account arrowhead-down arrowhead-up mobile-menu search sm-bold-x x-skinny-rounded x-skinny arrowhead-right social-facebook social-googleplus social-instagram social-linkedin social-pinterest social-qzone social-renren social-tencent social-twitter social-vkontakt social-weibo social-youku social-youtube

Please help us improve our website

Take our customer survey to evaluate your visit.

It should only take a few minutes to answer five quick questions. Just click the Launch survey button at the end of your visit to begin.

cancel
Showing results for 
Search instead for 
Did you mean: 
Welcome to the Open Devices community!

This is where we discuss open source and development related questions on our Open Devices initiative.

Porting kernel to custom ROM

Talent
Message 21 of 29
2,355 Views
Message 21 of 29

Re: Porting kernel to custom ROM

Thank you for taking the time to point it out

 

EDIT:

I read somewhere some binaries are needed, but I have included aries and shinano vendor blobs.

Am I missing something?

 

If anyone has link to useful information regarding the issue (article, guide,etc.) I would be grateful.

 

 

Regular
Message 22 of 29
2,321 Views
Message 22 of 29

Re: Porting kernel to custom ROM


@rancid wrote:

Thank you

 

I guess that will work for DU, which has a separate branch for AOSP and CAF.

But, for omnirom there is only one branch (android-5.1)

How do I make sure I am using AOSP HALs and not CAF, since I am using AOSP trees there should not be referencing any CAFs?

 

I am building userdebug but adb logcat always hangs on 'waiting on device',

will try dmesg hopefully it works.

 

Thank you for your advice


Standard AOSP HALs are the default on EVERY project I'm aware of.

 

In the case of Omni, the CAF repos aren't even synced unless a device pulls them in.  Even if they're present, they won't be used unless explicitly enabled with BoardConfig directives that are NOT in Sony's AOSP repos.

 

"no EGL implementation" makes me think that there's something wrong with your vendor blob repos.

 

Right now we have leo and (I believe) sirius working.

 

I need to sync a vanilla AOSP tree so I can determine whether some of the issues I'm having are common to vanilla AOSP or not, and also to potentially push some fixes.

 

Is there a list of known issues being worked on by device?  I get the following issues that I'm in the process of investigating, it would be useful to know if these have been observed by others on leo (D6603 specifically):

1) Periodic display corruption.  It looks similar to what happens in an MDP underrun, except instead, it appears as noise.  Always at the bottom of the screen physically, NOT affected by display rotation.  It usually appears and then "slides downward" until it disappears - I'll post some logs Wednesday if I don't make any progress on my own, I haven't seen anything in the logs so far indicating what's going on.

 

2) Screen recording not working - This is the logcat that results from:

adb logcat -c; adb shell screenrecord /sdcard/corruption.mp4; adb logcat -v time -d > recordlog.txt

 

http://paste.omnirom.org/view/0c8a4a50

 

This is output by adb shell:

Unable to get output buffers (err=-38)
Encoder failed (err=-38)

 

3) Bluetooth LPM causes a constant kernel wakelock - BTLowPower and <I'll update this later, it looked like it was one of the MSM serial channels>

 

4)  If modem firmware is pulled from a stock firmware, radio works including data (actually I haven't tested call audio yet, I should do that...), but it only supports HSPA, not LTE.  Have not had time to investigate this yet.

Highlighted
Developer World
Message 23 of 29
2,314 Views
Message 23 of 29

Re: Porting kernel to custom ROM

thanks for logs

is great to see you here

Regards

Alin

Regular
Message 24 of 29
2,298 Views
Message 24 of 29

Re: Porting kernel to custom ROM

Solved the LTE issues - needs an overlay of packages/services/Telephony (I'll upload this later, but it's pretty much copypastaing the overlay from any other LTE device)

 

Also solved the BT wakelock issues - There's a multitude of issues with the current BT config.  (For example, BOARD_BLUEDROID_VENDOR_CONF is not set in the shinano BoardConfig, so bluetooth/vnd_generic.txt is not actually used, but there are some other issues too - kernel defconfig needs CONFIG_BT_BCM4339 replaced with CONFIG_BT_MSM_SLEEP, and init.shinano.rc needs to set the permissions on the proper BT wake sys entries) - I'll try to push up the patches when I get home from work tonight.

 

Display corruption issue occurs even in recovery - indicating this may be a DSI config issue, so I've got an idea where to start.  I probably won't be able to touch this again until Wednesday.

 

However, once BT wakelock issues are solved, the device has a very bad tendency to enter SoD (sleep of death) states.  I forgot to see if ADB was still active in this case, another thing for Wednesday.  I have a question - on slightly older Xperias (like the original Z), holding power+VolUp would first give one vibration, at which point the device would reboot if you let go, and three vibrations to power off.  The Z3 goes straight to three vibrations to power off.  How do you reset a hung device (for example, to obtain a last_kmsg after an SoD)?

Talent
Message 25 of 29
2,290 Views
Message 25 of 29

Re: Porting kernel to custom ROM

Great news.

 

Omnirom boots.

It might have been the bootrec using event3 instead of event2.

I fixed bootrec and also merged  yesterdays sony upstream changes and it boots.

 

One thing about this commit, Fix copying of multiple dtb files.

I do not believe it is needed as kernel compiles without it, as it is already included in shinano-common Android.mk,

It is redundant unless of course you want to remove it from there due to organization.

 

UPDATE:

I believe it was booting directly to recovery before, due to bootrec.

I am unable to go to recovery, I get the flashing blackscreen issue when I try.

Regular
Message 26 of 29
2,280 Views
Message 26 of 29

Re: Porting kernel to custom ROM


@rancid wrote:

Great news.

 

Omnirom boots.

It might have been the bootrec using event3 instead of event2.

I fixed bootrec and also merged  yesterdays sony upstream changes and it boots.

 

One thing about this commit, Fix copying of multiple dtb files.

I do not believe it is needed as kernel compiles without it, as it is already included in shinano-common Android.mk,

It is redundant unless of course you want to remove it from there due to organization.

 

UPDATE:

I believe it was booting directly to recovery before, due to bootrec.

I am unable to go to recovery, I get the flashing blackscreen issue when I try.



For the recovery bootrec mess:

https://gerrit.omnirom.org/#/c/12818/ and https://gerrit.omnirom.org/#/c/12817/

 

For recovery to actually display:

https://gerrit.omnirom.org/#/c/12812/

Talent
Message 27 of 29
2,264 Views
Message 27 of 29

Re: Porting kernel to custom ROM


@Entropy512 wrote:
For recovery to actually display:

https://gerrit.omnirom.org/#/c/12812/


This does not seem to work for me.

I still get the flashing black screen when I try to go to recovery.

[60122.158386]  loop0: unknown partition table

EDIT:

Have been told that it is a kernel issue. I will test with 3.4 kernel and check results.

 

The keycheck works well, though.

 

Compiling OmniRom with 3.10 kernel and the addition of modem files from 23.1.A.0.690 everything works well, apart from the audio routing issues and delays caused by them.

 

But,

when I compile DU with a similar set up, I keep getting no sim card detected;

logcat: http://pastebin.com/du082phn

dmesg: http://pastebin.com/tebYMmMb

 

 

Talent
Message 28 of 29
2,186 Views
Message 28 of 29

Re: Porting kernel to custom ROM

Built on the 18th with latest binaries and aries - 23.1.A.0.690 (D5803) modem files:

 

Screenshot_1970-08-14-19-52-11.png

 

Recovery does not work, LED stays red while screen stays blank

Recovery dmesg: http://hastebin.com/luduzotase.xml

Talent
Message 29 of 29
2,119 Views
Message 29 of 29

Re: Porting kernel to custom ROM

OmniRom with Sony AOSP changes on May 18th, (Tested on Aries):

SIM Card not detected, I get No Service, ( Using SuperBenevolent ).

Dmesg: http://hastebin.com/vozupumuyu.xml

Logcat: http://hastebin.com/joduseseri.vhdl

 

TWRP does not boot, all I get is black screen which flashes sometimes,

Dmesg: http://hastebin.com/qobogayibe.xml

 

Tested with same Sony AOSP changes but with DU,

Same result No Service

Dmesg: http://hastebin.com/odekunalap.xml

Logcat: http://hastebin.com/vupuxohame.vhdl

 

TWRP Dmesg: http://hastebin.com/akozibazej.xml

 

It might be due to changes made on the 18th,

(UPDATE: AOSP Build with latest changes also has the above problem - See Bugs section for logs)

as AOSP Build from earlier in the morning has service:

 

Aosp(working),

Dmesg: http://hastebin.com/esudujejav.xml

Logcat: http://hastebin.com/quvatekako.coffee

Issue Fixed