maemo.org - Talk - [Android] Another Android port: NITdroid

maemo.org - Talk (https://talk.maemo.org/index.php)

- Alternatives (https://talk.maemo.org/forumdisplay.php?f=36)

- - [Android] Another Android port: NITdroid (https://talk.maemo.org/showthread.php?t=25410)

Re: Another Android port: NITdroid

Bundyo: any chance you could release the source to your miracle kernel?

Re: Another Android port: NITdroid

Yep, soon, sorry, doing something else and doesn't have the time to clean it up.

Re: Another Android port: NITdroid

Quote:

Originally Posted by solca (Post 254735)

I get the time at init from retutime but I never set it on shutdown

My clock is always about 2 hours 15 minutes off after a power cycle. It's keeping time OK, but just needs an offset to local time.

Quote:

Originally Posted by solca (Post 254735)

Hmm that's not good, does this work before 0.3.1 or it is some new issue now? can you describe it in more datils?

I had the problem since my first install of NITdroid (around 18th December). I'm thinking this may be a screen calibration problem. I found the pointer location app in DevTools and this shows that at the bottom of the screen the calibration is off by about 4mm. There is quite a bit of variation across screen with only the centre left edge having no error.

Is there a screen calibration app in Android?

Finally, I can't get 0.3.1 to return from screen blanking. Once it's gone blank pressing the keys or the screen has no effect. Eventually, it seems to power off as pressing the power key will produce a restart. Strangely, when adb is connected running logcat, it will recover. Here is the logcat trace, although I'm not sure it will be much use.

Code:

**** screen blanks ****

W/ActivityManager(  682): Unable to start service Intent { action=android.accounts.IAccountsService comp={com.google.android.googleapps/com.google.android.googleapps.GoogleLoginService} }: not found

E/LockPatternKeyguardView(  682): Failed to bind to GLS while checking for account

D/dalvikvm(  682): GC freed 15372 objects / 702696 bytes in 211ms



**** keys pressed ****



D/KeyguardViewMediator(  682): wakeWhenReadyLocked(26)

D/KeyguardViewMediator(  682): handleWakeWhenReady(26)

D/KeyguardViewMediator(  682): pokeWakelock(10000)

D/KeyguardViewMediator(  682): wakeWhenReadyLocked(82)

D/KeyguardViewMediator(  682): handleWakeWhenReady(82)

D/KeyguardViewMediator(  682): pokeWakelock(10000)

W/KeyCharacterMap(  682): Can't open keycharmap file

W/KeyCharacterMap(  682): Error loading keycharmap file '/system/usr/keychars/Internal_keypad.kcm.bin'. hw.keyboards.0.devname='Internal keypad'

W/KeyCharacterMap(  682): Using default keymap: /system/usr/keychars/qwerty.kcm.

bin

D/KeyguardViewMediator(  682): pokeWakelock(10000)

D/KeyguardViewMediator(  682): pokeWakelock(10000)



**** screen blanks ****



W/ActivityManager(  682): Unable to start service Intent { action=android.accounts.IAccountsService comp={com.google.android.googleapps/com.google.android.googleapps.GoogleLoginService} }: not found

E/LockPatternKeyguardView(  682): Failed to bind to GLS while checking for account

Re: Another Android port: NITdroid

Hey Guys, got android up and running on my 770 (thanks solca, bundyo, rzr), but adb bridge no go. USB Debugging is checked in Settings->Application->Development, but I think not registering correctly.

dmesg in ubuntu lappy:

Code:

[402380.417159] usb 4-1: USB disconnect, address 25

[402384.725097] usb 4-1: new full speed USB device using uhci_hcd and address 26

[402384.889675] usb 4-1: configuration #1 chosen from 1 choice

[402384.894245] scsi129 : SCSI emulation for USB Mass Storage devices

[402384.895605] usb-storage: device found at 26

[402384.895622] usb-storage: waiting for device to settle before scanning

[402390.894515] usb-storage: device scan complete

[402391.008182] usb 4-1: reset full speed USB device using uhci_hcd and address 26

[402391.280203] usb 4-1: reset full speed USB device using uhci_hcd and address 26

[402391.544648] usb 4-1: reset full speed USB device using uhci_hcd and address 26

adb shell and adb ppp -d (w/ or without su) fail, device not found

edit: fixed by killing previous adb processes

Re: Another Android port: NITdroid

battery management is wacky. First, it drains. Then, the icon disappears. Then, it "charges" (without being plugged in). Then it drains. Then, the icon disappears...

Same time problem here as well...

Re: Another Android port: NITdroid

The battery meter works as good as KDE 4:s battery applet in diablo, or gnome-power-manager in deblet, it shows only 0% - 25% - 50% -75% -100%

Re: Another Android port: NITdroid

Quote:

Originally Posted by meizirkki (Post 255087)

The battery meter works as good as KDE 4:s battery applet in diablo, or gnome-power-manager in deblet, it shows only 0% - 25% - 50% -75% -100%

Sounds like it works about as well as the meter did in Qtopia on the Sl5500. The biggest issue with the Sharp devices was the poor battery life and poorer battery indicator (full, 3/4 full, half full, empty).

Re: Another Android port: NITdroid

My manager tried the Android port for his HTC phone yesterday - OMFG!! NITdroid is "fully" working in comparison. You guys are doing an amazing job ;) Thanks!

Re: Another Android port: NITdroid

[QUOTE=gowen;254864]My clock is always about 2 hours 15 minutes off after a power cycle. It's keeping time OK, but just needs an offset to local time.[\QUOTE]

Most devices have 2 clocks, the real time clock (RTC) which is hardware and the internal system time "clock" which is just software. The hardware RTC keeps time even when the device is powered off, the internal is just a software clock that normally deviates slowly from the real time, so it needs to keep in sync normally in startup and shutdown, NITdroid misses the shutdown part as Android just shuts the hardware without running any scripts. My plan to solve this is to set the RTC clock when the user set it in the GUI.

Quote:

Originally Posted by gowen (Post 254864)

Yes, it is a calibration problem, Maemo solve it using tslib and shipping "default" calibration data for different touchscreen panels. Android doesn't use tslib nor calibration data as it is expected all shipping devices with Android will use capacitive touchscreen hardware which doesn't require calibration. My plan is to teach Android tslib and reuse the calibration data on Maemo partition, I'm working on it.

Quote:

Originally Posted by gowen (Post 254864)

Is there a screen calibration app in Android?

I don't think so.

Quote:

Originally Posted by gowen (Post 254864)

Finally, I can't get 0.3.1 to return from screen blanking. Once it's gone blank pressing the keys or the screen has no effect. Eventually, it seems to power off as pressing the power key will produce a restart. Strangely, when adb is connected running logcat, it will recover.

Other people with N800 reported the same problem, this works nicely on N810 and it seems to be a problem when the device entering retention mode but it can't get out of it. It's a kernel problem and there is no fix yet, other than disabling power management, for the next release I'll provide a kernel without pm for N800 users until a fix get's done.

Quote:

Originally Posted by Thesandlord (Post 255044)

battery management is wacky. First, it drains. Then, the icon disappears. Then, it "charges" (without being plugged in). Then it drains. Then, the icon disappears...

Battery metering is implemented in NITdroid (and I belive in Maemo too) by reading from the RETU ASIC. NITdroid uses the nice kcbatt tool to read the raw value of the current charge and speculatively knows if it's charging or discharing. Maemo uses other method to know if it's charging or discharging but the current charge I'm pretty sure it's read too from RETU.

The problem, is the value reported by RETU which deviates depending on which hardware is turned on (WiFi, LCD, backlights), so this value it's reported 'weirdly', I'm pretty sure that's the reason why Maemo doesn't show you the real charge in their battery applet.

If you don't belive me try this on NITdroid 'kcbatt -r -l 1' then connect/disconnect AC power, turn on/off LCD, WiFi, backlights, and you will see what I mean.

The only way to improve it is by observing and adding that logic to NITdroid so a better value is reported in the GUI or in the Settings/About section. Shame NITs doesn't use standard power class drivers and the hardware doesn't report a single absolute _real_ value.

BTW does somebody knows something about the RETU ASIC? it seems the kernel drivers just read/set registers, the real thing is obfuscated in the proprietary daemons.

I recently (Oct/2008) buy my N810 because Nokia say it was open, they even have great publicity on this, for their next tablet Nokia should put clearly which drivers are following standard Linux APIs and which don't and are implemented by binary only daemons in userspace as belive me this is _very_very_ important.

Anyway I'm happy how NITdroid is progressing thanks too you. :)

Re: Another Android port: NITdroid

solca:

What are the plans for audio? I know you have it turned off for now. Is there any way any of us can help you out with implementing sound? Maybe doing some research for you, etc?

Thanks