SanDisk Sansa c200v2, m200v4, clipv1, clipv2, clip+, and fuzev2

[wildmetal]

Greetings!
I have just made an acount on this forum to let you know of my troubles with rockbox on my v2 fuze and i can try diferent build's on my fuze to help find the bugs on v2.
The only build that doesn't give me the  "Bad checksum" on my v2 fuze is this http://download.rockbox.org/daily/sansafuzev2/rockbox-sansafuzev2.zip
And with this build i can browse the contents of the internal memory and the sd card but when i play a song it plays fine but when it should pass to the next song it crashes.
Also i have been trying some games and some of them crashes when exiting.

Cumps

[FlynDice]

It seems that r25502 "Bookmark.c cleanup, still no functional changes... yet" has broken playback from the uSD on my clip+.  I looked quickly and can't see quite why this would be yet.  It also prevents the player from booting with uSD installed.

Anyone else notice this?  Or am I just special again... 

[funman]

See http://www.anythingbutipod.com/forum/showthread.php?t=53702&page=8

Same build works differently on 2 Clip+ -> non-deterministic bug again?

Today I tried to use different transfer request sizes (DMA_S4 / DMA_S1) : no result.

I tried to panicf() from the ISR when the error bits were set but nothing happened, I could insert µSD and see the info in debug menu > disk info, but entering the filebrowser locked the Clip+

[keland44]

Greetings!
I have just made an acount on this forum to let you know of my troubles with rockbox on my v2 fuze and i can try diferent build's on my fuze to help find the bugs on v2.
The only build that doesn't give me the  "Bad checksum" on my v2 fuze is this http://download.rockbox.org/daily/sansafuzev2/rockbox-sansafuzev2.zip
And with this build i can browse the contents of the internal memory and the sd card but when i play a song it plays fine but when it should pass to the next song it crashes.
Also i have been trying some games and some of them crashes when exiting.

Cumps

i've noticed the same thing with the 0407 build for today maybe tonights build will be different hopefully

[FlynDice]

I was convinced I had solved the random freezups last night by not turning off the dma synchronization logic.  We disable this logic right now and for the v1's I think it's ok but for the v2's I think we probably should leave it enabled for the sd controller.  The manual says that if the dma controller and the peripheral are using the same clock you can disable this logic to get better throughput.  I think with the v2's though we are definitely on 2 different clocks.  Well, I added this change to what I was working with and got 8 straight hrs with no lockup last night.  I turned the unit off at that point, reverted to svn and started adding my changes part by part to see what the magic piece was and now I can't seem to get back there.  I figured it was either only configuring the FIFO once or the DMA sync logic but perhaps it  is a timing issue and some of the debug code I had in place was making it work.  Still investigating...  I have tried different watermark levels and different dma sizes and like funman have had similar results.

As far as SD writes go the problem to solve now is a FIFO over/underrun error whenever we try to write to the card.  We do get data into the FIFO as you can read the MCI_FIFO_COUNT to see how many bytes are there but I have not yet gotten to looking at  the MCI_TCBCNT & MCI_TBBCNT registers (card byte count & bus byte count) nor any sd card response values which may help.

Edit:

On further thought perhaps playing with the software hold enabled on the clip+ is what helps.  Testers?  please report your results good or bad!

[yelped]

Thanks for the update and "Keep up the great work"!

(Delete this post when you're done reading it)

[vadik]

Look at rbutil/mkamsboot , especially dualboot/dualboot.S, but I advice you don't modify this code because it's very easy to brick your Clip if you do so.

Thanks. I just wanted to flip the condition to boot OF by default, which is unlikely to brick the device. (Famous last words, I know.)

As far as SD writes go the problem to solve now is a FIFO over/underrun error whenever we try to write to the card.  We do get data into the FIFO as you can read the MCI_FIFO_COUNT to see how many bytes are there but I have not yet gotten to looking at  the MCI_TCBCNT & MCI_TBBCNT registers (card byte count & bus byte count) nor any sd card response values which may help.

Quite interesting.  I took a look at the code and the AS3525 datasheet (linked to from the SansaAMS wiki page), but is it the correct datasheet for the SoC inside Clipv2?  For one, the SD controller's base address is different (SD_MCI_BASE 0xC8020000 in as3525.h and the datasheet, SD_BASE 0xC6070000 in sd-as3525v2.c).  Are there other secret documents I have yet to find?

Edit: found the MMC/SD Controller Interface spec (I think this is the right model):
http://download.chipestimate.com/dscache/techdocs/PL180_TRM.pdf

[wildmetal]

Greetings!
I have tried the new build but i'm still getting the "Bad checksum" error on my v2 fuze 
Do i need to make a new patched OF for the build's to run?

Cumps

[FlynDice]

@vadik

One of the big differences in the v1 and v2 chips is the sd controller(s).  The as3525v1 has 2 pl180 controllers(1 paired with internal, 1 paired with uSD) that we have documentation for and understand quite well.  The v2's have a different controller and setup.  I think we have identified the controller (find it in one of my previous posts) which has both the internal and uSD attached to it.  Unfortunately all we have are some lunux patches, marketing sheets, and our own exprimenting to "deduce" how we need to use it.

[keland44]

just to report this http://download.rockbox.org/daily/sansafuzev2/rockbox-sansafuzev2-20100408.zip build works pretty well on my v2 fuze 4gb when i get home i'll test it on my other v2 fuze to see how it works
but the playback seems to be working pretty well on this one i haven't tested any of the other stuff

[vadik]

I think we have identified the controller (find it in one of my previous posts) which has both the internal and uSD attached to it.  Unfortunately all we have are some lunux patches, marketing sheets, and our own exprimenting to "deduce" how we need to use it.

Synopsys DesignWare® Mobile Storage Host Controller?  This seems to be a spec for hardware designers: https://www.synopsys.com/dw/doc.php/iip/DWC_mobile_storage/latest/doc/dwc_mobile_storage_db_eval.pdf .  It says:

1.2.6 Example Linux Demonstration Software Package Features
❖ DWC_mobile_storage controller-specific host-driver APIs
❖ DWC_mobile_storage controller-independent, SD_MMC_CEATA protocol-specific bus-driver APIs
Note: The software driver is available on request. Contact Synopsys support center for more information.

Seems worth a try.  (There are more documents at http://www.synopsys.com/dw/ipdir.php?c=dwc_mobile_storage , but you can only download two without signing up for an account.)

Oh, and can you point me to the Linux patches?

Now, something else I don't get.  What exactly is this AS3525v2?  (A quick google search brought lots of Rockbox references.)  Is it ARM926EJ-S?  (Not that I know the difference)  Is it a variant of AS3525 hacked by SanDisk or something different?  Is there a spec or at least a summary of differences between AS3525 and v2?

Thanks for answering...  And for writing Rockbox, while I'm at it.

[saratoga]

Seems worth a try. 

It has been tried.  Do a search for Synopsys in this thread. 

Now, something else I don't get.  What exactly is this AS3525v2?  (A quick google search brought lots of Rockbox references.)  Is it ARM926EJ-S?  (Not that I know the difference)  Is it a variant of AS3525 hacked by SanDisk or something different?  Is there a spec or at least a summary of differences between AS3525 and v2?

The ARM926EJ-S is a CPU, the AS3525 is a system on a chip.  The V2 (also called +) is some customized version that uses the ARM926EJ-S instead of an ARM922 CPU, has more memory, and some changed hardware (e.g. SD controller). 

[wildmetal]

just to report this http://download.rockbox.org/daily/sansafuzev2/rockbox-sansafuzev2-20100408.zip build works pretty well on my v2 fuze 4gb when i get home i'll test it on my other v2 fuze to see how it works
but the playback seems to be working pretty well on this one i haven't tested any of the other stuff

Greetings!
I have tried this build and the playback works well now, some games still crash.
Maybe this is a stupid question but i have been loocking for the builds for the v2 fuze and don't find them only for the v1 fuze maybe thats why i only get the "bad chesum" error, where in the download section is the v2 builds please, because i would like to try all the new builds for the v2 and report the bugs i found.

Cumps

[Xanikseo]

@wildmetal: All the daily builds for fuzev2 are here. The newest file is always http://download.rockbox.org/daily/sansafuzev2/rockbox-sansafuzev2.zip. Whatever the day, that file will always be the most up-to-date.

EDIT: There also seems to be a "current" build for the v2 here. The most up-to-date of all the options if I'm not mistaken.

BTW I'd just like to take the opportunity to thank all the rockbox developers , it's a great project, and I'm so pleased that you've started working on the fuze/clipv2 and clip+.

[wildmetal]

Thanks for your reply Xanikseo, i'm already testing a new build and for now the playback works great in the internal memory and on the sd card.