This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re[2]: Suggested patch to egcs-19981019/config.guess => auto
- To: <tege at matematik dot su dot se>, <bug-gnu-utils at prep dot ai dot mit dot edu>
- Subject: Re[2]: Suggested patch to egcs-19981019/config.guess => auto
- From: rodneybrown at pmsc dot com
- Date: Wed, 28 Oct 98 17:51:56 -0500
- Cc: <egcs at cygnus dot com>
We have /usr/lib/sched.models on HP-UX 9.04 - but we also have the unbundled C
compiler
which is why I thought it was shipped by default on HP-UX 9.
Neither the {/usr,/opt/langtools}/lib/sched.models recognized our new box which
/usr/bin/model (also new on 10) gives as 9000/800/K380 .
sysconf(_SC_CPU_VERSION) correctly sees it as CPU_PA_RISC2_0
Generalizing the C program from the 3050*:HI-UX:*:* case could allow
config.guess to correctly identify the HPPA version on both HI-UX
and HP-UX, without autoconf needing changes for new HP models.
sysconf(_SC_CPU_VERSION) works on both the HP-UX 9.04 box and the
HP-UX 10.20 box I have accessable.
HP defines their CPU_... constants in /usr/include/sys/unistd.h
On 9.04 CPU_{HP_MC680{2,3,4}0,PA_RISC{1_{0,1},2_0}}
On 10.20 CPU_{HP_MC680{2,3,4}0,PA_RISC{1_{0,1,2},2_0}}
In view of the definitions for Motorola processors I'd be surprised
if it didn't work on HP-UX 8 and maybe even on the 300 series boxes
the current test recognizes.
Worth doing?
I'd also suggest changing the triplet for the *9??*:MPE/iX:*.* case
from hppa1.0-hp-mpeix to hppa1.1-hp-mpeix, since hppa1.0 boxes would be
infeasibly expensive to keep on maintenance by now - so I'd expect
MPE/iX to be running only on HPPA1.1 & HPPA2.0 boxes. As at MPE/iX5.5
(C.55.00 C.05.08) /usr/include/unistd.h doesn't define _SC_CPU_VERSION so the
suggestion above won't work for MPE/iX
____________________Reply Separator____________________
Subject: Re: Suggested patch to egcs-19981019/config.guess => autocon
Author: Torbjorn Granlund <tege@matematik.su.se>
Date: 23/10/98 11:48
The sched.models file is new in HP-UX 10.
It would be unwise to make config.guess dependent on it,
at least unconditionally.
Torbjörn