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

[FlynDice]

Alright, once again calling all testers.

I especially would like to make sure this runs well on a clip.

This patch changes the assumption for MCLK on the internal pl180 controller from PCLK to IDE_CLK.

Please see http://www.rockbox.org/tracker/task/10838 and leave me some feedback with your results!

Thanks.

[daniel_at]

Hi hackers,

I also have problems with a recent SVN (r23905), Fuze V1 and microSD. Just updated the fw and bootloader and now my Fuze cant be switched on with an MicroSD card inserted (Hama 2GB - dont know further details).

The bootloader can load RB, the status-bar (battery, clock...) gets renderd, but instead of the menu the Rockbox-Bootupsplash stays and the device locks up. Sometimes led/background fades out and in at buttonpress, sometimes a full lock up.

If I insert the card while RB is running, I get instantaneously

Code: [Select]

*PANIC* microSD init failed: -9 (and sometimes i get -3; but mostly -9).
So I cant visit the debug-screen, with card inserted.

Without a microSD card inserted, everything works fine.

[Edit]
Just downdated to r23836: Booting with/without and/or later inserting now works. The debug screen says:

Code: [Select]

G20D REV 7.12
M=27 O=PH
Blocks: 0x003cd000
Speed: 25MBits/s
Taac: 50.0ms
Nsac: 0clk
R2W:*2

[/Edit]

[Edit 2]
Some further test: Still using bootloader r23836 but testing different RBs
r23900 is the last one that boots/works on insert
r23901 after this change to ata_sd_as3525.c RB fails again to boot with inserted microSD and Panics at insert

[/Edit 2]

bye,
Daniel
 

[FlynDice]

@ daniel_at:

The -9 is a failure during the init stage where the card is being deselected in order to read the csd register.  In the revision that breaks for you I had moved the point where we bring the cards up to operating speed to just after this point so we could look at the speed value from the csd register and set the card frequency appropriately.

If you can build can I get you to try adding a delay to line 336 of ata_sd_as3525.c:

335                  return -8;
336          mci_delay();
337    }

I'll attach a diff also if you'd rather do that.

I think this should work.  The problem you're running into is that the card is now running at 400kHz at this point as opposed to 31 MHz before.

Let us know how you make out!

[daniel_at]

Hello FlynDice,

Thx for the quick response - seems youre right. I added the delay, now booting and inserting is possible again, without crashing.

Further I noticed that in the debug screen, my microSd card is now listed with Speed: 50MBit/s (where it had only 25M with the previous build)

Filelisting and playing is also working

Thx.
Daniel

[FlynDice]

Still need someone to test FS#10838  on a clip &/or fuze please.

[funman]

FlynDice, looks like it is working on both Clip & Fuze

Do you think we'll have to upgrade the bootloaders for 3.5 release after all the modifications you made to SD driver ?

[BlastTyrant]

Still need someone to test FS#10838  on a clip &/or fuze please.

Just patched r23928 with the diff and tested it out for 20 minutes or so. I did 15 power on/off and let it play for around 15 minutes. I had no issues running it on my Fuze.

IDE is set to 50MHz. Actual shows 49MHz.
SD (internal ?) is running at 25MHz.
8GB cl6 Transcend uSD is running at 31MHz (50.0MBit/s).
1GB cl2 Sandisk uSD is running at 15MHz (25.0MBit/s).

[FlynDice]

funman

I don't think we would have to but quite honestly you would probably have a better feeling for that than I do.  I'm running a bootloader built off of r23799 right now but I was running off of one from August until just before that.  We would still be overclocking the internal card with the present bootloader but it seems to work fine.  I guess what it comes down to is if we want to spend the time to make it technically correct or is the fact that it "just works" good enough?

Once again , thanks all for the testing!

[saratoga]

We don't even know what the internal memory is speced for, so if its working I wouldn't worry too much about trying to conform to some spec that may not even apply.

[FlynDice]

Well, I have finally managed to get 4 bit widebus working  !!

I had to fly all night though so I'm going to get some shuteye and clean it up when I'm a little more coherent.

In order to head off kugel's question I'll post the speed tests now.  UNS = UNALIGNED_NUM_SECTORS


http://pastie.org/741378

[kugel.]

Nice, it shows some decent speed ups! But only for reads (create/write is still slow)? That seems strange.

SVN has gotten pretty slow?

BTW, my fuze stopped working. It just doesn't turn on anymore. The last time I used it was thursday. I turned it off normally after listening to radio.

[saratoga]

Nice, it shows some decent speed ups! But only for reads (create/write is still slow)? That seems strange.

The 1MB write is basically 4x faster, so its probably just a limitation of the internal memory controller for smaller writes, since those often involve lots of costly copying and rewriting of sectors.

BTW, my fuze stopped working. It just doesn't turn on anymore. The last time I used it was thursday. I turned it off normally after listening to radio.

Want my Clipv2 to play with

[kugel.]

Nice, it shows some decent speed ups! But only for reads (create/write is still slow)? That seems strange.

The 1MB write is basically 4x faster, so its probably just a limitation of the internal memory controller for smaller writes, since those often involve lots of costly copying and rewriting of sectors.

Where are those writes faster? They are only faster with a bigger buffer (64 UNS). And they're slower without compared to my tests I made a while back (see FS#10805)

BTW, my fuze stopped working. It just doesn't turn on anymore. The last time I used it was thursday. I turned it off normally after listening to radio.

Want my Clipv2 to play with

No thanks, I have a FuzeV2 rotting around myself

[DavidW]

For the last two days, I have been having trouble with the database on my Fuze V1.  It will initialize properly, but later I get a database is not ready error.  When I try to initialize again it fails with some write error to the internal flash disk.  (Sorry I can't give the exact error now I am not sure I want to intentionally cause this to happen again.)  If I run scandisk on the internal disk it seems to fix everything but a couple of time it left files I could not delete on the disk.  These files finally went away when on one occasion the entire .rockbox directory was corrupted. 

Can anyone tell me if this is an issue with my player or is it a software problem?  I am currently using r23989-091214.

Thanks,
David

[FlynDice]

@DavidW:

Sorry to hear about the problems.  A couple of days ago I put in the change for 4 bit widebus on the SD cards so that may be a possible culprit here but you would be the first reporting problems.  I'm working with an e280v2 and I have just initialized my database and rebooted to commit it and it is functioning normally for me.

I would try the following:

In the .rockbox folder you should find several files named database_(X).tcd.  X runs 0-8 and theres an idx also.  Delete those and try to reinitialize the database.

If that doesn't work I would try formatting the internal disk with the OF and then reinstalling.

If that fails then copy down your steps to reproduce this and any error messages and let us know.

Good Luck!