Date: Sat, 12 Mar 2005 19:37:07 +0100 From: Marius Strobl <marius@alchemy.franken.de> To: Scott Long <scottl@samsco.org> Cc: sparc64@freebsd.org Subject: Re: Blade 100 w/ 5.4-PRERELEASE console Message-ID: <20050312193707.G55534@newtrinity.zeist.de> In-Reply-To: <42332793.1030005@samsco.org>; from scottl@samsco.org on Sat, Mar 12, 2005 at 10:32:03AM -0700 References: <20050312062214.GH10159@hex.databits.net> <20050312181935.A90975@newtrinity.zeist.de> <42332793.1030005@samsco.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Mar 12, 2005 at 10:32:03AM -0700, Scott Long wrote:
> Marius Strobl wrote:
> > On Sat, Mar 12, 2005 at 12:22:14AM -0600, Will Andrews wrote:
> >
> >>Hello,
> >>
> >>David O'Brien asked me to email you about this issue. I have a
> >>Sun Blade 100 running FreeBSD since 5.0. Recently, it appears
> >>that a change was made to uart(4) which causes the console to
> >>break. With older RELENG_5 kernels (specifically, mine from
> >>November 2004 and the Feb_2005 SNAP001 version), the kernel sends
> >>messages to ttya (serial console). However, with the recent
> >>versions, you have to set {output,input}-device to "ttya" in the
> >>OpenBoot PROM, to get the kernel messages. I believe the system
> >>will still boot. But apparently with Solaris (and older FreeBSD)
> >>the kernel sent messages to the serial console without needing to
> >
> >
> > This should be fixed in sys/dev/uart/uart_cpu_sparc64.c rev. 1.18.
> > Could you please update and verify that you now get a serial console
> > on the Blade 100 with uart(4) when input-device=keyboard and
> > output-device=screen but no keyboard is plugged in?
> >
>
> Thanks. Once there is some verification from David or others I'll start
> 6-current snapshots, and I would imagine that 5.4-pre snapshots will
> start soon too.
>
> >
> >>mess with the OBP variables. I believe this is needless pain for
> >>users trying to install FreeBSD on their Blade 100s (and possibly
> >>other models).
> >
> >
> > No, this is a nit which easily can be worked around. A real pain is
> > that binutils are broken on sparc64 and users have to manually apply
> > a patch to the base src and recompile binutils in order to be able
> > to build a number of ports.
> >
> > Marius
> >
>
>
> So, um, where do we stand on this? Is it really a problem, is there an
> issue of lack of time, lack of resources, or what? Is it fixable by
> doing a vendor import of a few files?
>
See the thread:
Error: Illegal operands: There are only 32 single precision f registers; [0-31]
started by Kris on Nov 14 '04.
It can be fixed by updating src/contrib/binutils/opcodes/sparc-opc.c
from rev. 1.9 to rev 1.10 (or rev. 1.9.6.1 which is the merge of
rev. 1.10 to binutils-2_15-branch, depending on which branch one
wants to track) on the vendor branch.
AFAIK Kris also uses rev. 1.10 of that file on the package builder
boxes since November in order to be able to build packages for the
affected ports.
Marius
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050312193707.G55534>