Cowon D2

[shotofadds]

vodi, it's probably the same problem that we've been talking about earlier: the lack of single-bit error correction. I presume that SD cards are quite capable of generating single-bit errors as well 

Actually, that's almost certainly not the case. SD cards and the internal memory both use NAND flash chips for storage, but SD cards have error-correction and an FTL built into the hardware itself. The internal NAND does not have this, so Rockbox has to implement these functions itself in software, hence the potential for bugs!

I've also seen occasional problems when skipping tracks (music stops and the WPS becomes unresponsive except for pressing POWER to return to the menu). Not sure what's causing that, but it seems a fairly recent problem. I get the same behaviour if I use PictureFlow to change tracks, but this usually only happens if I control PictureFlow with the touchscreen rather than physical buttons. How strange.

[ezzieyguywuf]

I am interested in working on the touchscreen code. I think that what I'd like to do first is touch up the 3X3 stuff and then maybe get into the "absolute" touchscreen stuff. I have a little bit of coding experience, not tons, but I'm an engineer and like a good problem to solve 

Step one for me would be to get rather familiar with all things touch screen related. I've seen this (http://svn.rockbox.org/viewvc.cgi/trunk/apps/keymaps/keymap-touchscreen.c?view=markup ) source file. I understoood mainly what was going on. What I need next is for some of you guys to point me to the right files that actually use this keymap. I unsuccesfully searched for them last night.

Step two is I want to add some new keys to that key map. I think I saw some discussion on something like this waaaaay back. But, similar to the current D2 firmware, a swipe from top to bottom would scroll down etc... I think this could me implemented rather easily by defining a new key as Top-middle-bottom with some maximum time period in between.

Those two steps in themselves should keep me busy for a while, and I'm hoping that they'll get me familiar enough with Rockbox to really add some functionality to the touchscreen support.

So, waddaya got?

[alfchin]

I'am very sorry to trouble you but I think I'am facing a difficult situation while installing rockbox on some D2+,which Orginal Firmware is 3.11 when it comes from their manufactory,and which are sold in Korea.
I've used the same way to make the patched firmware of 3.11/2.11 as I did to those of 2.02/3.02.
However,it doesn't work on those new D2+.Also this kind of New D2+ can only flash the FW of 3.11/2.11,even with the help of TCC Tool.Neither does the bootloader with SD Driver work.
Could you please help me,please

[shotofadds]

Also this kind of New D2+ can only flash the FW of 3.11/2.11,even with the help of TCC Tool.Neither does the bootloader with SD Driver work.

So you're saying for example that the D2+ firmware 3.02 doesn't work on your player? I think this might be the same issue seen here: http://iaudiophile.net/forums/showthread.php?t=30037

Can you try shining a bright light at your D2+ and see if anything is visible with either FW 3.02 or Rockbox?

If so we can be sure there are different variants of the D2+ that have different LCD backlight handling. If that's the case, someone will need to sit down with a disassembly of 3.02/3.11 and work out what the difference in backlight handling is...

[alfchin]

Also this kind of New D2+ can only flash the FW of 3.11/2.11,even with the help of TCC Tool.Neither does the bootloader with SD Driver work.

So you're saying for example that the D2+ firmware 3.02 doesn't work on your player? I think this might be the same issue seen here: http://iaudiophile.net/forums/showthread.php?t=30037

Can you try shining a bright light at your D2+ and see if anything is visible with either FW 3.02 or Rockbox?

If so we can be sure there are different variants of the D2+ that have different LCD backlight handling. If that's the case, someone will need to sit down with a disassembly of 3.02/3.11 and work out what the difference in backlight handling is...

Thank you for  your help.
Here's the result:
When I flash to 3.11 patched firmware,and try to start Rockbox,however,there's nothing displaying on the screen with a bright light,just like it is powered of .Only when I used reset could I went into its orginal firmware .
When I tried  to flash 3.02 to the player,I can see the screen of the firmware update.However,when I tried to turn it on,I found its screen all blank,just like above.The computer can detect it as an normal D2+ when I connected it by USB cable.Then I have to flash 3.11 back,using the normal way.The firmware  update procedure  displayed nothing,and after it finished,d2+ can load 3.11,proving that the update was sucessful.
I guess that the new D2+s use a new LCD,which cannot drive it in the old way.

[shotofadds]

Ok, so we know there's a difference in LCD and/or backlight handling in some newer D2+ units. Firmware 3.11 works with these as well as older players, so it obviously has a mechanism to detect and work with both types. Someone needs to do some disassembly to work out how this is done in the new firmware.

I'd look into it myself, but I'm about to go away for 2 weeks and I already have plenty of other things to do before then!

[shotofadds]

Followers of this thread might be interested in a couple of recent changes:

r22283 re-enables the SD driver, since there have been no more reports of damaged cards.

r22284 implements ECC correction for files read from NAND, which should solve problems with tracks skipping unexpectedly when played from the internal memory.

[helloworld1]

Ok, so we know there's a difference in LCD and/or backlight handling in some newer D2+ units. Firmware 3.11 works with these as well as older players, so it obviously has a mechanism to detect and work with both types. Someone needs to do some disassembly to work out how this is done in the new firmware.

I'd look into it myself, but I'm about to go away for 2 weeks and I already have plenty of other things to do before then!

I can confirm that the backlight cannot be turned on in the new D2+ model. The 2.01 firmware cannot drive the backlight too. But 2.11 2.10 or 3.11 can drive the light. I am using the latest SVN to build the bootloader.

Under very very strong light, I can even see the rockbox logo. So LCD is working but the backlight.


EDIT:
How to dissemable the OF under Linux?, I am willing to help.

[shotofadds]

Ok, I spent a couple of hours looking into this today and it looks like some new D2+ units have an entirely different power management chip, instead of the PCF50606 used by older D2+'s and the 'classic' D2.

Looking at firmware 4.11+ (since that is what I use on my player) it shows some interesting changes almost straight away:

- Follow the branch at 0x200000F8 (LDR PC,=0x20009BD4)
- In older firmwares, the first call in this routine starts writing to the PCF50606 almost immediately (using I2C slave address 0x10)
- In 4.11, this now tests the value of GPIOA6 (used for backlight control on older D2s) and stores a flag depending on its value
- Depending on this flag, the next routine (at 0x200722E4) either calls the PCF50606 init routine (0x200BD1C8) or another init routine (0x200BDBE8)
- The new init routine makes numerous I2C writes to slave address 0xE6, with completely different register addresses to the PCF chip.

Similarly, the backlight control routine (0x20072320) checks this flag, and calls either the PCF routine (0x200BD5C4) or a new one (0x200BE19C). To control the backlight brightness, this new routine writes registers 0x28 and 0x29 to I2C address 0xE6.

I don't know what this new chip is - anyone fancy opening their D2+ to find out? Although having said that, this disassembly should give us enough information to get Rockbox working eventually...  it's just that I won't be around for the next couple of weeks to look into it.

[samokhov]

r22284 implements ECC correction for files read from NAND, which should solve problems with tracks skipping unexpectedly when played from the internal memory.

It seems to work great so far. I've had a track that only played a few seconds from the start. Now it plays the full 10 minutes. Thanks a ton!

[grantmasterflash]

Followers of this thread might be interested in a couple of recent changes:

r22283 re-enables the SD driver, since there have been no more reports of damaged cards.

r22284 implements ECC correction for files read from NAND, which should solve problems with tracks skipping unexpectedly when played from the internal memory.

Nice work as always, shotofadds.  Many thanks on your hard work.  With ECC enabled, I'm not getting any of the track skipping behavior.

I do notice an issue where tracks that are on my SDHC card get added to the database nine times (so for each track, there are nine entries for that track in the database view).  This doesn't happen for tracks on the internal memory, only the music on the SDHC card.  I've tried deleting the .rockbox folder entirely and re-initializing the db, and still see the same behavior after the db is built each time.

I tried 2 scenarios:

1. Used a 128mb SD card - Put only one track on it, deleted .rockbox dir from the card and put a new .rockbox dir in place, then booted the device and initialized the db.  No duplication of the track on the SD card.

2. Used a 16GB SDHC card - This card is almost completely filled with music and only has ~250mb of free space on the card.  Deleted .rockbox dir from the card and put a new .rockbox dir in place, then booted the device and initialized the db.  The db initialization takes a very long time and all the tracks on the SDHC card are duped 9 times in the db.

What I mean by "duped 9 times" is that if I navigate to Database -> Artist -> The Clientele -> Suburban Light -> I then see 9 track 1s, 9 track 2s, etc., for example,

01. I Had To Say This
01. I Had To Say This
<etc. and 7 more until 9 total are listed, then>
02. Rain
02. Rain
<etc. and 7 more until 9 total are listed>

Obviously I went to two extremes there (128mb vs. 16GB).  Maybe has something to do with the amount of music on a large card?

This is with rockbox built from svn yesterday afternoon, r22286-090812, on D2 16GB.

[shotofadds]

That's pretty bonkers, I don't know what could cause that...  maybe you could try to find if any particular track / combination of tracks triggers the problem?

[grantmasterflash]

That's pretty bonkers, I don't know what could cause that...  maybe you could try to find if any particular track / combination of tracks triggers the problem?

Agreed, very bonkers.  I'll play around with paring down the amount of files on the SDHC card and see if the dupe behavior stops at some point.

No big deal.  Music is playing great now and I can use the file browser until I figure out what's going on.

[Chronon]

There is an open bug at FS#8414 that has to do with duplicate entries in the database.  It has, so far, eluded a proper explanation.  It seems some of the database options may play a role in its appearance. . ..  Maybe you are seeing an instance of this bug.

[helloworld1]

Ok, I spent a couple of hours looking into this today and it looks like some new D2+ units have an entirely different power management chip, instead of the PCF50606 used by older D2+'s and the 'classic' D2.

Looking at firmware 4.11+ (since that is what I use on my player) it shows some interesting changes almost straight away:

- Follow the branch at 0x200000F8 (LDR PC,=0x20009BD4)
- In older firmwares, the first call in this routine starts writing to the PCF50606 almost immediately (using I2C slave address 0x10)
- In 4.11, this now tests the value of GPIOA6 (used for backlight control on older D2s) and stores a flag depending on its value
- Depending on this flag, the next routine (at 0x200722E4) either calls the PCF50606 init routine (0x200BD1C8) or another init routine (0x200BDBE8)
- The new init routine makes numerous I2C writes to slave address 0xE6, with completely different register addresses to the PCF chip.

Similarly, the backlight control routine (0x20072320) checks this flag, and calls either the PCF routine (0x200BD5C4) or a new one (0x200BE19C). To control the backlight brightness, this new routine writes registers 0x28 and 0x29 to I2C address 0xE6.

I don't know what this new chip is - anyone fancy opening their D2+ to find out? Although having said that, this disassembly should give us enough information to get Rockbox working eventually...  it's just that I won't be around for the next couple of weeks to look into it.

I crack open my 3-day D2+. Unfortunately I break the device's mic somehow.
I don't know where is the LCD controller chip. here lists most chips I can find on the main board.
PCF50635HN   (Is this one different from original?)
Samsung 919 (8G)
24020A
TCC7801
Samsung 840  K4M56323P I
MAX 1779
WM89859
8117 907A
D129C

The LCD screen have some text like this: ATOUCH A025 3DE4