"wd0e" - is it default?

I made a new kernel, and the system - when booting - still's
looking for
root partition in "wd0e", while my is
"wd0a".

Is this "wd0e" kind of "default setting"
in "generic" kernels? How could I
change it for steady (without hard-coding into kernel
config)?
-- 
				pozdrawiam / regards

						Zbigniew Baniewski

you have to update your bootstrap code to boot from wd0a.
the old code may be still looking for wd0e
try:
installboot -v -m i386 -o timeout=5,console=pc /dev/rwd0a
/usr/mdec/bootxx_ffsv1
(replace where appropriate)

regards, Zafer.




2007/5/11, Zbigniew Baniewski <zbispid.com.pl>:
> I made a new kernel, and the system - when booting -
still's looking for
> root partition in "wd0e", while my is
"wd0a".
>
> Is this "wd0e" kind of "default
setting" in "generic" kernels? How could I
> change it for steady (without hard-coding into kernel
config)?
> --
>                                pozdrawiam / regards
>
>                                                Zbigniew
Baniewski
>

On Fri, May 11, 2007 at 11:06:25PM +0200, Zafer Aydogan
wrote:

> you have to update your bootstrap code to boot from
wd0a.
> the old code may be still looking for wd0e
> try:
> installboot -v -m i386 -o timeout=5,console=pc
/dev/rwd0a 
> /usr/mdec/bootxx_ffsv1
> (replace where appropriate)

Believe it, or not - I'm still getting "booting
hd0e:netbsd" at boot-time.

I tried similar invocation earlier, just without "-m
i386" and "console=pc".

No idea, how can I suppress wd0a instead of wd0e. Perhaps
there was a bug in
"installboot" in 2.0?
-- 
				pozdrawiam / regards

						Zbigniew Baniewski

On Fri, May 11, 2007 at 11:35:17PM +0200, Zbigniew Baniewski
wrote:
> On Fri, May 11, 2007 at 11:06:25PM +0200, Zafer Aydogan
wrote:
> 
> > you have to update your bootstrap code to boot
from wd0a.
> > the old code may be still looking for wd0e
> > try:
> > installboot -v -m i386 -o timeout=5,console=pc
/dev/rwd0a 
> > /usr/mdec/bootxx_ffsv1
> > (replace where appropriate)
> 
> Believe it, or not - I'm still getting "booting
hd0e:netbsd" at boot-time.
> 
> I tried similar invocation earlier, just without
"-m i386" and "console=pc".
> 
> No idea, how can I suppress wd0a instead of wd0e.
Perhaps there was a bug in
> "installboot" in 2.0?

I'm fairly certain you got confused at some point.

Please post full output of fdisk and disklabel for that disk
and we'll
tell you what you did wrong.

Nothing to be ashamed of:  this stuff is much more
complicated than it
really needs to be and we all got stuck in a issue of that
kind at some
point.  Well, at least I know I have.

-- 
Quentin Garnier - cubecubidou.net - cubeNetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2,
2005.

On Fri, May 11, 2007 at 11:49:38PM +0200, Quentin Garnier
wrote:

> Please post full output of fdisk and disklabel for that
disk and we'll
> tell you what you did wrong.

Just take a look:


debian# fdisk /dev/wd0
Disk: /dev/wd0d
NetBSD disklabel disk geometry:
cylinders: 12592, heads: 16, sectors/track: 63 (1008
sectors/cylinder)
total sectors: 12692736

BIOS disk geometry:
cylinders: 790, heads: 255, sectors/track: 63 (16065
sectors/cylinder)
total sectors: 12692736

Partition table:
0: NetBSD (sysid 169)
    start 63, size 12692673 (6198 MB, Cyls 0-790/22/1),
Active
1: <UNUSED>
2: <UNUSED>
3: <UNUSED>


debian# disklabel /dev/wd0
# /dev/wd0d:
type: ESDI
disk: IBM-DHEA-36480  
label: fictitious
flags:
bytes/sector: 512
sectors/track: 63
tracks/cylinder: 16
sectors/cylinder: 1008
cylinders: 12592
total sectors: 12692736
rpm: 3600
interleave: 1
trackskew: 0
cylinderskew: 0
headswitch: 0           # microseconds
track-to-track seek: 0  # microseconds
drivedata: 0 

5 partitions:
#        size    offset     fstype [fsize bsize cpg/sgs]
 a:  12288000         0     4.2BSD   2048 16384 27560  #
(Cyl.      0 -  12190*)
 b:    404736  12288000       swap                     #
(Cyl.  12190*-  12591)
 c:  12692673        63     4.2BSD   2048 16384 27800  #
(Cyl.      0*-  12591)

-- 
				pozdrawiam / regards

						Zbigniew Baniewski

On Fri, May 11, 2007 at 06:29:00PM -0400, Thor Lancelot
Simon wrote:

> What's going on is that somehow you got an unused wd0e
partition in your
> disklabel whose first sector matches that of wd0a
(perhaps by using
> mbrlabel -- it defaults to partition 'e' for the first
MBR partition).
> Now that that's there, the bootblock can't tell which
one you actually
> want to boot from.

Ah, so... I was supposing, it can - it some way, which I
don't know -
distinguish between all the partitions, whether they are
empty, or not.

> The solution is to edit the disklabel and change the
"5 partitions" line
> to "4 partitions" (4, not 3, since you don't
want to lose partition 'd'
> which should be unused and which is the whole disk). 
Then all will work
> as you expect.

Yes, solved at last. Thanks a lot!
-- 
				pozdrawiam / regards

						Zbigniew Baniewski

On Sat, May 12, 2007 at 12:03:24AM +0200, Zbigniew Baniewski
wrote:
[...]
> 5 partitions:
> #        size    offset     fstype [fsize bsize
cpg/sgs]
>  a:  12288000         0     4.2BSD   2048 16384 27560 
# (Cyl.      0 -  12190*)
>  b:    404736  12288000       swap                    
# (Cyl.  12190*-  12591)
>  c:  12692673        63     4.2BSD   2048 16384 27800 
# (Cyl.      0*-  12591)

It's a tad dangerous to make a starts at 0, for several
reasons.

First, while I think our pbr code does a good job at passing
for a mbr,
it's still risky.  Then, such a setup is likely to confuse
the kernel
when looking for a root partition, because it starts outside
the NetBSD
slice.

-- 
Quentin Garnier - cubecubidou.net - cubeNetBSD.org
"You could have made it, spitting out benchmarks
Owe it to yourself not to fail"
Amplifico, Spitting Out Benchmarks, Hometakes Vol. 2,
2005.

On Fri, 11 May 2007 23:49:38 +0200
Quentin Garnier <cubecubidou.net> wrote:

> Nothing to be ashamed of:  this stuff is much more
complicated than it
> really needs to be and we all got stuck in a issue of
that kind at
> some point.  Well, at least I know I have.
> 
And I.  I regard the disk geometry/partitioning/booting
stuff as second
only to X11 configuration for making systems hard to
install.  They've
all gotten easier over the years, but there's still plenty
of room for
trouble.


		--Steve Bellovin, http://www.cs.columbi
a.edu/~smb

On Fri, 11 May 2007 18:29:00 -0400
Thor Lancelot Simon <tlsrek.tjls.com> wrote:

> > 5 partitions:
>  ^^^
>  ^^^ Notice how you have only three partitions listed
below?  But
> there are actually two more, of type "unused"
-- so disklabel does not
>      print them out for you by default.

Shouldn't the disklabel default option be changed to show
_all_
partitions? What are the pros and the cons of the present
status?

Regards,
--
Przemysław Pawełczyk <pp_o2o2.pl>
P2O2's WWW: http://pp.kv.net.pl (The
LearNBSD Project)
P2O2's WWW Forum: http://www.p2o2.fora.pl/

On Sat, May 12, 2007 at 01:37:33AM +0200, Quentin Garnier
wrote:

> First, while I think our pbr code does a good job at
passing for a mbr,
> it's still risky.  Then, such a setup is likely to
confuse the kernel
> when looking for a root partition, because it starts
outside the NetBSD
> slice.

But I understand, that when boot-process is successfull,
then I can make
assumption, that kernel wasn't been confused?
-- 
				pozdrawiam / regards

						Zbigniew Baniewski