Cowon D2

[Yeeze]

Ah I will try that!

[Yeeze]

Well now I have: *PANIC* Unknown NAND: 0x06 0x06 0x06 0x06 0x06

EDIT: You ment this little fix?

In the nand_chip_select() function (ata-nand-tcc780x.c) could you try changing the "if (chip & 1)" line to "if (!(chip & 1))" and tell me if your D2 still reports 1 bank * .... as it did before.

[shotofadds]

Yes, I meant that one - I guess it wasn't the problem... I need to think about it and get back to you (like I say, pesky 2Gb players   )

[carsten]

I'm still having the same problems with the new nand-driver. I never got beyond "unaligned read-buffer", most of the time I have a read error... :-/

edit: forgot to mention: 4gb d2

[stel]

Hi,

I've just been trying out andiators d2_patched.bin file with mixed results.
Firstly I tried andiators rockbox.zip - This gave me the black screen after mounting.
Then I tried shotofadds rockbox.iaudio (post#211) and got the same results as andiators(post#212) so I then compiled my own rockbox.zip and now I get the *PANIC* nand_read_raw() non_aligned input buffer message as soon as I disconnect USB

The above info probably doesn't help much but I'm willing to try any suggestions to help out.

I've got a 8Gb D2

[shotofadds]

If you are getting the "non-aligned buffer" error, then the updated ata-nand code (post #215) has not been built in successfully. The new version does not contain that error message - but you may need to alter it slightly in order to build without errors, as andiator reported.

Sorry about this mess - I will update SVN when I get a chance, but I don't have a lot of spare time this week (aka. none at all).  

[andiator]

I was playing with the adc code and I don't think that it causing the button problem. Yes, there is some jitter at the press or release, but it would cause a random behavior. And Yes, we still have to fix this in an extra thread.
The idea is to run a thread with sleep and waiting for() for the stabilized adc value to avoid the transition values from press to release and back

But the actual behavior is too systematic: it seems to happens only with plus and minus buttons and ONLY when you switch to the other buttons. So more investigation needed.

[shotofadds]

Well, probably not a new thread but a tick task (fixed 100Hz timer) which periodically requests a new reading for each of the four ADCs and exits immediately. An ADC IRQ handler would then retrieve the new values when ready, and if they are within some abritrary level of tolerance of the last value read, make the new value available to adc_read(). That is my intention, anyway.

[andiator]

For our three buttons the tick task is fine, but what about the touchpad? I don't think it's a good idea to do longer operations in timer task.

[shotofadds]

I was purely talking about the ADC driver using a tick task. The touchpad is accessed through the PCF50606 (via I2C), and yes that might need some thought about how to implement it correctly. I need to check how other targets do periodic I2C polling.

Edit: This discussion would probably best be held in IRC where devs who know best about such things can comment (but I think it can wait until I've fixed the ata-nand driver and ADCs first)...

[carsten]

If you are getting the "non-aligned buffer" error, then the updated ata-nand code (post #215) has not been built in successfully. The new version does not contain that error message - but you may need to alter it slightly in order to build without errors, as andiator reported.

Yes, I don't get the non-aligned buffer error anymore, but the read error given by the bootloader is still there. Now I got it to boot, but it's still a matter of luck... (and I definitely build it with the new driver...)

andiator: I've a little home-server running on a nslu2 with debian if you just need a little space for things like that, I can arrange that... 512kbits upstream though

[andiator]

Thanks carsten, but one hour before your post there was a parcel delivered to me - and what was inside? nslu2!   and my upload is 1024 Kbit.

But back to topic. Did you rebuild the bootloader with the newer ata-driver? And did you do the ../tools/mktccboot ... stuff?

I played again with buttons and copied/inserted the touchpad test code into button_read_device(). Well, the virtual buttons work normaly, so the problem has to be the adc, but I don't get it since the code is so simple.
I uploaded the new binaries for testing. This touchpad stuff is really only for testing. It will be done right in future, so don't say it works badly  
In bootloader screen simply tip on the screen to get further.

[stel]

I'm up and running now as described in the previous posts and the majority of the demo plugins work fine. The only thing I've come across so far that I don't think has been mentioned is that the lcd seems to cut off a couple of rows from the top and displays them at the bottom.
I'm really impressed, you've done some great work shotofadds.
 I probably won't be able to help much with the coding but I will certainly start taking a look at the weekend.

[shotofadds]

The only thing I've come across so far that I don't think has been mentioned is that the lcd seems to cut off a couple of rows from the top and displays them at the bottom.

Yep, that's a consequence of not fully understanding the LCD controller initialisation. I get the same here (but not every time) - it'll be fixed eventually.  

I'll hopefully have more time to work on this next week. My priorities at the moment are fixing the button/ADC code, getting the flash driver into a reasonable state and reducing the compiler warnings so we can get the D2 builds added to the Rockbox build table. Shouldn't take long...

EDIT: The updated NAND driver has now been committed so I've removed it from the previous post. "svn update" is your friend.  

[Llorean]

Any particular reason for decreasing the CPU frequency? Seems like it should stay at "max" until scaling works, to maximize the range of things that won't suffer from CPU speed limitations.