FP_REGS.  This was the first class s
 > elect
 > >  > ed
 > >  > >  > for the psuedo and a register %fr22 was selected from this class.
 > >  > > Does the choice of GENERAL_OR_FP_REGS make sense given the uses/sets 
 > of
 > >  > > the particular register?  [ I'm probably not going to have time to lo
 > ok
 > >  > > seriously at this today. ]
 > >  > 
 > >  > No.  I would say the class should be GENERAL_REGS.
 > > Agreed now that you've remined me that 9reg 66) is a pseudo for PA64 :-)
 > > 
 > > What I find curious is that we have the same cost (0) for 
 > > R1_REGS, GENERAL_REGS and FP_REGS for reg714, yet GENERAL_OR_FP_REGS has
 > > a cost of 7000+?!?  Weird.
 > 
 > Possibly because GENERAL_OR_FP_REGS doesn't match FP_REG_CLASS_P?
Unknown.  Do we actually pass super classes down to REGISTER_MOVE_COST?

jeffk


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