[ZenAurora]

kuschelganxta, on 31 July 2011 - 07:09 AM, said:

rh+ is right. I was just about writing a mini how to myself

I'm looking into the source right now and here's my first impression:
- The ARM Core seems to be a kind of Samsung's S3C2410/S3C6410 as there are heavy similarities (and code references)
- Power Management is able to shut down more clocks than now (APLL and MPLL Clocks (CPU Clocks))
- Clock scaling is not implemented. IF CPU Core and Power Mgmt is the same as Samsung Core, we may copy code from plat-s3c

Generally hard to read (and improve) without datasheet... :-/ FYI, S3C6410 Datasheet is here
However, I have to say the code itself is well-written and fast understandable.

My main goal first is to implement CPU clock scaling as my tab really gets hot on the core (besides wifi module...
I already took a look at the SPLs differences but was not able to find the part where clocks are set up.

so guys, till later

looking forward to it.. i guess it will really help with the battery life and heat

[ZenAurora]

atp_uestc, on 31 July 2011 - 09:30 PM, said:

rh+:
you can put in sdcard directory and press left key when boot and debug it.

Colin:
it's android kernel, but i think you can make a native kernel which load your cpio

merlot:
for b0 it's 0603,for other it's 0618

rh+:
after install GCC/ccache/toolchains/maketest.sh,and your enviroment,you can build success withou modify other.

kuschelganxta:
yes,it's samiliar,many register similar %. I will private releas a datasheet,but need removed the watermark(who tell me easy way?)

pepelillo:
nice to see Meego if you do.
zt-280 open source will be thinking about when zt-380 is selling.

It depends on the watermark.. Maybe u can email us like when u informed us about the source code.. If i can remove the watermark i'll send u the clean one back.

[merlot]

atp_uestc, on 31 July 2011 - 09:30 PM, said:

rh+:
you can put in sdcard directory and press left key when boot and debug it.

Colin:
it's android kernel, but i think you can make a native kernel which load your cpio

merlot:
for b0 it's 0603,for other it's 0618

rh+:
after install GCC/ccache/toolchains/maketest.sh,and your enviroment,you can build success withou modify other.

kuschelganxta:
yes,it's samiliar,many register similar %. I will private releas a datasheet,but need removed the watermark(who tell me easy way?)

pepelillo:
nice to see Meego if you do.
zt-280 open source will be thinking about when zt-380 is selling.

Thank you again atp_uestc!!

I've been trying to compile the sources without modificaction and I have almost get it.

But after compiling several .o object files it fails here

make[2]: *** There is no rule to build «drivers/memalloc/memalloc.bin», needed for «drivers/memalloc/built-in.o». Halt.
make[1]: *** [drivers/memalloc] Error 2

Any suggestion?

Thank you

[merlot]

merlot, on 01 August 2011 - 06:49 AM, said:

Thank you again atp_uestc!!

I've been trying to compile the sources without modificaction and I have almost get it.

But after compiling several .o object files it fails here

make[2]: *** There is no rule to build «drivers/memalloc/memalloc.bin», needed for «drivers/memalloc/built-in.o». Halt.
make[1]: *** [drivers/memalloc] Error 2

Any suggestion?

Thank you

drivers/memallloc is almost empty (just Kconfig and Makefile), while there is a memalloc.tgz in drivers, which have some source code, headers, etc.

Should I extract this?

Thank you

[kuschelganxta]

atp_uestc,
thanks for your offer to release datasheet. I highly appreciate this. Would be nice if you could PM this to me if public release is too risky for you.
Another question: where are the kernels located in flash? I can't see any info in MTD partition table

It's nice seeing you release sourcecode when next product is selling. It's not GPL-compliant (most chinese don't care... :-/ ) but at least it's kind of fair. Thank you!

Progress so far:
Implemented IMAP-Core into normal Linux-Kernel and got core and uarts initialised. Flash access not working sofar but I guess it's up to some stupid settings.

@all, merlot:
Kernel compiles clearly but you first have to extract /drivers/memalloc.tgz into /drivers/memalloc.

This post has been edited by kuschelganxta: 01 August 2011 - 09:01 AM

[cantalup]

kuschelganxta, on 01 August 2011 - 08:58 AM, said:

a...
@all, merlot:
Kernel compiles clearly but you first have to extract /drivers/memalloc.tgz into /drivers/memalloc.

could you post memalloc.tgz? or that file is in the kernel source code.?

[merlot]

cantalup, on 01 August 2011 - 11:11 AM, said:

could you post memalloc.tgz? or that file is in the kernel source code.?

Is in the source repository

[cantalup]

merlot, on 01 August 2011 - 03:55 PM, said:

Is in the source repository

got it, thanks!!!

[atp_uestc]

o,not
the source code of memalloc is old for 0503. and the newest 0603 is bin files,i don't has source code also.The git lost the bin files first,i upload it again.

the spec protect by NDA, i will think way to distribute it.

[merlot]

atp_uestc, on 01 August 2011 - 09:27 PM, said:

o,not
the source code of memalloc is old for 0503. and the newest 0603 is bin files,i don't has source code also.The git lost the bin files first,i upload it again.

the spec protect by NDA, i will think way to distribute it.

Perhaps is because this that the zImage I've been able to generate is different in size, and of course in MD5, from what we have for 0603. I'll check for new memalloc

Thank you

[merlot]

merlot, on 02 August 2011 - 02:11 AM, said:

Perhaps is because this that the zImage I've been able to generate is different in size, and of course in MD5, from what we have for 0603. I'll check for new memalloc

Thank you

I've got the new memalloc but I get the same different result

These are the original 0603 files

-rwxrwxrwx 1 root root 3477884 2011-07-29 17:01 zImage
-rwxrwxrwx 1 root root 3473264 2011-07-29 17:01 zImage_adb

And these are the compiled ones

-rw-r--r-- 1 root root 3195276 2011-08-02 09:25 zImage
-rw-r--r-- 1 root root 3195276 2011-08-02 09:25 zImage_adb

Always for B0_1024_576_GPS

Is that normal?

Thank you

This post has been edited by merlot: 02 August 2011 - 02:42 AM

[cantalup]

atp_uestc, on 01 August 2011 - 09:27 PM, said:

o,not
the source code of memalloc is old for 0503. and the newest 0603 is bin files,i don't has source code also.The git lost the bin files first,i upload it again.

the spec protect by NDA, i will think way to distribute it.

thanks for the clarification and your efforts .

[atp_uestc]

just press left key start and test it

[merlot]

atp_uestc, on 02 August 2011 - 08:26 PM, said:

just press left key start and test it

I guess it's safe to flash zImage and that I will be able to flash the previous one, isn't it?

Is it enough to put these files in zt-update?

zimage
zimage_adb
zimage_update

Or should I put also the system.tar.bz?

Thank you for your support

[rh+]

merlot, on 03 August 2011 - 01:23 AM, said:

I guess it's safe to flash zImage and that I will be able to flash the previous one, isn't it?

Is it enough to put these files in zt-update?

zimage
zimage_adb
zimage_update

Or should I put also the system.tar.bz?

Thank you for your support

do not flash .. be careful, although i do not tried myself yet,
i guess that left key press meant invocation of special debug mode to just test compiled image.

all safe, without replacing your firmware ...

so i think you do not need to include system.tar, nor zimage_update

This post has been edited by rh+: 03 August 2011 - 02:00 AM

[HAL9000]

rh+, on 03 August 2011 - 01:57 AM, said:

do not flash .. be careful, although i do not tried myself yet,
i guess that left key press meant invocation of special debug mode to just test compiled image.

I don't think there is any risk of bricking the device flashing a non-working kernel. You can always go back and flash a working kernel as long as you have not touched the SPL and/or boot code.

[rh+]

HAL9000, on 03 August 2011 - 09:10 AM, said:

I don't think there is any risk of bricking the device flashing a non-working kernel. You can always go back and flash a working kernel as long as you have not touched the SPL and/or boot code.

yes you are right, but i personally prefer not to reinstall my PAD on each second monday.
(and a lot apl)

maybe there is also limit on internal memory flash writes.
i guess that not more that 1000 is guaranted.

it is on you, i just suggest you to avoid unnecessary flash

[valdas_a]

usually it takes from 5k to 10k writes till failure, so if you are flashing every Monday there are 48 monday`s in year even if flash can take up to 1k writes i will last for 20 and a bit years I doubt if you will be using it after a year mine broke few days before source code was released And i can swear that is my worst spent money Well maybe some weekend i`ll take it apart and try to resurrect it. And if i cant fix it, well i will salvage parts from it and use it form my electronic projects

would be interesting to compile android 2.3 for this crap, and its pity that google wont release android 3 source.

[atp_uestc]

It's safe,but suggest debuging it first.

[merlot]

atp_uestc, on 04 August 2011 - 05:43 AM, said:

It's safe,but suggest debuging it first.

Thank ATP.

Debug?. That could be really good. But how can we do this?

I think you said before that the only way to debug is putting on SD and pressing left button?

"just press left key start and test it"

Or is there a zt-180 slate emulator or something like this?

Thank you so much again