HEADSUP: Alpha support is being retired in 7.0
John Baldwin
greatly aided the efforts to keep FreeBSD from being too i386-centric.
However, recently the Alpha port has not had any active development or
maintenance. As a result, the quality of the Alpha releases that the Project
provides are not on par with other supported architectures and is in fact
degrading. Unfortunately, as an architecture it has also been killed by its
creator.
After considering all of this, it is time to part with Alpha for 7.0 and
beyond. At this time it is still planned to provide 6.x releases for
FreeBSD/alpha. The code will still be around in CVS history if someone
suddenly shows up and fixes a bunch of bugs and/or the architecture is
revived. Users with Alpha systems are welcome to use existing releases of
FreeBSD/alpha or another BSD such as NetBSD/alpha. We would still like to
see bug fixes for FreeBSD/alpha on 6.x so that the final release is solid.
--
John Baldwin <j...@FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve" = http://www.FreeBSD.org
_______________________________________________
freebsd...@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-curre...@freebsd.org"
Kevin Oberman
> Date: Thu, 11 May 2006 14:24:03 -0400
> Sender: owner-free...@freebsd.org
>
> Alpha was the first non-x86 port that was added to FreeBSD, and as such it has
> greatly aided the efforts to keep FreeBSD from being too i386-centric.
> However, recently the Alpha port has not had any active development or
> maintenance. As a result, the quality of the Alpha releases that the Project
> provides are not on par with other supported architectures and is in fact
> degrading. Unfortunately, as an architecture it has also been killed by its
> creator.
I'd say to was killed by its owner. It's creator (corporate) no longer
exists and its creators (designers) are very sad at its demise. It was
an amazing design for its time and could have been a powerful force in
hardware with anything that resembled reasonable marketing.
(Sorry for the digression. The Alpha was very dear to me and, while not
a designer, I knew some of them.)
--
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: obe...@es.net Phone: +1 510 486-8634
Wilko Bulte
> > From: John Baldwin <j...@freebsd.org>
> > Date: Thu, 11 May 2006 14:24:03 -0400
> > Sender: owner-free...@freebsd.org
> >
> > Alpha was the first non-x86 port that was added to FreeBSD, and as such it has
> > greatly aided the efforts to keep FreeBSD from being too i386-centric.
> > However, recently the Alpha port has not had any active development or
> > maintenance. As a result, the quality of the Alpha releases that the Project
> > provides are not on par with other supported architectures and is in fact
> > degrading. Unfortunately, as an architecture it has also been killed by its
> > creator.
>
> I'd say to was killed by its owner. It's creator (corporate) no longer
> exists and its creators (designers) are very sad at its demise. It was
> an amazing design for its time and could have been a powerful force in
> hardware with anything that resembled reasonable marketing.
Interesting side note: quite a few of the Alpha folks moved to AMD.
And see what happened :)
> (Sorry for the digression. The Alpha was very dear to me and, while not
> a designer, I knew some of them.)
Same here :/
--
Wilko Bulte wi...@FreeBSD.org
_______________________________________________
freebsd...@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-curre...@freebsd.org"
--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-...@muc.de
Peter Jeremy
>exists and its creators (designers) are very sad at its demise. It was
>an amazing design for its time and could have been a powerful force in
>hardware with anything that resembled reasonable marketing.
My sentiments as well. When I first used it (in late 1998), I thought
the architecture looked weird but after using it for a while and
reading more about the design decisions, I came to the conclusion that
it was one of the better designed architectures around. There were a
few warts (requiring software assistance to fully support IEEE FP but
not supporting precise exceptions was the biggest IMHO) but DEC
actually considered the likely impact of future changes to technology,
rather than just band-aiding an existing architecture to meet the
current technology limitations/requirements.
--
Peter Jeremy
Peter Jeremy
>Of course, I regret that the Alpha branch is not continued in FreeBSD
>development because I intend to use this and some other Alpha boxes that I
>plan to buy for many years.
As has been said elsewhere in this thread, I don't believe anyone is
overjoyed at the axing of the Alpha branch. Unfortunately, given the
demise of the architecture, enthusiasm for the Alpha has diminished to
the point where it is holding back the FreeBSD project. I'll accept
my share of the responsibility for this - neither of my Alphas are
currently in working condition and I haven't been sufficiently
motivated to repair them, as a result, I haven't been following the
progress of FreeBSD/Alpha for about 18 months.
AFAIK, FreeBSD/Alpha will continue to be supported for the life of the
6.x branch - at least another 3 years. After which, you can always
run NetBSD on it.
> I like the Alpha architecture and I got the
>impression that most of the higher level problems like seg-faults in
>Mozilla on Alpha are caused by dirty programming techniques or bugs in the
>C compiler related to 64 Bit addressing. Cleaning up such problems would
>result in better code for the other architectures also.
Definitely. This is one reason why the Alpha branch wasn't axed in the
past. However there are now several other 64-bit architectures so this
is no longer a justification for maintaining the Alpha. The SPARC64
branch has the added bonus of being a big-endian architecture so it
also detects cases where code assumes little-endian addressing.
--
Peter Jeremy
Wilko Bulte
> On Thu, 2006-May-18 12:17:34 +0200, freebs...@infopuls.com wrote:
> >Of course, I regret that the Alpha branch is not continued in FreeBSD
> >development because I intend to use this and some other Alpha boxes that I
> >plan to buy for many years.
>
> As has been said elsewhere in this thread, I don't believe anyone is
> overjoyed at the axing of the Alpha branch. Unfortunately, given the
> demise of the architecture, enthusiasm for the Alpha has diminished to
> the point where it is holding back the FreeBSD project. I'll accept
> my share of the responsibility for this - neither of my Alphas are
> currently in working condition and I haven't been sufficiently
> motivated to repair them, as a result, I haven't been following the
> progress of FreeBSD/Alpha for about 18 months.
>
> AFAIK, FreeBSD/Alpha will continue to be supported for the life of the
> 6.x branch - at least another 3 years. After which, you can always
> run NetBSD on it.
One thing that I have been pondering is that for Alpha you will now have to
directly commit things onto RELENG_[56] without going to HEAD first.
After all, there is no buildable Alpha support in HEAD anymore.
I think this is a first in FreeBSD, which makes it interesting..
> past. However there are now several other 64-bit architectures so this
> is no longer a justification for maintaining the Alpha. The SPARC64
Well.. if you just count the problems with unaligned accesses etc
I wonder if SPARC64 fills that gap? I just don't know enough about
SPARC64 here, mind you.
> branch has the added bonus of being a big-endian architecture so it
> also detects cases where code assumes little-endian addressing.
--
Wilko Bulte wi...@FreeBSD.org
Peter Jeremy
has been helpfully and invisibly inserting a "Mail-Followup-To"
line that includes an address that won't work. Thanks to Wilko
for (inadvertently) drawing my attention to it and apologies for
any bounces. I hope I've fixed it.]
On Fri, 2006-May-19 11:41:54 +0200, Wilko Bulte wrote:
>One thing that I have been pondering is that for Alpha you will now have to
>directly commit things onto RELENG_[56] without going to HEAD first.
>
>After all, there is no buildable Alpha support in HEAD anymore.
>
>I think this is a first in FreeBSD, which makes it interesting..
There have been device drivers that have been lost in -current before
so it not a totally novel concept but this is the first architecture
to have been dropped.
>Well.. if you just count the problems with unaligned accesses etc
>I wonder if SPARC64 fills that gap? I just don't know enough about
>SPARC64 here, mind you.
SPARC definitely requires aligned accesses. I'm not sure what Sun
did with the SPARC64. The other option is to set the "alignment
checking" bit on i386.
--
Peter Jeremy
Joseph Koshy
> did with the SPARC64. The other option is to set the "alignment
> checking" bit on i386.
The AC bit only works for privilege level 3 (userland code) on the i386.
--
FreeBSD Developer, http://people.freebsd.org/~jkoshy
Dag-Erling Smørgrav
> There have been device drivers that have been lost in -current before
> so it not a totally novel concept but this is the first architecture
> to have been dropped.
There is at least one precedent: Sparc32.
DES
--
Dag-Erling Smørgrav - d...@des.no
Dag-Erling Smørgrav
> Peter Jeremy <peter...@optushome.com.au> writes:
> > There have been device drivers that have been lost in -current before
> > so it not a totally novel concept but this is the first architecture
> > to have been dropped.
> There is at least one precedent: Sparc32.
oh, and MIPS.
Dag-Erling Smørgrav
> d...@des.no (Dag-Erling Smørgrav) writes:
> > Peter Jeremy <peter...@optushome.com.au> writes:
> > > There have been device drivers that have been lost in -current before
> > > so it not a totally novel concept but this is the first architecture
> > > to have been dropped.
> > There is at least one precedent: Sparc32.
> oh, and MIPS.
...which doesn't really count because it was abandoned before any of
the kernel bits hit the tree, IIRC.
Wilko Bulte
> d...@des.no (Dag-Erling Smørgrav) writes:
> > d...@des.no (Dag-Erling Smørgrav) writes:
> > > Peter Jeremy <peter...@optushome.com.au> writes:
> > > > There have been device drivers that have been lost in -current before
> > > > so it not a totally novel concept but this is the first architecture
> > > > to have been dropped.
> > > There is at least one precedent: Sparc32.
> > oh, and MIPS.
>
> ...which doesn't really count because it was abandoned before any of
> the kernel bits hit the tree, IIRC.
I'm also not too sure on sparc32 in that sense?
--
Wilko Bulte wi...@FreeBSD.org
Dag-Erling Smørgrav
> I'm also not too sure on sparc32 in that sense?
AFAIK, it was axed before 2.0, so it's not in the current CVS tree.
DES
--
Dag-Erling Smørgrav - d...@des.no
Wilko Bulte
> Wilko Bulte <w...@freebie.xs4all.nl> writes:
> > I'm also not too sure on sparc32 in that sense?
>
> AFAIK, it was axed before 2.0, so it's not in the current CVS tree.
Ah, ok..
--
Wilko Bulte wi...@FreeBSD.org
John-Mark Gurney
> >Well.. if you just count the problems with unaligned accesses etc
> >I wonder if SPARC64 fills that gap? I just don't know enough about
> >SPARC64 here, mind you.
>
> SPARC definitely requires aligned accesses. I'm not sure what Sun
> did with the SPARC64. The other option is to set the "alignment
> checking" bit on i386.
We also now have arm and ppc that help fill the aligned access
requirements.. And I've done a bit of work with arm to try to make
our IP stack more friendly to alignment required machines, and ethernet
cards that can't handle special alignment...
--
John-Mark Gurney Voice: +1 415 225 5579
"All that I will do, has been done, All that I have, has not."