kernel-4.4 branch is ready

[Chih-Wei Huang]

Hi all,
I've pushed the kernel-4.4 branch into the SF git server.
It contains:

* AOSP experimental/android-4.4 branch
for Android specific features
* Intel drm-intel-next-queued branch
to support newer chips
(test ok on Surface 3)
* Some patches from LKML to support newer
peripherals or fix bugs.
(thanks to Peter's mint-17.3-for-surface-pro-3)
* Rob Herring's virtio-gpu patches
(I didn't tested yet)
* Other older patches applied by android-x86

Known issues:
* uvesafb (v86d) is broken -- not sure it's due to
the kernel 4.4 regression or missing necessary configs

Please help to test it:
(on the top of android-x86 tree)

cd kernel
git fetch x86 kernel-4.4
git checkout x86/kernel-4.4
cd ..

Then build the android-x86 image as usual.

Refer:
https://sourceforge.net/p/android-x86/kernel/ci/kernel-4.4/tree/

--
Chih-Wei
Android-x86 project
http://www.android-x86.org

[Jaap Jan Meijer]

Hi Chih-Wei,

I tested your work on my HP Omni10 (Bay Trail tablet) and it works great, thank you again for your great work!

I attached some Bluetooth patches to add missing ACPI ID's for bcm43241 chips and two fixes to get IOCTL working in mixed arch situations. Also there is a fw package I extracted from the Win10 bcm driver that should go into /lib/firmware/brcm. After that we could start using btattach for UART BT devices. Do you agree with this? Should I push to Sourceforge?


The biggest problem I was having with vanilla 4.4 where MMC issues even with a patch set from Adrian Hunter, I must have messed up somewhere.  I'm running your kernel now with Android from internal eMMC and SD-card and haven't encountered any I/O errors so far.
One more strange problem, on my tablet with brcmfmac these is no WiFi strength information from the connected network, other networks do have information. On my laptop with iwlwifi there are no problems. It started on kernel 4.2, not sure yet what thats about.

[spin877]

thanks great job.

I would like to report a problem 

1) Mouse does not work  vmware-Workstation 12 pro

2)  with the firmware of the network adapter rtl8723bs on (Bay Trail tablet) 

[  121.866155] init: Starting service 'wpa_supplicant'...

[  121.967330] rtl8723bs: accquire FW from file:rtlwifi/rtl8723bs_nic.bin

[  121.974317] rtl8723bs mmc1:0001:1: Direct firmware load for rtlwifi/rtl8723bs_nic.bin failed with error -2

[  121.974329] Request firmware failed with error 0xfffffffe

[  121.978213] init: Service 'wpa_supplicant' (pid 3334) exited with status 255

[  121.978237] init: Service 'wpa_supplicant' (pid 3334) killing any children in process group

[  126.980853] init: Starting service 'wpa_supplicant'...

[  127.018092] rtl8723bs: accquire FW from file:rtlwifi/rtl8723bs_nic.bin

[  127.018126] rtl8723bs mmc1:0001:1: Direct firmware load for rtlwifi/rtl8723bs_nic.bin failed with error -2

[  127.018131] Request firmware failed with error 0xfffffffe

[  127.021275] init: Service 'wpa_supplicant' (pid 3336) exited with status 255

[  127.021293] init: Service 'wpa_supplicant' (pid 3336) killing any children in process group

[  132.027654] init: Starting service 'wpa_supplicant'...

[  132.054851] rtl8723bs: accquire FW from file:rtlwifi/rtl8723bs_nic.bin

[  132.054885] rtl8723bs mmc1:0001:1: Direct firmware load for rtlwifi/rtl8723bs_nic.bin failed with error -2

[  132.054891] Request firmware failed with error 0xfffffffe

[  132.058628] init: Service 'wpa_supplicant' (pid 3342) exited with status 255

[  132.058691] init: Service 'wpa_supplicant' (pid 3342) killing any children in process group

[  142.003407] init: Starting service 'wpa_supplicant'...

[  142.038020] rtl8723bs: accquire FW from file:rtlwifi/rtl8723bs_nic.bin

[  142.038070] rtl8723bs mmc1:0001:1: Direct firmware load for rtlwifi/rtl8723bs_nic.bin failed with error -2

[  142.038075] Request firmware failed with error 0xfffffffe

[  142.041793] init: Service 'wpa_supplicant' (pid 3346) exited with status 255

[  142.041856] init: Service 'wpa_supplicant' (pid 3346) killing any children in process group

[Jaap Jan Meijer]

Op woensdag 27 januari 2016 16:58:30 UTC+1 schreef Jaap Jan Meijer:

The biggest problem I was having with vanilla 4.4 where MMC issues even with a patch set from Adrian Hunter, I must have messed up somewhere.  I'm running your kernel now with Android from internal eMMC and SD-card and haven't encountered any I/O errors so far.

I'm sorry, I spoke to soon. The MMC situation is slightly improved with the Hunter patches but still the complete MMC bus freezes up after a while. First ADMA errors and than I/O errors and timeouts. Kernel 4.0.9 seems to do work better with MMC.


I'm now testing some workarounds. With DMA (sdhci.debug_quirks=0x40) seems to prevent freezing. Also I read about disabling 1.8V mode, will try that as well.

[Mario Holzinger]

Tested marshmallow 32bit with kernel 4.4 (onńly include my kernel patch for GSL1680 I2C touchscreen support)

sorry can not sync lollipop sources atm, drm cound not be pulled from git repo

On my "Microsoft Surface Pro 2" Wifi still mentioned "Autentication failed" so no wireless connection can be established (works with JJ Meijer's kernel repo) ==> Marvell 88W8797 (mwifiex)

On my "TrekStor Surftab Wintron 7.0" Wifi with RTL8723BS wirks well (also my GSL1680 Touchscreen I've patched)

I know marshmallow is not the best to report back but it is confusing why Marvell 88W8797 wont work. it seems that ther is someting in android-x86 kernel that breaks my Wifi.

I I find time between getting my 7 inch Tablet working better (need to implement nearly everry HW support on my own :-) ) I will share my sollution

thanks from my side

Mario

[Povilas Staniulis]

Error -2 is ENOENT (aka. File not found).
Do you have rtlwifi/rtl8723bs_nic.bin in /lib/firmware ?

--
You received this message because you are subscribed to the Google Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send an email to android-x86...@googlegroups.com.
To post to this group, send email to andro...@googlegroups.com.
Visit this group at https://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

[Povilas Staniulis]

From what I read on the T100 group, kernel 4.4 still freezes with MMC errors.
Some devs there suggest disabling intel cstates on Baytrail to improve stability: intel_idle.max_cstate=0 (or 1)
This does increase power drain significantly though.

It seems that cstates somehow mess up the MMC bus...

[spin877]

Il giorno giovedì 28 gennaio 2016 01:54:10 UTC+1, Povilas Staniulis ha scritto:

Error -2 is ENOENT (aka. File not found).
Do you have rtlwifi/rtl8723bs_nic.bin in /lib/firmware ?

 I've added and everything is back to work

[Chih-Wei Huang]

2016-01-27 23:58 GMT+08:00 Jaap Jan Meijer <jjmei...@gmail.com>:
> Hi Chih-Wei,
>
> I tested your work on my HP Omni10 (Bay Trail tablet) and it works great,
> thank you again for your great work!
>
> I attached some Bluetooth patches to add missing ACPI ID's for bcm43241
> chips and two fixes to get IOCTL working in mixed arch situations. Also
> there is a fw package I extracted from the Win10 bcm driver that should go
> into /lib/firmware/brcm. After that we could start using btattach for UART
> BT devices. Do you agree with this? Should I push to Sourceforge?

Looks good. Please push them.

About the 0001-Add-missing-ACPI-IDs-for-bcm43241.patch,
has it been submitted to the upstream?

> The biggest problem I was having with vanilla 4.4 where MMC issues even with
> a patch set from Adrian Hunter, I must have messed up somewhere. I'm
> running your kernel now with Android from internal eMMC and SD-card and
> haven't encountered any I/O errors so far.

I haven't made a lot of testing yet.

> One more strange problem, on my tablet with brcmfmac these is no WiFi
> strength information from the connected network, other networks do have
> information. On my laptop with iwlwifi there are no problems. It started on
> kernel 4.2, not sure yet what thats about.

[Chih-Wei Huang]

2016-01-28 4:27 GMT+08:00 Mario Holzinger <sandma...@gmail.com>:
> Tested marshmallow 32bit with kernel 4.4 (onńly include my kernel patch for
> GSL1680 I2C touchscreen support)
> sorry can not sync lollipop sources atm, drm cound not be pulled from git
> repo
>
> On my "Microsoft Surface Pro 2" Wifi still mentioned "Autentication failed"
> so no wireless connection can be established (works with JJ Meijer's kernel
> repo) ==> Marvell 88W8797 (mwifiex)

Hmm, what patches does it have?

> On my "TrekStor Surftab Wintron 7.0" Wifi with RTL8723BS wirks well (also my
> GSL1680 Touchscreen I've patched)

Please submit the patch.

> I know marshmallow is not the best to report back but it is confusing why
> Marvell 88W8797 wont work. it seems that ther is someting in android-x86
> kernel that breaks my Wifi.
> I I find time between getting my 7 inch Tablet working better (need to
> implement nearly everry HW support on my own :-) ) I will share my sollution

[Chieh-Min Wang]

I was trying to build the 4.4 kernel, however, I encountered a build error. Any idea how can I resolve this?

Thanks in advance :).

[Mario Holzinger]

Am Donnerstag, 28. Januar 2016 03:38:08 UTC+1 schrieb Chih-Wei Huang:

2016-01-28 4:27 GMT+08:00 Mario Holzinger <sandma...@gmail.com>:
> Tested marshmallow 32bit with kernel 4.4 (onńly include my kernel patch for
> GSL1680 I2C touchscreen support)
> sorry can not sync lollipop sources atm, drm cound not be pulled from git
> repo
>
> On my "Microsoft Surface Pro 2" Wifi still mentioned "Autentication failed"
> so no wireless connection can be established (works with JJ Meijer's kernel
> repo) ==> Marvell 88W8797 (mwifiex)

Hmm, what patches does it have?

> On my "TrekStor Surftab Wintron 7.0" Wifi with RTL8723BS wirks well (also my
> GSL1680 Touchscreen I've patched)

Please submit the patch.

My tests are with no modifications except the integration of https://github.com/onitake/gslx680-acpi for my tablet touchcontroller

 

> I know marshmallow is not the best to report back but it is confusing why
> Marvell 88W8797 wont work. it seems that ther is someting in android-x86
> kernel that breaks my Wifi.
> I I find time between getting my 7 inch Tablet working better (need to
> implement nearly everry HW support on my own :-) ) I will share my sollution

 

I do not want to provide a patch yet as long as it is such a dirty one like it is at the moment

 

Integration of the complete Github repo into drivers/staging + adapt driver/staging/Kconfig, driver/staging/Makefile and enable the module to build (as module) during complete clean android build

I need to place a firmware extracted from the windows Touchscreen driver (tools included in onitake's repo)

Autostart kernel module at boot time with init.sh

regards Mario 

[Mario Holzinger]

only for testing here is the Patch to implement GSLx680 touch controllers.

but be aware that it is only working if the correkt firmware is aso included in device/generic/common/firmware and if tha module is loaded manually or by startup script

The patch has included my firmware for Trekstor Surftab Wintron 7.0 extracted from the Windwos 10 driver firmware

it is only a alpha version because I don't know at the moment how I shell calibrate the touchscreen.

Maybe I find a way to hack the firmware to ready to use. (that would be the best sollution)

under ubuntu I used xinput calibrator but on android the built in calibration do not work for me atm.

[Chih-Wei Huang]

Thank you for the patch.
However it can't be applied to kernel-4.4 cleanly.

Please provide a patch for kernel-4.4.

[zhongtian xie]

maybe lack of bcm wireless driver, Generally it download automatic ， but sometimes maybe something wrong , and you can download it to kernel/drivers/net/wireless/wl/,

vim kernel/drivers/net/wireless/wl/build.mk

=========================================================================================================================
# parts of build/core/tasks/kernel.mk

WL_ENABLED := $(if $(wildcard $(WL_PATH)),$(shell grep ^CONFIG_WL=[my] $(KERNEL_CONFIG_FILE)))
WL_ARCH_CHANGED := $(if $(shell file $(WL_PATH)/lib/wlc_hybrid.o_shipped | grep -s $(if $(filter x86,$(TARGET_KERNEL_ARCH)),80386,x86-64)),,FORCE)
WL_SRC := $(WL_PATH)/hybrid-v35$(if $(filter x86,$(TARGET_KERNEL_ARCH)),,_64)-nodebug-pcoem-6_30_223_271.tar.gz
$(WL_SRC):
        @echo Downloading $(@F)...
        $(hide) curl http://www.broadcom.com/docs/linux_sta/$(@F) > $@

$(WL_PATH)/Makefile : $(WL_SRC) $(wildcard $(WL_PATH)/*.patch) $(WL_ARCH_CHANGED) $(KERNEL_ARCH_CHANGED)
        $(hide) tar zxf $< -C $(@D) --overwrite && \
                patch -p5 -d $(@D) -i wl.patch && \
                patch -p1 -d $(@D) -i linux-recent.patch

$(INSTALLED_KERNEL_TARGET): $(if $(WL_ENABLED),$(WL_PATH)/Makefile)

==========================================================================================================================

--

[Chieh-Min Wang]

Thanks, I was using the old build/core/tasks/kernel.mk for kernel 3.18, and it always downloads the old version of bcm driver that's what cause the problem.

[Chieh-Min Wang]

I build kernel 4.4 for android 4.4, however, it failed to start. Can I use kernel 4.4 for android 4.4?

On Wednesday, January 27, 2016 at 5:02:00 PM UTC+8, Chih-Wei Huang wrote:

[Jaap Jan Meijer]

Hi Chih Wei,

I just noticed 32-bit binder is not enabled is not enabled for x64 kernel. Did you maybe forget it or do you plan to switch to x64 user space as well? I haven't tried x64 user space myself, is it usable?

[Jaap Jan Meijer]

> I attached some Bluetooth patches to add missing ACPI ID's for bcm43241
> chips and two fixes to get IOCTL working in mixed arch situations. Also
> there is a fw package I extracted from the Win10 bcm driver that should go
> into /lib/firmware/brcm. After that we could start using btattach for UART
> BT devices. Do you agree with this? Should I push to Sourceforge?

Looks good. Please push them.

About the 0001-Add-missing-ACPI-IDs-for-bcm43241.patch,
has it been submitted to the upstream?

The ioctl stuff is pushed upstream now. For the missing ACPI ID's I still have to create a patch they will be able to apply. During my quest I noticed there are a lot of missing ID's so I hope they will just push it at once or find a better solution, some have been pushed already.

http://www.driveridentifier.com//scan/driver_file_detail.php/driver-detail/262C118A233E4B45BC8EB76D27B63F9A/1811458/md5/1583152406/ACPI%5CBCM2E54?scanid=262C118A233E4B45BC8EB76D27B63F9A&inf_file_id=1811458&md5=md5&item_id=1583152406&hardware_id=ACPI\BCM2E54&friendly_pc_model=broadcom-serial-bus-driver-over-uart-bus-enumerator


 

> The biggest problem I was having with vanilla 4.4 where MMC issues even with
> a patch set from Adrian Hunter, I must have messed up somewhere.  I'm
> running your kernel now with Android from internal eMMC and SD-card and
> haven't encountered any I/O errors so far.

I haven't made a lot of testing yet.

Alright, I hope you will find some time to test, I'll try get some time to do some testing myself.

[Chih-Wei Huang]

2016-01-30 21:38 GMT+08:00 Jaap Jan Meijer <jjmei...@gmail.com>:
> Hi Chih Wei,
>
> I just noticed 32-bit binder is not enabled is not enabled for x64 kernel.

We don't need it.

To be clearer, that patch is only used to support
32-bit userspace without binder version 8,
i.e., android version prior to kitkat (included).

Since I have no plan to use kernel-4.4 with
kitkat-x86, I'm not going to apply that patch.

> Did you maybe forget it or do you plan to switch to x64 user space as well?

Already have -- the android_x86_64 target.

> I haven't tried x64 user space myself, is it usable?

We have fixed several critical 64-bit issues.
(thanks to the patches of Paulo and Jide)
IMO it's quite stable now.
Surely we will continue fixing 64-bit issues
if more are found.

[Jaap Jan Meijer]

Op zo 31 jan. 2016 16:19 schreef Chih-Wei Huang <cwh...@android-x86.org>:

>
> 2016-01-30 21:38 GMT+08:00 Jaap Jan Meijer <jjmei...@gmail.com>:
> > Hi Chih Wei,
> >
> > I just noticed 32-bit binder is not enabled is not enabled for x64 kernel.
>
> We don't need it.
>
> To be clearer, that patch is only used to support
> 32-bit userspace without binder version 8,
> i.e., android version prior to kitkat (included).
>
> Since I have no plan to use kernel-4.4 with
> kitkat-x86, I'm not going to apply that patch.
>

Thanks for the explanation. I'm still on 32 bit user space for my cm builds for which I do need to enable it. The only reason I switched to x64 kernel is boot time on by tablet (for some reason there is a long delay with x86).

> > Did you maybe forget it or do you plan to switch to x64 user space as well?
>
> Already have -- the android_x86_64 target.
>
> > I haven't tried x64 user space myself, is it usable?
>
> We have fixed several critical 64-bit issues.
> (thanks to the patches of Paulo and Jide)
> IMO it's quite stable now.
> Surely we will continue fixing 64-bit issues
> if more are found.
>

Sounds good, I'll do some testing as soon as I get cm buildable from the Android-x86 repo.

[Chih-Wei Huang]

2016-01-31 23:57 GMT+08:00 Jaap Jan Meijer <jjmei...@gmail.com>:
>>
>> To be clearer, that patch is only used to support
>> 32-bit userspace without binder version 8,
>> i.e., android version prior to kitkat (included).
>>
>> Since I have no plan to use kernel-4.4 with
>> kitkat-x86, I'm not going to apply that patch.
>
> Thanks for the explanation. I'm still on 32 bit user space for my cm builds
> for which I do need to enable it. The only reason I switched to x64 kernel
> is boot time on by tablet (for some reason there is a long delay with x86).

No, you don't need it unless you're still
working on kitkat. Are you?

For lollipop or later, what you need is
TARGET_USES_64_BIT_BINDER := true
for 64-bit kernel.

[Milan Parikh]

Any chance you can post your build for the Surface 3 (non-pro)? I don't have a Linux machine to build it myself, or I wouldn't bother you.

[Chih-Wei Huang]

2016-02-01 6:49 GMT+08:00 Milan Parikh <mila...@gmail.com>:
> Any chance you can post your build for the Surface 3 (non-pro)? I don't have
> a Linux machine to build it myself, or I wouldn't bother you.

Try android-x86-6.0_20160129.iso
https://sourceforge.net/projects/android-x86/files/Testing/

[Milan Parikh]

Thank you! I'll try it later today.

[Chris Rogers]

There seems to be a regression on he Dell Venue 8 Pro. The power button is no longer recognized at all. Could be an IRQ conflict? Also, is the sleep behavior set to mem or freeze? Baytrail doesn't have mem support, so it wouldn't properly sleep. I suppose the real solution to this is to build for my own device.

[JJ Meijer]

Hi,

For lollipop or later, what you need is
TARGET_USES_64_BIT_BINDER := true
for 64-bit kernel.

That worked, thanks!

There seems to be a regression on he Dell Venue 8 Pro. The power button is no longer recognized at all. Could be an IRQ conflict? Also, is the sleep behavior set to mem or freeze? Baytrail doesn't have mem support, so it wouldn't properly sleep. I suppose the real solution to this is to build for my own device.

Can you please supply output of "cat /proc/interrupts"? Is pata_legacy in there? If so you have the same IRQ conflict as on my HP Omni10. Chih-Wei, can't we just remove this obsolete (I think) module? Otherwise we can just blacklist it I guess.
 

[Jianlong Liu]

From my brief experiment with 4.5 RC2 (Mauro's branch), it's been fixed (I think, pretty sure I didn't comment the module out). Some other things weren't working though, so I went back to 4.4.

[Jianlong Liu]

And yes, it's the same IRQ conflict (I checked a while ago and removed the module).

On Monday, February 1, 2016 at 5:29:42 PM UTC-5, JJ Meijer wrote:

[Chih-Wei Huang]

2016-02-02 6:29 GMT+08:00 JJ Meijer <jjmei...@gmail.com>:
>
> Can you please supply output of "cat /proc/interrupts"? Is pata_legacy in
> there? If so you have the same IRQ conflict as on my HP Omni10. Chih-Wei,
> can't we just remove this obsolete (I think) module? Otherwise we can just
> blacklist it I guess.

OK, let's remove it if nobody needs it.

[Daniel N]

kernel4.4 work great for my t90chi.

Finally it is able to boot up, wifi, blacklight, touchscreen is working

bt, power button, sound is not working.

and sometimes it will be freeze.

Besides, the touchscreen will not work after will times reboot.

Chih-Wei Huang於 2016年2月2日星期二 UTC+8下午7時10分19秒寫道：

[spin877]

missing firmware rtlwifi/rtl8723bs_nic.bin

I added in the kernel module

try adding these two files in

/drivers/staging/rtl8723bs/hal/

[Chih-Wei Huang]

The firmware should not be put into the kernel directly.

I've pushed the rtl8723bs firmware into
the firmware dir. Please sync to get it.

[spin877]

Il giorno martedì 16 febbraio 2016 07:30:20 UTC+1, Chih-Wei Huang ha scritto:
> 2016-02-16 14:15 GMT+08:00 spin877 <sfuma...@gmail.com>:
> > missing firmware rtlwifi/rtl8723bs_nic.bin
> > I added in the kernel module
> >
> > try adding these two files in

Ok.

Ok . But I have just imitated the "master " you. joke :-)
Look at version 4.0.9

https://sourceforge.net/p/android-x86/kernel/ci/android-4.0/tree/drivers/staging/rtl8723bs/hal/HalHWImg8723B_FW.c

[Mario Holzinger]

At first thx for all your great work.

I'v tested most of my devices with kernel-4.4 and whant to share what I need to patch so far

Microsoft Surface Pro 2

- seems to wotk out of the box great (also my patch that was allready integreated for wifi is working great)

- bluetooth Yes

- rotation Yes

- Type Cover Yes

...

Trekstore Surtab Wintron 7.0

- Wifi RTL8723BS updated sources from github (kernel) and add binaries in (device/generic/common/firmware/rtlwifi) ==> see patches

- Touchscreen GSL1680 I2C is working with integrating the acpi kernel module ==> see patches

- rotation is working out of the box but is misaligned 90° clockwise ==> need to investigate further where to add on offset for my tablet

[spin877]

For rotation 90° i modified

https://sourceforge.net/p/android-x86/hardware_libsensors/ci/kitkat-x86/tree/iio-sensors.cpp

Line 244 245 to



data[i].acceleration.y = -scale * read_sysfs_int("in_accel_x_raw");
data[i].acceleration.x = scale * read_sysfs_int("in_accel_y_raw");

Swap x-y If it does not work try -scale to scale

[Hypo Turtle]

Patch for 180 rotation here: http://forum.xda-developers.com/showpost.php?p=65181882&postcount=2549

If you take spin877's notes and add in a prop option for 90 and 270 I guess this can be added upstream.

[Cemehtypnahov]

that if you make an application requires Root rights, which will replace the patch for rotation, in the system folder? because ordinary users can not be integrated this file in the system.


среда, 27 января 2016 г., 17:02:00 UTC+8 пользователь Chih-Wei Huang написал:

[Adrian Elston]

AMD KABINI SUPPORT????????????

[Mario Holzinger]

Thanks again, for all the tips.

here is a patch for hardware/libsesnors/ to xchange all possibilities of axis and invert all axis

with this I can mod /system/build.prop (needs root access)

to support my tablets accelerometers perfect

default is untoched as it was before :-)

best regards

Mario

[Mario Holzinger]

I know I'm not an professional coder, but maybe some of you can optimize the code a bit.

this patch + the 2 lines in /system/build.prop

hal.sensors.iio.accel.axis=1

hal.sensors.iio.accel.inv=7

makes my Trekstor Surftab Wintron 7.0 tablet act as expected :-)

[Chih-Wei Huang]

2016-02-20 20:05 GMT+08:00 Mario Holzinger <sandma...@gmail.com>:
> At first thx for all your great work.
>
> I'v tested most of my devices with kernel-4.4 and whant to share what I need

> to patch so far0001-add-binaries-for-RTL8723BS-Wifi-BT-Chipset.patch

> Microsoft Surface Pro 2
> - seems to wotk out of the box great (also my patch that was allready
> integreated for wifi is working great)
> - bluetooth Yes
> - rotation Yes
> - Type Cover Yes
> ...
>
> Trekstore Surtab Wintron 7.0
> - Wifi RTL8723BS updated sources from github (kernel) and add binaries in
> (device/generic/common/firmware/rtlwifi) ==> see patches

The device/generic/common part has already merged.

> - Touchscreen GSL1680 I2C is working with integrating the acpi kernel module
> ==> see patches

Again, the device/generic/common part has contained
patches you need. Please sync your codebase before
preparing the patches.

I forgot the kernel part patch.
Will cherry-pick from kernel-4.0 later.

Anyway, thanks for the patches.

> - rotation is working out of the box but is misaligned 90° clockwise ==>
> need to investigate further where to add on offset for my tablet

[Chih-Wei Huang]

2016-02-22 2:54 GMT+08:00 Mario Holzinger <sandma...@gmail.com>:
> Thanks again, for all the tips.
>
> here is a patch for hardware/libsesnors/ to xchange all possibilities of
> axis and invert all axis

Thanks a lot for the patch.

I would suggest you implement it in a 3x3 conversion matrix
in_accel_{x,y,z}_raw => acceleration.{x,y,z}

The conversion matrix could be described by a property
containing 9 elements of array, say

hal.sensors.iio.accel.matrix=1,0,0,0,1,0,0,0,1

> with this I can mod /system/build.prop (needs root access)
> to support my tablets accelerometers perfect
>
> default is untoched as it was before :-)

[Chih-Wei Huang]

Since Google has pushed an official porting
of kernel 4.4 to the android-4.4 branch,
I've re-merged and rebased our branch to it.

Be aware of the force update on repo sync
and rebase your local patches if any.

[Chih-Wei Huang]

2016-02-23 18:22 GMT+08:00 Chih-Wei Huang <cwh...@android-x86.org>:
> 2016-02-20 20:05 GMT+08:00 Mario Holzinger <sandma...@gmail.com>:
>> - Touchscreen GSL1680 I2C is working with integrating the acpi kernel module
>> ==> see patches
>
> Again, the device/generic/common part has contained
> patches you need. Please sync your codebase before
> preparing the patches.
>
> I forgot the kernel part patch.
> Will cherry-pick from kernel-4.0 later.

All patches for GSL1680 should have been merged.
Please sync and test to see if anything still missed.

[Hypo Turtle]

Would something like this work (and deal with inverse);

template<> int Sensor<ID_GYROSCOPE>::readEvents(sensors_event_t *data, int cnt)
{
    static float scale = read_sysfs_float((*nodes)[0]);
    int ret = SensorBase::readEvents(data, cnt);
    // TODO: read orientation from the properties
    char rota[6] = property_get("hal.sensors.iio");
    // or propery_get("hal.sensors.iio", rota, " ,x, ,y, ,z");
    for (int i = 0; i < ret; ++i) {
        data[i].gyro.x = rota[1]scale * read_sysfs_int("in_anglvel_rota[2]_raw");
        data[i].gyro.y = rota[3]scale * read_sysfs_int("in_anglvel_rota[4]_raw");
        data[i].gyro.z = rota[5]scale * read_sysfs_int("in_anglvel_rota[6]_raw");
        data[i].gyro.status = SENSOR_STATUS_ACCURACY_HIGH;
    }
    return ret;
}

With hal.sensors.iio = -,x,-,y,-,z
or hal.sensors.iio = ,x,,y,,z

[Jon West]

There are a couple patches still needed to solve issues with the multitouch on the Surface Pro 3

https://github.com/shvr/fedora-surface-pro-3-kernel/commit/9033a2d97a2c40a4260928b609908f65d2edfbe0

https://github.com/shvr/fedora-surface-pro-3-kernel/blob/master/HID-multitouch-Add-MT_QUIRK_NOT_SEEN_MEANS_UP-to-MT_.patch

https://github.com/shvr/fedora-surface-pro-3-kernel/commit/3602b1729fe4e0a38f28828cb80d2c2765eb9a62

https://github.com/shvr/fedora-surface-pro-3-kernel/blob/master/HID-multitouch-Ignore-invalid-reports.patch

[Chih-Wei Huang]

2016-03-02 9:22 GMT+08:00 Jon West <electr...@gmail.com>:
> There are a couple patches still needed to solve issues with the multitouch
> on the Surface Pro 3
>
> https://github.com/shvr/fedora-surface-pro-3-kernel/commit/9033a2d97a2c40a4260928b609908f65d2edfbe0
> https://github.com/shvr/fedora-surface-pro-3-kernel/blob/master/HID-multitouch-Add-MT_QUIRK_NOT_SEEN_MEANS_UP-to-MT_.patch
>
> https://github.com/shvr/fedora-surface-pro-3-kernel/commit/3602b1729fe4e0a38f28828cb80d2c2765eb9a62
> https://github.com/shvr/fedora-surface-pro-3-kernel/blob/master/HID-multitouch-Ignore-invalid-reports.patch

Thank you for the willing to help but
please check our kernel repository first
otherwise you are wasting our time to
check them again.

[Jon West]

I am sorry then. I am still getting the hang of things in this group. 

[youling 257]

the new remix os for  pc，4.4.2 kernel，has GSLx680 ，on my bay tablet，it isn't work well，i will upload picture.

在 2016年1月29日星期五 UTC+8上午1:15:07，Mario Holzinger写道：

only for testing here is the Patch to implement GSLx680 touch controllers.

but be aware that it is only working if the correkt firmware is aso included in device/generic/common/firmware and if tha module is loaded manually or by startup script

The patch has included my firmware for Trekstor Surftab Wintron 7.0 extracted from the Windwos 10 driver firmware

it is only a alpha version because I don't know at the moment how I shell calibrate the touchscreen.

Maybe I find a way to hack the firmware to ready to use. (that would be the best sollution)

under ubuntu I used xinput calibrator but on android the built in calibration do not work for me atm.

[youling 257]

the new remix os for pc，i use its kernel、firmware、modules，copy to Android x86 6.0.1

[Mario Holzinger]

sorry for highjacking the thread again.

what do you think about the redefined patch for hardware/libsensors/iio-sensors.cpp

hal.sensors.iio.accel.matrix=-000+000-

first sensor inverted for x

second sensor for y

third sensor inverted for z 

hal.sensors.iio.gyro.matrix=+000+000+

first sensor for x

second sensor for y

third sensor for z 

defines a matrix each for every aspect of the sensors in one property

+ = positive sensor value to be used

- = inverted sensor value to be used

0 = ignor sensor value for this axis

the first 3 characters stands for the first enumerated sensor were to be used

e.g.

-00 = use first sensor inverted for x axis

0+0 = use first sensor for y axis

the second 3 characters are for the 2nd sensor

and the last 3 are for the 3rd sensor

best regards

Mario

PS: I have no Gyro so this could not be tested by my selfe, and also other sensors can easily ported with this code blocks

+ //hal.sensors.iio.accel.matrix=-000+000- ==> default configuration

+ char cmatrix[PROPERTY_VALUE_MAX];

+ int imatrix[9];

+

+ property_get("hal.sensors.iio.accel.matrix", cmatrix, "-000+000-" );

+

+ for (int i = 0, j = 0; i < PROPERTY_VALUE_MAX; i++) 

+ {

+ if (cmatrix[i] == '+') {imatrix[j] =  1; j++;}

+ if (cmatrix[i] == '-') {imatrix[j] = -1; j++;}

+ if (cmatrix[i] == '0') {imatrix[j] =  0; j++;}

+ }

+

+ for (int i = 0; i < ret; ++i) 

+ {

+ if (imatrix[0] != 0) { data[i].acceleration.x = imatrix[0] * scale * read_sysfs_int("in_accel_x_raw"); }

+ if (imatrix[1] != 0) { data[i].acceleration.x = imatrix[1] * scale * read_sysfs_int("in_accel_y_raw"); }

+ if (imatrix[2] != 0) { data[i].acceleration.x = imatrix[2] * scale * read_sysfs_int("in_accel_z_raw"); }

+ if (imatrix[3] != 0) { data[i].acceleration.y = imatrix[3] * scale * read_sysfs_int("in_accel_x_raw"); }

+ if (imatrix[4] != 0) { data[i].acceleration.y = imatrix[4] * scale * read_sysfs_int("in_accel_y_raw"); }

+ if (imatrix[5] != 0) { data[i].acceleration.y = imatrix[5] * scale * read_sysfs_int("in_accel_z_raw"); }

+ if (imatrix[6] != 0) { data[i].acceleration.z = imatrix[6] * scale * read_sysfs_int("in_accel_x_raw"); }

+ if (imatrix[7] != 0) { data[i].acceleration.z = imatrix[7] * scale * read_sysfs_int("in_accel_y_raw"); }

+ if (imatrix[8] != 0) { data[i].acceleration.z = imatrix[8] * scale * read_sysfs_int("in_accel_z_raw"); }

+ data[i].acceleration.status = SENSOR_STATUS_ACCURACY_HIGH;

[youling 257]

need gslx680 patch for https://sourceforge.net/p/android-x86/kernel/ci/kernel-4.8/tree/ 

在 2016年2月20日星期六 UTC+8下午8:05:29，Mario Holzinger写道：

[Mario Holzinger]

what did you mean?

there is already a working version of the gslx680 driver included

there is a updated version available at https://github.com/onitake/gslx680-acpi

for android-x86 you could apply the attachment that replaces the current version with the latest one from onitake

I sucessfully use the updated driver on ma android builds but the essential thing is that you need a propper silead_ts.fw from https://github.com/onitake/gsl-firmware

or selfmade one that fits as good as possible to your device.

with a good calibrated firmware you can also avoid recalibration under android. mine fits perfectly on my Trekstor 7.0

regards

Mario

[youling 257]

0001-updated-gslx680-driver-from-https-github.com-onitake.patch , only update https://sourceforge.net/p/android-x86/kernel/ci/kernel-4.4/tree/drivers/staging/gslx680/

now ,kernel-4.8 branch is ready, didn't has https://sourceforge.net/p/android-x86/kernel/ci/kernel-4.8/tree/drivers/staging/gslx680, need patch
在 2016年9月4日星期日 UTC+8下午9:15:42，Mario Holzinger写道：

[Chih-Wei Huang]

2016-09-04 21:59 GMT+08:00 youling 257 <youli...@gmail.com>:
> https://sourceforge.net/p/android-x86/kernel/ci/kernel-4.4/tree/drivers/staging/gslx680/
> now ,kernel-4.8 branch is ready, didn't has

Well, I didn't say kernel-4.8 is ready.
Nobody said that.

The branch was pushed to sf.net
just for GPL compliance.
It's the kernel source I used to
build the 6.0-rc2-for-s3 iso.
It's not officially created or maintained
for android-x86 porting.
You are free to try it but
don't expect it too much.

What's the next kernel for porting
is not decided.
I'm still waiting the next LTS kernel
but it's yet to be decided.

> https://sourceforge.net/p/android-x86/kernel/ci/kernel-4.8/tree/drivers/staging/gslx680,
> need patch

[Mario Holzinger]

as soon as i get kernel-4-8 compiled properly I will support you with a patch.

Than Chih-Wei Huang can decide if he likes to merge it.

I also play a bit to compile the 4.8 branch outside the android tree (and kernel master branch from Linus Torvalds) but without success so far

best regards

Mario

btw. I also tried to mod my manifests/default.xml to get kernel-4.8 automatically when syncing but it fails due to the currently facing syncing issues from sourceforge (on my side)

[youling 257]

https://groups.google.com/d/msg/android-x86/nseLSalOtGI/Gh9UcPa1CQAJ , only compile vmlinuz and modules success, rename vmlinuz to kernel

make android-x86_defconfig ，time make -j8，time make -j8 targz-pkg -i

在 2016年9月5日星期一 UTC+8上午2:01:41，Mario Holzinger写道：

[Andrew Watts]

If I am correct its only after kernel 4.5 that skylake's power saving feature's are implemented properly right?

I ran 6.0 RC-2 on my Core M 6y30... When I looked at the cpu it was (claimed) to be running fast - up to 2Ghz 

and 4 cores (I assume this was the kernel misrecognising Hyperthreading on the dual core)

Although it wasn't running hot.

Also power use didn't seem to be an issue.

I downloaded a pre-build 7.0 image... Now power consumption there is an issue!

[Mario Holzinger]

ok found out that my implemented gslx680_ts_acpi driver in kernel/drivers/staging/gslx680 is now obsolete and replaced by silead_ts wich comes directly with the kernel

the disadvantage is that the firmware is not able to be loaded anymore because it is especially created for the gslx680_ts_acpi module

my luck is that my win10 firmware can be loaded and is working is I rename it as mssl1680.fw however the calibration is a mess.

I will provide a patch as soon as I find out how to fix the calibration directly in the win10.fw

regards

Mario

[dgdn]

did location of wl-driver change sometime in the last week by any chance????

https://git.solus-project.com/packages/broadcom-sta/commit/?h=broadcom-sta-6.30.223.271-88

getting the following that was working fine on dec 2??

[ 99% 48020/48180] build kernel/drivers/net/wireless/wl/Makefile

FAILED: /bin/bash -c "tar zxf kernel/drivers/net/wireless/wl/hybrid-v35_64-nodebug-pcoem-6_30_223_271.tar.gz -C kernel/drivers/net/wireless/wl --overwrite && patch -p5 -d kernel/drivers/net/wireless/wl -i wl.patch && patch -p1 -d kernel/drivers/net/wireless/wl -i linux-recent.patch"

gzip: stdin: not in gzip format

tar: Child returned status 1

tar: Error is not recoverable: exiting now

[ 99% 48020/48180] host C: bison <= external/bison/src/InadequacyList.c

external/bison/src/InadequacyList.c:67:57: warning: unused parameter 'i' [-Wunused-parameter]

                                      ContributionIndex i)

[Jon West]

No, their zip is malformed or incomplete. Search around for it to grab a good zip and place it in the wl dir. You can disable the curl too so it doesn't try to grab another. This is only temporary until Broadcom realizesthey posted a bad zip.

[dgdn]

that maybe BUT....I pulled the sources from the new link, added to my wl folder and everything built fine.....

[Chih-Wei Huang]

2016-12-09 5:28 GMT+08:00 dgdn <rbg...@gmail.com>:
> did location of wl-driver change sometime in the last week by any chance????
>
> https://git.solus-project.com/packages/broadcom-sta/commit/?h=broadcom-sta-6.30.223.271-88

Thank you for pointing it out.
I've updated to the new URL.

[dgdn]

ok thank you.....will resync and test within the next few days.....in the mean time

more recent test builds are here:

https://drive.google.com/drive/folders/0B3Vjym21YjOTTERmbk1MZjJ6Ylk?usp=sharing

[youling 257]

4.9 is longterm，4.9.13

在 2016年9月5日星期一 UTC+8上午12:35:36，Chih-Wei Huang写道：