headers
Valeriy E. Ushakov | 4 Sep 2005 03:32
Picon

Please, test an hpcfb fix

I've fixed (in current) an issue with hpcfb not passing r/g/b widths
and positions to rasops correctly for 16-bit framebuffers (arguably,
rasops15 should use sane defaults).  If you have a machine with 16-bit
framebuffer, please, test that console colors are working correctly
for you.

The fix is in sys/dev/hpc/hpcfb.c rev 1.33

If you have options WSDISPLAY_CUSTOM_OUTPUT you can simply try

    $ wsconsctl -dw msg.default.fg=red	# etc

Please, send replies directly to me.  I will request pull up to 3.0,
and we'd better fix any fallout before the release.

On hpcsh it fixes the issue with very dark console colors (another
bug, incorret endianness setting, contributed to that effect).

Thanks.

SY, Uwe
--

-- 
uwe <at> ptc.spbu.ru                         |       Zu Grunde kommen
http://www.ptc.spbu.ru/~uwe/            |       Ist zu Grunde gehen
Permalink | Reply | 0 comments |
headers
Andy Ruhl | 5 Sep 2005 15:52
Picon

Re: Please, test an hpcfb fix

On 9/3/05, Valeriy E. Ushakov <uwe <at> ptc.spbu.ru> wrote:
> I've fixed (in current) an issue with hpcfb not passing r/g/b widths
> and positions to rasops correctly for 16-bit framebuffers (arguably,
> rasops15 should use sane defaults).  If you have a machine with 16-bit
> framebuffer, please, test that console colors are working correctly
> for you.
> 
> The fix is in sys/dev/hpc/hpcfb.c rev 1.33
> 
> If you have options WSDISPLAY_CUSTOM_OUTPUT you can simply try
> 
>     $ wsconsctl -dw msg.default.fg=red  # etc
> 
> Please, send replies directly to me.  I will request pull up to 3.0,
> and we'd better fix any fallout before the release.
> 
> On hpcsh it fixes the issue with very dark console colors (another
> bug, incorret endianness setting, contributed to that effect).
> 
> Thanks.

Sorry to reply publicly, but the current kernel still doesn't boot on
my hpcmips based NEC Mobilepro 880 due to pr 30590. I don't know how
to fix it other than the workaround provided in the pr submitted by
others, which I'm not sure is something that is a viable workaround to
be doing. If I did know how to fix it I would try.

Andy
(Continue reading)

Permalink | Reply | 0 comments |
headers
Valeriy E. Ushakov | 5 Sep 2005 15:58
Picon

Re: Please, test an hpcfb fix

On Mon, Sep 05, 2005 at 06:52:29 -0700, Andy Ruhl wrote:

> Sorry to reply publicly, but the current kernel still doesn't boot on
> my hpcmips based NEC Mobilepro 880 due to pr 30590. I don't know how
> to fix it other than the workaround provided in the pr submitted by
> others, which I'm not sure is something that is a viable workaround to
> be doing. If I did know how to fix it I would try.

Have you tried asking Chuck (chs <at> n.o), the auhtor of the offending
1.16 of sys/arch/mips/mips/mipsX_subr.S?

SY, Uwe
--

-- 
uwe <at> ptc.spbu.ru                         |       Zu Grunde kommen
http://www.ptc.spbu.ru/~uwe/            |       Ist zu Grunde gehen
Permalink | Reply | 0 comments |
headers
Florian Cramer | 7 Sep 2005 12:17
Picon
Picon
Favicon

Installation troubles with NetBSD 2.0.2 on z50

I previously ran NetBSD 1.6.1 from a 512 MB CF card and decided to
upgrade to NetBSD 2.0.2 on a 1 GB card with a fresh install. However,
the installation process hangs when the install kernel formats the root
file system, i.e. at the command

/sbin/newfs -O 1 -b 8192 -f 1024 /dev/rwd0a

At this point, the status indicator remains at "in progress" forever (I
tried to wait up to 24 hours). The error is fully reproducible.
Rebooting and trying again, with the same or a different 1GB CF card
always yields the same result.

I am desperate at this point and would be greatly thankful for
suggestions.

Florian

--

-- 
http://cramer.netzliteratur.net
Permalink | Reply | 5 comments |
headers
Florian Cramer | 7 Sep 2005 14:05
Picon
Picon
Favicon

Re: Installation troubles with NetBSD 2.0.2 on z50

Am Mittwoch, 07. September 2005 um 12:17:12 Uhr (+0200) schrieb Florian Cramer:

> /sbin/newfs -O 1 -b 8192 -f 1024 /dev/rwd0a

OK, it seems that I tracked down the source of the problem: Apparently,
the geometry values which the installer calculates aren't correct. If I
run newfs without parameters, it works fine. Unfortunately, there is no
way to bypass newfs in the installation routine. Does anyone have a
suggestion for the correct disk geometry data, or how to retrieve them?

-F

--

-- 
http://cramer.netzliteratur.net
Permalink | Reply | 4 comments |
headers
Andy Ruhl | 7 Sep 2005 16:38
Picon

Re: Installation troubles with NetBSD 2.0.2 on z50

On 9/7/05, Florian Cramer <cantsin <at> zedat.fu-berlin.de> wrote:
> Am Mittwoch, 07. September 2005 um 12:17:12 Uhr (+0200) schrieb Florian Cramer:
> 
> > /sbin/newfs -O 1 -b 8192 -f 1024 /dev/rwd0a
> 
> OK, it seems that I tracked down the source of the problem: Apparently,
> the geometry values which the installer calculates aren't correct. If I
> run newfs without parameters, it works fine. Unfortunately, there is no
> way to bypass newfs in the installation routine. Does anyone have a
> suggestion for the correct disk geometry data, or how to retrieve them?
> 
> -F

Can you post what type of disk it is, and what NetBSD thinks the geometry is?

If possible, can you attach the disk to another (hopefully not broken)
NetBSD system and compare results? Does it newfs on another system?

I have very little experience in this area, but looking at what the OS
thinks the disk is might jog something.

Andy
Permalink | Reply | 3 comments |
headers
Florian Cramer | 7 Sep 2005 18:07
Picon
Picon
Favicon

Re: Installation troubles with NetBSD 2.0.2 on z50

Am Mittwoch, 07. September 2005 um 07:38:06 Uhr (-0700) schrieb Andy Ruhl:

> Can you post what type of disk it is, and what NetBSD thinks the geometry is?

I tried two 1024 MB CF flash cards, one by Sandisk, the other a no-name
brand with a Toshiba chip inside. Since I got identical problems with
both of them, I sticked with the SanDisk.

> If possible, can you attach the disk to another (hopefully not broken)
> NetBSD system and compare results? Does it newfs on another system?

It does newfs on the other system, yes, without problems,
On that NetBSD system - actually a Thinkpad booted with a NetBSD
1.6 boot CD - I get the folling fdisk information for the card:

NetBSD disklabel disk geometry:
cylinders: 678400 heads: 4 sectors/track: 0 (0 sectors/cylinder)

BIOS disk geometry:
cylinders: 497 heads: 64 sectors/track: 64 (4032 sectors/cylinder)

disklabel /dev/sd0 says:
bytes/sector: 512
sectors/track: 63
tracks/cylinder: 1008
total sectors: 10000944

During the (unsuccessful) newfs run in the install routine, it says:
/dev/rwd0a: 915.5 MB (1874912 sectors) block size 8192, fragment size
1024 using 21 cylinder groups of 43.50 MB, 5581 blks, 10816 inodes
(Continue reading)

Permalink | Reply | 2 comments |
headers
Andy Ruhl | 7 Sep 2005 18:18
Picon

Re: Installation troubles with NetBSD 2.0.2 on z50

On 9/7/05, Florian Cramer <cantsin <at> zedat.fu-berlin.de> wrote:
> Am Mittwoch, 07. September 2005 um 07:38:06 Uhr (-0700) schrieb Andy Ruhl:
> 
> > Can you post what type of disk it is, and what NetBSD thinks the geometry is?
> 
> I tried two 1024 MB CF flash cards, one by Sandisk, the other a no-name
> brand with a Toshiba chip inside. Since I got identical problems with
> both of them, I sticked with the SanDisk.
> 
> > If possible, can you attach the disk to another (hopefully not broken)
> > NetBSD system and compare results? Does it newfs on another system?
> 
> It does newfs on the other system, yes, without problems,
> On that NetBSD system - actually a Thinkpad booted with a NetBSD
> 1.6 boot CD - I get the folling fdisk information for the card:
> 
> NetBSD disklabel disk geometry:
> cylinders: 678400 heads: 4 sectors/track: 0 (0 sectors/cylinder)
> 
> BIOS disk geometry:
> cylinders: 497 heads: 64 sectors/track: 64 (4032 sectors/cylinder)
> 
> disklabel /dev/sd0 says:
> bytes/sector: 512
> sectors/track: 63
> tracks/cylinder: 1008
> total sectors: 10000944
> 
> During the (unsuccessful) newfs run in the install routine, it says:
> /dev/rwd0a: 915.5 MB (1874912 sectors) block size 8192, fragment size
(Continue reading)

Permalink | Reply | 1 comment |
headers
Florian Cramer | 7 Sep 2005 18:32
Picon
Picon
Favicon

Re: Installation troubles with NetBSD 2.0.2 on z50

Am Mittwoch, 07. September 2005 um 09:18:34 Uhr (-0700) schrieb Andy Ruhl:

> If I'm reading this correct, those sector counts are WAY off, if both
> are total sector counts. The one from the thinkpad seems even more
> wrong? Strange. 

Well, the one from the Thinkpad is a Disklabel sector count, but the
disklabel was generated during installation in the z50.

> Are you plugging the card directly into the z50, and
> using some kind of card reader on the thinkpad?

Yes. 

-F 

--

-- 
http://cramer.netzliteratur.net
Permalink | Reply | 0 comments |
headers
Valeriy E. Ushakov | 12 Sep 2005 14:21
Picon

Re: Please, test an hpcfb fix


Did anyone have a chance to test it?

On Sun, Sep 04, 2005 at 05:32:51 +0400, Valeriy E. Ushakov wrote:

> Date: Sun, 4 Sep 2005 05:32:51 +0400
> From: "Valeriy E. Ushakov" <uwe <at> ptc.spbu.ru>
> To: port-hpcarm <at> netbsd.org, port-hpcmips <at> netbsd.org
> Subject: Please, test an hpcfb fix
> 
> I've fixed (in current) an issue with hpcfb not passing r/g/b widths
> and positions to rasops correctly for 16-bit framebuffers (arguably,
> rasops15 should use sane defaults).  If you have a machine with 16-bit
> framebuffer, please, test that console colors are working correctly
> for you.
> 
> The fix is in sys/dev/hpc/hpcfb.c rev 1.33
> 
> If you have options WSDISPLAY_CUSTOM_OUTPUT you can simply try
> 
>     $ wsconsctl -dw msg.default.fg=red	# etc
> 
> Please, send replies directly to me.  I will request pull up to 3.0,
> and we'd better fix any fallout before the release.
> 
> On hpcsh it fixes the issue with very dark console colors (another
> bug, incorret endianness setting, contributed to that effect).
> 
> Thanks.
(Continue reading)

Permalink | Reply | 0 comments |

Gmane