Kernel 4.0 is released for testing

[Chih-Wei Huang]

Hi devs,
I have pushed the android-4.0 branch of kernel
based on the vanilla kernel 4.0 to the git server.
To test it, fetch it from the server and checkout it:

cd kernel
git fetch x86 android-4.0
git checkout x86/android-4.0
(or git checkout FETCH_HEAD)
cd ..

Then build the iso_img as usual.

m -jX iso_img

I only tested it with the lollipop-x86 branch
for x86 and x86_64 targets.
It should work with kitkat-x86 iso_img as well,
but may not work with efi_img 64-bit kernel
unless you manually modify the binder to 32-bit.

Enjoy it!


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

[Mauro Rossi]

I was looking forward to check latest drm 2.4.60, which bring some specific nouveau multithreading code, with latest 4.0 kernel, having specific nouveau mmap corrections.

Now those checks are possibile.
Thanks a lot!

M.

[pstglia]

Thanks a lot!!!

[Meerio]

I know i'm annoying, but if anyone with a build system want to create the isos i'll test nouveau on my system ( I've 32bit atom and ion, but i think many will come here wanting the various configurations between kitkat/lollipop and 32/64 bit)

[Edgardo Gho]

I have this problem since kernel 3.18.8 (the last working kernel is 3.18.2 on my system) and now 4.0 has the same issue. Computer boots , finds the system, switches to graphical mode (intel framebuffer seems to load, i915 loads) but the boot animation never appears, and there is just a blank screen. If I go to Alt+F1 terminal I get the terminal and dmesg shows nothing weird. If I start with nomodeset it works (terribly-unusable) so I know there is nothing wrong with the install and I'm guessing its a kernel (maybe drm or i915) issue.

I'm running on a Celeron CPU 1037U.

Kernel 3.18.2 works great (Android 5.0.2).

[The Black Death]

Hey All,

Lollipop 5.1.0 With the new 4.0 Kernel (From RBG) has Been Added to the Downloads Section at:

http://www.android-x86.net

TBD

[pstglia]

Hi Edgardo Gho,

Can you attach logcat and dmesg outputs? Maybe we can find some hint what's wrong looking those

[Edgardo Gho]

I will,

I'm trying to bisect the different versions.

Currently:

5.1 + K4.0 -> Black screen on boot up (works with nomodeset)

5.1 + K3.18.8 --> Same as above

5.1 + K3.10.52 --> Same as above

I will try 5.1 + K3.18.2 (which was the working one on 5.0.2) and if it fails will revert back to 5.0.2 and try Kernel 4 and 3.18.8.

5.1 runs great on an AMD system I have and its nice to see email working on lollipop for the first time.

[Edgardo Gho]

Here are the two log files (dmesg and logcat).

Is there a way to repo sync with 5.0.2 ? I would like to checkout 5.0.2 and test kernel 4 and see if that works.

[Edgardo Gho]

I tested an old 5.0.2 source folder I had and basically replaced /kernel with the 4.0 kernel checked out for 5.1. It works fine. So 5.0.2 + Kernel 4.0 works on my computer. I just can't make 5.1 + any kernel on the computer.

[rbg]

has any of the work you and mrossi did yet been committed to lollipop git???

my meld diff for example on mesa tells me no it hasn't.....

On Tuesday, April 14, 2015 at 1:17:51 PM UTC-4, pstglia wrote:

[midi jari]

Sorry, doesn't work good :/

[Jianlong Liu]

Thanks, both 4.4 and 5.1 work great on my Venue 8 Pro. Most of the issues I encountered before with 3.18.x are still present though.

[pstglia]

has any of the work you and mrossi did yet been committed to lollipop git???

my meld diff for example on mesa tells me no it hasn't.....

As far as I know, lollipop is still synced with mesa 9.2.0. But except for small changes, 10.4.2 can be used with lollipop without problems (last time I tried on my hw I had no graphical problems)

[Chih-Wei Huang]

2015-04-15 4:17 GMT+08:00 Edgardo Gho <edgar...@gmail.com>:
> I tested an old 5.0.2 source folder I had and basically replaced /kernel
> with the 4.0 kernel checked out for 5.1. It works fine. So 5.0.2 + Kernel
> 4.0 works on my computer. I just can't make 5.1 + any kernel on the
> computer.

Please try the steps:

cd frameworks/native/
git revert -n 6747be9fa
cd ..

The rebuild the image to see if it fix the black screen issue.

That Google commit seems broke i965 driver of mesa 9.2.
(i915/radeon seems fine)
So I applied the above workaround to lollipop-x85 5.0 before.

But I didn't apply it to the 5.1 branch since
I haven't tested mesa 9.2 for a long time.
I'm testing mesa 10.5/10.6 which seems doesn't
need this workaround.

[Mauro Rossi]

Hi,

For those who want to try mesa 10.4.7/10.5.3 on kitkat-x86 build:

https://github.com/maurossi/mesa/tree/10.4.7

https://github.com/maurossi/mesa/tree/10.5.3

Regarding lollipop-x86, the lastest mesa I've tested is the following:

https://github.com/maurossi/mesa/tree/lollipop-x86_multiarch

Not much of my contribution, it's just a collection of various patches and some changes to Android.mk files to build multiarch (but having completely disabled ASM optimizations) and some LOCAL_INCLUDES that were missing.

Upcoming Chih-Wei patches are better than my git, which I use as a lab and playground.

How to download the git for lollipop-x86:

cd [lollipop-x86]

cd external/mesa

git remote add maurossi https://github.com/maurossi/mesa

git fetch maurossi lollipop-x86_multiarch

git checkout maurossi/lollipop-x86_multiarch

be aware that last time I built 5.0.2 there was a patch needed for chromium_org with gallium drivers for radeon and nvidia chipsets (more details in this forum), but now it is not possible anymore to apply the patch to chromium_org.

I don't know yet if the sign-in problem was corrected with lollipop 5.1, but Intel chipsets were unaffected, probably due to a more compliant GLES implementation.

Mauro

[Edgardo Gho]

Worked like a charm.

thanks!

[rbg]

thanks...is their a list of all the other "small" changes that need to be made to get 10.4.2 working on lollipop???

[Chih-Wei Huang]

2015-04-16 3:29 GMT+08:00 rbg <rbg.hp...@gmail.com>:
> thanks...is their a list of all the other "small" changes that need to be
> made to get 10.4.2 working on lollipop???

Do you mean mesa 10.4.2?
I'm not sure whether if it needs that workaround.
You may with or without it and post the result.

[midi jari]

Well, I used these mesa drivers and it worked, the workaround you provided didn't have any effect. I used the original kernel provided, I will remake it again with kernel 4.0 :)

[Chih-Wei Huang]

For the Baytrail / T100 support,
I've applied all series of patches for
the lost IRQ for brcmfmac to the android-4.0.
(byt-fix branch)
Thanks to the suggestion of Povilas Staniulis.
The backlight patches are also applied.
But I still can't get it work on T100 yet.

For the mmc issues, I still can't cleanly apply the mmc patches
by Adrian Hunter to kernel 4.0.
(the re-tuning patches can't be applied)
Have tried to contact the author but no help yet.

Any thought?

[Povilas Staniulis]

On 2015.04.16 13:36, Chih-Wei Huang wrote:
> For the Baytrail / T100 support,
> I've applied all series of patches for
> the lost IRQ for brcmfmac to the android-4.0.
> (byt-fix branch)
> Thanks to the suggestion of Povilas Staniulis.
> The backlight patches are also applied.
> But I still can't get it work on T100 yet.
>
> For the mmc issues, I still can't cleanly apply the mmc patches
> by Adrian Hunter to kernel 4.0.
> (the re-tuning patches can't be applied)
> Have tried to contact the author but no help yet.
>
> Any thought?
>

I got it working with kitkat branch.
Can upload an image for those interested.
Didn't test stability, but it boots and appears to work fine.

* Graphics, Backlight, Sound, Wireless: working

* Tablet buttons (Volume, home, power): working

* Suspend: screen turns off, but not sure if suspend really works

* eMMC/SD: detected, didn't test stability yet

* Bluetooth: can't get to work because there's no hciattach command in
Android

* Sensors: mpu6500 detected, but Android doesn't work with it.
(do we need userspace driver for this or just enable some option ?)

Also, Android's logcat doesn't work due to missing /dev/log/main (where
did it go ?)

As for the MMC patches:

Support maximum DMA latency request.patch
drivers/mmc/host/sdhci.h: Line: Hunk at line 1348 fails due to removed
comment, the code is the same, so I applied it by hand.
drivers/mmc/host/sdhci.h: Fails due to moved header file, the original
header was moved to include/linux/mmc/sdhci.h

Add pm_qos_cancel_request_lazy.patch - different offsets, but applies fine.

sdhci_pci Fix device hang on Intel BayTrail.patch, sdhci_apci Fix device
hang on Intel BayTrail.patch - these apply fine, with offsets.

Chih-Wei has already applied the other patches.

[T100TA User]

Please do. Much obliged!

[Povilas Staniulis]

On 2015.04.16 15:28, T100TA User wrote:
> Please do. Much obliged!
>
>
Here you are:
http://raspis.ddscentral.org/pub/downloads/android-x86/Android-x86-Kitkat_Baytrail-K4.0_Apr16.rar

Extract the archive to a USB flash drive.

Sorry for the delay, my upload speed is dog slow.

[Jianlong Liu]

I don't see how you patched correctly. We need the MMC retune series of patches added first (and maybe some other prerequisite patches).

E.g., if we read hunk #2, it says its applied to sdhci_request, which is clearly missing "err = mmc_retune(mmc);" and everything else (although I don't know if any of the modifications depend on the retune stuff).

[Povilas Staniulis]

On 2015.04.16 22:06, Jianlong Liu wrote:
> I don't see how you patched correctly. We need the MMC retune series
> of patches added first (and maybe some other prerequisite patches).
> E.g., if we read hunk #2, it says its applied to sdhci_request, which
> is clearly missing "err = mmc_retune(mmc);" and everything else
> (although I don't know if any of the modifications depend on the
> retune stuff).
>

Of course, for pushing to git, it would be best to find and apply other
mmc patches too.

The patches which I've applied were meant to fix a specific problem
which affected system stability - SD controller freezing when entering
deep c-states.
They did apply fine, and did not break anything (SD devices still work
fine), so I assume they aren't dependent on other patches.

I was able to copy around 1.2GB data to SD card and read it back without
any system freezes, which I couldn't do previously.

[Jianlong Liu]

I'm failing on the patches for sdhci.h on patch 11 of the retuning series.

I compared with the latest 4.0 upstream, and for some reason we are missing this patch that merged include/linux/mmc/sdhci.h into drivers/mmc/host/sdhci.h. The retuning patches (V5) then apply correctly I think (with some offsets).

Not sure how the series of four need to be updated if using V5 (should work with V3 directly).

On Thursday, April 16, 2015 at 3:36:34 AM UTC-7, Chih-Wei Huang wrote:

For the Baytrail / T100 support, 
I've applied all series of patches for 
the lost IRQ for brcmfmac to the android-4.0. 
(byt-fix branch) 
Thanks to the suggestion of Povilas Staniulis. 
The backlight patches are also applied. 
But I still can't get it work on T100 yet. 

For the mmc issues, I still can't cleanly apply the mmc patches 
by Adrian Hunter to kernel 4.0. 
(the re-tuning patches can't be applied) 
Have tried to contact the author but no help yet. 

Any thought? 

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

Ah, I missed that you mentioned missing the patch that merged the headers!

[midi jari]

I tried again and I tried you workaround, it worked, thanks :)

[T100TA User]

Thank you so much! Been using it for a while on my T100 with a dualboot solution (rEFInd), seems the freezing problem is now gone. Wi-Fi is also fast and stable.
Like you said, the screen turns off but suspend doesn't seem to be working. Anyway, this is definitely the best KitKat image for the T100 so far.

[Chih-Wei Huang]

2015-04-16 20:23 GMT+08:00 Povilas Staniulis <wdmo...@gmail.com>:
> On 2015.04.16 13:36, Chih-Wei Huang wrote:
>>
>> For the Baytrail / T100 support,
>> I've applied all series of patches for
>> the lost IRQ for brcmfmac to the android-4.0.
>> (byt-fix branch)
>> Thanks to the suggestion of Povilas Staniulis.
>> The backlight patches are also applied.
>> But I still can't get it work on T100 yet.
>>
>> For the mmc issues, I still can't cleanly apply the mmc patches
>> by Adrian Hunter to kernel 4.0.
>> (the re-tuning patches can't be applied)
>> Have tried to contact the author but no help yet.
>>
>> Any thought?
>>
> I got it working with kitkat branch.
> Can upload an image for those interested.
> Didn't test stability, but it boots and appears to work fine.
>
> * Graphics, Backlight, Sound, Wireless: working

Interesting. Are you testing on ASUS T100 as mine?

I can't get sound works for kitkat-x86 on T100 with kernel 4.0 yet.
Everything seems ok as before, but just no sound.
Since there is no logcat, it's hard to debug...

For lollipop-x86 with the same kernel 4.0,
the sound driver can't be loaded automatically
(another very strange thing I can't understand)
But if I manually modprobe them, it works.
(snd-soc-sst-baytrail-pcm & snd-soc-sst-byt-rt5640-mach)

For backlight,
So what's your /sys/class/backlight shows?
Mine only shows acpi_video0 & acpi_video1.
But changing the brightness in them doesn't work.
I also noticed some pwn error in dmesg.

> * Tablet buttons (Volume, home, power): working

Not work for me yet... strange...

Did you change any kernel config?

> * Suspend: screen turns off, but not sure if suspend really works

Same.

> * eMMC/SD: detected, didn't test stability yet
>
> * Bluetooth: can't get to work because there's no hciattach command in
> Android

I don't think hciattach is needed.
The problem is we have changed bluedroid to support usb hci
that disable bt serial support (see commit 2001b2fc).
However, T100 uses serial port (/dev/ttyS1 or ttyS3??) for bt as I known.
We need to patch bluedroid to support both usb hci and serial port.

> * Sensors: mpu6500 detected, but Android doesn't work with it.
> (do we need userspace driver for this or just enable some option ?)

Need a sensor hal, of course.
I'm working on it.

> Also, Android's logcat doesn't work due to missing /dev/log/main (where did
> it go ?)

The android logger driver was removed from kernel 4.0
because Android lollipop no longer uses the logger driver
(migrated to userspace logd)
However, kitkat still needs the logger driver.

I've tried to revert some patches (a0a23bbce),
but still can't get it back in kitkat yet.
Need to look into deeper what else changed.

> As for the MMC patches:
>
> Support maximum DMA latency request.patch
> drivers/mmc/host/sdhci.h: Line: Hunk at line 1348 fails due to removed
> comment, the code is the same, so I applied it by hand.
> drivers/mmc/host/sdhci.h: Fails due to moved header file, the original
> header was moved to include/linux/mmc/sdhci.h
>
> Add pm_qos_cancel_request_lazy.patch - different offsets, but applies fine.
>
> sdhci_pci Fix device hang on Intel BayTrail.patch, sdhci_apci Fix device
> hang on Intel BayTrail.patch - these apply fine, with offsets.
>
> Chih-Wei has already applied the other patches.

[Chih-Wei Huang]

2015-04-17 4:16 GMT+08:00 Jianlong Liu <lvlr...@gmail.com>:
> I'm failing on the patches for sdhci.h on patch 11 of the retuning series.
> I compared with the latest 4.0 upstream, and for some reason we are missing
> this patch that merged include/linux/mmc/sdhci.h into

Thank you for pointed it out.
That patch was merged after the kernel 4.0 release.
(so should be in 4.1-rc)

Anyway, by merging the mmc branch from 4.1-rc,
I'm able to apply the V5 re-tuning patches now. But the
'Support maximum DMA latency request via PM QOS'
patch still has minor conflict.
Will try to resolve it later.

[Jianlong Liu]

For me (Dell Venue 8 Pro), apparently there's a regression somewhere that made my volume buttons not work. Power button and Windows key button still work though.

My tablet also only has acpi_video0/1 in /sys/class/backlight.

Suspend also isn't working for me yet, and backlight still stays on even though screen blanks when pressing the power button.

I applied retuning V5, followed by drive strength for eMMCs (no modifications needed if I remember correctly). After that, I applied the 4 patches for latency, with some changes needed for hunks #2 and #3, since some stuff changed between V3 and V5, and the latency patches were for V3.

Not sure if the patches for drive strength are needed, but patch #4 for latency mentioned it in the first hunk, so I used them (although that patch #4 can be modified to not reference drive strength).

[Mario Holzinger]

Thanks Chih-Wei!

That commit did the trick on my Surface Pro 2

Now I am able to debug further

- Wifi is broken (Authentification Problem)

- new GApps included because my 5.0.2 ones force to close

great. :-)

[Chih-Wei Huang]

2015-04-17 14:56 GMT+08:00 Jianlong Liu <lvlr...@gmail.com>:
> For me (Dell Venue 8 Pro), apparently there's a regression somewhere that
> made my volume buttons not work. Power button and Windows key button still
> work though.
> My tablet also only has acpi_video0/1 in /sys/class/backlight.

Can you control the brightness via it?

> Suspend also isn't working for me yet, and backlight still stays on even
> though screen blanks when pressing the power button.
>
> I applied retuning V5, followed by drive strength for eMMCs (no
> modifications needed if I remember correctly). After that, I applied the 4
> patches for latency, with some changes needed for hunks #2 and #3, since
> some stuff changed between V3 and V5, and the latency patches were for V3.
> Not sure if the patches for drive strength are needed, but patch #4 for
> latency mentioned it in the first hunk, so I used them (although that patch
> #4 can be modified to not reference drive strength).

Hmm, I don't understand what patch #4 you mentioned.
To align the work, here are the patches I tried to apply

the re-tuning patches
6215651 New [V5,01/15] mmc: host: Add facility to support re-tuning
6215661 New [V5,02/15] mmc: core: Enable / disable re-tuning
6215671 New [V5,03/15] mmc: core: Add support for re-tuning
before each request
6215681 New [V5,04/15] mmc: core: Check re-tuning before retrying
6215701 New [V5,05/15] mmc: core: Hold re-tuning during switch commands
6215691 New [V5,06/15] mmc: core: Hold re-tuning during erase commands
6225411 New [V6,07/15] mmc: core: Hold re-tuning while bkops ongoing
6215721 New [V5,08/15] mmc: mmc: Comment that callers need to
hold re-tuning if the card is put to sleep
6215731 New [V5,09/15] mmc: core: Separate out the mmc_switch
status check so it can be re-used
6215741 New [V5,10/15] mmc: core: Add support for HS400 re-tuning
6215751 New [V5,11/15] mmc: sdhci: Change to new way of doing re-tuning
6215761 New [V5,12/15] mmc: sdhci: Flag re-tuning is needed
on CRC or End-Bit errors
6215771 New [V5,13/15] mmc: block: Check re-tuning in the recovery path
6215781 New [V5,14/15] mmc: block: Retry errored data
requests when re-tuning is needed
6215791 New [V5,15/15] mmc: core: Don't print reset warning
if reset is not supported

follow by driver strength patches

5790621 New [V2,01/11] mmc: core: Reset driver type to default
5790531 New [V2,02/11] mmc: core: Allow card drive strength
to be different to host
5790541 New [V2,03/11] mmc: core: Simplify card drive strength mask
5790551 New [V2,04/11] mmc: core: Add 'card' to drive
strength selection callback
5790601 New [V2,05/11] mmc: core: Factor out common code in
drive strength selection
5790591 New [V2,06/11] mmc: core: Record card drive strength
5790521 New [V2,07/11] mmc: mmc: Read card's valid driver strength mask
5790581 New [V2,08/11] mmc: mmc: Add driver strength selection
5790561 New [V2,09/11] mmc: sdhci: Add a callback to select
drive strength
5790611 New [V2,10/11] mmc: sdhci-pci: Add support for drive
strength selection for SPT
5790571 New [V2,11/11] mmc: sdhci-pci: Enable HS400 for some
Intel host controllers

then the maximum DMA latency patches

6079041 New [RFC,1/4] PM / QoS: Add
pm_qos_cancel_request_lazy() that doesn't sleep
6079051 New [RFC,2/4] mmc: sdhci: Support maximum DMA latency
request via PM QOS
6079021 New [RFC,3/4] mmc: sdhci-acpi: Fix device hang on
Intel BayTrail
6079001 New [RFC,4/4] mmc: sdhci-pci: Fix device hang on Intel BayTrail

These patches could be found and applied by pwclient:

pwclient list -w adrian...@intel.com -p linux-mmc

All patches applied well until 6079051.
(checking how to resolve it...)


For those who don't know pwclient yet, please refer to

https://patchwork.kernel.org/help/pwclient/

I just learned to use it, a very convenient way to apply patches
from lkml.

[Jianlong Liu]

I meant 6079001, but since you're using drive strength patches, it'll patch fine. And nope, can't control backlight with either of them.

[Chih-Wei Huang]

2015-04-17 16:07 GMT+08:00 Jianlong Liu <lvlr...@gmail.com>:
> I meant 6079001, but since you're using drive strength patches, it'll patch
> fine. And nope, can't control backlight with either of them.

OK.
I've pushed all patches into the android-4.0 branch.
Please help to test it.

[Ravid Maduni]

can someone make an .img or .iso for testing? i can test for both intel gpu (intel hd graphic 4000) and nouveau gpu (nvidia gt 220)

בתאריך יום שני, 13 באפריל 2015 בשעה 20:58:04 UTC+3, מאת Chih-Wei Huang:

Hi devs,
I have pushed the android-4.0 branch of kernel
based on the vanilla kernel 4.0 to the git server.
To test it, fetch it from the server and checkout it:

cd kernel
git fetch x86 android-4.0
git checkout x86/android-4.0
(or git checkout FETCH_HEAD)
cd ..

Then build the iso_img as usual.

m -jX iso_img

I only tested it with the lollipop-x86 branch
for x86 and x86_64 targets.
It should work with kitkat-x86 iso_img as well,
but may not work with efi_img 64-bit kernel
unless you manually modify the binder to 32-bit.

Enjoy it!

[Mario Holzinger]

You can use my debug build that boots on my Surface Pro 2

https://drive.google.com/file/d/0B43N_UcsJhpjbkFDX2dzR09PWTg/view?usp=sharing

[Mario Holzinger]

Switched back to 3.18.8

4.0 seems to break my WiFi Authentification. :-(
WPA2-PSK

any one else facing the same issue?

[Povilas Staniulis]

On 2015.04.17 07:33, Chih-Wei Huang wrote:
> Interesting. Are you testing on ASUS T100 as mine?

Mine is Asus T100TA-DK024H

> I can't get sound works for kitkat-x86 on T100 with kernel 4.0 yet.
> Everything seems ok as before, but just no sound.
> Since there is no logcat, it's hard to debug...

> For lollipop-x86 with the same kernel 4.0,
> the sound driver can't be loaded automatically
> (another very strange thing I can't understand)
> But if I manually modprobe them, it works.
> (snd-soc-sst-baytrail-pcm & snd-soc-sst-byt-rt5640-mach)

Strange, sound works just fine for me right after booting.
Sometimes (rarely) error -110 (timeout) shows up in dmesg, but nothing
major.

> For backlight,
> So what's your /sys/class/backlight shows?
> Mine only shows acpi_video0 & acpi_video1.
> But changing the brightness in them doesn't work.
> I also noticed some pwn error in dmesg.

Very strange. Brightness control works just fine for me. Maybe this is
related to kernel config options ?

/sys/class/backlight shows:
intel_backlight

/sys/class/backlight/intel_backlight shows:
actual_brightness
bl_power
brightness
device
max_brightness
power
type
uevent

>> * Tablet buttons (Volume, home, power): working
> Not work for me yet... strange...
>
> Did you change any kernel config?

I did change CONFIG_GPIO_CRYSTAL_COVE to y (default m) as suggested by
Ubuntu guys.
I also set CONFIG_PINCTRL_BAYTRAIL to y (default n), but don't know if
this is really needed.

>> * Suspend: screen turns off, but not sure if suspend really works
> Same.

Well, at least the screen turns off...

>> * Bluetooth: can't get to work because there's no hciattach command in
>> Android
> I don't think hciattach is needed.
> The problem is we have changed bluedroid to support usb hci
> that disable bt serial support (see commit 2001b2fc).
> However, T100 uses serial port (/dev/ttyS1 or ttyS3??) for bt as I known.
> We need to patch bluedroid to support both usb hci and serial port.

Ok, will wait for this.

* Sensors: mpu6500 detected, but Android doesn't work with it.
(do we need userspace driver for this or just enable some option ?)

> Need a sensor hal, of course.
> I'm working on it.

Thanks, only wanted to check if it's just me, or there's no driver.

> The android logger driver was removed from kernel 4.0
> because Android lollipop no longer uses the logger driver
> (migrated to userspace logd)
> However, kitkat still needs the logger driver.
>
> I've tried to revert some patches (a0a23bbce),
> but still can't get it back in kitkat yet.
> Need to look into deeper what else changed.
>

Would be really nice to get this back. Although not needed for normal
use, it's hard to debug anything in userspace without logcat.

[Ravid Maduni]

Tried that on my both devices I mentioned earlier. The device with the gt 220, doesn't even getting to the grub (perhaps because it's not uefi) but in my intel hd graphic 4000 it  does, but the live cd doesn't bring up the terminal or the android itself. Same with the other builds here of kernel 4.0.

p.s, I have no idea how to create the necessary .log file

בתאריך יום שישי, 17 באפריל 2015 בשעה 15:47:00 UTC+3, מאת Mario Holzinger:

[midi jari]

I always have this problem and it's not kernel related (at most, since I got it too when I transferred the install from an external source to my internal HDD cuz I have GPT), so try this fix here: https://groups.google.com/forum/#!topic/android-x86/34y2_QJkCTk (just chmod 755 to dhcpd-run-hooks in /system/etc/dhcpd)

[midi jari]

ok, I made an iso with the HD4000 fix and Kernel 4.0 for anyone who want to try out: https://mega.co.nz/#!QYgUVZBD!pf_Q38WkjMOnsXCiAnt6J0zrIq090k5RkI4PgCj0K_w
The only problem I have here is the Google Apps, they dont show up after I manually installed them from CM12's gapps zip. if anyone have any way or ideas is welcomed :)

[midi jari]

Hi, how could you implement Gapps? I'm looking forward for this too, I tried from cm12 and it didn't work

[Mario Holzinger]

GApps integration like I did on every other build.

search the web for a GApps package compatible with your Android version (e.g.: lollipop 5.1)

open the package and place it in 

out/target/product/x86/system/

and start a build again to automatically integrat t in your image

2. different verson of GApps I've found so far.

1.) Package allready in the correct folder structure

2.) Package that needs manually relocate the pacages to fit into the system folder (just search in the package in all folders if you find folders which are also existing in systems when you build without GApps)

[midi jari]

yeah I tried that, but it didn't work, they are still not loaded, if you use a specific Gapps then link them to me :)
So what I did is copy the system folder from GApps' zip to the out folder so it merges with the built system folder, then I ran make -j4 iso_img. but then I didn't see any Gapps in the ISO (and yes, I checked the built system folder in target/x86 folder, it has the Gapps), does make filter what should be included?

[Ali Razavi]

Hi
I want Install lollipop-x86 with kernel 4 on my fresh PC, I  download android_x86-5.1.0_r3-20150413.iso. but when install this on my pc, google play (and very of other apps like openvpn connect and ...) does not install !!!
please help me about correct Partition Configuration,and correct installation for this installation.

My PC configuration:
CPU: Intel J1900
HDD: 500GB
RAM: 2GB
VGA: Intel HD
LAN: RTL8111F

I have Android 4.4.4 x86 on similar PC and this work very good, but I want update the kernel and android version.
thank you

[Raymond Lai]

Same here, on a Livefan F8C, using grub installed into EFI, situation much like T100TA User.

Wifi (bcm4330) finally started to scan with settings (brcmfmac4330-sdio.txt) copied from installed Windows 8.1, but should require some more tuning to get rid of brcmf_cfg80211_del_key errors (seen in dmesg).

Touchscreen (i2c_hid + hid_multitouch) still failed to initialize properly intermittently, but had been there from the beginning, and a cold boot seemed to fix the problem, so not really a big problem.

Hope there will be one more Kitkat release with kernel 4.0 before completely move to Lollipop...

2015年4月17日金曜日 11時42分30秒 UTC+8 T100TA User:

[pstglia]

Hi Rbg /Chih-Wei, how are you?

Em quarta-feira, 15 de abril de 2015 23:52:11 UTC-3, Chih-Wei Huang escreveu:

2015-04-16 3:29 GMT+08:00 rbg <rbg.hp...@gmail.com>:
> thanks...is their a list of all the other "small" changes that need to be
> made to get 10.4.2 working on lollipop???

Do you mean mesa 10.4.2?
I'm not sure whether if it needs that workaround.
You may with or without it and post the result.

 Sorry for the delay answering. Basically I had to do 2 changes to allow compiling Mesa 10.4.2 on Lollipop (32 bits target):

1) Added a path to libsync on 2 Android.mk files:

native_android.cpp (gallium) and platform_android.c (dri) have a include to libsync <sync/sync.h>

However, this file is on a different location on Lollipop:

* Kitkat: Located at system/core/include/

* Lollipop: Located at system/core/libsync/include

So i've added the new path to Android-mk files in order to allow compiling on both versions

2) Removed a check for "__i386__" define:

Lollipop defines "__i386__" for 32 bits targets (check core/clang/TARGET_x86.mk). This makes libpciaccess include a supposed unwanted file that produces a linkage error on when creating libdrm_intel.so:

external/libpciaccess/src/linux_sysfs.c:796: error: undefined reference to 'iopl'

collect2: error: ld returned 1 exit status

make: *** [out/target/product/x86/obj/SHARED_LIBRARIES/libdrm_intel_intermediates/LINKED/libdrm_intel.so] Error 1

So, I removed this check.

Attached the patches I applied.

off-topic: why that "flappy-bird like" game hidden on settings is so difficult? I could score a maximum of 5 :D

[rbg]

Not doing bad thanks.....have you noticed with lollipop that VLC seems to display

videos ok??

....I'm assuming this( Mesa 10.4.2 on Lollipop) is based on your 10.4 KitKat git contributions???

[pstglia]

Not doing bad thanks.....have you noticed with lollipop that VLC seems to display

videos ok??

Interesting. I'll check it out

 

....I'm assuming this( Mesa 10.4.2 on Lollipop) is based on your 10.4 KitKat git contributions???

It's the very same used currently on Kitkat. In fact, I removed 9.2.0 mesa dir and did a local "cp -R" from my local Kitkat git repo 

Also copied external/drm and external/libpciaccess

[Povilas Staniulis]

I've re-added the Android logger to kernel 4.0.

Patch attached.

Add ANDROID_LOGGER=y to kernel config to enable.

I've reversed this patch:

https://lkml.org/lkml/2015/1/26/672

And added this line to /drivers/staging/android/Makefile:

obj-$(CONFIG_ANDROID_LOGGER)            += logger.o

[Chih-Wei Huang]

OK. I think I missed this line.
Thanks!

[Chih-Wei Huang]

Hi Povilas,
I spent some time to test the image you released.
As you said, the sound, vol up/down, backlight work,
while my kernel does not. Very strange...

Could you send me the full kernel config
for comparison?

Is there any other patch you applied?

> --
> 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 http://groups.google.com/group/android-x86.
> For more options, visit https://groups.google.com/d/optout.

[Povilas Staniulis]

On 2015.04.20 19:37, Chih-Wei Huang wrote:
> Hi Povilas,
> I spent some time to test the image you released.
> As you said, the sound, vol up/down, backlight work,
> while my kernel does not. Very strange...
>
> Could you send me the full kernel config
> for comparison?
>
> Is there any other patch you applied?
>
>
>

Hello Chih-Wei,

The kernel in that image was based on older git revision. I redid the
kernel against the current revision.
Updated image:
http://raspis.ddscentral.org/pub/downloads/android-x86/Android-x86-Kitkat_Baytrail-K4.0_Apr20.rar

Regarding the kernel:
IIRC, I changed these 2 options (maybe there are more, see attached
configuration).
CONFIG_GPIO_CRYSTAL_COVE=y
CONFIG_PINCTRL_BAYTRAIL=y

I modified kernel/firmware/Makefile to add firmware files for sound and
wireless to the final image.
(patch attached, you should already have the firmware files, but if you
need them, let me know).

I re-added the Android logger.

Regarding backlight:
Intel patches which you've committed didn't work. There were 2 entries
in /sys/class/backlight: acpi_video0 and acpi_video1, but neither of
them actually worked.
I have manually applied the old force_backlight_pmic patch against the
current git source to get backlight working (patch attached).
The patch is Bay Trail specific and perhaps would have to be redone to
not affect other devices (e.g. by adding a config option) to be suitable
for committing to git.

Also, I found a few bugs:
There seems to be a bug in i915 which causes strange blinking in Play
Store app pages.
I see errors like this in dmesg:
[i915]] *ERROR* Atomic update failure on pipe A (start=<number>
end=<number>)

Second, the clock display does not update in launcher, but works fine in
lock screen.
(not sure if this is kernel related)

[Chih-Wei Huang]

2015-04-21 3:26 GMT+08:00 Povilas Staniulis <wdmo...@gmail.com>:
> On 2015.04.20 19:37, Chih-Wei Huang wrote:
>>
>> Hi Povilas,
>> I spent some time to test the image you released.
>> As you said, the sound, vol up/down, backlight work,
>> while my kernel does not. Very strange...
>>
>> Could you send me the full kernel config
>> for comparison?
>>
>> Is there any other patch you applied?
>>
>>
>>
> Hello Chih-Wei,
>
> The kernel in that image was based on older git revision. I redid the kernel
> against the current revision.
> Updated image:
> http://raspis.ddscentral.org/pub/downloads/android-x86/Android-x86-Kitkat_Baytrail-K4.0_Apr20.rar
>
> Regarding the kernel:
> IIRC, I changed these 2 options (maybe there are more, see attached
> configuration).
> CONFIG_GPIO_CRYSTAL_COVE=y
> CONFIG_PINCTRL_BAYTRAIL=y

OK, I saw the gap.
By adding them the audio and vol buttons work.
I can see how it affects the vol buttons,
but can't see why the audio is affected.
The audio just works in kernel 3.18.x
even without these configs.
Need to dig into deeper...

The reason I didn't set CONFIG_PINCTRL_BAYTRAIL=y
is it makes the camera led be always on.
So I'm reluctant to add it...
Any idea to fix it?

> I modified kernel/firmware/Makefile to add firmware files for sound and
> wireless to the final image.
> (patch attached, you should already have the firmware files, but if you need
> them, let me know).

I don't think it's necessary.
All of them are in device/generic/common/firmware/
of android-x86 tree.
Anyway, thank you for pointing it out.

> I re-added the Android logger.
>
> Regarding backlight:
> Intel patches which you've committed didn't work. There were 2 entries in
> /sys/class/backlight: acpi_video0 and acpi_video1, but neither of them
> actually worked.

Same for me.
The patches are applied to Fedlet kernel:

https://www.happyassassin.net/cgit/kernel/commit/?h=baytrail&id=99ec7a3a1710e8d24e4bd6cd17ac8b0945fbc66c

So I thought it should work. But seems it's not for T100.
Will try to contact the author for help.

> I have manually applied the old force_backlight_pmic patch against the
> current git source to get backlight working (patch attached).
> The patch is Bay Trail specific and perhaps would have to be redone to not
> affect other devices (e.g. by adding a config option) to be suitable for
> committing to git.

OK. I knew that patch but reluctant to use it
because it generates some kernel warnings I dislike.

My plan is still to to make Shobhit Kumar's patch work
for T100. I think it's the right direction to go and
acceptable by the upstream:

http://lists.freedesktop.org/archives/intel-gfx/2015-March/061908.html

> Also, I found a few bugs:
> There seems to be a bug in i915 which causes strange blinking in Play Store
> app pages.
> I see errors like this in dmesg:
> [i915]] *ERROR* Atomic update failure on pipe A (start=<number>
> end=<number>)
>
> Second, the clock display does not update in launcher, but works fine in
> lock screen.
> (not sure if this is kernel related)

No idea yet...

[Jianlong Liu]

I don't think it's supposed to be working yet. He mentions (in the bug report for the Venue 8 Pro) that once there is a proper fix, brightness control should work (meaning it doesn't yet?).

On Monday, April 20, 2015 at 8:33:22 PM UTC-7, Chih-Wei Huang wrote:

Same for me.
The patches are applied to Fedlet kernel:

https://www.happyassassin.net/cgit/kernel/commit/?h=baytrail&id=99ec7a3a1710e8d24e4bd6cd17ac8b0945fbc66c

So I thought it should work. But seems it's not for T100.
Will try to contact the author for help.

My plan is still to to make Shobhit Kumar's patch work
for T100. I think it's the right direction to go and
acceptable by the upstream:

http://lists.freedesktop.org/archives/intel-gfx/2015-March/061908.html

[Povilas Staniulis]

There's another patch that should be reversed for KitKat:

[PATCH 1/2][RFS] staging: Remove the Android alarm-dev driver

http://www.spinics.net/lists/kernel/msg1908029.html

Lollipop no longer uses the /dev/alarm device, but I believe Kitkat still does.

This may be the cause of the time display problem in Kitkat.

[Jianlong Liu]

Hm. With your image, my Venue 8 Pro boots up fine, but when I'm trying to compile (using your patches and kernel config), it seems that as soon as I use CONFIG_GPIO_CRYSTAL_COVE=y, I boot into a blank screen with no Android logo. But as soon as I change it into the default =m, it boots fine.

Any ideas? I tried both with the stock mesa 10.4.2 and 10.6-devel.

[Povilas Staniulis]

On 2015.04.21 21:01, Jianlong Liu wrote:
> Hm. With your image, my Venue 8 Pro boots up fine, but when I'm trying
> to compile (using your patches and kernel config), it seems that as
> soon as I use CONFIG_GPIO_CRYSTAL_COVE=y, I boot into a blank screen
> with no Android logo. But as soon as I change it into the default =m,
> it boots fine.
> Any ideas? I tried both with the stock mesa 10.4.2 and 10.6-devel.
>
>

Strange. No problems on my T100TA.
It could be that I forgot some changes. I can upload the entire kernel
source if that will help you.

[Jianlong Liu]

It definitely would if it's not too much trouble!

[Povilas Staniulis]

--

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 http://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

OK, uploading it, will post a link when done.

[Ramana Kumar]

@Povilas

newbie question...  do I need to convert rar file to iso and then use rufus to build a bootable usb?  are there any instructions to do it?

Thanks

Rockmeister

On Thu, Apr 16, 2015 at 10:24 AM, Povilas Staniulis <wdmo...@gmail.com> wrote:

On 2015.04.16 15:28, T100TA User wrote:

Please do. Much obliged!

Here you are:
http://raspis.ddscentral.org/pub/downloads/android-x86/Android-x86-Kitkat_Baytrail-K4.0_Apr16.rar

Extract the archive to a USB flash drive.

Sorry for the delay, my upload speed is dog slow.

[midi jari]

What rar file?

You received this message because you are subscribed to a topic in the Google Groups "Android-x86" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/android-x86/sSG1e_dLf7o/unsubscribe.
To unsubscribe from this group and all its topics, send an email to android-x86...@googlegroups.com.

[Povilas Staniulis]

On 2015.04.22 02:54, Ramana Kumar wrote:
> @Povilas
> newbie question... do I need to convert rar file to iso and then use
> rufus to build a bootable usb? are there any instructions to do it?
> Thanks
> Rockmeister

The only instruction is: Extract the archive to a USB flash drive.

[Ramana Kumar]

@Pavilas

Thanks a bunch

Rockmeister

[Povilas Staniulis]

On 2015.04.21 23:53, Jianlong Liu wrote:

--

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 http://groups.google.com/group/android-x86.
For more options, visit https://groups.google.com/d/optout.

Here's the kernel from my last build:
http://raspis.ddscentral.org/pub/downloads/android-x86/dev/kernel_wgithist_Apr21.tar.xz

Backup of the git repo and my changes.

This kernel based on current git revision (except for the last 2 commits).
My patches are in the "patches" directory.

A build based on this kernel (and all my other Android x86 files) can be found here:
http://raspis.ddscentral.org/pub/downloads/android-x86/

[Jianlong Liu]

Thanks!

I tested all 3 of your images, and only the April 16 one works. April 20/21 and compiling using your uploaded kernel didn't change anything :( (boots to black screen).

Not sure what happened between the April 16 and the April 20 images to make the tablet boot to black screen.

It's just the display not coming up though, apparently the touchscreen and the virtual terminal work (touch sounds register, and Alt+F1 followed by typing in poweroff works to turn it off, I just have to do it blind, since the screen is blank in both cases).

[Сергей Трофимов]

Could you compile the driver accelerometer and wifi under your kernel.
accelerometer - https://github.com/hadess/iio-sensor-proxy
wifi - https://github.com/hadess/rtl8723bs
Or tell me how to compile wifi driver for your kernel.

среда, 22 апреля 2015 г., 6:02:35 UTC+3 пользователь Povilas Staniulis написал:

[Povilas Staniulis]

On 2015.04.22 08:47, Jianlong Liu wrote:
> Thanks!
> I tested all 3 of your images, and only the April 16 one works. April
> 20/21 and compiling using your uploaded kernel didn't change anything
> :( (boots to black screen).
> Not sure what happened between the April 16 and the April 20 images to
> make the tablet boot to black screen.
> It's just the display not coming up though, apparently the touchscreen
> and the virtual terminal work (touch sounds register, and Alt+F1
> followed by typing in poweroff works to turn it off, I just have to do
> it blind, since the screen is blank in both cases).
>
>

The April 16th image does not contain byt-fixes and mmc-fixes (mmc,
brcmfmac, intel patches), except for two small patches to fix mmc and
brcmfmac freezing.
Maybe those Intel patches broke something on your device...

[Povilas Staniulis]

On 2015.04.22 13:30, Сергей Трофимов wrote:
> Could you compile the driver accelerometer and wifiunder your kernel.
> accelerometer - https://github.com/hadess/iio-sensor-proxy
That's not the driver, that's just a proxy which depends on GNOME.
Android needs a hal driver for accelerometer.

> wifi - https://github.com/hadess/rtl8723bs
> Or tell me how to compile wifi driver for your kernel.

Build it against the kernel I've uploaded.

Change these in the Makefile
KVER := "4.0.0-android-x86+"
KSRC := path to kernel build dir
MODDESTDIR:= path to modules

[rbg]

FWIW:

from my testing....

April 16 kernel 4.0 was working fine for my lollipop 5.1.0_r3 builds I was testing with...

patches introduced after the 16th which seem to be T100 specific break those builds, so

I had rolled back those patches to continue my testing....BUT I'm NOT testing on T100...

am testing on basic Intel machines...

[IamBackX5]

Hi, of cource this is not priority problem, but some people who test your image on T100 says that they have a problems with video playback...

Translate: In Youtube lags HD-video even 720p. 480p seems to be OK. The problem is not WiFi, this is video playback - it loads quickly, but playing lag. 

Original message: В Youtube тормозят HD-видео, даже 720p. 480p вроде нормально. Проблема не в вайфае, а в самом воспроизведении видео - оно загружается быстро, но вопроизведение лагает.

Just for info, nothing more... then you or Chih-Wei Huang will have a time it will be fixed, i know it :)

четверг, 16 апреля 2015 г., 20:24:20 UTC+5 пользователь Povilas Staniulis написал:

On 2015.04.16 15:28, T100TA User wrote:
> Please do. Much obliged!
>
>
Here you are:
http://raspis.ddscentral.org/pub/downloads/android-x86/Android-x86-Kitkat_Baytrail-K4.0_Apr16.rar

Extract the archive to a USB flash drive.

[Jianlong Liu]

Eh, I reverted to the commit on April 13 (a2c27b6fcaad51114faffe4284098dd0a4764b45), and things are fine when compiling with CONFIG_GPIO_CRYSTAL_COVE=y.

As soon as I apply the 9 patches by Shobhit Kumar, black screen.

Eh.

[IamBackX5]

And i forgot three more problems:

Original message: 

Иногда пропадает звук на несколько секунд, особенно под нагрузкой, во время установки приложений.
При подключении наушников звук идёт и в наушниках, и из динамиков.
Планшет не выключает экран при закрытии крышки, но "просыпается" при открытии. 

 

Translate: 

 Sometime sound is gone, especially under load during application installation. 

 Then headsets plugged-in, sound plays from headsets and from speakers

 Then T100 installed to dock, if you close it, display still works, but if display is off and you open tablet, then display on.

And thanks to you and Chih-Wei Huang for working with T100 ;)

среда, 22 апреля 2015 г., 22:41:41 UTC+5 пользователь IamBackX5 написал:

[rbg]

yes checking my notes...I only had to roll back

CONFIG_PWM_CRC=y

and things were ok....remember doing a quick read on CONFIG_PWM_CRC and it had to have stuff rolled up with

i915 driver so just arbitrated it to that issue and figured they'd fix it eventually

[Povilas Staniulis]

On 2015.04.22 22:49, IamBackX5 wrote:

And i forgot three more problems:

Original message: 

Иногда пропадает звук на несколько секунд, особенно под нагрузкой, во время установки приложений.
При подключении наушников звук идёт и в наушниках, и из динамиков.
Планшет не выключает экран при закрытии крышки, но "просыпается" при открытии. 

 

Translate: 

 Sometime sound is gone, especially under load during application installation. 

 Then headsets plugged-in, sound plays from headsets and from speakers

 Then T100 installed to dock, if you close it, display still works, but if display is off and you open tablet, then display on.

This is from 4pda...
I'll post the link so others who can read Russian (or know how to use Google Translate) can follow the thread.
http://4pda.ru/forum/index.php?showtopic=611095&st=680

I know about the sound issue. Probably related to system load.
No idea how to fix...

About the display, probably the device reacts to lid close as input so turns the display on.
Ideally the device should go to standby when lid is closed, but standby doesn't work yet.

And thanks to you and Chih-Wei Huang for working with T100 ;)

среда, 22 апреля 2015 г., 22:41:41 UTC+5 пользователь IamBackX5 написал:

Hi, of cource this is not priority problem, but some people who test your image on T100 says that they have a problems with video playback...

Translate: In Youtube lags HD-video even 720p. 480p seems to be OK. The problem is not WiFi, this is video playback - it loads quickly, but playing lag. 

Original message: В Youtube тормозят HD-видео, даже 720p. 480p вроде нормально. Проблема не в вайфае, а в самом воспроизведении видео - оно загружается быстро, но вопроизведение лагает.

Just for info, nothing more... then you or Chih-Wei Huang will have a time it will be fixed, i know it :)

--

No idea about this...

[Jianlong Liu]

Didn't work for me :(. Commenting it out while leaving the Crystal Cove option =y still boots to black screen.

[rbg]

so you must be using a different video driver than I am.....mine is i915 as I recall

what's the driver for Intel-SOC??

I still have Crystal Cove option=y in my defconfig and changed CONFIG_PWM_CRC=n

and works fine on 3 different Intel machines...no black screens...

and in double checking again I said April 16 kernel was working fine but as you found out it

was the April 13 kernel that was working fine...the April 16 one is the one that caused me issues so I rolled back

to the April 13 one...

[hombre]

sound works with this kernel?

среда, 22 апреля 2015 г., 13:30:42 UTC+3 пользователь Сергей Трофимов написал:

[Jianlong Liu]

Mine should be i915 as well (Atom Z3740D).

Ah, it might be the sequence of 9 patches by Shobhit Kumar that's causing your problems. It's starting to look like it from my end, although I haven't tried explicitly stating CONFIG_PWM_CRC=n.

[rbg]

no I think we're on different pages.....I only made my initial comment to back up what you had said

about "I tested all 3 of your images, and only the April 16 one works." and I should have just left it at that....sorry....

the only difference between my apr16 defconfig pulled from android-x86 git for kernel 4 and my apr13 defconfig pulled from

android-x86 git for kernel 4.0 is the CONFIG_PWM_CRC=y....which I removed from the apr16 defconfig to get my build working.....

I now realize you appear to be using the kernel from Povilas which has about 7 differences from apr13/apr16 defconfig that I have.....

I have NOT used that kernel.....it all appears to be geared to T100.....and some of the stuff has since been merged into android-x86 4.0

but in using it my build would break again so I been sticking with the apr13 one that works...no present need for T100 changes if they are

causing issues to my testing.....

[Chih-Wei Huang]

2015-04-22 23:50 GMT+08:00 rbg <rbg.hp...@gmail.com>:
> FWIW:
> from my testing....
>
> April 16 kernel 4.0 was working fine for my lollipop 5.1.0_r3 builds I was
> testing with...
>
> patches introduced after the 16th which seem to be T100 specific break those
> builds, so
> I had rolled back those patches to continue my testing....BUT I'm NOT
> testing on T100...
> am testing on basic Intel machines...

What was broken?
I didn't see any regression on Intel's machine.
I've tested the kernel on different Intel's devices
including Atom N450 (i915) and
IvyBridge, Haswell and Baytrail (i965).
So far so good, though the backlight control
doesn't work on T100 as my expected.

What devices did you tested and what errors did you see?

[Jianlong Liu]

I definitely should have been more clear. Other than that one time using Povilas's kernel, I've been using the one on git.android-x86.org (with certain commits reverted for testing).

I guess the issues we're having are different after all?

By the way, just out of curiosity, on the builds you had the black screens, does the machine still work? As in, if you blindly switch via Alt+F1, type in "poweroff", does it turn off?

Mine does, so as far as I can tell, it's only the screen that doesn't work on my tablet. Tapping the screen at where the soft buttons are supposed to be also produces touch sounds.

[rbg]

asus PBZ77-M motherboard , 4GB mem, intel core i3-3225

    Integrated Intel Graphics HD 4000

asus P5G41-M motherboard , 3GB mem, intel core 2 duo E7500

    Integrated Intel Graphics Media Accelerator(Intel GMA X4500)

HP T60G laptop, 3GB mem, Intel Pentium Dual-Core T4300

    Intel Graphics Media Accelerator 4500MHD

and as I remember only the i3 had issues....black screen on bootup

[rbg]

yes machine still works...went to debug mode and looked through dmesg & logcat to try and see what was going on....

there was a revert that Chih-Wei spoke of depending on what version of mesa your using....don't remember where but

it's in the forums here...something put in 5.0.2 but not carried to 5.1.0........

[Tom Briden]

For anyone else struggling to get this booting in an efi image (so 64bit kernel/32bit userspace), change the android_x86_64_defconfig as so...

diff --git a/arch/x86/configs/android-x86_64_defconfig b/arch/x86/configs/android-x86_64_defconfig
index bfe5b73..6f9ff3e 100644
--- a/arch/x86/configs/android-x86_64_defconfig
+++ b/arch/x86/configs/android-x86_64_defconfig
@@ -5551,7 +5551,7 @@ CONFIG_THUNDERBOLT=m
 #
 CONFIG_ANDROID=y
 CONFIG_ANDROID_BINDER_IPC=y
-# CONFIG_ANDROID_BINDER_IPC_32BIT is not set
+CONFIG_ANDROID_BINDER_IPC_32BIT=y
 
 #
 # Firmware Drivers