Date: Sat, 9 Mar 2002 09:52:13 +0100 From: Cliff Sarginson <csfbsd@raggedclown.net> To: FreeBSD-questions@FreeBSD.org Subject: Terrible problems with A7V-E mobo/AMD 1200 Mhz duron [Long message] Message-ID: <20020309085213.GB870@raggedclown.net>
next in thread | raw e-mail | index | archive | help
Hello,
After 2 days of trying I am about to give up on this.
I am trying to successfuly get 4.5 Release/Stable to do
what its should on the following h/ware:
Asus A7V-E with 512MB PC133
Duron 1200 (*not* overclocked)
IDE at UDMA-100
300 Watt power supply
PCI card very old S3 Trio, but works fine
PCI card RTl8139, works fine
PCI sound card (old ESS)
I can install it. Configure it and run it using a CD ISO image.
When I try to build-world it gave SIG 11 errors.
So I did the following in various combinations:
- Changed the RAM
- Reduced the mem speed from 133 to 100
- Disabled Level 1/2 caching
- Checked of course fan speed/CPU temp (all good)
None of this had any effect.
So next I ran buildworld twice with make -k.
The SIG 11 occurs always on *exactly* the same files, after a complaint
about end of file found before an end of line.
So. Maybe the CD image was crap.
Re-installed all sources via NFS from a good working Rel 4 Stable
repository (on my own network). This builds without problem on other
machines.
Ran buildworld again.
Exactly the same problem.
Since the files it was failing on were non-critical I forced a make -k
all the way through. Rebuilt a generic kernel etc. etc..all according to
the rules and rebooted a new kernel and userland.
Kernel boots, programs work.
So, I think, maybe there was a GCC bug it was hitting.
Having rebuilt that as above, I tried to again rebuild world,
and a new kernel.
Now it SIG 11's immediately as soon as it starts compiling, both the
world and kernel.
A memtest86 showed an error, on both the new ram and old ram.
However after reading the author of memtest86's page, he said that it
can produce false positivies, in particular it can try and access
non-existant memory on some tests. This is exactly what I think
the errors were, they were tests of block moves, but the size of the
data being moved was reported as zero. These were identical on old and
new, and occur right at the end of the test (99%).
So I installed Linux on the system (SuSE 7.3 with a 2.4.10 kernel), and
it compiles it's kernel and runs without any complaint, both it's
generic kernel, and one specifically for Duron processors. It mentions
in dmesg that it is loading a work-around for a known VIA chipset
problem (but that has been known about for ages I think).
I checkd for BIOS revisions on Asus, but there is only one and that
didn't sound relevant.
Anybody any ideas on this ?
I am going to try and install NetBSD on it this morning, to see what it
says.
I have run out of ideas.
I am sending this to current, if someone thinks it may elicit more
response on another list please forward it for me, or tell me and I will
resubmit it.
Oh yes, all my kernels were GENERIC, not specific for any processor.
Thanks.
--
Regards
Cliff Sarginson -- <csfbsd@raggedclown.net>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020309085213.GB870>