Cowon D2

[vodi]

I don't see much improvements, because i didn't had very much problems before, just one or two times a "bad checksum" - error.

I also don't use very much Rockbox on my D2 because i miss write access

And i don't think, its too much work to build a new bootloader when you also build the rockbox.zip from SVN.

[gevaerts]

I had some issues reading a file. This may be due to filesystem corruption, as fsck.vfat said the free cluster summary was wrong (now fixed).

In general it's a lot better thatn it was though

[kfoltman]

shotofadds: I have the "2" (European, non-DAB) 2GB version here. I've played with Rockbox a little, and didn't run into easily spotted flash issues (in form of "bad checksum" messages or bursts of noise during playback). The sound is fine to my ears. Volume control works too. The display is fine - no abnormal lines or any visible shift.

One problem: the touchscreen stopped working after a couple of minutes of playing with different plugins. Also, I got "undefined instruction" when changing to another song (hopefully not flash-related?). Unfortunately the address is mostly obscured by "NOW PLAYING" message on yellow background (reading white on yellow isn't that easy). Looks like 000D40DC or similar. I'll try to test more later.

Edit: s/DRAM/NAND flash/ - I never thought I'll ever confuse those two, but I did.
Edit2: I get "undefined instruction" semi-regularly, the address displayed seems to be different each time. I've also started getting the LCD lines problem, but that was after shutting down the original firmware during load stage two times. It kept showing those two lines consistently after that (reboot after reboot), until I started the original firmware again. Then it stopped happening and I couldn't reproduce it ever since.
Edit3: does Rockbox bootloader (or the loaded binary) do any checksum verification on the code read from NAND Flash?

[wahamir]

I would love to try it, but since i'm not used to rockbox development and don't have much time ...

It would be apreciated if someone could post a compiled bootloader or just explain me how to do it ....

Thanks

[Llorean]

This thread is for development. If you're not planning on developing, please wait until a user version is available.

[shotofadds]

@Llorean,

I did specifically ask whether the lack of a pre-compiled bootloader was holding anyone back from testing, so that response was entirely justified. As I'm sure you know, user testing can be incredibly helpful even in early stages of development.

If there are people who would find it useful, I see no problem providing a mktccboot binary (with associated bootloader.bin), especially given that we already provide a pre-built rockbox.zip and tcctool, too.

@kfoltman,

The undefined instruction errors are rather interesting - I haven't seen those during playback, even after extended periods of use (eg. I left it to run for over 24hrs recently without a hitch). Does this happen with any particular audio formats, or when changing to a particular song?

It's quite annoying that the error text is obscured by the viewport background, as the handler really should clear the screen beforehand. I'll see if I can fix that. If you're able to read the hex address next time, do write it down and also post which revision (r19???) you're running, since the address is only useful with this information.

I've seen the problem where the touchscreen stops responding before, and I think I have a workaround (I hope to do some more work in that area over the next couple of weeks).

The bootloader does verify the rockbox.d2 checksum before starting Rockbox, that's why the old version used to fail on bootup with "Bad Checksum" errors rather often...

[luck3r]

NAND seems to have improved for me, didn't get any errors since I've built r19189.
It used to make problems with high-bitrate (low compression) FLAC files (sometimes the playback was interrupted by extremely loud artifacts, kinda sounded like /dev/urandom), seems be gone now, but I'll keep testing.
The "shifted lines" problem also disappeared.

BTW, I like how the D2 gets warm when using Rockbox, as it's getting really cold outside, so I use it to prevent my hands from freezing off. Nice side-effect

[gevaerts]

It used to make problems with high-bitrate (low compression) FLAC files (sometimes the playback was interrupted by extremely loud artifacts, kinda sounded like /dev/urandom), seems be gone now, but I'll keep testing.

Those were probably caused by NAND issues as well I guess

[luck3r]

I was able to repoduce the problem where the touch screen stops working.

1. go to Settings->Theme Settings->Colours
2. go to Background Colour and change it to #080000
3. go there again, this time changing it to #000000

[gevaerts]

I was able to repoduce the problem where the touch screen stops working.

1. go to Settings->Theme Settings->Colours
2. go to Background Colour and change it to #080000
3. go there again, this time changing it to #000000

Also reproducible here

[shotofadds]

What symptoms are you seeing? Does the touchscreen stop responding completely?

My local version doesn't do that (although it does include some touchscreen-related changes), but there's definitely something funny about that screen regardless. Touching midleft/midright on the colour picker screen initially works as expected (ie. increment/decrement the coloured bar) but as soon as Select is pressed the behaviour changes, and next time the screen is entered pressing Left/Right will also exit the screen.

Very weird. Something must be causing the touchscreen "button" driver to get rather confused...

[luck3r]

It stops responding completely.

[buk]

Hi,
I'm new here and I first want to congratulate you for your work, I really enjoy Rockbox.
Since rev 19228, I have no sound. I fixed the problem by changing line 46 of wmcodec-telechips.c to :
#if defined(HAVE_WM8731) || defined(HAVE_WM8751) || defined(HAVE_WM8985)
I'm not sure it's the best and correct way to fix it but it works.

I will try to look deeper into the code and maybe one day I'll be able to help you guys (not sure as I never code hardware stuff but it's a good way to learn ).

[FranzD2]

Hi all! 

I just posted on FlySpray a patch that seems to solve the problem of the touchscreen driver, i've introduced an error compensation mechanism on the values read from the pcf50606 and some other stuff.
I think that the wrong values were due to some touchscreen voodoo (eg signal rebound) and does not depend on how the software is made, so the only way is to use a compensation algorithm.
It works fine for me.

Please try this fix and let me know...

[sikon]

Hi! I want to try new build and i can write some comments about it. But there is a one problem, where can i download compiled builds(i have 2.57 version)