Jump to content
PaulOBrien

Advent Vega kernel source code now available!

Recommended Posts

@cass & eduardo

Putting the ppp thingy aside, in our ICS build, was it built using LTE based network or GSM? because, if it was GSM, the messaging apps MMS.apk should be able to send out sms and receive call. I did not manage to send out sms even when i got the tablet to register to network. I dont know if i read correctly but i read somewhere that the RILD binary have 2 different version. one for LTE and one for GSM. From the log also, i see some LTE logs being recorded.

Its not built to specifically use anything, the profile its using is Wingray so thats the wifi only Xoom, this should not matter though id not susect as long as we have the ril in place ... if you can connect to a network try using AT commands to dial a number .. you should find how somewhere in google .. that will at least isolate if its a gui issue or other...

also a full logcat -b radio may help

Edited by Cass67

Share this post


Link to post
Share on other sites

fix posted above ;) -- Let's hope!

Unfortunately not it seems :(

guess i need to add the emergency_restart() function ...

Edited by Cass67

Share this post


Link to post
Share on other sites

Eduardo,

The suspend problem does seem familiar; maybe this may help as a reminder.

Billy..

Posted 24 July 2011 - 12:11 PM

Hi Eduardo , great work.

1. wlan under android solved

2. pwm issue solved.

3. battery status works under android

When CONFIG_CONSOLE_EARLYSUSPEND is set, pwm backlight control doesn´t work Set instead CONFIG_FB_EARLYSUSPEND , then pwm control works as expected

The git repo is up to date.

This post has been edited by rebel1: 24 July 2011 - 04:40 PM

We have that set already ..

Cheers

Cass

Share this post


Link to post
Share on other sites

Unfortunately not it seems :(

guess i need to add the emergency_restart() function ...

We will have to continue investigation ... :S - But, according to some logs i already have, this mmc_delayed_work wakelock was preventing suspension... Now we need to figure out if there is something else ...

Share this post


Link to post
Share on other sites

2nd try: I backported from the tuna kernel (3.0) the wakelock handling for the mmc stack ... So we exactly duplicate the fix ... I will also try it later.. To be honest, i think the tuna kernel wakelock handling of the mmc was reengineered, and should be working ...

suspend_v3.rar

Share this post


Link to post
Share on other sites

2nd try: I backported from the tuna kernel (3.0) the wakelock handling for the mmc stack ... So we exactly duplicate the fix ... I will also try it later.. To be honest, i think the tuna kernel wakelock handling of the mmc was reengineered, and should be working ...

On first try it does not boot ... ill look again in a bit ...

Share this post


Link to post
Share on other sites

On first try it does not boot ... ill look again in a bit ...

No problem, i will also try it a little bit later ... I am curious to know why the kernel is not entering suspend...

Share this post


Link to post
Share on other sites

No problem, i will also try it a little bit later ... I am curious to know why the kernel is not entering suspend...

Hehe yeah stuff like this keeps it interesting eh :-)

Share this post


Link to post
Share on other sites

No problem, i will also try it a little bit later ... I am curious to know why the kernel is not entering suspend...

Ok tried after i did a make clean and it appears to sleep then wake up ... but it switches off again ... it continues to do so when i press the button .. so on the face if it we have success :)

Just have to ensue the keyboard repeat is off and debounce is set to 50 as i see this behaviour with these in place ... ill update in a bit

edit indeed, thats what it was ... suspend appears to work .. woot woot !

can you change your code to be

.rep = false, /* auto repeat enabled */

.gpio = SHUTTLE_KEY_POWER,

.active_low = true,

.debounce_interval = 50,

in board-shuttle-keyboard.c so i dont have to do it every update, i get stung all the time ;)

Nice work again !!

edit - hmm more tests needed .. resumed a few times without issue then after leaving it for a while its hung again :(

edit - grrr where the hell in this gpio short long kb do i put the emergency_restart() cant make it restart when it hangs ... can make it restart all other times though so thats something :)

Cheers

Cass

Edited by Cass67

Share this post


Link to post
Share on other sites

Ok tried after i did a make clean and it appears to sleep then wake up ... but it switches off again ... it continues to do so when i press the button .. so on the face if it we have success :)

Just have to ensue the keyboard repeat is off and debounce is set to 50 as i see this behaviour with these in place ... ill update in a bit

edit indeed, thats what it was ... suspend appears to work .. woot woot !

can you change your code to be

.rep = false, /* auto repeat enabled */

.gpio = SHUTTLE_KEY_POWER,

.active_low = true,

.debounce_interval = 50,

in board-shuttle-keyboard.c so i dont have to do it every update, i get stung all the time ;)

Nice work again !!

edit - hmm more tests needed .. resumed a few times without issue then after leaving it for a while its hung again :(

edit - grrr where the hell in this gpio short long kb do i put the emergency_restart() cant make it restart when it hangs ... can make it restart all other times though so thats something :)

Cheers

Cass

It still is the same issue... the mmc stack is holding sometimes the wakelock, preventing the device to suspend... I have new logs ... Of course that all the suspend patches have contributed, we are pretty close to make it work :)

Share this post


Link to post
Share on other sites

It still is the same issue... the mmc stack is holding sometimes the wakelock, preventing the device to suspend... I have new logs ... Of course that all the suspend patches have contributed, we are pretty close to make it work :)

Cool but for my sanity .. where does the restart function go in the shortlong file ? cant hit it right !

Share this post


Link to post
Share on other sites

Ok tried after i did a make clean and it appears to sleep then wake up ... but it switches off again ... it continues to do so when i press the button .. so on the face if it we have success :)

Just have to ensue the keyboard repeat is off and debounce is set to 50 as i see this behaviour with these in place ... ill update in a bit

edit indeed, thats what it was ... suspend appears to work .. woot woot !

can you change your code to be

.rep = false, /* auto repeat enabled */

.gpio = SHUTTLE_KEY_POWER,

.active_low = true,

.debounce_interval = 50,

in board-shuttle-keyboard.c so i dont have to do it every update, i get stung all the time ;)

Nice work again !!

edit - hmm more tests needed .. resumed a few times without issue then after leaving it for a while its hung again :(

edit - grrr where the hell in this gpio short long kb do i put the emergency_restart() cant make it restart when it hangs ... can make it restart all other times though so thats something :)

Cheers

Cass

I suspect it is actually suspending... :S -- when suspended, no interrupts are serviced, until the system resumes... We would actually need a hw reset, rather than a sw reset now :(

Share this post


Link to post
Share on other sites

I suspect it is actually suspending... :S -- when suspended, no interrupts are serviced, until the system resumes... We would actually need a hw reset, rather than a sw reset now :(

Hmm s*** .. whats the options now ... backout the last round of patches at least and work again towards making it suspend all bar mmc ? im slowly liking the idea of making .36 the base and backporting to that :)

Share this post


Link to post
Share on other sites

Well... the wakelock problem is fixed. Now, the kernel actually suspends, but, seems not to return... :S - At least, attached the required patches for the wakelock fixes...

edit - Yes, we still have that option ... backporting to .36 ...

patch2.6.39.rar

Edited by ejtagle

Share this post


Link to post
Share on other sites

Well... the wakelock problem is fixed. Now, the kernel actually suspends, but, seems not to return... :S - At least, attached the required patches for the wakelock fixes...

edit - Yes, we still have that option ... backporting to .36 ...

Ok .. patches compile and boot fine ... i also think i have a workaround for the problem for the moment .. change the sleep mode to power collapse instead of power collapse suspend as i think it defaults to... seems to work fine, just have to test the affect on battery but hey as far as i can see it at the moment a potential slight detriment to the battery when sleeping is better then nothing at all :)

add

pm.sleep_mode=1

to the build.prop

appears to recover very nicely each and every time, even after a long time

edit 2 .. i give up ... worked all day perfectly .. survived reboots and now ive reflashed its not working .. think i need to go sleep some more .. something im missing here :)

Cheers

Cass

Edited by Cass67

Share this post


Link to post
Share on other sites

Ok .. patches compile and boot fine ... i also think i have a workaround for the problem for the moment .. change the sleep mode to power collapse instead of power collapse suspend as i think it defaults to... seems to work fine, just have to test the affect on battery but hey as far as i can see it at the moment a potential slight detriment to the battery when sleeping is better then nothing at all :)

add

pm.sleep_mode=1

to the build.prop

appears to recover very nicely each and every time, even after a long time

edit 2 .. i give up ... worked all day perfectly .. survived reboots and now ive reflashed its not working .. think i need to go sleep some more .. something im missing here :)

Cheers

Cass

Sometimes, you need a full rebuild to fix things.. seems that not doing clean build starts accumulating garbage ... :S

Share this post


Link to post
Share on other sites

Sometimes, you need a full rebuild to fix things.. seems that not doing clean build starts accumulating garbage ... :S

Yeah .. it was indeed a false alarm .. at least i know now why and im not going mad :) had flashed a new build this morning with the new kernel patches and added the GalleryGoogle.apk and all was well, suspend was still broke at this point .. i added the pm sleep entry when i was looking to some tunings and reboot .. hadn't noticed the Gallery app had started spinning in the background

W/PrefetchHelper( 1204): space not enough: 908951552, stop sync

D/PicasaSyncManager( 1204): sync complete: SyncResult: stats []

Its been keeping a wakelock open i presume and suspend really was just powering off the screen.. Must have been like that for hours, across many reboots. It must have eventually just gave up coincidently when i did an update.zip of the kernel again when someone on TR mentioned it did not work for him ... reboot and it was broke again ... funny old world !! but now i have reproduced the issue im happy ...

Edited by Cass67

Share this post


Link to post
Share on other sites

Right now, regarding suspension, i don't know what to do... nvidia2.6.39 boots but has the SOD (Suspend of death!) ... moto2.6.39 does not boot for us... the tuna kernel does not boot for us... 2.6.36 is the only one working ... What to do ? - Start backporting drivers from 2.6.39 to the 2.6.36 ? -- Perhaps just the netfilter subsystem ? .. try to fix 2.6.39 ? .. It will be hard , as we have no menas to force a reboot when suspended ... I have tried to use the tegra watchdog to reset the chip, but it does not work ... Very, very strange...

Debugging this will be very hard... :(

Besides, we still have the omx acceleration issues... I would try to use the .36 kernel just to know if the omx starts working, and the suspension starts to work... That would be something very interesting to try...

Edited by ejtagle

Share this post


Link to post
Share on other sites

Right now, regarding suspension, i don't know what to do... nvidia2.6.39 boots but has the SOD (Suspend of death!) ... moto2.6.39 does not boot for us... the tuna kernel does not boot for us... 2.6.36 is the only one working ... What to do ? - Start backporting drivers from 2.6.39 to the 2.6.36 ? -- Perhaps just the netfilter subsystem ? .. try to fix 2.6.39 ? .. It will be hard , as we have no menas to force a reboot when suspended ... I have tried to use the tegra watchdog to reset the chip, but it does not work ... Very, very strange...

Debugging this will be very hard... :(

Besides, we still have the omx acceleration issues... I would try to use the .36 kernel just to know if the omx starts working, and the suspension starts to work... That would be something very interesting to try...

Catch 22 we are in then i think .... The .36 as i remember it had no suspend issues under ICS, this is true .. but it had the green flash problem. I think in order to use the .36 we need to do something similar to Pershoots patches so we can use the current NV libs which removes the green flashes, but it seems with Pershoot's words on his kernel efforts it not resolve OMX acceleration as he cannot play HQ video either. Im sure i also ported the quota2 netfilter to .36 with ease the other one we required xt_guid i didnt spend much time with. Don't think that part will be difficult.

Saying that, i was playing around the BoardConfig last night trying to get some legacy options to be in action with 4.0.3 to try to get the older libs to work, first time i succeeded in not making it boot past netd (copybits issue).. Currently with my latest build i have some legacy stuff in there but it does not resolve the omx issue we have now ... maybe there is others we can compile for our chipset that may help.

Its up to you mate what we do, your the brains behind this kernel operation ;)

Cheers

Cass

Edited by Cass67

Share this post


Link to post
Share on other sites

Catch 22 we are in then i think .... The .36 as i remember it had no suspend issues under ICS, this is true .. but it had the green flash problem. I think in order to use the .36 we need to do something similar to Pershoots patches so we can use the current NV libs which removes the green flashes, but it seems with Pershoot's words on his kernel efforts it not resolve OMX acceleration as he cannot play HQ video either. Im sure i also ported the quota2 netfilter to .36 with ease the other one we required xt_guid i didnt spend much time with. Don't think that part will be difficult.

Saying that, i was playing around the BoardConfig last night trying to get some legacy options to be in action with 4.0.3 to try to get the older libs to work, first time i succeeded in not making it boot past netd (copybits issue).. Currently with my latest build i have some legacy stuff in there but it does not resolve the omx issue we have now ... maybe there is others we can compile for our chipset that may help.

Its up to you mate what we do, your the brains behind this kernel operation ;)

Cheers

Cass

we dont have the drivers for omx, so no youtube HQ at the moment. unless a way can be hacked in to get HC 3.2 libraries working on ICS for HD/full HW video decoding. waiting on nvidia to do a release as they promised. maybe the TF prime ICS image will also help.

everything you need is in my github for netfilter, gadget, touchscreen and video, for ICS, on .36.

Share this post


Link to post
Share on other sites

We have that set already ..

Cheers

Cass

I know just brought back some memories.

I tried running the Vega for a couple of days with your ICS build (well gave it to the kids) and it does lock up every once and a while if not connected to the USB. The screen stays on but is totally unresponcive.

I turned on the Dev tools included with ICS to get what little info I could but apart from the odd message telling the app is not reponding to you wish to close or wait. Nothing useful.

Billy..

If there is anything you would like me to test feel free to ask. Be aware very limited time at present.

Share this post


Link to post
Share on other sites

we dont have the drivers for omx, so no youtube HQ at the moment. unless a way can be hacked in to get HC 3.2 libraries working on ICS for HD/full HW video decoding. waiting on nvidia to do a release as they promised. maybe the TF prime ICS image will also help.

everything you need is in my github for netfilter, gadget, touchscreen and video, for ICS, on .36.

The HC3.2 libs for acceleration can be used, and they should work, except omx from HC3.2, that need an extra exported function in libmedia that was removed in ICS... But, we could even mix libraries to fix that... All the HC3.2 libs, except the libnvomx.so, that we should take from the google supplied ones for wingrey/stingrey tablets. That should work! ...

Pershoot: Could you try to push into your device the libnvomx.so from the wingrey propietary libs released by google and check if OMX starts to work, as it should... You need to restart the device after pushing them, and they should work... Humm... Could you ?

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.