x86 vs. SPARC BIOSes (was: Small servers (was Re: WTT: 1.5G of PC2700 for 1G of PC100))

[Andrew Gaylard]

On Thu, May 8, 2008 at 4:07 AM, Shannon Hendrix <shannon@...>
wrote:

> On May 5, 2008, at 16:15 , Ethan O'Toole wrote:
>
>  I don't mind the PC stuff really. I mean, once a server is setup and
>> deployed you don't really have to deal with it that much. Even on Sun and
>> SGI boxes you don't really mess with the boot prom stuff once it's up and
>> running.
>>
>
> The issue comes at 0400 when you have to drive out to a server, or even fly
> out to it, all because they could not spend $1.50 in quantity to have a real
> console.
>

I agree.  And for me, it can take *days * to travel to some of my machines.
And transporting the video data requires links that are fast & fat (by our
standards, anyway).  Obviously, since these boxes are in remote places, the
cost of such links is prohibitive.  And the rest of the time, we don't
*need* video.  So a serial BIOS is essential.

However, the reality is that x86 boxes are built to run Windows, and Windows
assumes a keyboard, mouse, and hi-res bitmapped display.  So the x86 BIOS
writers can assume the same I/O will be available.  This is a major reason
why I favour SPARC machines -- they were never designed under this
assumption.

Some of the things that drive me nuts about x86 BIOSes are:

   - There's no command-line, so there's no way to script a setup.  OK, you
   can capture the settings with dmidecode, but how do you enter them
   programmatically? (The OS hasn't booted the first time yet, remember.)
   - There's no standard way to enter the menu system.  It could be <DEL>,
   or <F11> or who knows what.
   - The menus contain all kind of stuff that you don't understand (what
   does "optimised settings" mean?  And why wouldn't I want it?)
   - BIOSes on motherboards, video boards, RAID cards (don't get me started
   on RAID cards...) all assume that they can clear the screen, wiping out that
   vital error that appeared for a brief, tantalising, moment.  Why can' they
   just scroll?
   - How do you tell an x86 box the equivalent of "setenv auto-boot? false",
   so that it comes up in the BIOS, where you can run "probe-scsi", "test-all",
   and then "boot"?

And here are some of the things that I love about the OBP:

   - emacs keybindings
   - devalias: "boot disk" can point whereever.
   - diag-mode
   - FORTH: you can actually *program* the darn thing without booting an OS.
   Or even having disks or CDROMs plugged in.  And F-code is portable to other
   CPUs, so it should suit the peripheral manufacturers too.
   - Plugin-cards can extend the boot environment with their own F-code.  So
   the basic PROM can be lean & mean. Video cards, SCSI cards, etc.can do what
   they need to.
   - TFTP net-booting (actually, HP's ILO feature where it boots from a
   http:// URL is pretty neat)
   - OBP is designed to be upgradeable, without needing to boot an OS.

IEEE 1275-1994 must be the most under-rated feature of the SPARC boxen.

OK, enough now.  Back to work.  Sorry, I get carried away...

Andrew
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Ethan O'Toole]

> However, the reality is that x86 boxes are built to run Windows, and Windows
> assumes a keyboard, mouse, and hi-res bitmapped display.  So the x86 BIOS
> writers can assume the same I/O will be available.  This is a major reason
> why I favour SPARC machines -- they were never designed under this
> assumption.

How come no one has hacked together a "command line" BIOS?

In the modern days, once a machine is setup I can't think of needing to go
into the BIOS from remote. The OS handles things once it starts booting,
and all of the Unixes have serial bios support.

>   - BIOSes on motherboards, video boards, RAID cards (don't get me started
>   on RAID cards...) all assume that they can clear the screen, wiping out that
>   vital error that appeared for a brief, tantalising, moment.  Why can' they
>   just scroll?

I think they all use direct VGA writes too.

>   - How do you tell an x86 box the equivalent of "setenv auto-boot? false",
>   so that it comes up in the BIOS, where you can run "probe-scsi", "test-all",
>   and then "boot"?

Well, in defense of the PC probe-scsi would be facilitated by the SCSI add
in card. I think EISA had some sort of system where each card provided an
extension to the BIOS or something... but I've seen it on Suns where an
add-in card couldn't be probed.

> OK, enough now.  Back to work.  Sorry, I get carried away...

What about a port of OBP to PC?




--
05 REM Signature
10 PRINT " Ethan O'Toole "
20 PRINT " FLICKR ", " http://www.flickr.com/photos/ethanotoole "
30 PRINT " YOUTUBE ", " www.youtube.com/user/telmnstr "
40 PRINT " HOMEPAGE ", " users.757.org/~ethan "
RUN
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Sevan / Venture37]

>> However, the reality is that x86 boxes are built to run Windows, and
Windows

>> assumes a keyboard, mouse, and hi-res bitmapped display.  So the x86 BIOS
>> writers can assume the same I/O will be available.  This is a major reason
>> why I favour SPARC machines -- they were never designed under this
>> assumption.
>
> How come no one has hacked together a "command line" BIOS?
>
> In the modern days, once a machine is setup I can't think of needing to go
> into the BIOS from remote. The OS handles things once it starts booting,
> and all of the Unixes have serial bios support.

what about linuxbios/coreboot
http://www.coreboot.org/Welcome_to_coreboot
_________________________________________________________________
Be a Hero and Win with Iron Man
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Bill Bradford]

On Thu, May 08, 2008 at 09:55:25AM -0400, Ethan O'Toole wrote:
> What about a port of OBP to PC?

It already exists in a way - see NetApp filers.

Bill

--
Bill Bradford
Houston, Texas
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Andrew Gaylard]

On Thu, May 8, 2008 at 3:58 PM, Sevan / Venture37 <venture37@...>
wrote:

>
> what about linuxbios/coreboot
> http://www.coreboot.org/Welcome_to_coreboot
>

Has anyone tried this?   I've been looking at it, but it looks a bit
"young".
I've got a pile of HP DL380s that I'm setting up, and I'd hate to brick any
of them.  I've also tried http://www.openbios.org/Welcome_to_OpenBIOS
but my machine hung when trying to boot it (ISTR it could be booted by
grub, but it's been a while since I tried it).

If a vendor offered coreboot+openfirmware, I'd definitely buy their HW.
(Sun? HP?  Are you guys listening?!)

Andrew
(who's now spent all day messing around with ILO, and is getting grumpy)
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Sevan / Venture37]

> On Thu, May 8, 2008 at 3:58 PM, Sevan / Venture37
> wrote:
>
>>
>> what about linuxbios/coreboot
>> http://www.coreboot.org/Welcome_to_coreboot
>>
>
> Has anyone tried this?   I've been looking at it, but it looks a bit
> "young".

I looks like the pcengines alix (successor to the wrap) boards are supported
so it should be pretty cheap to get a test system to play about with.


Sevan / Venture37
_________________________________________________________________
Great deals on almost anything at eBay.co.uk. Search, bid, find and win on
eBay today!
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Joshua Boyd]

On Thu, May 08, 2008 at 09:06:30AM -0500, Bill Bradford wrote:
> On Thu, May 08, 2008 at 09:55:25AM -0400, Ethan O'Toole wrote:
> > What about a port of OBP to PC?
>
> It already exists in a way - see NetApp filers.

Or the XO.
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Nate-30]

> How come no one has hacked together a "command line" BIOS?

http://openbios.info/OpenBIOS

- Nate
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Mark Benson-2]

Does EFI not open this type of possibility?

--
Mark Benson

My Blog:
<http://markbenson.org/blog>
Visit my Homepage: <http://homepage.mac.com/markbenson>

"Never send a human to do a machine's job..."
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Sevan / Venture37]

> Does EFI not open this type of possibility?

As beautifully demonstrated on Macs? ;)
_________________________________________________________________
Be a Hero and Win with Iron Man
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Mark Benson-2]

On 8 May 2008, at 19:27, Sevan / Venture37 wrote:
>> Does EFI not open this type of possibility?
>
> As beautifully demonstrated on Macs? ;)

You meant he ones with no serial ports? :)

I meant in general, because it's extensible as OpenBoot/OpenFirmware is.

Itanium64 machines use EFI too, is there any decent and proper console  
redirection on those?

--
Mark Benson

My Blog:
<http://markbenson.org/blog>
Visit my Homepage: <http://homepage.mac.com/markbenson>

"Never send a human to do a machine's job..."
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Ian Finder]

On Thu, May 8, 2008 at 2:35 PM, Mark <md.benson@...> wrote:
> On 8 May 2008, at 19:27, Sevan / Venture37 wrote:
>>>
>>> Does EFI not open this type of possibility?
>>
>> As beautifully demonstrated on Macs? ;)
>
> You meant he ones with no serial ports? :)
>

Yeah, If only Mac servers had serial management ports like, umm, I dunno, these:
http://www.apple.com/xserve/design.html

- F
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Ian Finder]

*edit --

But before I get too sarcastic here, I don't think the serial
management port drops you to any sort of pre-boot configuration-- it
only allows you to get a OS X console or stripped down network
configuration wizard *post-boot*, which is why the Xserve used to
require the use of a PCI slot for a video card, and now has one
onboard.

I agree with all of you that this continuing trend of needing a video
card and local keyboard to change preboot settings-- whether on an
OpenBoot, EFI, or PC BIOS platform, is broken by design-- not to
mention ridiculous given how easy it'd be for a manufacturer to
implement such a facility in their preboot environment. And on a
high-end rackmount server like the Xserve, it's inexcusable.

What I'd like to find is a server with an always available embedded
supervisor processor that provides ethernet and serial connections for
configuration, and unlike some SPs I've seen (like in the SunFire
V20zs) has the ability to change eeprom-stored boot parameters that
are used by the normal processor's boot sequence. It could also attach
via an internal serial port, and make the port available over telnet,
providing the functionality of those little ethernet to serial
embedded boxes.

-F

On Thu, May 8, 2008 at 10:37 PM, Ian Finder <ian.finder@...> wrote:

> On Thu, May 8, 2008 at 2:35 PM, Mark <md.benson@...> wrote:
>> On 8 May 2008, at 19:27, Sevan / Venture37 wrote:
>>>>
>>>> Does EFI not open this type of possibility?
>>>
>>> As beautifully demonstrated on Macs? ;)
>>
>> You meant he ones with no serial ports? :)
>>
>
> Yeah, If only Mac servers had serial management ports like, umm, I dunno, these:
> http://www.apple.com/xserve/design.html
>
> - F
>

--
 Ian Finder
 224.659.4204
 ian.finder@...
 finder | solutions - IT on your terms
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Shannon Hendrix]

On May 8, 2008, at 09:55 , Ethan O'Toole wrote:

>> However, the reality is that x86 boxes are built to run Windows,  
>> and Windows
>> assumes a keyboard, mouse, and hi-res bitmapped display.  So the  
>> x86 BIOS
>> writers can assume the same I/O will be available.  This is a major  
>> reason
>> why I favour SPARC machines -- they were never designed under this
>> assumption.
>
> How come no one has hacked together a "command line" BIOS?
>
> In the modern days, once a machine is setup I can't think of needing  
> to go into the BIOS from remote. The OS handles things once it  
> starts booting, and all of the Unixes have serial bios support.

...which does you no good in the frequent situation where a machine  
will not boot.

One major problem with PC BIOS is that it changes things even when you  
don't tell it to.

For example, a momentary error condition can cause PC BIOS to alter  
the drive boot order.  This can happen after a power failure, or even  
just a temporary drive failure on some RAID BIOS.

Your machine will not come up until you enter BIOS and fix the drive  
order.

Of course, one must ask: why can't the BIOS just take your settings  
and not change them if you didn't ask?

That's just one example.

> I think they all use direct VGA writes too.

The other day you said they didn't.... :)

> Well, in defense of the PC probe-scsi would be facilitated by the  
> SCSI add in card. I think EISA had some sort of system where each  
> card provided an extension to the BIOS or something... but I've seen  
> it on Suns where an add-in card couldn't be probed.

In that case, the add-in card was broken.

As far as I know, probe-scsi just sends standard SCSI commands, so it  
should work with any compliant adapter.

>> OK, enough now.  Back to work.  Sorry, I get carried away...
>
> What about a port of OBP to PC?

...or even just a command line interface that can talk serial or video.

Of course, the manufacturers would have to start following standards...


--
Shannon Hendrix
shannon@...
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Shannon Hendrix]

On May 8, 2008, at 23:53 , Ian Finder wrote:

> What I'd like to find is a server with an always available embedded
> supervisor processor that provides ethernet and serial connections for
> configuration, and unlike some SPs I've seen (like in the SunFire

[ snip ]

That's what a Data General MV20000 supermini I worked on had.  It had  
a small computer that ran the big one, and you could talk to it  
without booting the big one.

Very useful.

It's 2008, 20 years later... surely we have the ability to do this so  
cheaply it should not matter.

Normal users will never see it, so what?

--
Shannon Hendrix
shannon@...
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Jonathan Katz-2]

On Tue, May 13, 2008 at 12:44 PM, Shannon Hendrix <shannon@...>
wrote:

> That's what a Data General MV20000 supermini I worked on had.  It had a
> small computer that ran the big one, and you could talk to it without
> booting the big one.
>
> Very useful.


"Mainframe power in the size of a pizza box!" I remember that ad campaign
from when i was a little kid... Was never 100% sure what DG box was being
referenced in the ad, though. Anyone remember?

-Jon
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue

[Carl R. Friend]

    On Tue, 13 May 2008, Jonathan Katz wrote:

> "Mainframe power in the size of a pizza box!" I remember that ad campaign
> from when i was a little kid... Was never 100% sure what DG box was being
> referenced in the ad, though. Anyone remember?

    That was one of the later Motorola quad-88k AViiON machines.  I,too,
forget the exact model number.  I had one of the printed pizza-box
marketing promos for a while until a plumbing leak in the basement
did it in.

+------------------------------------------------+---------------------+
| Carl Richard Friend (UNIX Sysadmin)            | West Boylston       |
| Minicomputer Collector / Enthusiast            | Massachusetts, USA  |
| mailto:crfriend@...                        +---------------------+
| http://users.rcn.com/crfriend/museum           | ICBM: 42:22N 71:47W |
+------------------------------------------------+---------------------+
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue