wm8850 netbook success except networking

[D. J. Bernstein]

I'm co-managing a crypto benchmarking project (http://bench.cr.yp.to)
and as part of that find myself installing Debian/Ubuntu/Arch/... on
devices with many different CPUs. A while ago I tried to buy

http://www.amazon.com/RUNN-701-Android-NandFlash-Supports-Flash10-1/dp/B00D5DT8IE

from a low-cost vendor and somehow ended up receiving a newer CPU in the
same case---I guess this isn't surprising for those of you familiar with
the Chinese netbook market.

The device was shipped with Android 4.0.3, kernel 3.0.8, build ICS
Ver1.1.0-20120801.042457. It identifies itself as WM8850-mid/PID7_8198C.
Android CPU Checker says "VIA WM8850 1.2GHz. MID Tablet with EKEN/AGPTek
firmware. Screen size 7.0 inches. Frequency states not supported. NEON
capable. RAM memory module: 512 Mb DDR. Mali-400 MP (dual core GPU)."

This is an easy version of Android to jailbreak via adb, so normally I'd
just run a chrooted Debian, but I don't see an easy way to run adb on
this device (there's no obvious USB slave port). So I went through the
steps shown below to create a bootable Debian SD card for the device.
The part that took me the longest to find documented online was that
wmt_scriptcmd needs to be mkimage output, not just a text file; anyway,
I've tried to spell out all the steps in detail.

There's one serious problem with the Debian installation, namely that
none of the built-in physical networking devices show up in ifconfig -a.
I wasn't really expecting the Ethernet chip (Via Rhine) to work with
this kernel, but I was expecting wireless (rtl8192cu---see attached
Android dmesg) to work, and it's not immediately obvious to me what to
look for in the kernel output to diagnose the problem.

I tried adding a ridiculous number of device drivers (after manually
adding "select MIGHT_HAVE_PCI" under "config ARCH_VT8500" in
arch/arm/mach-vt8500/Kconfig) in case I had missed something relevant.
The only things that I went out of my way to disable were I2C and the
vt* sensors, since those consistently produce an early kernel crash. I
can still boot into Android by simply removing the SD card, and I'm
happy to collect more information, run test programs, etc. if someone
has an idea of what to try next. I should note that I _can_ plug in and
use an rtl8150 USB stick, so USB and networking aren't totally busted.

Attachments: Android dmesg (old-dmesg); steps to create SD card
(new-sdcard); latest kernel configuration (new-config); resulting dmesg
(new-dmesg).

---Dan

[Alexey Charkov]

Hi Dan!

Normally, the only extra step required to enable WiFi (apart from
having the correct drivers and firmware) is to enable power to the
built-in adapter by driving one of the "external" GPIO's high. You can
typically determine which exactly by looking at u-boot variables. Or
just export all of them (0 to 7) through the sysfs interface and try
setting each of them to 'output', 'high' until you see a new USB
device.

As for the built-in Rhine adapter, you may want to try the patch
series that I've recently posted to netdev:

http://news.gmane.org/find-root.php?message_id=1390823503%2d24087%2d1%2dgit%2dsend%2demail%2dalchark%40gmail.com

You'd also need to modify your device tree file after applying these
patches to get it working.

Note that PCI drivers would not work directly without modification, as
these chips don't have any real PCI bus. WonderMedia kernels provide
an in-kernel emulation layer that makes PCI drivers work with platform
devices, but that is not considered good practice for the mainline
kernel (so drivers need to be updated to support platform-bus
configuration instead).

Best,
Alexey

2014-02-01 D. J. Bernstein <vt8...@box.cr.yp.to>:

> --
> 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.

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

Hallo Dan,

I have that same netbook (WM8850mid, 7" screen, 512MB-Ram, no OTG-USB port, unfortuntely green as well ...)

1. Root Access: You can get root access to the device using the same mechanism you use to boot debian. A tool called "RootBurner_v0.1_MaPan" did the trick for me.

2. WiFi: To activate your WiFi-Hardware you use the pin controller as described by Alexey. For my device bringing up pin 6 works:
- echo 6 > /sys/class/gpio/export
- echo out > /sys/class/gpio/gpio6/direction
- echo 1 > /sys/class/gpio/gpio6/value
dmesg should then show your wifi hardware. Mine shows a Realtek 802.11n WLAN NIC with a Product Id 8179. Select driver and firmware accordingly.


3. Eth: I could not get Alexeys Rhine-driver to work unfortunately (Hi there!) - undoubtedly my fault ...

Cheers,
   gothian

[Alexey Charkov]

2014-02-02 <st.go...@gmail.com>:

> 3. Eth: I could not get Alexeys Rhine-driver to work unfortunately (Hi
> there!) - undoubtedly my fault ...

Could you please post the .dts you use? I suspect that the problem is
somewhere around the "rhine,revision" property (I'm also currently
re-working the patch to get rid of it, as it's 0x84 for all of our
devices anyway).

Best,
Alexey

[D. J. Bernstein]

st.go...@gmail.com writes:
> - echo 6 > /sys/class/gpio/export

Hmmm. This gives me an imprecise external abort:

[ 52.350000] pinctrl-wm8850 d8110000.pinctrl: request pin 6 (extgpio6) for gpio-wmt:6
[ 52.350000] Unhandled fault: imprecise external abort (0x1406) at 0x014073ec
[ 52.350000] Internal error: : 1406 [#1] ARM
[ 52.350000] Modules linked in: uio_pdrv_genirq
[ 52.350000] CPU: 0 PID: 1608 Comm: bash Not tainted 3.13.0-rc8+ #15
[ 52.350000] task: 9ba99880 ti: 9b36a000 task.ti: 9b36a000
[ 52.350000] PC is at gpiod_get_direction+0x5c/0xa4
[ 52.350000] LR is at wmt_gpio_get_direction+0x20/0x54
[ 52.350000] pc : [<802504b0>] lr : [<8024f648>] psr: 20000013
[ 52.350000] sp : 9b36be80 ip : 9b36be68 fp : 9b36be94
[ 52.350000] r10: 9b3e2688 r9 : 9b87e908 r8 : 80854e08
[ 52.350000] r7 : 00000000 r6 : 80b1c998 r5 : 9b9e7634 r4 : 80b1c998
[ 52.350000] r3 : 00000012 r2 : 806ddbbc r1 : f8110000 r0 : 00000001
[ 52.350000] Flags: nzCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user
[ 52.350000] Control: 10c5387d Table: 1b2b8059 DAC: 00000015
[ 52.350000] Process bash (pid: 1608, stack limit = 0x9b36a238)
[ 52.350000] Stack: (0x9b36be80 to 0x9b36c000)
[ 52.350000] be80: 8024f628 20000013 9b36bebc 9b36be98 8025115c 80250460 80b1c950 00000000
[ 52.350000] bea0: 80b1c998 00000002 9b36bf78 00000000 9b36bee4 9b36bec0 80253244 80251070
[ 52.350000] bec0: 800f5b54 00000006 801fd928 802989b0 9b3e2680 00000002 9b36bef4 9b36bee8
[ 52.350000] bee0: 802989d8 80253204 9b36bf44 9b36bef8 801563e8 802989bc 9b36bf44 9b36bf08
[ 52.350000] bf00: 800f9758 800f9cd4 9b19f9c0 9b3e2140 00000000 00000000 9b36bf44 9b280480
[ 52.350000] bf20: 00000002 013ff408 9b36bf78 00000000 9b36a000 00000002 9b36bf74 9b36bf48
[ 52.350000] bf40: 800f9c10 8015628c 9b36bf74 9b36bf58 00000000 00000000 9b280480 013ff408
[ 52.350000] bf60: 00000000 00000002 9b36bfa4 9b36bf78 800f9ff0 800f9b64 00000000 00000000
[ 52.350000] bf80: 76ee35e0 00000002 013ff408 00000004 8000ebc4 00000000 00000000 9b36bfa8
[ 52.350000] bfa0: 8000e9a0 800f9fb0 76ee35e0 00000002 00000001 013ff408 00000002 00000000
[ 52.350000] bfc0: 76ee35e0 00000002 013ff408 00000004 7eaafabc 000ad08c 00000000 0141e038
[ 52.350000] bfe0: 00000002 7eaafa40 76e5296d 76e8ab9c 40000010 00000001 00000000 00000000
[ 52.350000] [<802504b0>] (gpiod_get_direction+0x5c/0xa4) from [<8025115c>] (gpiod_request+0xf8/0x160)
[ 52.350000] [<8025115c>] (gpiod_request+0xf8/0x160) from [<80253244>] (export_store+0x4c/0xd4)
[ 52.350000] [<80253244>] (export_store+0x4c/0xd4) from [<802989d8>] (class_attr_store+0x28/0x34)
[ 52.350000] [<802989d8>] (class_attr_store+0x28/0x34) from [<801563e8>] (sysfs_write_file+0x168/0x1fc)
[ 52.350000] [<801563e8>] (sysfs_write_file+0x168/0x1fc) from [<800f9c10>] (vfs_write+0xb8/0x190)
[ 52.350000] [<800f9c10>] (vfs_write+0xb8/0x190) from [<800f9ff0>] (SyS_write+0x4c/0x80)
[ 52.350000] [<800f9ff0>] (SyS_write+0x4c/0x80) from [<8000e9a0>] (ret_fast_syscall+0x0/0x30)
[ 52.350000] Code: 0a000012 e12fff32 e3500000 da000007 (e10f3000)
[ 52.350000] ---[ end trace 698e864bc310a9c9 ]---

The gpio6 directory doesn't appear, and trying the export again produces
"device busy". I've tried all pins between 0 and 7, same effect. I also
noticed "setenv wmt.gpo.wifi 6:1:6:d8110040:d8110080:d81100C0" in
various people's wmt_scriptcmd but this doesn't seem to help. Current
.config is attached.

---Dan

[Alexey Charkov]

2014-02-02 Alexey Charkov <alc...@gmail.com>:

Just in case: the updated patches are attached (not yet shared on the
mailing list).

These no longer require setting the "rhine,revision" property (only
"via,vt8500-rhine" as the 'compatible' string).

Best,
Alexey

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

Hello Alexey,

here is the dts-file I used.
Thank you for the updated patches. I will update the dts file and patch accordingly. Which kernel should I use? 3.13.1 ok?

Cheers,
   gothian

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

sorry, Dan, beats me ...
Attached you'll find the .config I am using - maybe you'll give that a spin ...

On Sunday, February 2, 2014 3:22:27 PM UTC+1, D. J. Bernstein wrote:

st.go...@gmail.com writes:
> - echo 6 > /sys/class/gpio/export

Hmmm. This gives me an imprecise external abort:

  ...

...

[andy.chernyak]

Hi Alexey,

I have tried platform rhine patches you have just sent to this group,
but can't make them work on WM8850 netbook. I am using following dts:

/include/ "wm8850.dtsi"
/ {
(...)
ethernet@d8004000 {
compatible = "via,vt8500-rhine";
reg = <0xd8004000 0x100>;
interrupts = <10>;
rhine,revision = <0x84>;
};
};

I have also tried both enabling/disabling CONFIG_VIA_RHINE_MMIO and
removing "rhine,revision" from dts. Either way, it fails with following
error at either boot (CONFIG_VIA_RHINE=y):

[ 0.720000] via-rhine: probe of d8004000.ethernet failed with error -22

... or module load time:

[ 45.110000] via_rhine: v1.10-LK1.5.1 2010-10-09 Written by Donald Becker
[ 45.110000] via-rhine: probe of d8004000.ethernet failed with error -22

Is there any way to debug this further?

Regards,
Andy.

[Alexey Charkov]

On 8 Feb 2014 17:48, "andy.chernyak" <andy.c...@gmail.com> wrote:
>
> Hi Alexey,
>
> I have tried platform rhine patches you have just sent to this group, but can't make them work on WM8850 netbook. I am using following dts:
>
> /include/ "wm8850.dtsi"
> / {
> (...)
>         ethernet@d8004000 {
>                 compatible = "via,vt8500-rhine";
>                 reg = <0xd8004000 0x100>;
>                 interrupts = <10>;
>                 rhine,revision = <0x84>;
>         };
> };

Have you tried putting the ethernet node inside the soc node? Basically, replace the "/" on your second line with "/soc/". That's my configuration at least...

As to further debugging, I can only suggest putting some printk calls in between functions within platform initialization path to determine which one is failing.

Best,
Alexey

[andy.chernyak]

Moving ethernet dt entry under soc did the trick, thanks a lot!

One small correction: "/soc/ { }" does not seem to be valid syntax, or at least I failed to make it compile with 3.14.0-rc1-git. So here is what I ended up with:

/dts-v1/;
/include/ "wm8850.dtsi"
#include <dt-bindings/pwm/pwm.h>

/ {
        model = "Wondermedia WM8850-W70v2 Tablet";

        backlight {
                compatible = "pwm-backlight";
                pwms = <&pwm 0 50000 PWM_POLARITY_INVERTED>;

                brightness-levels = <0 40 60 80 100 130 190 255>;
                default-brightness-level = <5>;
        };

        soc {

                ethernet@d8004000 {
                        compatible = "via,vt8500-rhine";
                        reg = <0xd8004000 0x100>;
                        interrupts = <10>;
                        rhine,revision = <0x84>;
                };
        };

};

&fb {                                                                          
(...)

...and it works as a charm now.

Regards,
Andy.

--

[wktin...@gmail.com]

I appled 3 patches to the kernel 3.13 of linux-vtwm
and tried compiling. Here is what I get :

  LD      vmlinux.o
  MODPOST vmlinux.o
  GEN     .version
  CHK     include/generated/compile.h
  UPD     include/generated/compile.h
  CC      init/version.o
  LD      init/built-in.o
drivers/built-in.o: In function `rhine_remove_one_pci':
/home/linux-vtwm/drivers/net/ethernet/via/via-rhine.c:2350: undefined reference to `pci_iounmap'
make[1]: *** [vmlinux] Error 1
make: *** [sub-make] Error 2

Are there newer patch versions and how/from where do I get them ?

[Alexey Charkov]

2014-02-20 15:29 GMT+04:00 <wktin...@gmail.com>:
> I appled 3 patches to the kernel 3.13 of linux-vtwm
> and tried compiling. Here is what I get :
>
> LD vmlinux.o
> MODPOST vmlinux.o
> GEN .version
> CHK include/generated/compile.h
> UPD include/generated/compile.h
> CC init/version.o
> LD init/built-in.o
> drivers/built-in.o: In function `rhine_remove_one_pci':
> /home/linux-vtwm/drivers/net/ethernet/via/via-rhine.c:2350: undefined
> reference to `pci_iounmap'
> make[1]: *** [vmlinux] Error 1
> make: *** [sub-make] Error 2

Did the patches apply cleanly? Also, which SoC are you compiling for -
is it WM8650? Guess I'd need to reproduce the behaviour on my end to
figure it out.

> Are there newer patch versions and how/from where do I get them ?

Not yet... This looks like something that may require fixing.

Best,
Alexey

[wkt]

Yes : I config-ed for the wm8650.

Yes : I repeated the 0001-patch erroneously 2 times.

         Dont know whether this does any harm.

         Need I restore the.c.orig version and repeat ?

[Alexey Charkov]

2014-02-20 16:05 GMT+04:00 wkt <wktin...@gmail.com>:
> Yes : I config-ed for the wm8650.
>
> Yes : I repeated the 0001-patch erroneously 2 times.
> Dont know whether this does any harm.
> Need I restore the.c.orig version and repeat ?

Yes, please try to start from a clean tree first of all, just in case...

Thanks,
Alexey

[wktin...@gmail.com]

As you told me i did a "git checkout <path-to-via-rhine.c>" call
and then reapplied the 3 patches.
There appeared 2 prompting messages which I answered
with default (n) - didnt understand the meaning of them...

Got this (again) after make zImage ( also when doing make clean before )

Using /home/linux-vtwm as source for kernel
  GEN     /home/wkt8650nbvtwm/build/kernel/Makefile
  CHK     include/generated/uapi/linux/version.h
  CHK     include/generated/utsrelease.h
make[2]: `include/generated/mach-types.h' is up to date.
  CALL    /home/linux-vtwm/scripts/checksyscalls.sh
  CHK     include/generated/compile.h
  CHK     kernel/config_data.h
  CC      drivers/net/ethernet/via/via-rhine.o
/home/linux-vtwm/drivers/net/ethernet/via/via-rhine.c: In function ‘rhine_init_one_pci’:
/home/linux-vtwm/drivers/net/ethernet/via/via-rhine.c:1101: warning: label ‘err_out_unmap’ defined but not used
/home/linux-vtwm/drivers/net/ethernet/via/via-rhine.c:1035: warning: unused variable ‘quirks’
/home/linux-vtwm/drivers/net/ethernet/via/via-rhine.c:1031: warning: unused variable ‘i’
  LD      drivers/net/ethernet/via/built-in.o
  LD      drivers/net/ethernet/built-in.o
  LD      drivers/net/built-in.o
  LD      drivers/built-in.o
  LINK    vmlinux

  LD      vmlinux.o
  MODPOST vmlinux.o
  GEN     .version
  CHK     include/generated/compile.h
  UPD     include/generated/compile.h
  CC      init/version.o
  LD      init/built-in.o
drivers/built-in.o: In function `rhine_remove_one_pci':
/home/linux-vtwm/drivers/net/ethernet/via/via-rhine.c:2350: undefined reference to `pci_iounmap'
make[1]: *** [vmlinux] Error 1
make: *** [sub-make] Error 2

[Alexey Charkov]

This looks like you somehow have CONFIG_PCI defined, but PCI itself is
not compiled in. Are you sure that your tree is otherwise clean?

Otherwise, pci_iounmap should be defined as a no-op stub that the
compiler should optimize away completely.

[Alexey Charkov]

Note that I've just built a zImage for 8500/8505/8650 successfully, no
linking problems. I'm using mainline git version at commit
4675348e78fab420e70f9144b320d9c063c7cee8 (dated Feb 12) with just the
Rhine patches applied.

[wktin...@gmail.com]

1. "you somehow have CONFIG_PCI defined, but PCI itself is not compiled in."

 

    How can this happen ?

 

2. "I'm using mainline git version at commit 4675348e78fab420e70f9144b320d9c063c7cee8 "

 

    Dont understand what that means nor what ( command ) I should do now.

    I am not educated in GIT.

[wktin...@gmail.com]

Meanwhile I tried and got this :

git fetch origin && git merge 4675348e78fab420e70f9144b320d9c063c7cee8

fatal: '4675348e78fab420e70f9144b320d9c063c7cee8' does not point to a commit

On the GitHub site linux-wmt/linux-vtwm I also cant find this commit hash value.

[Alexey Charkov]

I actually referred to mainline, i.e. Linus Torvalds' repo.

Best,
Alexey

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

to access "mainline", i.e. Torvalds' current repo, try the following:
 
        git clone https://kernel.googlesource.com/pub/scm/linux/kernel/git/torvalds/linux

Cheers,
   Gothian

[wktin...@gmail.com]

I now got the git source of Thorvald that Gothian wrote above.

This source compiles without error also with my 8650 version of .config

When I patch the 3 patches I again get the error mentioned before.
I set/added BTW now CONFIG_PCI=n ( correct ? )

When I say : git show 4675348e78fab420e70f9144b320d9c063c7cee8
i get the info about this commit

Must I somehow switch to this commit ? How if yes ?

[Alexey Charkov]

2014-02-24 16:39 GMT+04:00 <wktin...@gmail.com>:
> I now got the git source of Thorvald that Gothian wrote above.
>
> This source compiles without error also with my 8650 version of .config
>
> When I patch the 3 patches I again get the error mentioned before.
> I set/added BTW now CONFIG_PCI=n ( correct ? )

Any errors while patching? Can you post your .config and/or
pre-processed source for via-rhine? (make <path...>/via-rhine.i)

It's a bit difficult to help without further information ;-)

Best,
Alexey

[wktin...@gmail.com]

I think now because of error messages ( e.g. Hunk xx failed )
I cant apply the 3 patches to the Torvalds version of via-rhine.c
( only to the vtwm version ) Maybe the cause are TABs in the
patch files but BLANKS in the Torvalds version (?).

I add the 2 .i files : one before the patch the other thereafter.

I have not the faintest idea what my error is.

[mmaso...@googlemail.com]

Hi,

Was the imprecise external abort issue ever resolved? I find myself in the same situation.

Any help much appreciated.

Thanks

[Alexey Charkov]

Hi there,

What is the kernel version you are using?

I've recently rebased against current master and it seemed to work just fine on APC Rock.

Haven't yet pushed the changes to Github, as there was some breakage in serial flash due to API changes.

Cheers,

Alexey

--
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/d/optout.

[mmaso...@googlemail.com]

Thanks for the response on this 3 year old thread Alexey - much appreciated!

Using 3.13.0-rc8-37588-gc4386ef-dirty, built using the awesome instructions provided by D. J. Bernstein here:

https://groups.google.com/forum/#!topic/vt8500-wm8505-linux-kernel/rWjZu4PWmiM

Looking at dmesg after booting into Android tells me I have a Realtek RTL8188EUS (0bda:8179) on board. Any recommendations you have for getting this up and running will be warmly received.

[Alexey Charkov]

2017-06-27 19:14 GMT+03:00 mmason75123 via VT8500/WM8505 Linux Kernel
<vt8500-wm8505...@googlegroups.com>:

> Thanks for the response on this 3 year old thread Alexey - much appreciated!
>
> Using 3.13.0-rc8-37588-gc4386ef-dirty, built using the awesome instructions provided by D. J. Bernstein here:
>
> https://groups.google.com/forum/#!topic/vt8500-wm8505-linux-kernel/rWjZu4PWmiM

Try the testing-alchark branch from Github - just in case. That is if
you have GCC 5 or earlier around to build it (I didn't, so had to
rebase).

Will only be able to access my machine where the repo is sometime
during the weekend.

> Looking at dmesg after booting into Android tells me I have a Realtek RTL8188EUS (0bda:8179) on board. Any recommendations you have for getting this up and running will be warmly received.

Last time I checked, there was a driver in staging for that one, and
it worked for me. Though you still need to do the usual tricks with
GPIO pins to get it to power on - try exporting the first dozen via
sysfs and toggling them to 0 and 1, and watching lsusb. Looking at
u-boot environment might also help identify which one you need.

Cheers,
Alexey

[mmaso...@googlemail.com]

Brilliant, thanks again - I'll give the other branch a go and see if it sorts out the GPIO issues. After that I'm sure there will be some fun finding a working driver!

Kind regards

[mmaso...@googlemail.com]

Can confirm building on the new branch has solved the GPIO issue.

lsusb is now showing:

Bus 001 Device 003: ID 0bda:8179 Realtek Semiconductor Corp

Having a bit of trouble compiling a driver for it... but I'll try and work through this before taking up any more of your time.

Thanks!