Re: gdb/1891: info all-registers or info reg register don't show double values of left (64-bit) floating registers

["John David Anglin" <dave at hiauly1 dot hia dot nrc dot ca>]

The following reply was made to PR gdb/1891; it has been noted by GNATS.

From: "John David Anglin" <dave@hiauly1.hia.nrc.ca>
To: kettenis@gnu.org (Mark Kettenis)
Cc: danglin@gcc.gnu.org, gdb-gnats@sources.redhat.com
Subject: Re: gdb/1891: info all-registers or info reg register don't show double values of left (64-bit) floating registers
Date: Wed, 9 Mar 2005 17:24:25 -0500 (EST)

 > I guess this feature was lost in the recent modernization of the HP-UX
 > stuff.  The preferred way to handle this nowadays in GDB is to use
 > something called pseudo-registers.  But this requires that
 > single-precision and double-precision registers have different names.
 > The natural way to this for PA-RISC is to use "frN" as the name of the
 > full-width double-precision register and use "frNR" and "frNL" as
 > names for the single-precision halves.  This is consistent with the
 > "HP Assembler Reference Manual", but it wouldn't exactly help people
 > who expect the gdb 5.0 behaviour.
 
 The above naming convention seems logical to me.  I don't care much
 about the behavior of previous gdb versions as they are all broken
 in one way or another.  Just need some way to see the registers in
 both float and double modes.  The main problem is consistency with
 gcc and gas.  I believe that they  treat "frN" and "frNL" as equivalent.
 Whether the register is a float or double depends on the context in
 which the name is used.  It might take some work to make the
 disassembler put out the correct name.
 
 Dave
 -- 
 J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
 National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)