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

[FlynDice]

I spoke with topik today in irc who was experiencing this problem with a fuze and an 8gb class 2 sandisk card.

For those experiencing this problem would you first try removing the uSD card and see if rockbox runs ok without the card inserted.

If it does would you then  insert the card and see if it inits properly(you can browse files and play music from it).

Would you then go to system/debug/View disk info and see what the Speed value is, press select to get microSD 1(uSD card)

You should see either 25.0 or 50.0  MBit/s.

Then would you report back here with player type and sd card info, manufacturer, size, speed class(small number in circle), and speed you got from the debug page.

Sorry for your problems!  Thanks for your help!

[epithetless]

It's hit-or-miss as to whether my microSDHC card freezes the player (either during bootup with it already inserted or when inserting it after bootup), but when it does work, the details are as follows:

Speed (from debug screen): 25.0MBit/s
Manufacturer: SanDisk
Size: 8GB
Speed Class: 2*

[*My apologies for mis-stating it as class 6 earlier; I was thinking of a different card at the time.]

I should also add that the audio playback halting/freezups I've experienced have occurred with music that's stored on the microSDHC card, not the internal memory (so far, at least) -- Which leads me to think that problems exist with the current microSD implementation even beyond the initialization stage.

[FlynDice]

@epithetless

If you can build would you try this patch?

http://pastie.org/728675

Thanks.

[epithetless]

@epithetless

If you can build would you try this patch?

http://pastie.org/728675

Thanks.

Unfortunately, I haven't worked my way up to compiling and patching builds yet, and I'm slammed with other work at the moment, so it will be a while before I have the time to sit down and sort through it.

Can someone else assist FlynDice with this or compile and share a build utilizing his patch?

[BlastTyrant]

I'm getting the Fuze freezing issue also. I'm able to reproduce it fairly well.

First some basic info:
2GB Fuze
8GB Transcend microSD Card
Class 6
microSD speed from debug menu: 25.0MBit/s
Currently running r23854-091205
Previously running r23781 - didn't have this issue when running this build.

I'll first report that it doesn't freeze at all when the microSD card is removed.

Specifics on reproducing the freeze (if I'm not currently having the freezing issue):

• Connect Fuze to computer and boot into OF
• Add or remove music from player
• Let OF refresh the media list
• Boot player like normal (into Rockbox)

At this point RB will usually freeze up on every 2nd boot on the boot screen where it displays "Ver. r23854-091205" right after it displays the "Boot Ver. 1.0RC" part of the boot screen.

Sometimes it will freeze up on every boot for 3-4 boots in a row.

When the player freezes a hard shut down is required.

*note that this is when the microSD card is in the player



The only thing that seems to stop the freezing problem 100% is to remove the microSD card when I boot the player.

It seems that the player is less likely to freeze on boot if I go to resume playback and listen to music for at least 2 minutes.



I'm going to try to read up on getting set up with a build environment either tonight or tomorrow so that I can test FlynDice's patch. Hopefully that will go smoothly

If anyone needs any more information let me know and I'll try my best to accommodate you.

[7o9]

The patch is now commited as r23870.

[FlynDice]

I have committed the patch above that moves the boost from SD identification freq to SD operating freq to after the switch to HS timings happens.  I'm not sure why this all worked fine at 62 MHz and now at the lower freq problems arise.  This seems to get the cards through the init process for now though so it's a first step.

When the cards get switched to HS timings we can actually see that from the info on the view disk info debug page.  If the switch to HS timings is successful the speed will show 50 MBits/s.  If we get this we can run the card at 31 MHz and be within the spec.  If not we are running slightly over spec, the limit being 25 MHz for a standard speed card.  My problem here is that we run PCLK at 62 MHz and we can only divide this by 1 or even integers (1,2,4,6...) so our next lower frequency is 15.5 MHz.  Using 15.5 I get data crc failures presently during writes on both internal and uSD cards and I'm not quite sure why.

Right now I'm looking at the data crc failures. I think this is the preferred solution.  If unable to overcome that then I think lowering PCLK to ~50 MHz solves our overclocking problem but may have other negatives associated with it.

Anyway, please yell loudly if you are still experiencing problems & thanks to those testing & helping out!

EDIT: Divider is even ints not power of 2...

[epithetless]

Sadly, I'm still experiencing some system freezes during startup with my microSDHC card inserted, when inserting it with Rockbox already on, and when playing music from the card.

As before, the "View disk info" screen under the debug menu lists the card's speed as 25.0Mbit/s. (To refresh, it's an 8GB Class 2 SanDisk microSDHC card.)

Thanks for your work on this, FlynDice!

[BlastTyrant]

I have committed the patch above...

I just put r23879 on my Fuze. The freezing issue is completely resolved for me now. I've booted it about 30 times since putting the new build on and it didn't freeze once.

MicroSD card is still running at 25MBits/s.

[epithetless]

Well, I haven't had any issues since my last post, so perhaps they were just a fleeting residual effect of resuming playback immediately after updating to the newer build. Thanks again, FlynDice.

[FlynDice]

Glad to hear things are working good now.

Could I get some tests on this:   http://pastie.org/732553
This patch runs the internal at 15.5(sort of see below..), standard uSD at 15.5, and HS uSD at 31

I think I have solved the problem with crc failures on sd writes at 15.5 MHz.  For some reason we need a small delay after we send the write command to the controller before we start the transfer when writing to the cards at 15.5 MHz.

I am now almost certain that the internal controller uses the IDE clock as it's MCLK.  For some reason It needs both  IDE_CLK and NAF_CLK enabled to function which is where my confusion came from several weeks ago.  With this thought in mind it should be possible to run the internal cards at 25 MHz, High speed uSD cards at 31 MHz and standard speed uSD cards at 15.5 MHz.  That would be all cards running at their max frequency within the specs.  We could choose to lower that later if tests show an advantage power wise.

[BlastTyrant]

Glad to hear things are working good now.

Could I get some tests on this:   http://pastie.org/732553
This patch runs the internal at 15.5(sort of see below..), standard uSD at 15.5, and HS uSD at 31
...

OK, I finally got set up on my Ubuntu install to make Rockbox builds.

I applied your patch to r23895 and put it on my Fuze.

I didn't experience any problems with it when I was testing it for about 15 minutes. I booted it multiple times with different cards and tested playback. I'll leave it on there for a little while. Let me know if you need a more extensive test.

Here's the info that I got while running the patched build:

With my 8GB Class 6 Transcend uSD:
HW Info*:
   SD: 15MHz
   uSD: 31MHz
Disk info:
   Speed: 25.0MBit/s

I also tested it with a 2GB Class 2 Kingston uSD and got the same results.

*The 'Actual' column reading was the same as the 'Set' column reading while it was filling the buffer. I didn't know if you needed me to include this, but I figured it couldn't hurt.

[kugel.]

Shouldn't class 6 cards be 50MBit/s? Those are rated for over 10MB usually (which we can't handle anyway but we do manage to be faster than 25MBit/s if the cards allow it).

[FlynDice]

@BlastTyrant:

The speed value is a maximum value which we aren't actually running the card at. I believe your card(s) is indeed successfully switching to 50.0.  The initial screen which appears for View disk info shows [microSD 0], which is the internal SD card, which is a v1 and does indeed not switch to 50.0.  Press the select button while in the View disk info page and it will cycle to  [microSD 1] which is the uSD.

Since your card is running at 31 MHz, the code checked on the speed value reported in the card's CSD register and found it to be 50.0 so it divided PCLK by 2 to get 31 MHz.  Otherwise it would have divided by 4 to run at 15.5.

EDIT: ---------->  Thanks to all for taking the time to test for me!

[BlastTyrant]

@BlastTyrant:
...Press the select button while in the View disk info page and it will cycle to  [microSD 1] which is the uSD....

Well, I feel kind of stupid now 

Yes, when I press the select button it brings up microSD 1 and it shows that the speed is in fact 50.0MBit/s

Sorry about the confusion, and I'm glad that I can help.