proposal for changes in config.guess
Peter Farkas
Peter.Farkas at Sun.COM
Sat Dec 9 00:39:46 CET 2006
Hello developers,
I am proposing two small changes in config.guess. I am referring to
the version which is in GMP 4.2.1 (2006-05-04).
The first change is about the call to sysinfo in the lines 544-563,
specifically line 550. The following output from diff shows the
proposed change:
% diff -C 5 config.guess.orig config.guess.new
*** config.guess.orig Fri Dec 8 14:53:56 2006
--- config.guess.new Fri Dec 8 15:06:31 2006
***************
*** 545,555 ****
# sysinfo has been seen living in /bin or in /usr/kvm
# cpu0 is a "SuperSPARC Model 41 SPARCmodule" CPU
# cpu0 is a "75 MHz TI,TMS390Z55" CPU
#
if test -z "$exact_cpu"; then
! for i in sysinfo /usr/kvm/sysinfo; do
if $SHELL -c $i 2>/dev/null >conftest.dat; then
if grep 'cpu0 is a "SuperSPARC' conftest.dat >/dev/null; then
exact_cpu=supersparc
break
elif grep 'cpu0 is a .*TMS390Z5.' conftest.dat >/dev/null; then
--- 545,555 ----
# sysinfo has been seen living in /bin or in /usr/kvm
# cpu0 is a "SuperSPARC Model 41 SPARCmodule" CPU
# cpu0 is a "75 MHz TI,TMS390Z55" CPU
#
if test -z "$exact_cpu"; then
! for i in /bin/sysinfo /usr/kvm/sysinfo; do
if $SHELL -c $i 2>/dev/null >conftest.dat; then
if grep 'cpu0 is a "SuperSPARC' conftest.dat >/dev/null; then
exact_cpu=supersparc
break
elif grep 'cpu0 is a .*TMS390Z5.' conftest.dat >/dev/null; then
The reason for the change is the following: sysinfo should not be
called without a full path, by relying on the user's path. A utility
with this name exists (see http://www.MagniComp.com/sysinfo ) and it
may be in the user's path). This sysinfo is a GUI based, interactive
utility, and it may prevent config.guess from running smoothly.
By the way, I believe that the sysinfo from SunOS has been deleted
from recent versions of Solaris (2.8 and following), so this section
of the script may be completely obsolete.
The second change is about adding a case to the statements following
the call to uname in the lines 647-655. The following output from
diff shows the proposed change:
% diff -C 3 config.guess.orig config.guess.new
*** config.guess.orig Fri Dec 8 14:53:56 2006
--- config.guess.new Fri Dec 8 15:06:31 2006
***************
*** 644,655 ****
rm -f conftest.dat
fi
! # sun4m and sun4d are v8s of some sort, sun4u is a v9 of some sort
#
if test -z "$exact_cpu"; then
case `uname -m` in
sun4[md]) exact_cpu=sparcv8 ;;
! sun4u) exact_cpu=sparcv9 ;;
esac
fi
;;
--- 644,655 ----
rm -f conftest.dat
fi
! # sun4m, sun4d are v8 of some sort, sun4u, sun4v are v9 of some sort
#
if test -z "$exact_cpu"; then
case `uname -m` in
sun4[md]) exact_cpu=sparcv8 ;;
! sun4[uv]) exact_cpu=sparcv9 ;;
esac
fi
;;
The reason for the change is the following: The cases after the call
to "uname -m" in config.guess do not cover the case sun4v, which is
the architecture for systems based on the recently released
UltraSPARC-T1. Here is the "uname -a" output from a UltraSPARC-T1
based system:
% /usr/bin/uname -a
SunOS perftech-014 5.10 Generic_118833-16 sun4v sparc SUNW,Sun-Fire-T200
% /usr/sbin/prtdiag
System Configuration: Sun Microsystems sun4v Sun Fire T200
System clock frequency: 200 MHz
Memory size: 32760 Megabytes
========================= CPUs
===============================================
CPU CPU
Location CPU Freq Implementation Mask
------------ ----- -------- ------------------- -----
MB/CMP0/P0 0 1200 MHz SUNW,UltraSPARC-T1
MB/CMP0/P1 1 1200 MHz SUNW,UltraSPARC-T1
MB/CMP0/P2 2 1200 MHz SUNW,UltraSPARC-T1
MB/CMP0/P3 3 1200 MHz SUNW,UltraSPARC-T1
MB/CMP0/P4 4 1200 MHz SUNW,UltraSPARC-T1
. . . . . . . . . . . . . . . . . . . .
Thank you,
Peter Farkas
Staff Engineer
Sun Microsystems, Inc.
More information about the gmp-devel
mailing list