marshmallow-x86 branch is ready

[Chih-Wei Huang]

Hi devs,
The marshmallow-x86 branch based on
Android 6.0 is ready to download.
See http://www.android-x86.org/getsourcecode

I only fixed all building errors of the
32-bit target android_x86 so
you can build the iso_img.
It's not even bootable to home.
64-bit target is not tested yet, either.
But it's useful for whose who want
to work on the marshmallow-x86 porting.

About the git server status.
It seems the server is overloading.
The errors happened most are out of memory,
especially on downloading big repositories
like kernel or frameworks/base.

For developers who really need to download
the source, please download the AOSP first

repo init -u https://android.googlesource.com/platform/manifest -b
android-6.0.0_r1
repo sync -c

then download marshmallow-x86 to the same directory:

repo init -u git://git.android-x86.org/manifest -b marshmallow-x86
repo sync

In this way only differences between AOSP and
marshmallow-x86 will be download.
You can also download marshmallow-x86 from
an old lollipop-x86 directory. It only downloads
the differences as well.
DO NOT make a fresh download.
It will probably never succeed.

If you are not a developer, please DO NOT download
the source code from git.android-x86.org now.
Please leave the resource to the real developers.
Just use the unofficial mirror posted in the forum.

I'll setup the an official mirror soon.


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

[fgdn17]

???

fgdn17@ubuntu-14LTS:~/MARSH-SYNC-6.0.0_r1$ 

fgdn17@ubuntu-14LTS:~/MARSH-SYNC-6.0.0_r1$ repo init -u http://git.android-x86.org/manifest -b marshmallow-x86

fatal: repository 'http://git.android-x86.org/manifest/' not found

fgdn17@ubuntu-14LTS:~/MARSH-SYNC-6.0.0_r1$ date

Mon Oct 12 22:22:50 EDT 2015

was something missed???

[David Brubaker]

Thanks!!

[Geek Till It Hertz]

So i finally got the time to compile it and it compiled well enough, although booting it was a different story.

The iso itself booted well upto the point where the android logo should have appeared, instead i just got a static white cursor..... then nothing. 

Upon running it in DEBUG mode it kept looping this output:

[Povilas Staniulis]

Looks like the SurfaceFlinger is crashing with SIGABRT.
Can you get to the shell (IIRC, Ctrl+Alt+F1) or the boot is just stuck at init phase ?

--
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.

[Geek Till It Hertz]

I can get to the shell in DEBUG mode, its only after it asks me to exit the second time that the said error occurs.

[Chih-Wei Huang]

2015-10-17 20:42 GMT+08:00 Geek Till It Hertz <sahaj...@gmail.com>:
> I can get to the shell in DEBUG mode, its only after it asks me to exit the
> second time that the said error occurs.

I've said t's just the very early stage porting.
It's just buildable.
The boot to home porting is not finished yet.
I'm still working on it.

You may try to help,
or just wait another announcement.

[rbg]

wow.....what use to build in 3-31/2 hours now takes 11-12 hours....what am I missing here...???

Have things changed by google that much??? This jack & jill thing takes forever even if I change

.jack's SERVER_NB_COMPILE from 4 to 2 doesn't help much.....any tricks to speed it up short of getting a newer

system which isn't going to happen......3ghz dual core cpu with 3gb memory...has been reasonable since

2011 time frame with builds times. Any help / tricks / hacks???

On Saturday, October 10, 2015 at 3:51:13 PM UTC-4, Chih-Wei Huang wrote:

[Bradford Lilly]

On Saturday, October 10, 2015 at 3:51:13 PM UTC-4, Chih-Wei Huang wrote:

I've looked into this a little (and about to be a lot more). I'm full well aware this isn't working yet, but figured I'd post some of what I'm seeing here for other / debugging. It seems there are a couple things going on:

* grub doesn't install. I noticed that grub-installer wasn't present on the install medium
* To get around this, I first installed android-x86-4.4, opted NOT to format, then did the upgrade
* The upgrade did not modify the old grub (as expected), so during boot you need to modify the grub options to point to the correct kernel / initrd
* I'm running this in a vm, so I need to add qemu=1 to the kernel line
* I boot to debug mode, set an ip address, then let android try to start.
* When I see the "ANDROID" logo rolling on the screen, adb connect / logcat
* Logcat indicates that it's having trouble with EGL. Not a surprise. I see the following:
11-17 21:02:17.898 1088 1088 I SurfaceFlinger: EGLSurface: 5-6-5-0, config=0x0
11-17 21:02:17.899 1088 1088 D SurfaceFlinger: Set power mode=2, type=0 flinger=0xb70e2000
11-17 21:02:17.932 1088 1089 E SurfaceFlinger: ro.sf.lcd_density must be defined as a build property
11-17 21:02:17.932 1102 1106 D libEGL : Emulator without GPU support detected. Fallback to software renderer.
11-17 21:02:17.932 1102 1106 D libEGL : loaded /system/lib/egl/libGLES_android.so

I think the take away there is the config=0x0. I know that there was a commit in previous version (8d7c775a37640af2c4081ad03b23b27184deba19) from Andy Ross that resolved this (at least for me) in the past. Unfortunately, it looks like the EGLConfig algorithm has changed (83835359e51ddb8be37cea9bf4bb32f9390d82b7), so this might need to be tweaked.

Like I said, in the next few weeks I'm going to be digging into this full bore. Is there a working group for this port (or a topic dedicated to it)? I'd be happy to work with you to get this done.

[Chih-Wei Huang]

2015-11-18 5:34 GMT+08:00 Bradford Lilly <bradford.r...@gmail.com>:
> On Saturday, October 10, 2015 at 3:51:13 PM UTC-4, Chih-Wei Huang wrote:
>
> I've looked into this a little (and about to be a lot more). I'm full well aware this isn't working yet, but figured I'd post some of what I'm seeing here for other / debugging. It seems there are a couple things going on:
>
> * grub doesn't install. I noticed that grub-installer wasn't present on the install medium

It's surprised. I don't expect the installer need any change
since it doesn't depend on the android release.
But I admit I didn't test it yet.

> * To get around this, I first installed android-x86-4.4, opted NOT to format, then did the upgrade
> * The upgrade did not modify the old grub (as expected), so during boot you need to modify the grub options to point to the correct kernel / initrd
> * I'm running this in a vm, so I need to add qemu=1 to the kernel line

This is not necessary, or we have a bug to be fixed.

> * I boot to debug mode, set an ip address, then let android try to start.
> * When I see the "ANDROID" logo rolling on the screen, adb connect / logcat
> * Logcat indicates that it's having trouble with EGL. Not a surprise. I see the following:
> 11-17 21:02:17.898 1088 1088 I SurfaceFlinger: EGLSurface: 5-6-5-0, config=0x0
> 11-17 21:02:17.899 1088 1088 D SurfaceFlinger: Set power mode=2, type=0 flinger=0xb70e2000
> 11-17 21:02:17.932 1088 1089 E SurfaceFlinger: ro.sf.lcd_density must be defined as a build property
> 11-17 21:02:17.932 1102 1106 D libEGL : Emulator without GPU support detected. Fallback to software renderer.
> 11-17 21:02:17.932 1102 1106 D libEGL : loaded /system/lib/egl/libGLES_android.so
>
> I think the take away there is the config=0x0. I know that there was a commit in previous version (8d7c775a37640af2c4081ad03b23b27184deba19) from Andy Ross that resolved this (at least for me) in the past. Unfortunately, it looks like the EGLConfig algorithm has changed (83835359e51ddb8be37cea9bf4bb32f9390d82b7), so this might need to be tweaked.

Basically speaking, it's not ported yet.

For graphic porting, we have two major tasks:
* For supported GPUs, enable Mesa support
* For non-supported GPUs, enable the software renderer.

Currently I only focus on the Mesa porting.
But it's blocked (totally broken) because
Google removed the workaround 10194508.
I've listed more details in another post.
Due to my limited knowledge for these topics
I can hardly move forward. Unfortunately
no Mesa developer is willing to help so far.

For the software renderer, the key point is to patch
the framework to enable or disable OpenGL renderer
at runtime. It's not done yet.
If you are interesting in vm, you can take this part.

> Like I said, in the next few weeks I'm going to be digging into this full bore. Is there a working group for this port (or a topic dedicated to it)? I'd be happy to work with you to get this done.

Thank you for the willing to help.
Except you and me I'm not aware of anyone is working on it.
You can create another topic for a more specific question,
of course.

[stavros daliakopoulos]

https://github.com/karthick111/android_frameworks_native-1/commit/96fa12aadda21979e7bf4fe02c344aebb9e49276

 

[Geek Till It Hertz]

With the recent bunch of commits, i synced my repo now and this is what i got.

[PS. I know this is a WIP, i am not asking for anything just a friendly update of the progress on marshmallow-x86]

I was able to get to the android boot animation while running it on virtual box, i selected to normal live boot option.

On both of my bare-bone x86 machines (C2D and i3-2120) were able to get into the android boot animation in Vesa mode only.

I have also attached the debug mode output, it automatically went into the debug console.

On Sunday, October 11, 2015 at 1:21:13 AM UTC+5:30, Chih-Wei Huang wrote: