Re: Linux for Wondermedia wm8850

[Alexey Charkov]

Sorry, wrong address for the group.

On 17 Apr 2013 09:01, "Alexey Charkov" <alc...@gmail.com> wrote:

On 16 Apr 2013 20:37, "Wolfgang Tintemann auf Windows 7 mit Windows Mail" <wktin...@online.de> wrote:
>
> Hello Alexey Charkov,
>  
> sorry that I dare to molest you.
>  
> I read in the net about you dealing with Wondermedia devices
> ( tablets, wm8650 etc. ).
>  
> I own a wm8850 NETBOOK and started to try to cross-compile a
> 3.0.8 kernel on a x86 machine under Linux Mint for it.

Hi Wolfgang,

From what I read, it seems that you are trying to compile the sources that WonderMedia has just released recently. Unfortunately, I'm not sure if those are in a decent enough condition to actually build a kernel, as you might have been the first one to try:-)

> 1. do you know about the progress in porting a Linux kernels
>     to wm8850 by cross-compiling ?
>     I am only interested in starting it from the SD card - no flashing.

Well, cross-compiling actually has nothing to do with porting. Porting is about creating necessary code for hardware support, while cross-compiling is about building the kernel image on a machine with different CPU architecture (like your x86 box).

From what I know, latest 3.9-rc kernels from Linus should work just fine on wm8850, thanks to the work that Tony Prisk puts into this port. There are also building instructions on Tony's wiki. Some features will still be missing, like sound and internal NAND flash, for example. But you'll be able to run normal Linux distros and browse the Internet at least.

> 2. if you have the time please have a look at this error messag(es)
>     and see if you can give me a helpful hint :
>  
>   CC      arch/arm/vfp/vfpmodule.o
> make[1]: *** No rule to make target `arch/arm/vfp/entry.o', needed by `arch/arm/vfp/vfp.o'.  Stop.
>  
> or
>  
>   CC      arch/arm/kernel/elf.o
> make[2]: *** No rule to make target `arch/arm/kernel/entry-armv.o', needed by `arch/arm/kernel/built-in.o'.  Stop.

Either the sources are broken, or there is something wrong with the way you build them. Note that typically during cross-compilation you'd also need to set the CROSS_COMPILE variable, either in kernel config or on make command line.

>     Does that mean that the source of the kernel is in bad shape
>     or do I make errors in the make ARCH=arm xconfig dialog ?

You can also try 'make mrproper' on the kernel tree before building, to ensure that there are no leftovers from other builds.

Note that I've also Cc'd our mailing list on this reply. Usually it makes sense to post such general questions there, so that you have more than one potential respondent. It also lets future googlers find the reply there :-)

Best,
Alexey

>  
> MfG:  W. Tintemann, BRDeutschland
> ------------------------------------------------------------------------------------------------ 
> In hundert Jahren
>  
> Was wird in hundert Jahren Sein?
> Was wird mit uns geschehen?
> Man weiß es nicht, denn niemand hier
> kann in die Zukunft sehen.
> Doch etwas kann man jetzt schon tun:
> Man kann die Zukunft denken.
> Und tut man dies, dann kann man sie
> vielleicht zum Guten lenken.
>  
> Paul Maar
>  

[Tony Prisk]

 Did you do 'make ARCH=arm uImage'?

The wiki says you should use 'make ARCH=arm zImage', so it shouldn't need a LOADADDR. The reason you need to use zImage is so you can append the compiled DTB to the end of the zImage, and then use mkimage to create the uzImage.

5) Build the image
cat arch/arm/boot/zImage arch/arm/boot/wm8850-w70v2.dtb > arch/arm/boot/zImage_w_dtb

• replace wm8850-w70v2.dtb with the appropriate file from step 4

mkimage -A arm -O linux -T kernel -C none -a 0x8000 -e 0x8000 -n "My Linux" -d arch/arm/boot/zImage_w_dtb ~/uzImage.bin

This command line specifies the LOADADDR and ENTRY points.


> Also I dont know whether I have to generate/install
> modules any more : this step isnt mentioned in the
> Build Wiki article (?).

The wiki only covers the VIA/Wondermedia specifics, not how to config all the kernel options, or any of the 'basic' kernel building stuff.
If you selected to have modules in your kernel config, you still need to build the modules and put them on the root filesystem.

Regards
Tony P

[Tony Prisk]

On Thursday, April 18, 2013 9:47:06 PM UTC+12, wktin...@gmail.com wrote:

OK, I did that : have now an uzImage.bin

What concerns the modules I have to look into the config dialog.

Do I understand right that the -a 0x8000 and -e 0x8000 in the
mkimage have an effect on how I have to write my wmt_scriptcmd.txt
file ?

Until now there was a bootm 0 command at the end. Has that to  be
modified to bootm 8000 or bootm 0x8000 or the decimal value ?
I also remember a fatload command which might be modified somehow ?

Example scriptcmd file:

fatload mmc 0 0 /script/uzImage.bin
setenv bootargs earlyprintk console=ttyWMT0,115200n8 mem=512M root=/dev/sda1 rootdelay=5
bootm 0


The fatload path will need to be changed to whereever you put the uzImage.bin file on your SDCARD, otherwise this should be all you need.
If you want to use the LCD panel (tablet/notebook) rather than UART0, then remove the 'console=ttyWMT0,115200n8' portion.

Regards
Tony P

[Peter Vasil]

On Fri, Apr 19, 2013 at 12:34 PM, wkt <wktin...@gmail.com> wrote:
> make ARCH=arm INSTALL_MOD_PATH=/home/wkt8850nb modules_install
>
> cp: cannot stat `/home/kernel-3.9-rc7/modules.order': No such file or
> directory
> make: *** [_modinst_] Error 1>
>
> After a new kernel cross-compilation I tried the above command.
>
> Did I miss something my xconfig dialog ?
> As I saw there are many M option settings : so I did expect modules in the
> INSTALL_MOD_PATH - but there is only an empty lib/modules/3.9.0-rc7 dir
>
> --
> You received this message because you are subscribed to the Google Groups
> "VT8500/WM8505 Linux Kernel" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to vt8500-wm8505-linux...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

Looks like you didn't actually build the modules - this usually
happens when you only call "make uImage" or "make zImage". You need
something like "make modules" before you can do modules_install...
Regards,
Peter

[Tony Prisk]

On Saturday, April 20, 2013 7:57:48 AM UTC+12, wktin...@gmail.com wrote:

Thank you, forgot that...

I now tried a cross-compiling make ARCH=arm ... modules command and get
these messages at the end of stage 1

  HOSTLD  scripts/mod/modpost
  Building modules, stage 2.
  MODPOST 1804 modules
ERROR: "__udivdi3" [sound/soc/codecs/snd-soc-wm8974.ko] undefined!
ERROR: "__umoddi3" [sound/soc/codecs/snd-soc-wm8974.ko] undefined!
ERROR: "__udivdi3" [sound/soc/codecs/snd-soc-wm8940.ko] undefined!
ERROR: "__umoddi3" [sound/soc/codecs/snd-soc-wm8940.ko] undefined!
ERROR: "__udivdi3" [sound/soc/codecs/snd-soc-wm8753.ko] undefined!
ERROR: "__umoddi3" [sound/soc/codecs/snd-soc-wm8753.ko] undefined!
ERROR: "__udivdi3" [sound/soc/codecs/snd-soc-wm8510.ko] undefined!
ERROR: "__umoddi3" [sound/soc/codecs/snd-soc-wm8510.ko] undefined!
ERROR: "__sync_fetch_and_and_4" [drivers/staging/line6/line6usb.ko] undefined!
ERROR: "__sync_fetch_and_or_4" [drivers/staging/line6/line6usb.ko] undefined!
make[1]: *** [__modpost] Error 1
make: *** [modules] Error 2

BTW : I cant imagine that 1804 modules are necessary for a wm8850 device... (?)
E.g. there were myriads of file sytem (fs) modules...

Is it better to take/try another kernel as 3.9-rc7 ? Perhaps a 3.8 one ?

Or just unselect those modules :)
The ones you show are all sound codecs, and we don't have a an I2S or AC97 driver so you may as well unselect all the sound subsystem stuff - it won't do anything but take up space :)

Regards
Tony P

[Tony Prisk]

On Friday, April 19, 2013 10:38:47 PM UTC+12, Peter Vasil wrote:

On Fri, Apr 19, 2013 at 12:34 PM, wkt <wktin...@gmail.com> wrote:
> make ARCH=arm INSTALL_MOD_PATH=/home/wkt8850nb modules_install
>
> cp: cannot stat `/home/kernel-3.9-rc7/modules.order': No such file or
> directory
> make: *** [_modinst_] Error 1>
>
> After a new kernel cross-compilation I tried the above command.
>
> Did I miss something my xconfig dialog ?
> As I saw there are many M option settings : so I did expect modules in the
> INSTALL_MOD_PATH - but there is only an empty lib/modules/3.9.0-rc7 dir
>
> --
> You received this message because you are subscribed to the Google Groups
> "VT8500/WM8505 Linux Kernel" group.
> To unsubscribe from this group and stop receiving emails from it, send an

> email to vt8500-wm8505-linux-kernel+unsub...@googlegroups.com.

> For more options, visit https://groups.google.com/groups/opt_out.
>
>

Looks like you didn't actually build the modules - this usually
happens when you only call "make uImage" or "make zImage". You need
something like "make modules" before you can do modules_install...
Regards,
Peter

Peter,

Would you mind posting a config for Wolfgang to use as a starting point?
My config is extremely minimal and wouldn't help him to build a functioning system - I go for the minimalist approach :)

Regards
Tony P

[Peter Vasil]

On Fri, Apr 19, 2013 at 10:38 PM, Tony Prisk
<sent...@digitalfantasy.co.nz> wrote:
> On Friday, April 19, 2013 10:38:47 PM UTC+12, Peter Vasil wrote:
>>
>> On Fri, Apr 19, 2013 at 12:34 PM, wkt <wktin...@gmail.com> wrote:
>> > make ARCH=arm INSTALL_MOD_PATH=/home/wkt8850nb modules_install
>> >
>> > cp: cannot stat `/home/kernel-3.9-rc7/modules.order': No such file or
>> > directory
>> > make: *** [_modinst_] Error 1>
>> >
>> > After a new kernel cross-compilation I tried the above command.
>> >
>> > Did I miss something my xconfig dialog ?
>> > As I saw there are many M option settings : so I did expect modules in
>> > the
>> > INSTALL_MOD_PATH - but there is only an empty lib/modules/3.9.0-rc7 dir
>> >
>> > --
>> > You received this message because you are subscribed to the Google
>> > Groups
>> > "VT8500/WM8505 Linux Kernel" group.
>> > To unsubscribe from this group and stop receiving emails from it, send
>> > an

>> > email to vt8500-wm8505-linux...@googlegroups.com.

>> > For more options, visit https://groups.google.com/groups/opt_out.
>> >
>> >
>>
>> Looks like you didn't actually build the modules - this usually
>> happens when you only call "make uImage" or "make zImage". You need
>> something like "make modules" before you can do modules_install...
>> Regards,
>> Peter
>
>
> Peter,
>
> Would you mind posting a config for Wolfgang to use as a starting point?
> My config is extremely minimal and wouldn't help him to build a functioning
> system - I go for the minimalist approach :)
>
> Regards
> Tony P
>

> --
> You received this message because you are subscribed to the Google Groups
> "VT8500/WM8505 Linux Kernel" group.
> To unsubscribe from this group and stop receiving emails from it, send an

> email to vt8500-wm8505-linux...@googlegroups.com.

> For more options, visit https://groups.google.com/groups/opt_out.
>
>

Sure, no problem :-) Download this file and copy it to .config in
kernel build directory:
http://relghuar.net/.config.wm8850
After you copy it, call make menuconfig, change support to wm8750 and
save it again to sync it with your sources, I have some local changes
inside (platform via-rhine and device-tree gpio-charger for instance).
Btw. I've checked my build, there's less than 500 module binaries...

[Tony Prisk]

Sure, no problem :-) Download this file and copy it to .config in
kernel build directory:
http://relghuar.net/.config.wm8850
After you copy it, call make menuconfig, change support to wm8750 and
save it again to sync it with your sources, I have some local changes
inside (platform via-rhine and device-tree gpio-charger for instance).
Btw. I've checked  my build, there's less than 500 module binaries...

Thanks Peter,

By comparison, I have 0 modules and a 1.7Mb compiled kernel :)

Regards
Tony P

[Peter Vasil]

On Sat, Apr 20, 2013 at 7:41 PM, wkt <wktin...@gmail.com> wrote:

> change support to wm8750 : ??? why that - I have a wm8850...

Whoops, sorry, forget I said anything :-) I've noticed emails from
someone new saying he had an apc board with wm8750, must have confused
you with him.

>
> I browsed casually through the .config - and found fs ext2/3 activated what
> I think could be important for boot test from ect2/3 partition on SD card
> (?).

Well, depends on what filesystem exactly you have. I'm using ext4 as
my rootfs, but I always compile ext2/3 into all my kernels as well
(along with a few others) in case I need to access some flash disk or
sd card with such a system.

>
> 500 modules sounds almost fine ;)

Actually it's only half of that :-/ I forgot about my temporary
install location inside kernel directory, so I counted each module
twice.

>
> I dream also of a mini .config ( special edition for my wm8850 needs )
> as a minimum e.g. support keyboard, touchpad, XVGA screen, UDB sticks, SD
> card,
> WiFi ( also e.g. my RALINK stick ), Ethernet LAN port, sound chip,
> microphone, camera, Bluetooth stick, only most nec. fs ( ext2/3/4, MS,
> Android ),
> and may be some more ( am not firm in hardware ).
>

Quite a few of my modules are for USB hardware - network devices,
serial adapters, printers, sound cards... I use a lot of gadgets. You
can safely remove most of them. There are also many network modules -
full xt/ip tables, conntrack, netfilter, dial-up support, some of
these things can be removed as well, if you're careful to keep things
required for basic wireless/ethernet.
Regards,
Peter

[Andy Chernyak]

Looks like it's userspace hanging. Try to boot with "init=/bin/sh"

On 04/23/2013 04:35 PM, wkt wrote:

Above my best result so far -/

2 remarks :

1. if I use rootfstype explicitly in bootargs then the screen remains dark.

2. my impression according to my tests is that a boot from mmcblk0p2 does not work.

But may be I make some other errors - especially I am unsure about what content /etc/fstab

should have ( and what definitely not ). Also interesting would be the so called "populating of /dev"

with mknod commands : what would be required. E.g. I have no /dev/input in my disro.

[Sean Lee]

Here is the instruction for emdebian grip.

http://www.emdebian.org/grip/

wktin...@gmail.com於 2013年4月24日星期三UTC+8上午12時23分10秒寫道：

I grabbed a Debian non-GUI rootf from the Android app "Complete Installer Linux"

and made some modifications on it ( /dev, fstab, hostname, /etc/network/interfaces, /lib/modules )

emdebian - sounds wonderful. Where can I get it from ?

Am Dienstag, 23. April 2013 18:01:10 UTC+2 schrieb Sean Lee:

Hi wkt,

Where do you get your rootfs?

My WM8850 netbook can boot to login prompt with uzImage.bin (Peter's config) and emdebian rootfs built for WM8650.

The rootfs is placed in ext3 filesystem in SD card. (mmcblk0p2) 

Although there are still some issues have to be fixed. It always shows me "Unable to determine your tty name".

If you don't have an available rootfs, how about trying to build yours with emdebian?

Hope this helpful.

Thank you, Peter! Your config is really helpful for me. :)

wkt於 2013年4月23日星期二UTC+8下午10時35分30秒寫道：

[Sean Lee]

I write an instruction for Emdebian Grip.

This is been tested for WM8850 Netbook.

It is not only for WM8850 but also other ARM system.

http://seanlee99.blogspot.tw/2013/04/build-your-emdebian-grip-for-arm.html

(May I post my blog link here? Some forums don't like people do this. :P )

Sean Lee於 2013年4月24日星期三UTC+8上午12時42分24秒寫道：

[Sean Lee]

Hi wkt,

I am so glad that you have a good result after following the steps.

My kernel config is provided by Peter mentioned in this thread before.

I can't do this without his config. (no any secret of my steps. Ha.)

(I take your advice to change font color for more comfortable to read. :) )

wkt於 2013年4月24日星期三UTC+8下午9時18分58秒寫道：

Wonder(media)ful good news Sean,

I followed the steps on your blog and all works perfectly.
2nd boot successfully completed ! Great work.

Perhaps someone also interested in configuring now this
core emdebian could open a thread where one could discuss
whether there is away to get now e.g. X11 and network or
even sound running...

Could you post a download of your .config file of your kernel
on your blog ? Or is this file protected/private only ?
I think I could then study it and make the necessary modules (?) and
perhaps then start my other Debian with GUI from "Complete Linux Installer"
Android app.

[Peter Vasil]

On Fri, May 3, 2013 at 12:26 AM, wkt <wktin...@gmail.com> wrote:
> Meanwhile I did a new kernel compilation.
>
> I now have a emdebian system where I have WiFi with the Ralink USB stick
> and startx leads to a GUI screen. I also installed gdm display manager.
> 1. i can enter username/password in gdm but the keys typed are not
> visible - no correct echoing
> 2. the GUI destop is glaring intensely and the menu letters/texts for
> instance
> are hardly visible - as also the icon texts. I really dont know which
> changes
> in the make xconfig lead now to the progress in starting "successfully"
> the
> startx command now.

Hi, this glaring/invisible graphics strongly suggests an incorrect
contrast setting, it has been discussed a few times in this group. You
should find "contrast" file (I think that's its exact name) somewhere
in your /sys and check the value there. From what I've seen correct
value for most people is somewhere around 10-16. For some time there
has been a too high default value in kernel sources (128 I think)
which at least on my machine caused any even slightly brighter color
to show with maximum intensity, while dark colors were actually almost
black. It should be possible to fix this in runtime by writing correct
value back into the contrast file.
Regards,
Peter

>
> I can give the Xorg.0.log file content and that of /etc/X11/xorg.conf
> if that would be of any use.
> 3. the touchpad seems to work - not on every button I tried though but
> mostly.

[Peter Vasil]

On Fri, May 3, 2013 at 12:45 AM, <wktin...@gmail.com> wrote:
> @Sean
>
> Please tell me ( the URL ? ) where you got your kernel source from.
>

Hi, by combining
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git and
git://server.prisktech.co.nz/git/linuxwmt.git (or alternatively my
git://git.relghuar.net/git/linuxwmt.git - though it hasn't been
updated in a while) you can build your own kernel with most of latest
features. Here http://relghuar.net/.config.wm8850 you can download
build configuration for my WM8850 netbook, it would probably not be a
perfect match, but may help you as a starting point. Before building
you should always check the relevant DTS files (wm8850.dtsi and one of
wm8850-???.dts in your case) and verify default values for brightness,
contrast and similar things.
Regards,
Peter

> I cant understand why the startx command leads to a perfect GUI
> when I run your kernel.

[Tony Prisk]

On Saturday, May 4, 2013 12:29:28 AM UTC+12, wktin...@gmail.com wrote:

Hi Peter,

your tip with the contrast file solved my problem :
on my device/embedian Linux I ran this

sudo echo -n 26 > /sys/devices/soc.0/d8051700.fb/contrast

Now the XFCE GUI looks fine and is fully usable.

Dont know how to handle the "git". I will have a look at the URLs to see
whether I understand whats to do.
What I lack at the moment is LAN eth0 detection and the internal WiFi.
Do your URLs handle these problems already ?

I am quite content now with the state of my system and can go on building it up.

The contrast default should be fixed in the current testing branch.
The velocity ethernet controller is supported in the current testing branch.
'Internal WIFI' is usually a USB wifi stick - different models have different sticks.

I have finally gotten around to installing XFCE on my APC8750. Everything looks fine and with apt + ethernet, things are much easier to install now :)

Regards
Tony P

PS:
If you have the git tools installed, you can simply use:
git clone git://github.com/linux-wmt/linux-vtwm

@Peter:
My local server is no longer the 'official' repo. It is now hosted on github, due to concerns about things 'disappearing' should I leave.
I still update my local copy, but it may disappear at some stage. Better to rely on the github repo.

[Tony Prisk]

On Sunday, May 5, 2013 8:19:58 AM UTC+12, wkt wrote:

Sorry, I am still studying the basics of kernel building.
So I cant deal with professional tools like "git" at this moment.
Will have a look later...

Under "internal WiFi" I understood not an USB stick I plug
manually into one ( of the 2 on my device ) external USB ports
but the one in the device itself : under wm8650 I always
worked with a module called rt3070sta.ko ( or rt5370sta.ko ??? )

The one in the device is a just a USB wifi module with the plastic shell removed.

With my kernel which runs the Emdebian quite well I also
run an Ubuntu 12.04 LTS system.
Here I have a special sproblem when I boot getting these messages :
>Unknown filesystem type devpts
>mount /dev/pts status 32
>Filesystem could not be mounted : /dev/pts
>Plymouth command failed
>Disconnected from Plymouth
and then it boots its way. BUT : if I open the Terminal under the GUI
i get a queer message and no prompt - so I suspect the cause might be
the messages... (?)

Did I miss something in my kernel config ( as I dont get these messages
with the kernel of Sean Lee blog - also the Terminal works with his )
In the www I found a hint to set CONFIG_UNIX98_PTYS=y ( is set in my config )
and to look under Filesystems->Pseudo filesystems for "Virtual memory filesystem
support" - bot this config option isnt shown in my xconfig.

Another problem is a complete garbage on my ALT+Fx consoles. Messages and
file contents are not sequentially scrollable but former shown lines come back.
Also lines are not cleared to the end.

 

The scrolling problem sounds like it could be the 'accelerated raster ops'. This currently isn't support on wm8850.
In drivers->graphics->framebuffers, you can select the wm8505 framebuffer driver - ensure that 'accelerated raster ops' is unchecked.

Regards
Tony P

 

[Tony Prisk]

On Tuesday, May 7, 2013 3:00:50 AM UTC+12, wkt wrote:

I ran as recommended above

git clone git://github.com/linux-wmt/linux-vtwm

Then I took my test .config file and when I compiled the kernel I was

prompted to activate the Velocity driver - what I did.

The kernel compilation came to a halt with the error messages already

reported in the thread about Velocity networking :

CC drivers/net/ethernet/via/via-velocity.o
drivers/net/ethernet/via/via-velocity.c: In function ‘velocity_suspend’:
drivers/net/ethernet/via/via-velocity.c:3170: error: expected ‘;’ before ‘else’
drivers/net/ethernet/via/via-velocity.c: In function ‘velocity_resume’:
drivers/net/ethernet/via/via-velocity.c:3249: error: implicit declaration of function ‘pci_get_devdata’
drivers/net/ethernet/via/via-velocity.c:3249: warning: assignment makes pointer from integer without a cast
drivers/net/ethernet/via/via-velocity.c: In function ‘velocity_platform_suspend’:
drivers/net/ethernet/via/via-velocity.c:3300: error: ‘state’ undeclared (first use in this function)
drivers/net/ethernet/via/via-velocity.c:3300: error: (Each undeclared identifier is reported only once
drivers/net/ethernet/via/via-velocity.c:3300: error: for each function it appears in.)
drivers/net/ethernet/via/via-velocity.c: At top level:
drivers/net/ethernet/via/via-velocity.c:3309: warning: initialization from incompatible pointer type
drivers/net/ethernet/via/via-velocity.c:3310: warning: initialization from incompatible pointer type
drivers/net/ethernet/via/via-velocity.c:3336: warning: initialization from incompatible pointer type
make[4]: *** [drivers/net/ethernet/via/via-velocity.o] Error 1
make[3]: *** [drivers/net/ethernet/via] Error 2
make[2]: *** [drivers/net/ethernet] Error 2
make[1]: *** [drivers/net] Error 2

make: *** [drivers] Error 2

Maybe I dont fully understand how to work with git...

There was a mistake in the original (v1/v2) patches in the PM code, which I fixed in v3+, but I couldn't reapply v3+ to
the testing branch because it was rebased onto net-next which required other changes.  When I rebase for 3.10-rc1,
this problem will be fixed correctly with the v6 patch.

I have applied some quick fixes to resolve the 'testing' version.

You can pull the changes with a 'git pull' now and it should be fine - I have compiled tested the changes this morning.

Regards
Tony Prisk

[Tony Prisk]

On Wednesday, May 8, 2013 6:06:24 AM UTC+12, wkt wrote:

Tony, you wrote :

In drivers->graphics->framebuffers, you can select the wm8505 framebuffer driver - ensure that 'accelerated raster ops' is unchecked.

in make xconfig I find   drivers->graphics->framebuffers->wm8505 framebuffer driver

BUT :  where is  "accelerated raster ops" ??? 

If you are using the current testing branch select the following under
 Device Drivers->Graphics Support->Support for frame buffer devices->

[ ] VIA VT8500 framebuffer support
[*] Wondermedia WM8xxx-series frame buffer support
[ ]     Enable VGA output + Disable LCD output
[*] VT8500/WM8xxx accelerated raster ops support

If you don't have these options, something else is wrong - probably an old version.

Regards
Tony Prisk

[Tony Prisk]

On Saturday, May 11, 2013 3:59:29 AM UTC+12, wktin...@gmail.com wrote:

Hi Tony,

I Have now the options below in my make xconfig tree.

I am somewhat confused now as you wrote "ensure that 'accelerated raster ops' is unchecked"

( what I did !!! ) but below you have the option checked. Whats correct ?

I did a new kernel compilation and now also the velocity.c was compiled : fine !

But if I boot my system the screen turns first glaringly white and the gets dimmed down

to a dark blue/grey color... No console messages any more...

Sorry - 'VT8500/WM8xxx accelerated raster ops support' should be unchecked, not checked.

[Tony Prisk]

There is no Rhine driver support in any of the current development kernels or mainline for the vt8500 platform - only Velocity is supported at the moment.
You will no find anything with lspci as there is no PCI bus on these platforms. Wondermedia wrote a fake pci bus driver so they didn't have to add platform support to the pci drivers that already existed. We don't do that :)

In regards to the glaring screen:
Do you get any output on serial? (if you have serial available)
Do you have a console=/dev/ttyWMT0 in your bootargs? If so, you won't get console messages unless you have a console=/dev/fb as well (not sure about this but I presume it works like that).
Have you mapped the console to the framebuffer in Kconfig?
  Device Drivers->Graphics Support->Console display driver support->Framebuffer Console support
  Device Drivers->Graphics Support->Console display driver support->Map the console to the primary display device

Regards
Tony P

[Peter Vasil]

Hi everyone,

On Sun, May 12, 2013 at 2:27 AM, Tony Prisk
<sent...@digitalfantasy.co.nz> wrote:
> There is no Rhine driver support in any of the current development kernels
> or mainline for the vt8500 platform - only Velocity is supported at the
> moment.
> You will no find anything with lspci as there is no PCI bus on these
> platforms. Wondermedia wrote a fake pci bus driver so they didn't have to
> add platform support to the pci drivers that already existed. We don't do
> that :)

Like I've already mentioned a few times, I have an ugly
hacked-together version of rhine driver that works little more than
halfway on my wm8850 netbook (poor reception). I've already started to
read through Tony's velocity patches, most of the changes should also
be appliable to rhine. Not sure about the timeframe (month? two?) but
I definitely plan to fix rhine for VIA platform same way Tony fixed
velocity.

>
> In regards to the glaring screen:
> Do you get any output on serial? (if you have serial available)
> Do you have a console=/dev/ttyWMT0 in your bootargs? If so, you won't get
> console messages unless you have a console=/dev/fb as well (not sure about
> this but I presume it works like that).
> Have you mapped the console to the framebuffer in Kconfig?
> Device Drivers->Graphics Support->Console display driver
> support->Framebuffer Console support
> Device Drivers->Graphics Support->Console display driver support->Map the
> console to the primary display device
>
> Regards
> Tony P

I'm pretty sure it should be console=/dev/tty0 (or any other number)
for on-screen messages. I've already tried that before and it worked
quite fine. Of course Framebuffer Console support and mapping in
kernel config is a must...

Regards,
Peter

[Tony Prisk]

On Tuesday, May 14, 2013 5:16:48 AM UTC+12, wkt wrote:

@Peter

I would be glad if sometime the Rhine driver would run/work.

I personally am in absolutely no hurry.

I will check the framebuffer settings as said by Tony.

in the bootargs I use(d) /dev/tty1 as console.

May i ask about sound ? What is the state of a driver ?

This is what i miss most... hearing a little bit of music/radio... :) 

No sound support at all :\ It has always been low on my list of things to do.

My plan for the next few weeks/months is to sort out the current bugs (serial/raster ops)
and then look at the NAND driver (for a third time). If I can get back to the old state of
even having no ECC support I will probably post it in the hope someone else can sort it
out :)

Regards
Tony P

[Alexey Charkov]

2013/5/16 <wktin...@gmail.com>:
> Besides that I would like to know why no sound driver is available. Is it
> kept secret/proprietory by Wondermedia ?
> Isnt Wondermedia interested in making Linux run on their machines ? The
> driver must be available ( or not ? )
> as my Android plays sound.

If you looked at WonderMedia code, you'd understand why :)

Their drivers are mostly available in source form and under GPL, but
the code is typically very ugly, does not conform to Linux kernel
standards, does not use standard kernel infrastructure and manages
different hardware versions in very ugly ways (if at all). As such, it
is unsuitable for inclusion into the mainline Linux kernel and
requires substantial effort to rework (which takes time and devotion).

Furthermore, the code currently in mainline has a different (i.e.
modern) implementation for architecture-specific stuff (e.g.
pin-muxing, GPIO's, device trees vs. board files, clock implementation
using designated frameworks). WonderMedia drivers need porting to make
use of that

Best,
Alexey

[Tony Prisk]

On Monday, June 3, 2013 2:59:33 AM UTC+12, wkt wrote:

1. I am a little bit disappointed about the ugly source state of the 

   Wondermedia software. But on the other side I see no sense in

   porting any source into the Linux kernels if e.g. sound will never

   run on the devices...

This is a little chicken-and-egg. Nothing will work if it's not ported, and if its not ported it won't be supported.
Sound will work fine - it just needs to be rewritten for the current state of the kernel. If sound is that important, feel free to port the code yourself - otherwise, I guess you will have to wait until someone else does it.
 

2. I wanted to have access to the Android partitions in flash memory

   from my Linux versions.

   So i recompiled the 3.9.0-rc7 kernel after config-ing the MTD stuff

   as far as I found information in the net about this technology.

   But after booting no flash to be seen in /proc/mtd and there not

   even mtdblockN or mdtN or mtdrN devices to be found in /dev.

   So : does this work at all or where could be a solution of it ? 

 

There is currently no NAND MTD driver for these SoCs. An old one is floating around somewhere which can be made to work (I have previously patched it to work on the newer NAND devices) but there are issues around the ECC/OOB code and for some reason I deleted my work. This is the next 'major' component on my list of things to do.

Regards
Tony P

[wkt]

I got a interesting hint which could relieve the problems with SOUND and LAN ( eth0 ).

I was told there are small USB adapters on the market where one can plug in a headset

or a LAN cable and put it into one of the USB ports of the wm8850 netbook.

Can the kernel software handle such adapters ?

If yes what have I to change in the .config ? Other changes necessary ?

[Tony Prisk]

These are USB audio and USB network adapters - and yes they will 'solve'
the problem.
Like any device, they will work if there is support for them in the
kernel - find the chipset they are based on and check if there is a USB
version of the driver in the respective subsystem drivers list :)

Regards
Tony P

[wkt]

I want to tell that I have now a kernel which supports my WLAM, LAN, Sound - USB Adapters.

LAN Adapter : Kontron DM9601 Fast Ethernet Adapter ( 0fe6:9700 )

Sound Adapter : C-Media Electronics Audio Adapter ( 018c:000e )

For Sound I activated in .config ( info from net article ):

CONFIG_SOUND=y primary sound card support
CONFIG_SND=y
CONFIG_SND_TIMER=y
CONFIG_SND_PCM=y
CONFIG_SND_HWDEP=m
CONFIG_SND_RAWMIDI=m
CONFIG_SND_SEQUENCER=y WaveTable functionality
CONFIG_SND_OSSEMUL=y run OSS applications with ALSA
CONFIG_SND_MIXER_OSS=y run OSS applications with ALSA
CONFIG_SND_PCM_OSS=y run OSS applications with ALSA
CONFIG_SND_USB_AUDIO=m

CONFIG_USB_HID=m
CONFIG_USB_HIDINPUT=y

So the only drawback : netbook has only 2 USB ports... ;)

[gamec...@gmail.com]

 Sean Lee, 

Can I ask something. I have got WM8850. I want to setup linux but not starting with wmt_scriptcmd and uzImage.bin from fat32. What can I do? Can you help me?

[Alexey Charkov]

Hi there,

While I'm not Sean Lee, you can technically install Linux on WM8850 without scriptcmd ;-)

You'll need root access, and you'll have to write kernel image directly to NAND and then manually overwrite U-Boot configuration in serial flash from Android (to pass a correct command line to your kernel).

Please note though that this way is more risky, as you won't have the option to just remove your SD card and get a known-working system anymore. So, if something goes wrong and you reboot, you can get a brick.

You'll also need to manually apply patches from Tony's repo on Github if you want to access either NAND or serial flash from under a kernel different from WonderMedia's blob. Otherwise you won't be able to update the kernel or cmdline ever after you reboot (unless you use serial console or scriptcmd).

Just one question though. Why doesn't the scriptcmd method suit your needs?

Best,
Alexey

On 9 Jan 2014 12:55, <gamec...@gmail.com> wrote:

 Sean Lee, 

Can I ask something. I have got WM8850. I want to setup linux but not starting with wmt_scriptcmd and uzImage.bin from fat32. What can I do? Can you help me?

--

[Theodore Kilgore]

Since I see that at least one other person is in this situation:

I have a similar problem. Mine is an Iview 706-NB 7-inch netbook. I have
wondered if it is blocked and needs to be "rooted" or something like that.
Does anyone know about such things?

Theodore Kilgore


On Thu, 9 Jan 2014, gamec...@gmail.com wrote:

> �Sean Lee,�Can I ask something. I have got WM8850. I want to setup linux but

> not starting with wmt_scriptcmd and uzImage.bin from fat32. What can I do?
> Can you help me?
>

[Theodore Kilgore]

On Thu, 9 Jan 2014, Alexey Charkov wrote:

>
> Hi there,
>
> While I'm not Sean Lee, you can technically install Linux on WM8850 without
> scriptcmd ;-)
>
> You'll need root access,

[snip]

Aha. I guess I should have read the rest of my mail. What, exactly, is one
supposed to do in order to get root access?

Theodore Kilgore

[Alexey Charkov]

2014/1/9 Theodore Kilgore <kil...@banach.math.auburn.edu>:

Incidentally, one would need to use a scriptcmd file to modify the
boot sequence so that the machine boots into some different root FS or
initrd where you already have root access. Then you can use that to
mount the /system partition read-write and install /system/xbin/su
(owned by root and with permissions 6755) and
/system/apps/Superuser.apk from e.g. this archive:

http://download.clockworkmod.com/superuser/superuser.zip

Alternatively, one would need to find some privilege escalation
vulnerability in the stock Android image that would allow to gain root
access without booting into anything else first - but this is
obviously more tricky.

Last time I needed to root a WMT-based Android system (WM8710 set-top
box in my case) I compiled a minimal root filesystem with Buildroot
(http://buildroot.uclibc.org/download.html) and then used a serial
cable to boot into it and install the necessary files. For more common
configurations there are surely some pre-packaged solutions available
on the Internet.

Best,
Alexey

[Theodore Kilgore]

Thanks, Alexey. I can't try any of this right away, but some of what you
say below leaves me still a bit puzzled. Please read my comments and
questions in-line, below.

Theodore Kilgore

On Fri, 10 Jan 2014, Alexey Charkov wrote:

> 2014/1/9 Theodore Kilgore <kil...@banach.math.auburn.edu>:
> >
> >
> > On Thu, 9 Jan 2014, Alexey Charkov wrote:
> >
> >>
> >> Hi there,
> >>
> >> While I'm not Sean Lee, you can technically install Linux on WM8850 without
> >> scriptcmd ;-)
> >>
> >> You'll need root access,
> >
> > [snip]
> >
> > Aha. I guess I should have read the rest of my mail. What, exactly, is one
> > supposed to do in order to get root access?
>
> Incidentally,

Incidentally, or necessarily?

one would need to use a scriptcmd file to modify the
> boot sequence

This right here is where things are not working, I am guessing? Or???
What I mean is, the machine does not do anything when I stick a card in
with a scriptcmd, and a kernel, and a filesystem on it, and turn it on. It
just ignores the presence of the SD card until it boots into Android as
usual, and then on some occasions it even prompts me with a generous offer
to format my card for me!

so that the machine boots into some different root FS or
> initrd where you already have root access.

But this is what it is not doing. Catch 22?

Then you can use that to
> mount the /system partition read-write and install /system/xbin/su
> (owned by root and with permissions 6755) and
> /system/apps/Superuser.apk from e.g. this archive:
>
> http://download.clockworkmod.com/superuser/superuser.zip
>

And, said Archimedes, "If I had a place to put my lever, I could move the
earth." Sorry to write something which might make you think I am
impertinent, but your instructions more or less start in the middle and
seem to assume that I was able to do something which I could not do in the
first place. Again, it did not appear to matter at all what I put on the
SD card. The little machine simply ignored whatever is there while it is
booting.

> Alternatively, one would need to find some privilege escalation
> vulnerability in the stock Android image that would allow to gain root
> access without booting into anything else first - but this is
> obviously more tricky.

I guess so, yes.

>
> Last time I needed to root a WMT-based Android system (WM8710 set-top
> box in my case) I compiled a minimal root filesystem with Buildroot
> (http://buildroot.uclibc.org/download.html) and then used a serial
> cable to boot into it and install the necessary files. For more common
> configurations there are surely some pre-packaged solutions available
> on the Internet.

I certainly hope so, and I also thank you for the links above which I will
check into when I get time. Cheers.

>
> Best,
> Alexey

[Alexey Charkov]

2014/1/10 Theodore Kilgore <kil...@banach.math.auburn.edu>:

>
> Thanks, Alexey. I can't try any of this right away, but some of what you
> say below leaves me still a bit puzzled. Please read my comments and
> questions in-line, below.

Sure, it's a tricky one. See the replies in-line.

> Theodore Kilgore
>
> On Fri, 10 Jan 2014, Alexey Charkov wrote:
>
>> 2014/1/9 Theodore Kilgore <kil...@banach.math.auburn.edu>:
>> >
>> >
>> > On Thu, 9 Jan 2014, Alexey Charkov wrote:
>> >
>> >>
>> >> Hi there,
>> >>
>> >> While I'm not Sean Lee, you can technically install Linux on WM8850 without
>> >> scriptcmd ;-)
>> >>
>> >> You'll need root access,
>> >
>> > [snip]
>> >
>> > Aha. I guess I should have read the rest of my mail. What, exactly, is one
>> > supposed to do in order to get root access?
>>
>> Incidentally,
>
> Incidentally, or necessarily?

Not necessarily, as there are different ways to divert the boot
process from the standard sequence. Scriptcmd is usually the simplest,
but one can also directly issue u-boot commands from serial console,
re-flash u-boot configuration directly in SPI flash using an external
in-circuit programmer, or find some security hole in WonderMedia's
Android build that would give local root access.

> one would need to use a scriptcmd file to modify the
>> boot sequence
>
> This right here is where things are not working, I am guessing? Or???
> What I mean is, the machine does not do anything when I stick a card in
> with a scriptcmd, and a kernel, and a filesystem on it, and turn it on. It
> just ignores the presence of the SD card until it boots into Android as
> usual, and then on some occasions it even prompts me with a generous offer
> to format my card for me!

Different u-boot builds look for the scriptcmd file in different
locations. I've seen /script/scriptcmd, /script/wmt_scriptcmd,
/scriptcmd and /wmt_scriptcmd off the top of my head. Just try
different locations out of curiosity.

I would also recommend to make the first FAT32 partition relatively
small (say, 32MB) and make sure that the scriptcmd file is the first
thing that gets written there after formatting - just in case.

> so that the machine boots into some different root FS or
>> initrd where you already have root access.
>
> But this is what it is not doing. Catch 22?

Yes, sort of. If neither of the known scriptcmd locations work for you
and if you can't use any of the other methods to reconfigure u-boot
(serial console or SPI programmer), then it's indeed Catch 22 (as you
would need root access to reconfigure u-boot from a running Android
system to gain root access on that Android system).

> Then you can use that to
>> mount the /system partition read-write and install /system/xbin/su
>> (owned by root and with permissions 6755) and
>> /system/apps/Superuser.apk from e.g. this archive:
>>
>> http://download.clockworkmod.com/superuser/superuser.zip
>>
> And, said Archimedes, "If I had a place to put my lever, I could move the
> earth." Sorry to write something which might make you think I am
> impertinent, but your instructions more or less start in the middle and
> seem to assume that I was able to do something which I could not do in the
> first place. Again, it did not appear to matter at all what I put on the
> SD card. The little machine simply ignored whatever is there while it is
> booting.

You may have some bizarre build of u-boot there, or some buggy
hardware that doesn't initialize properly to make that happen.

Just wondering: have you tried different SD cards? It may happen that
your U-boot can't read cards of particular size/clock
frequency/whatever else it doesn't like. Its SD driver is not quite
the same as what Android uses.

>> Alternatively, one would need to find some privilege escalation
>> vulnerability in the stock Android image that would allow to gain root
>> access without booting into anything else first - but this is
>> obviously more tricky.
>
> I guess so, yes.

There are some "one-click root" solutions in the wild that are
applicable to a wide range of devices, so one of those might work for
you as well. I personally used z4root on several phones and tablets
pretty successfully in the past - not sure if it would work on your
device, though.

Best,
Alexey

>> Last time I needed to root a WMT-based Android system (WM8710 set-top
>> box in my case) I compiled a minimal root filesystem with Buildroot
>> (http://buildroot.uclibc.org/download.html) and then used a serial
>> cable to boot into it and install the necessary files. For more common
>> configurations there are surely some pre-packaged solutions available
>> on the Internet.
>
> I certainly hope so, and I also thank you for the links above which I will
> check into when I get time. Cheers.
>
>>
>> Best,
>> Alexey
>

[st.go...@gmail.com]

In addition to /script/scriptcmd, /script/wmt_scriptcmd ... there is also:
/FirmwareInstall/autorun.1.wmt
This works on some wm8850 7" and 10" android notebooks

Cheers,
   gothian

On Thursday, January 9, 2014 10:00:42 PM UTC+1, Alexey Charkov wrote:

2014/1/10 Theodore Kilgore <kil...@banach.math.auburn.edu>: