headers
Valeriy E. Ushakov | 14 Apr 2003 05:17
Picon

current kernel unbootable + fix

I tried to upgrade my kernel from 1.6L to current and found that the
kernel will panic very early on startup.  After some staring at the
ddb I think that the problem is that link_set_* sections are linked
incorrectly.

When kernel is linked, ld issues warnings that there are no suitable
memory region for link_set_* sections and put them at 0x0.  This
happens because sys/arch/hpcsh/conf/shl-elf.x defines memory region
"ram" and explicitly places output .text and .data sections into that
region.

But link_set_* sections are not mentioned in the script and when ld
copies those sections from its input to output it doesn't consider
"ram" region as suitable for those sections.

I fixed the problem with adding (rwx) attributes to the "ram" section.
The resulting kernel boots fine.  I'm not sure if it's a proper fixm
though.  Can someone with more ld clue look into this?  Jason?

SY, Uwe
--

-- 
uwe <at> ptc.spbu.ru                         |       Zu Grunde kommen
http://www.ptc.spbu.ru/~uwe/            |       Ist zu Grunde gehen
Permalink | Reply | 2 comments |
headers
Jason Thorpe | 15 Apr 2003 18:56

Re: current kernel unbootable + fix


On Sunday, April 13, 2003, at 08:17  PM, Valeriy E. Ushakov wrote:

> When kernel is linked, ld issues warnings that there are no suitable
> memory region for link_set_* sections and put them at 0x0.  This
> happens because sys/arch/hpcsh/conf/shl-elf.x defines memory region
> "ram" and explicitly places output .text and .data sections into that
> region.

Yah.  But link_set_* should be sorted with .text or .rodata by the 
linker... hm...

> But link_set_* sections are not mentioned in the script and when ld
> copies those sections from its input to output it doesn't consider
> "ram" region as suitable for those sections.

Right, you can't explicitly mention link_set_* in the linker script, 
because if you do, the start-of-section/end-of-section special marker 
symbols go away, which breaks actually *using* the link sets :-)

> I fixed the problem with adding (rwx) attributes to the "ram" section.
> The resulting kernel boots fine.  I'm not sure if it's a proper fixm
> though.  Can someone with more ld clue look into this?  Jason?

Hm.  That solution will work.  Look at 
sys/arch/evbarm/conf/ldscript.evbarm for another way that the problem 
can be solved.

Sadly, the linker script syntax doesn't deal well with these 
"anonymous" sections that link sets use.  When I attempted to deal with
(Continue reading)

Permalink | Reply | 1 comment |
headers
Valeriy E. Ushakov | 15 Apr 2003 19:31
Picon

Re: current kernel unbootable + fix

On Tue, Apr 15, 2003 at 09:56:02 -0700, Jason Thorpe wrote:

> On Sunday, April 13, 2003, at 08:17  PM, Valeriy E. Ushakov wrote:
> 
> > When kernel is linked, ld issues warnings that there are no suitable
> > memory region for link_set_* sections and put them at 0x0.  This
> > happens because sys/arch/hpcsh/conf/shl-elf.x defines memory region
> > "ram" and explicitly places output .text and .data sections into that
> > region.
[...]
> > I fixed the problem with adding (rwx) attributes to the "ram" section.
> > The resulting kernel boots fine.  I'm not sure if it's a proper fixm
> > though.  Can someone with more ld clue look into this?  Jason?
> 
> Hm.  That solution will work.  Look at 
> sys/arch/evbarm/conf/ldscript.evbarm for another way that the problem 
> can be solved.
> 
> Sadly, the linker script syntax doesn't deal well with these 
> "anonymous" sections that link sets use.  When I attempted to deal with 
> this in ldd itself, my changes to make the linker scripts more friendly 
> to these sections were soundly rejected.  So, I think we're stuck with 
> annoying work-arounds such as yours and mine.

So I committed a sligthly modified fix.  I think dreamcast needs this
fix too.   Not sure about coff scripts (does link_set_* work with coff?).

Thanks.

Module Name:    src
(Continue reading)

Permalink | Reply | 0 comments |
headers
Valeriy E. Ushakov | 23 Apr 2003 06:39
Picon

Jornada keyboard layouts

I'm interested what keyboard layouts exist for jornadas.  You can find
layout code as part of the product id at the bottom of the unit.
E.g. mine is "F1262A ABB", where "ABB" is HP's code for "European
English" layout.

Layouts we have data for are:

ABA - US English
ABB - EU English
ABD - German
ABJ - Japanese

If you have a jornada with a keyboard layout not listed above, can you
please send me the model code and the description of the layout (a
free form description or/and a photo).

I plan to commit hpcsh keymaps for US and German layouts (they have
some important chars not on the primary layer) shortly.

I'm especially interested in the Hungarian layout.  If it has
different layout code but its primary layer is the same as the ABB
<http://snark.ptc.spbu.ru/~uwe/netbsd/jornada/photo/jornada680-intl.jpg>
I'm going to recycle the current "Hungarian" entry for European.  The
intention is to handle layout differences that are only in the AltGr
layer via wsconsctl(8).

Thanks.

SY, Uwe
--

--
(Continue reading)

Permalink | Reply | 4 comments |
headers
Robert Solomon | 25 Apr 2003 04:45

Re: Jornada keyboard layouts

Mines says F1262A. It is a 32MB US market? 680.
I'll scan the keyboard at work and send off the scan as soon as
I get a chance.

Bob

On Wed, 23 Apr 2003, Valeriy E. Ushakov wrote:

> I'm interested what keyboard layouts exist for jornadas.  You can find
> layout code as part of the product id at the bottom of the unit.
> E.g. mine is "F1262A ABB", where "ABB" is HP's code for "European
> English" layout.
>
> Layouts we have data for are:
>
> ABA - US English
> ABB - EU English
> ABD - German
> ABJ - Japanese
>
> If you have a jornada with a keyboard layout not listed above, can you
> please send me the model code and the description of the layout (a
> free form description or/and a photo).
>
> I plan to commit hpcsh keymaps for US and German layouts (they have
> some important chars not on the primary layer) shortly.
>
> I'm especially interested in the Hungarian layout.  If it has
> different layout code but its primary layer is the same as the ABB
> <http://snark.ptc.spbu.ru/~uwe/netbsd/jornada/photo/jornada680-intl.jpg>
(Continue reading)

Permalink | Reply | 2 comments |
headers
Valeriy E. Ushakov | 25 Apr 2003 04:58
Picon

Re: Jornada keyboard layouts

On Thu, Apr 24, 2003 at 22:45:44 -0400, Robert Solomon wrote:

> Mines says F1262A. It is a 32MB US market? 680.

No 3-letter code?  (like ABA, ABU, ABB, ABF, ABD ...)

You can also check the code by doing a soft reset like this:

. press 's' or 'd' key and do a soft reset
. "System Diagnostic" window will come up after the reset
. press the "Build" button
. a "Time Stamp" dialog will come up with ROM info, it will say, e.g.

    Country: USA  ROM Option: ABA

It's that ROM option string that I'm interested in.

If it's ABA and your keyboard loooks like this:

  http://snark.ptc.spbu.ru/~uwe/netbsd/jornada/photo/jornada680-us2.jpg

then I already have the info ;).

But thanks for the offer anyway!

SY, Uwe
--

-- 
uwe <at> ptc.spbu.ru                         |       Zu Grunde kommen
http://www.ptc.spbu.ru/~uwe/            |       Ist zu Grunde gehen
(Continue reading)

Permalink | Reply | 1 comment |
headers
Robert Solomon | 25 Apr 2003 05:29

Re: Jornada keyboard layouts

The number ends in only A, but "Time Stamp" reports Country: USA
ROM Option: ABA.

On Fri, 25 Apr 2003, Valeriy E. Ushakov wrote:

> On Thu, Apr 24, 2003 at 22:45:44 -0400, Robert Solomon wrote:
>
> > Mines says F1262A. It is a 32MB US market? 680.
>
> No 3-letter code?  (like ABA, ABU, ABB, ABF, ABD ...)
>
> You can also check the code by doing a soft reset like this:
>
> . press 's' or 'd' key and do a soft reset
> . "System Diagnostic" window will come up after the reset
> . press the "Build" button
> . a "Time Stamp" dialog will come up with ROM info, it will say, e.g.
>
>     Country: USA  ROM Option: ABA
>
> It's that ROM option string that I'm interested in.
>
> If it's ABA and your keyboard loooks like this:
>
>   http://snark.ptc.spbu.ru/~uwe/netbsd/jornada/photo/jornada680-us2.jpg
>
> then I already have the info ;).
>
> But thanks for the offer anyway!
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Valeriy E. Ushakov | 27 Apr 2003 23:48
Picon

Re: Jornada keyboard layouts

On Wed, Apr 23, 2003 at 08:39:12 +0400, Valeriy E. Ushakov wrote:

> Layouts we have data for are:
> 
> ABA - US English
> ABB - EU English
> ABD - German
> ABJ - Japanese
> 
> If you have a jornada with a keyboard layout not listed above, can you
> please send me the model code and the description of the layout (a
> free form description or/and a photo).
> 
> I plan to commit hpcsh keymaps for US and German layouts (they have
> some important chars not on the primary layer) shortly.

I have committed US and German keymaps, so these keyboards should
"just work" now, including very early "break into ddb before
autoconfiguration" case.

I also have data/pictures for french keyboard but an option for french
should be added to hpcboot first.

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
Valeriy E. Ushakov | 27 Apr 2003 23:54
Picon

hd64461 video ac^Wdeceleration

I wonder if hd64461 video chip is really that bad at blits?  I tried
compiling with HD64461VIDEO_HWACCEL and scrolling performace
deteriorated significantly.  An unscientific test of

    $ time jot 1000

went from 7 (rasops) seconds to 17 (with hwaccel)!

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
Christian Claveleira | 28 Apr 2003 23:46

Jornada 680 screen and power management

Any hope to have the touch screen supported ?
And a suspend mode ?

--

-- 

Christian
Permalink | Reply | 1 comment |

Gmane