Anyone working on linux Alpha?
I'm talking with the Debian porters for Alpha architecture, who are having
trouble getting 6.5beta2 to work properly.
Is there anyone using this setup? If so, do you get a good regression test
result? (They don't.)
This patch was necessary for them to get 6.5beta2 to build at all:
======================================================================
diff -ruN pgsql/src/config.sub postgresql-patched/src/config.sub
--- pgsql/src/config.sub Tue Apr 20 09:01:00 1999
+++ postgresql-patched/src/config.sub Tue Jun 8 17:21:32 1999
@@ -152,7 +152,7 @@
tahoe | i860 | m32r | m68k | m68000 | m88k | ns32k | arc | arm \
| arme[lb] | pyramid | mn10200 | mn10300 | tron | a29k \
| 580 | i960 | h8300 | hppa | hppa1.0 | hppa1.1 | hppa2.0 \
- | alpha | alphaev5 | alphaev56 | we32k | ns16k | clipper \
+ | alpha | alphaev5 | alphaev56 | alphapca5[6,7] | we32k | ns16k | clipper \
| i370 | sh | powerpc | powerpcle | 1750a | dsp16xx | pdp11 \
| mips64 | mipsel | mips64el | mips64orion | mips64orionel \
| mipstx39 | mipstx39el \
@@ -176,7 +176,7 @@
| mips-* | pyramid-* | tron-* | a29k-* | romp-* | rs6000-* \
| power-* | none-* | 580-* | cray2-* | h8300-* | i960-* \
| xmp-* | ymp-* | hppa-* | hppa1.0-* | hppa1.1-* | hppa2.0*-* \
- | alpha-* | alphaev5-* | alphaev56-* | we32k-* | cydra-* \
+ | alpha-* | alphaev5-* | alphaev56-* | alphapca5[6,7]-* | we32k-* | cydra-* \
| ns16k-* | pn-* | np1-* | xps100-* | clipper-* | orion-* \
| sparclite-* | pdp11-* | sh-* | powerpc-* | powerpcle-* \
| sparc64-* | mips64-* | mipsel-* \
diff -ruN pgsql/src/template/linux_alpha postgresql-patched/src/template/linux_alpha
--- pgsql/src/template/linux_alpha Tue Jun 16 06:04:00 1998
+++ postgresql-patched/src/template/linux_alpha Tue Jun 8 17:21:32 1999
@@ -1,5 +1,5 @@
AROPT:crs
-CFLAGS:-O2 -mieee
+CFLAGS:-mieee
SHARED_LIB:-fpic
ALL:
SRCH_INC:======================================================================
This is required because of problems with egcs:
Is this patch suitable for forwarding upstream? Do all linux Alphas
have to disable -O2?Yes. At least all that use egcs as a primary compiler (gcc 2.7.x never
worked for us, so unless they're using gcc 2.8.x, it won't work with -O2).
...and...
The patch I mailed you DEFINITELY needs to be included upstream if
possible. All Alphas running linux and using egcs REQUIRE it to be
compiled without optimisation since it hoses the spinlock code (I'm
looking into it with the guy that wrote the spinlock code for Alphas).
And the alphapca56,7 machines are fairly common now. Mention to them that
it might be helpful to use a config.sub from a more recent autoconf or
libtool since it includes another two or three new Alpha types (at least
the Debian sources do...I make sure of that :-)
--
Vote against SPAM: http://www.politik-digital.de/spam/
========================================
Oliver Elphick Oliver.Elphick@lfix.co.uk
Isle of Wight http://www.lfix.co.uk/oliver
PGP key from public servers; key ID 32B8FAA1
========================================
"Fear not, for I am with thee; be not dismayed,
for I am thy God. I will strengthen thee and I will
help thee; yea, I will uphold thee with the right hand
of my righteousness." Isaiah 41:10
diff -ruN pgsql/src/template/linux_alpha postgresql-patched/src/template/linux_alpha--- pgsql/src/template/linux_alpha Tue Jun 16 06:04:00 1998 +++ postgresql-patched/src/template/linux_alpha Tue Jun 8 17:21:32 1999 @@ -1,5 +1,5 @@ AROPT:crs -CFLAGS:-O2 -mieee +CFLAGS:-mieee SHARED_LIB:-fpic ALL: SRCH_INC:
======================================================================
This is required because of problems with egcs:
Is this patch suitable for forwarding upstream? Do all linux Alphas
have to disable -O2?Yes. At least all that use egcs as a primary compiler (gcc 2.7.x never
worked for us, so unless they're using gcc 2.8.x, it won't work with -O2)....and...
The patch I mailed you DEFINITELY needs to be included upstream if
possible. All Alphas running linux and using egcs REQUIRE it to be
compiled without optimisation since it hoses the spinlock code (I'm
looking into it with the guy that wrote the spinlock code for Alphas).
And the alphapca56,7 machines are fairly common now. Mention to them that
it might be helpful to use a config.sub from a more recent autoconf or
libtool since it includes another two or three new Alpha types (at least
the Debian sources do...I make sure of that :-)
I have just removed optimization on linux/alpha for this reason. Can I
use -O optimization on this platform instead of -O2?
--
Bruce Momjian | http://www.op.net/~candle
maillist@candle.pha.pa.us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026
On Tue, 8 Jun 1999, Bruce Momjian wrote:
I have just removed optimization on linux/alpha for this reason. Can I
use -O optimization on this platform instead of -O2?
I'll test this and find out. In most cases with egcs, -O or -O0 is
sufficient, but it's unknown in this situation.
I'll get back to you in less than an hour as to whether or not -O will
work. I agree that any optimisation would be better than none.
C
well, right? I compile alot on an alphapca56-* machine and the compile
will fail without it unless I add a --host= configure option otherwise.
Plus, there are alot of these Alphas out there now, so I wanted to make
sure all of them were covered. The alphapca57-class should also cover the
new DS20's, btw, so we're set for the future as well...The full patch is attached. I see that there is (in beta3) now a comment
in src/template/linux_alpha about removing -O2, but it is not, in fact,
removed.
Done. Removed -O2 for linux_alpha and made it -O. I can not make the
other changes because that is generated by GNU autoconf.
--
Bruce Momjian | http://www.op.net/~candle
maillist@candle.pha.pa.us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026Import Notes
Reply to msg id not found: 199906091006.LAA01855@linda.lfix.co.uk | Resolved by subject fallback
Bruce Momjian <maillist@candle.pha.pa.us> writes:
Done. Removed -O2 for linux_alpha and made it -O. I can not make the
other changes because that is generated by GNU autoconf.
config.sub and config.guess are not auto-generated.
What we have already done a couple of times is to apply suggested
fixes to our current copies of these files, but to ask the submitter
to send the bug report to the GNU folks as well (autoconf@gnu.org
and bug-gnu-utils@prep.ai.mit.edu) so that it gets into future GNU
releases. We do not want to maintain a divergent version of these
files.
I think someone remarked a few days ago that they had access to a
more up-to-date version of the config scripts than was present in
the last Autoconf release (2.13). Maybe we should try to grab the
latest GNU sources of these files?
regards, tom lane
Import Notes
Reply to msg id not found: YourmessageofWed9Jun1999083908-0400199906091239.IAA06840@candle.pha.pa.us | Resolved by subject fallback
On Wed, 9 Jun 1999, Tom Lane wrote:
Bruce Momjian <maillist@candle.pha.pa.us> writes:
Done. Removed -O2 for linux_alpha and made it -O. I can not make the
other changes because that is generated by GNU autoconf.config.sub and config.guess are not auto-generated.
What we have already done a couple of times is to apply suggested
fixes to our current copies of these files, but to ask the submitter
to send the bug report to the GNU folks as well (autoconf@gnu.org
and bug-gnu-utils@prep.ai.mit.edu) so that it gets into future GNU
releases. We do not want to maintain a divergent version of these
files.I think someone remarked a few days ago that they had access to a
more up-to-date version of the config scripts than was present in
the last Autoconf release (2.13). Maybe we should try to grab the
latest GNU sources of these files?
Considering that we are pretty much requiring the use of 2.13, this does
make sense...
Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy
Systems Administrator @ hub.org
primary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org