Cowon D2

[cybergrind]

power off code in SDK:

Code: [Select]

..removed..

I tried do almost all of this (except IO_CKC_SetClock), nothing good :\

Question: can we software-disconnect USB?

p.s. some info:
ARM926EJ-S (includes ARM Jazelle technology which enables the direct execution of Java bytecode in hardware)

[shotofadds]

Please do not post copyrighted code here, or in the tracker, or anywhere else - it'll be removed.

It may be the case that power-off does not work because we are running code in usb-boot mode. Until code runs from flash we have no way of knowing. This is something still to be investigated.

[kalessin]

So... I will (hopefully) try to get the toolchain to work this weekend, since my brandnew D2 is begging for a working Rockbox. Don't know how much I will actually be able to contribute, apart from testing (unfortunately I have not much experience on embedded platforms), but we will see. First of all (before I put anything on it) I will read up on D2 desaster recovery, though

[shotofadds]

Since you won't be flashing anything to your D2 at the moment (everything is done in usb-boot mode using tcctool), there shouldn't be any disasters to recover from.

I'm frustratingly close to a first working version (ie. booting to the menu with working buttons) but unfortunately I managed to leave my D2 at work and won't be able to make any progress for a few days. Oops

Getting the toolchain to work should simply be a matter of checking out a current SVN build and running rockboxdev.sh - check the wiki for instructions. Hope it goes well!

ps. Experience on embedded platforms is mostly irrelevant, I had none before I started this port.

[luck3r]

Since I've checked out revision 16680, the screen of my D2 stays black with normal build, uploading the bootloader just loads the regular Cowon firmware after a short delay.  Is this normal or am I doing something wrong?

[shotofadds]

Apologies for that, the last commit message read:

Code: [Select]

r16646: Cowon D2: Make the bootloader functional (usage instructions to appear on the CowonD2Info wikipage shortly). Also re-enable IRAM by reducing Tremor's ICODE usage.

Of course, I forgot to update the wiki with the necessary instructions - I'll do that later today.

EDIT: I've added some instructions to the wiki page. Let me know if anything is unclear.

[Yeeze]

Wow really nice progress!!

I lost Internet for the last week, but now I'm back online =)

So if there is anything to be tested with a 2GB player I can do it!

And a while back I thought about writing the touchscreen driver, but I must say that I did not get anywhere...
This might be due to my short time of 1 and 1/2 year of programming experience...
I looked through the code, but I only hardly understand anything of the code for threads and stuff which, probably will be needed to have interrupts working... (well that was my guess)

But as soon as it comes to plugins and higher level functions, etc. I want to contribute =)

One question about cygwin...

I came really close to compiling but at one point I get:

$ make
make[1]: `rdf2binary' is up to date.
make[1]: `convbdf' is up to date.
make[1]: `codepages' is up to date.
make[1]: `scramble' is up to date.
make[1]: `mktccboot' is up to date.
make[1]: `bmp2rb' is up to date.
CC target/arm/tcc780x/ata-nand-tcc780x.c
target/arm/tcc780x/ata-nand-tcc780x.c: In function 'nand_test':
target/arm/tcc780x/ata-nand-tcc780x.c:438: warning: implicit declaration of function 'printf'
target/arm/tcc780x/ata-nand-tcc780x.c:493: warning: implicit declaration of function 'reset_screen'
target/arm/tcc780x/ata-nand-tcc780x.c:553:6: warning: #warning function not implemented
target/arm/tcc780x/ata-nand-tcc780x.c: At top level:
target/arm/tcc780x/ata-nand-tcc780x.c:38: warning: 'next_yield' defined but not used
CONVBDF
CC sysfont.c
/cygdrive/l/home/yeeze/myStuff/programming/rockbox-dev/rockbox/tools/convbdf: /cygdrive/l/home/yeeze/myStuff/programming/rockbox-dev/rockbox/tools/convbdf: cannot execute binary file
make[1]: *** [/cygdrive/l/home/yeeze/myStuff/programming/rockbox-dev/rockbox/buildc/firmware/sysfont.o] Error 126
make: *** [build] Error 2

Does anybody know why this happens? Is it a known issue?

[andiator]

I'm using cygwin too and don't have this problem.
Anyway, tools/convbdf.exe is compiled at some point, not shure when. Perhaps with tools/rockboxdev.sh?
Try to run it (convbdf.exe) in the cygwin console, it should output some help information. If not, you have to rebuild it. Besides, mine is 55.014 bytes big.
"make veryclean" and than "make tools" could help, I think.

[Yeeze]

Yeah that did the trick.... Thx andiator

[shotofadds]

The D2 build has had some pretty substantial updates over the course of the holiday weekend: buttons should now work, meaning the menus are usable (mostly..see caveat), plugins are now built (thanks, andiator) and some of them are usable - eg. fire, plasma, starfield demos and the Invadrox space-invaders game all work without any modification.

I'd appreciate some feedback from you guys - first read the wiki page, build the bootloader and main builds from source, install and let me know if you have trouble booting to the menu (please don't report any of the "Known Issues" from the wiki page).

If you're stuck with "Bad Checksum" or "File Not Found" errors in the bootloader, copy a few files to the D2 (anything will do, bigger is better), boot the OF a few times, and try again. I'll post an updated ATA/NAND driver later this week that should resolve some of these issues.

Finally, if you're not confident building from source yourself, the port probably isn't mature enough to be of much interest yet. There are no binary downloads available at this point.

Have fun....   8-)

[cybergrind]

I have an another error:
Sometimes boot's started,and thought few seconds lcd off's (i have not time to read any)

[andiator]

For two days I already tried this and had one time "Bad checksum" and three times "Read failed".
With today build is like cybergrind sais:

While holding a key, you can see some lines. Mine sais ...(1850Mb) and I have about 3.74 Gb.
Releasing the key shows some more line and the lcd goes off so fast, you can't read the last lines.

[shotofadds]

As I already mentioned, the "Bad Checksum" and such errors are due to the ATA driver, and I'll post an improved version later.

But if it has passed that stage, you should see the Rockbox logo and then the main menu, or USB logo if the cable is still inserted. I'll check this out later (I'm at work at the moment, sorry...)

Could you try inserting the following two lines before the rc = kernel_entry() line in bootloader/telechips.c? You should then be able to read any text output:

Code: [Select]

while (!button_read_device()) {};
while (button_read_device()) {};

Thanks for your help.

[cybergrind]

last messages:

Code: [Select]

Model name: d2
Loading rockbox.iaudio
Sum: 2D05E29

Then I press key, and LCD off's

[shotofadds]

Ok, that means the load was successful, but booting the actual Rockbox image fails. Could you try 'make clean' and build the Normal build again? Also make sure you are building the Bootloader and Normal build in separate directories (eg. d2_boot and d2_main).

If that still fails, could you find somewhere to upload your main rockbox.iaudio file and its corresponding apps/rockbox.map file? (not here though, the files will be too large to attach)