|Latest product reviews|
Which tool do I need to use to flash the single huge stock recovery img:
ST10216-2A_V05.img - 495 021 284 bytes
This must not be done with RKAndroidTool, right!?
I don't even know how to do this with my stock recovery partition, as it seems to expect the img file to reside in the external SD, but have a specific name in order to find it...!?
Sorry for my newbie questions...
Will also provide dmesg and logcat output for the Bluetooth chip once I have restored the correct (non-standard) MTD layout and reinstalled CM 10.1...
No, do not change anything. Goto here: http://crewrktablets.arctablet.com/?p=1913
Below the sensor fixes the generic stock recovery is linked. Just change it in folder rockdev/Images. Rename old recovery.img in CWM.img before. After booting into Android flash CWM, just tick parameter and recovery. You can load CWM.img by clicking on the empty box at the end of the kernel line.
Pls notice the new model fix in post#15.
- Oma -
The following users say thank you to Oma7144 for this useful post:awl
Thanks a million for your kind help!
OK, I noticed that the new "model fix" patch now additionally includes a Bluetooth kernel module! Great!
In the meantime, I have successfully recovered my old proper MTD partition setup by flashing the stock img (ST10216-2A_V05.img, see above) using RKBatchTool 1.8.
So I will now (maybe tomorrow, as it's getting close to midnight...) try to install CM 10.1 again from step 1, but keeping the stock recovery img. Hopefully this will keep the Ventos MTD layout - otherwise now I at least know how to restore it...
Thanks again & have a good night!
bad news: With the updated model fix from post 15, things are even worse: Neither WLAN nor Bluetooth now works, and we have a lot of error messages about OutOfMemory and OpenGL renderer in the catlog.
For WLAN, the issue seems to be:
02-13 19:14:28.740 E/wpa_supplicant(1815): wlan0: Unsupported driver 'wext'
for Bluetooth, Idn't even find any error messages, but it's simply not working...
Please find both catlog and dmesg output with this setup attached.
As I will only have time to look into this again after the weekend, please take your time - no need to hurry. And in case it looks like there is no easy remedy, I think we'll simply stay with the stock 4.1.1 ROM for this device...
Thanks again & best regards,
Ok, flash the temp model fix again and let me have the logs.
- Oma -
final update from me before the weekend...
I have finally managed to create dmesg and catlog output after trying to activate Bluetooth: see attachments.
This has been quite difficult, as typically, just a few seconds after trying to unsuccessfully activate Bluetooth for the first time, the kernel crashes: The tablet becomes hung/unresponsive, and I get kicked out of "adb shell" with this message:
error: protocol fault (no status)
Also, I noticed that with the original stock Android 4.1.1, the WLAN driver (compiled statically into the kernel) states in dmesg log:
MT5931 SDIO WiFi driver (Powered by Rockchip,Ver 2.07) init.
while with your CM 10.1 (but the exact same 3.0.8+ kernel), the WLAN driver states on init (from mt5931.ko):
MT5931 SDIO WiFi driver (Powered by Rockchip,Ver 1.08) init.
Many thanks again in advance & have a nice weekend,
Hm, wifi driver inline ... never seen this on a rk30board.
Flash system.img again (untick all other), so we have a clean start. Download again and try: http://crewrktablets.arctablet.com/?wpfb_dl=2450
- Oma -
Bad news: There are even two issues that need to be resolved with the most recent model fix from post #22:
(1) Even though the MT5931 driver must be included statically in the kernel:
in the kernel config, and I do not have any mt5931.ko on Trekstor's Android 4.1.1 stock ROM, with the most recent model fix on CM10.1, I still cannot activate WLAN either.
(2) Also, Bluetooth still cannot be activated. I think that the model fix still contains a wrong Bluetooth driver. While you seem to have included Bluetooth drivers for MT6622 (system\etc\firmware\MTK_MT6622_E2_Patch.nb0 and libbluetooth_mtk.so), the Ventos tablet has BCM4329 as its Bluetooth chip:
in the kernel config file and the dmesg file for the Trekstor 4.1.1 stock ROM says:
[BT_RFKILL]: bcm4329 module has been initialized,rc=0x0
This also means that the stock kernel should already include the Bluetooth driver (as well as the WLAN driver). Still, CM10.1 does not seem to detect both drivers...!?
Do you have any idea why neither WLAN nor Bluetooth do work with the CM10.1 ROM even though the kernel should include the drivers?
In the meantime, I try to set up a 3.0.8 kernel compile environment for the source provided by Trekstor and will try to compile both WLAN and Bluetooth as kernel modules. I'll report back on my success...
Many thanks one more time!
OK - now it's even worse: I fear that I have bricked my tablet...
And I was so happy, because I had succeeded to recompile the kernel from the source as provided by Trekstor and had created a kernel.img file from this kernel Image as well as the missing mt5931.ko module in version 2.07.
But after I flashed this new kernel.img (created by me) into the tablet, I got fatally stuck:
The tablet now does no longer boot in normal mode, but only in recovery mode, and unfortunately, I currently don't have CWM on the recovery partition, but only the stock recovery image with its limited capabilities...
I have tried to use "update rkimage from external storage" with the ST10216-2A_V05.img (renamed to update.img in the root directory of the external SD card), but no luck:
"Find and check firmware... Cannot found firmware image or invalid image. Update rkimage complete."
Is there a way to enter bootloader mode using the hard keys on the Ventos (I only know how to enter the recovery partition, which seems unable to recover from this mistake...)? If not, then I am really horribly stuck...
Do you have any idea whether it is possible to revive this tablet, or can I now dispose of it as it is unrecoverably bricked now?
If you are able to help me revive my tablet, I promise you will receive a well-deserved first donation...
Thanks a million in advance!
You need to open the tab to start it in mask-rom mode. In principle as described here:
- Oma -
Sorry, that won't work: I cannot open the Trekstor without breaking it - it is completely sealed...
But I seem to have found the solution: It turns out that the device now registers in Windows with device ID 18d1:d001, and I was missing the driver for this, but I managed to override Windows to accept the generic driver from the Android SDK.
So I can now connect via adb to my device when in stock recovery mode - this did work before, but with a Trekstore USB vendor ID...
Seems that I'll be back on track shortly...
And bad luck one more time...
Now CWM recovery is back in my recovery partition, but the tablet is only recognized by Windows as "mass storage", such that I am both unable to connect with adb any more, and also still unable to boot into a proper kernel (even though I thought that I had written the original working kernel back).
Is there any chance that I can use a CWM updater script to update the kernel partition?
Currently trying to do something like this:
but it looks like this does not work for whatever reason.
What is the correct "partition" name to be used as the second parameter to write_raw_image()?
Doesn't work this way. At least your kernel does crash your power management chip, which looks for key combinations. Remember what I said about corrupted sources. The kernel in recovery is also gone now.
I don't believe that the tab is sealed, because this would increase production costs. Use a scalpel or an old sharpened credit card to open it.
- Oma -
OK, so finally, I'm back on track without opening the tablet:
What was needed was to clean up, deinstall and reinstall the Windows adb drivers in order to deal with the perturbing number of different device IDs that stock ROM, stock recovery, CWM recovery, CWM ROM and bootloader are using: a nightmare! The Windows tool "USBDeview" was a tremendous help with this - otherwise I would have failed...
But with the correct driver, I was able to connect again with adb while CWM recovery was running, and switch the device back to bootloader mode using "adb reboot recovery" in oder to flash again...
So I have now tried once again to install the kernel that I had successfully compiled before, and this time, I succeeded. I still don't know what went wrong when I tried this for the very first time, but now it worked anyway.
In the meantime, I have found out that indeed from booting with the stock 4.1.1 ROM and examining locgcat output that indeed, my Bluetooth chip is MT6622 and not BCM432x, even though there also is a log entry stating that the tablet also seemingly successfully activates BCM4329, but it doesn't really do this...
I am now all set for further experiments with CM10.1: Even though I am now able to manually successfully insmod mt5931.ko (in my newly compiled 2.07 version) as well as the old wlan.ko from the stock ROM successfully on top, I have not yet managed to activate WLAN again using this configuration. Maybe I need to rebuild wlan.ko myself as well...!?
And you are indeed correct in terms of the vendors providing corrupt kernel sources, or - as in my case - kernel sources that only compile without changes on a particular gcc 4.3 version, but fail with any later gcc...
Let's see whether I will be able to sort something out - I will try to do so for some more hours until I either succeed or consider this an interesting, but frustrating and futile experience. I will keep this thread updated...
Thanks anyway for your help & best regards,
Hello again, Oma,
do you already know this git repository for a tablet-centric rk3066 3.0.36 kernel?
Starting from this source tree, and using "my" Ventos 10.1 config file from 3.0.8 (and adding selections for new modules using "make oldconfig"), I indeed got a compile-clean 3.0.36 kernel for RK3066 relatively quickly, but unfortunately, it does not (yet) link properly, but "only" gives two undefined symbols:
arch/arm/mach-rk30/built-in.o: In function `machine_rk30_board_init':
:(.init.text+0x1318): undefined reference to `__rk_platform_add_display_devices'
drivers/built-in.o: In function `at18_probe':
:(.devinit.text+0x2178): undefined reference to `kmem_cache_alloc_trace'
Do you have any idea where these missing symbold could be found?
Would it be worthwhile that I try and start looking deeper into this as a starting point?
Thanks again & best regards,
Most Users Ever Online: 749
Currently Online: fzsman
Currently Browsing this Page:
Devices in use: Desktop (74), Phone (10)
Guest Posters: 43
Moderators: globula_neagra, exelletor, JochenKauz, Oma7144, cracktech