[9fans] Dum-Bass question

[9fans] Dum-Bass question

[W B Hacker]

Fans,

Online docs (http://planb.lsub.org/magic/man2html/8/9load) describing Plan9's 
boot process, say, in part:

...in order to find configuration information, 9load searches all units on 
devices fd and sdCn, in that order, for a file called plan9\plan9.ini or 
plan9.ini  (see plan9.ini(8)) on a partition named dos or 9fat. If one is found, 
searching stops and the file is read into memory at physical address 0x1200 
where it can be found later by any loaded bootfile. Some options in plan9.ini 
are used by 9load...

NB: These may be the wrong docs... Google dasn't *quite* read minds....)


Elsewhere, instructions and examples are publshed for *creating* a plan9.ini file.

My questions are:

- Does a CD-install of 'native' Plan9 to a combined fossil/venti HDD partition 
(eminently bootable, runnable) create a default 'plan9.ini'?

- If so, *where*, and how can I cat or Sam it from a running system?

- If not, how and where *does* Plan9 derive the functionally equivalent 
information it needs to complete the boot process?

(i.e. - given that it is operating the hardware as expected, what goes on 
if/as/when the above search *fails*?)

Objective of the exercise (at this point) is to confirm that plan9.ini or 
functional equivalent, effects the published default of disabling the second 
core in a Core-D P4, and, if so, attempt to re-enable same and see what, if 
anything, breaks.

A mere link to more appropriate docs may be all I need...

Perplexed (after ls'ing all over Hell and at least half of Texas..)

Bill Hacker

Re: [9fans] Dum-Bass question

[Uriel]

glendy# 9fat:
glendy# cd /n/9fat
glendy# lc
9LOAD      9pc        9pccpuf    9pccpuf.gz 9pcf       9pxeload   PLAN9.INI
glendy# lookman 9fat
man 2 authsrv # authsrv(2)
man 4 dossrv # dossrv(4)
man 8 9load # 9load(8)
man 8 booting # booting(8)
man 8 prep # prep(8)
man 8 update # update(8)

Best wishes

uriel

On 3/24/07, W B Hacker <wbh@conducive.org> wrote:
> Fans,
>
> Online docs (http://planb.lsub.org/magic/man2html/8/9load) describing Plan9's
> boot process, say, in part:
>
> ...in order to find configuration information, 9load searches all units on
> devices fd and sdCn, in that order, for a file called plan9\plan9.ini or
> plan9.ini  (see plan9.ini(8)) on a partition named dos or 9fat. If one is found,
> searching stops and the file is read into memory at physical address 0x1200
> where it can be found later by any loaded bootfile. Some options in plan9.ini
> are used by 9load...
>
> NB: These may be the wrong docs... Google dasn't *quite* read minds....)
>
>
> Elsewhere, instructions and examples are publshed for *creating* a plan9.ini file.
>
> My questions are:
>
> - Does a CD-install of 'native' Plan9 to a combined fossil/venti HDD partition
> (eminently bootable, runnable) create a default 'plan9.ini'?
>
> - If so, *where*, and how can I cat or Sam it from a running system?
>
> - If not, how and where *does* Plan9 derive the functionally equivalent
> information it needs to complete the boot process?
>
> (i.e. - given that it is operating the hardware as expected, what goes on
> if/as/when the above search *fails*?)
>
> Objective of the exercise (at this point) is to confirm that plan9.ini or
> functional equivalent, effects the published default of disabling the second
> core in a Core-D P4, and, if so, attempt to re-enable same and see what, if
> anything, breaks.
>
> A mere link to more appropriate docs may be all I need...
>
> Perplexed (after ls'ing all over Hell and at least half of Texas..)
>
> Bill Hacker
>
>
>
>
>

Re: [9fans] Dum-Bass question

[Devon H. O'Dell]

2007/3/24, W B Hacker <wbh@conducive.org>:
> Fans,
>
> Online docs (http://planb.lsub.org/magic/man2html/8/9load) describing Plan9's
> boot process, say, in part:
>
> ...in order to find configuration information, 9load searches all units on
> devices fd and sdCn, in that order, for a file called plan9\plan9.ini or
> plan9.ini  (see plan9.ini(8)) on a partition named dos or 9fat. If one is found,
> searching stops and the file is read into memory at physical address 0x1200
> where it can be found later by any loaded bootfile. Some options in plan9.ini
> are used by 9load...
>
> NB: These may be the wrong docs... Google dasn't *quite* read minds....)
>
>
> Elsewhere, instructions and examples are publshed for *creating* a plan9.ini file.
>
> My questions are:
>
> - Does a CD-install of 'native' Plan9 to a combined fossil/venti HDD partition
> (eminently bootable, runnable) create a default 'plan9.ini'?

Yes, it does. You need to run 9fat: at an rc prompt (including the
colon) and the 9fat partition will be available under /n/9fat

> - If so, *where*, and how can I cat or Sam it from a running system?

Note that since namespaces are local, not global, you won't be able to
modify plan9.ini from an acme / sam session external to the terminal
from which you ran 9fat:.

> - If not, how and where *does* Plan9 derive the functionally equivalent
> information it needs to complete the boot process?
>
> (i.e. - given that it is operating the hardware as expected, what goes on
> if/as/when the above search *fails*?)
>
> Objective of the exercise (at this point) is to confirm that plan9.ini or
> functional equivalent, effects the published default of disabling the second
> core in a Core-D P4, and, if so, attempt to re-enable same and see what, if
> anything, breaks.

Yeah, by default *nomp=yes is set. If you want to enable SMP, remove
that line, and it should work. Though Plan 9 does periodically have
issues with various APICs.

> A mere link to more appropriate docs may be all I need...
>
> Perplexed (after ls'ing all over Hell and at least half of Texas..)
>
> Bill Hacker

--dho

Re: [9fans] Dum-Bass question

[W B Hacker]

Uriel wrote:
> glendy# 9fat:
> glendy# cd /n/9fat
> glendy# lc
> 9LOAD      9pc        9pccpuf    9pccpuf.gz 9pcf       9pxeload   PLAN9.INI

*snip*

Magical!

I've got the rest in hand...

Many thanks!

Bill Hacker

Re: [9fans] Dum-Bass question

[W B Hacker]

Devon H. O'Dell wrote:
> 2007/3/24, W B Hacker <wbh@conducive.org>:

*snip*

> 
> Yes, it does. You need to run 9fat: at an rc prompt (including the
> colon) and the 9fat partition will be available under /n/9fat
> 
>> - If so, *where*, and how can I cat or Sam it from a running system?
> 
> Note that since namespaces are local, not global, you won't be able to
> modify plan9.ini from an acme / sam session external to the terminal
> from which you ran 9fat:.
>

ACK.

*snip*
> 
> Yeah, by default *nomp=yes is set. If you want to enable SMP, remove
> that line, and it should work. Though Plan 9 does periodically have
> issues with various APICs.
> 

Standby one....

Bill

Re: [9fans] Dum-Bass question

[Devon H. O'Dell]

2007/3/24, W B Hacker <wbh@conducive.org>:
> Devon H. O'Dell wrote:
> > 2007/3/24, W B Hacker <wbh@conducive.org>:
>
> *snip*
>
> >
> > Yes, it does. You need to run 9fat: at an rc prompt (including the
> > colon) and the 9fat partition will be available under /n/9fat
> >
> >> - If so, *where*, and how can I cat or Sam it from a running system?
> >
> > Note that since namespaces are local, not global, you won't be able to
> > modify plan9.ini from an acme / sam session external to the terminal
> > from which you ran 9fat:.
> >
>
> ACK.

Though I should note that you could certainly, in acme, middle click
on 9fat:, then right click select /n/9fat/plan9.ini and that would
obviously work too :)

> *snip*
> >
> > Yeah, by default *nomp=yes is set. If you want to enable SMP, remove
> > that line, and it should work. Though Plan 9 does periodically have
> > issues with various APICs.
> >
>
> Standby one....
>
> Bill

Look forward to hearing whether it works.... It didn't like the last
P3 SMP box I tried.

--dho

Re: [9fans] Dum-Bass question

[W B Hacker]

Devon H. O'Dell wrote:
> 2007/3/24, W B Hacker <wbh@conducive.org>:
>> Devon H. O'Dell wrote:
>> > 2007/3/24, W B Hacker <wbh@conducive.org>:
>>
>> *snip*

Done.

Up and running, reporting both CPU at beginning of boot mesages.

Asus P5VDC-MX, Inel Pentium D 915 Dual-Core, 2 GB DDR2 800.

More as I stress it.... but looks golden from C3-800 up.

*snip*

> 
> Though I should note that you could certainly, in acme, middle click
> on 9fat:, then right click select /n/9fat/plan9.ini and that would
> obviously work too :)
> 

'Obviously?' Thanks, but if it is all the same, I would rather have Irish 
Alzheimer's.  Easier to remember....

;-)

>> *snip*
>> >
>> > Yeah, by default *nomp=yes is set. If you want to enable SMP, remove
>> > that line, and it should work. Though Plan 9 does periodically have
>> > issues with various APICs.
>> >
>>
>> Standby one....
>>
>> Bill
> 
> Look forward to hearing whether it works.... It didn't like the last
> P3 SMP box I tried.
> 

Intel's 'HAL' idea of SMP in P3 days (or earlier - my first was an Asus with 
twin Pentium 90 MHz) was somewhat *bent*. Warp on one CPU was 40% faster than 
NT4 on two, and that held on P1 200 & PentiumPro 200 as well.

Bill

Re: [9fans] Dum-Bass question

[lucio]

> Intel's 'HAL' idea of SMP in P3 days (or earlier - my first was an Asus with 
> twin Pentium 90 MHz) was somewhat *bent*. Warp on one CPU was 40% faster than 
> NT4 on two, and that held on P1 200 & PentiumPro 200 as well.

How much can you tell me about this early equipment?  I have such a
box and would dearly like to address its quirks as it is the only MP
device I own.  Even Linux (and NetBSD, but that's easier to forgive)
barfs on it, but SCO Unix used to cope with it.

++L

Re: [9fans] Dum-Bass question

[Kris Maglione]

[-- Attachment #1: Type: text/plain, Size: 541 bytes --]

On Sat, Mar 24, 2007 at 05:31:09PM -0400, Devon H. O'Dell wrote:
>Though I should note that you could certainly, in acme, middle click
>on 9fat:, then right click select /n/9fat/plan9.ini and that would
>obviously work too :)

The fact that it obviously would work notwithstanding, it doesn't sork. 
You would need to run 'Local 9fat:' instead. Also, you could run 'plumb 
Local 9fat:' and have it available everywhere.

-- 
Kris Maglione

Any improbable event which would create maximum confusion
if it did occur, will occur.

[-- Attachment #2: Type: application/pgp-signature, Size: 194 bytes --]

Re: [9fans] Dum-Bass question

[erik quanstrom]

one way to use equipment that does not boot plan 9 is to run
a fileserver kernel on it.  if you already have a cpu/auth server,
this works out pretty well.  my home fileserver is a dual processor
pIII-based machine with a 440gx chipset and an 8169-based nic.
obviously the fs doesn't use the second processor, but it's not
cpu-bound anyway.

- erik

On Sun Mar 25 00:55:16 EDT 2007, lucio@proxima.alt.za wrote:
> > Intel's 'HAL' idea of SMP in P3 days (or earlier - my first was an Asus with 
> > twin Pentium 90 MHz) was somewhat *bent*. Warp on one CPU was 40% faster than 
> > NT4 on two, and that held on P1 200 & PentiumPro 200 as well.
> 
> How much can you tell me about this early equipment?  I have such a
> box and would dearly like to address its quirks as it is the only MP
> device I own.  Even Linux (and NetBSD, but that's easier to forgive)
> barfs on it, but SCO Unix used to cope with it.
> 
> ++L

Re: [9fans] Dum-Bass question

[lucio]

> one way to use equipment that does not boot plan 9 is to run
> a fileserver kernel on it.

Sadly, the machine in question has an on-board Adaptec SCSI adapter.
I think mowing lawns would be more rewarding.

++L

Re: [9fans] Dum-Bass question

[erik quanstrom]

assuming that this is an unsupported adaptec adapter, there
is no law requiring its use.  doesn't this machine have pata?

- erik

Re: [9fans] Dum-Bass question

[W B Hacker]

lucio@proxima.alt.za wrote:
>> Objective of the exercise (at this point) is to confirm that plan9.ini or 
>> functional equivalent, effects the published default of disabling the second 
>> core in a Core-D P4, and, if so, attempt to re-enable same and see what, if 
>> anything, breaks.
> 
> try plan9.ini(8) specially for the above, the flag you're looking for
> is "*nomp" (the "*" at the beginning stops it being passed to the user
> environment/user shell; I disapprove of all these *no...  flags, but
> that was the wisdom then).
> 
> ++L
> 

Thanks, I had all that well in hand ahead of time.

I just could not find the 'plan9.ini' involved, as it didn't readily show up 
directly in listings.

Issuing the simple '9fat:' with colon was all it took.

That must be documented somewhere - but I had missed it.

Bill