Andy, Henry, Ulf, and the rest,
I've posted the patch that I'm using for my SD/MMC support. It is a new driver based on some of the code from the original Atmel driver that uses the MMC framework. I've tested it on a at91sam9g45 (ES and production chips) and on an ek board and our own board.
I hope this can help you guys out and I hope this can get mainlined for others to enjoy.
If the patch is not in your email, here is a link to it in the archives: http://lists.denx.de/pipermail/u-boot/2010-April/070816.html
--Rob
On Fri, Apr 23, 2010 at 6:18 PM, Andy Fleming afleming@gmail.com wrote:
On Fri, Apr 23, 2010 at 6:21 PM, Albin Tonnerre albin.tonnerre@free-electrons.com wrote:
On Fri, 23 Apr 2010 16:58 -0500, Andy Fleming wrote :
On Thu, Apr 22, 2010 at 7:51 PM, Rob Emanuele rob@emanuele.us wrote:
Hi Henry & U-Boot Community,
I've been experiencing the same errors and frustration you have.
So I've been looking at this code and these patch sets for a day or two now. I've done that in conjunction with reading the SD card spec: http://www.sdcard.org/developers/tech/sdcard/pls/
I've come to the conclusion that this code as it stands will not work with any card that conforms to the SD Physical Layer Simplified Specification Version 2.0. This includes all SDHC cards and some non-HC cards that conform to version 2.0. I have a few 1GB cards that work just fine with the atmel_mci.c code on a 'G45 as it was in rev 95c44ec485b46ffb43dbdaa299f1491a500fdadf .
If your SD card is newer, you'll see in the for loop in "sd_init_card" in atmel_mci.c time out. In the 2.0 spec, you need to perform a CMD8 (SEND_IF_COND) first to see if your card is a 2.0 card. In CMD8 you tell the card the voltages you support and if you support HC cards. Once you send it the right data there, then ACMD41 will not have its BUSY bit set. That's all well and good, but additionally the CSD register is in a new format and that needs updating before any of this will work.
The best solution is to use the MMC framework, which *does* do all of these things that you suggest. It should be fairly straightforward to port the atmel_mci driver to this framework. If you see something lacking, feel free to mention it, or modify the framework. :)
I did port the atmel_mci driver to the MMC framework and posted the results on this mailing list a few months back. However, some people apparently experienced issues I have never been able to reproduce, and got few review. I recently adapted the AT91 SD/MMC support patch and the atmel_mci port to use the new C structures access, I'll repost it in a couple days in case anyone's interested.
Yeah, I see that now. I'm catching up from being in various other project quagmires. Sadly, I can't apply your patch if people are running into problems with it, but I'd far prefer it.
I also don't have such a board, though. If someone could apply Albin's patches, and try to identify why it's not working, I'd be very appreciative. :)
Andy