[U-Boot-Users] drivers MMCplus for at91sam9x

[Ken.Fuchs at bench.com]

Pierre Savary wrote:

> Then my MMC 4GB works with my Linux kernel but if I can't 
> load my kernel
> (located on the first part of this MMC) ... it's not really 
> interesting :(

> So, somebody does already use MMC v4 with U-boot???

Assuming a AT91SAM926x MCI controller:

Probably.  Given an MMC v2.x compliant U-Boot driver, it
would be relatively easy to add support for an MMC v4.x chip.
If there is an MMC v4.x compliant Linux driver, it could be
ported to U-Boot (Many, if not most, U-Boot drivers are ports
from the corresponding Linux driver.)  Alternatively, if both
an MMC v2.x compliant U-Boot driver and MMC v4.x compliant Linux
driver exist, the port would be trivial; just copy the MMC v4.x
Linux code into the corresponding places in the MMC v2.2 U-Boot
driver until it supports your MMC v4.x chip.

Such an MMC v4.x U-Boot AT91SAM926x MCI driver should be able to
support any v4.x MMC chip, at least via a one bit MMC bus.  (I
remain unconvinced that the AT91SAM926x MCI controller can support
a 4 bit MMC bus; only a working MCI driver using a 4 bit MMC bus
verified by a logic analyzer or similar compliance/debugging device
would be convincing enough.)

If the Linux AT91SAM926x MCI driver supports a 4 bit MMC bus, then
the U-Boot AT91SAM926x MCI driver can borrow the Linux driver's
code and support a 4 bit bus also.

Sincerely,

Ken Fuchs

> -----Message d'origine-----
> De : u-boot-users-bounces at lists.sourceforge.net
> [mailto:u-boot-users-bounces at lists.sourceforge.net] De la 
> part de Pierre
> Ossman
> Envoyé : samedi 12 avril 2008 11:28
> À : Ken.Fuchs at bench.com
> Cc : pierre.savary at kerlink.fr; u-boot-users at lists.sourceforge.net
> Objet : Re: [U-Boot-Users] drivers MMCplus for at91sam9x
> 
> On Fri, 11 Apr 2008 13:54:13 -0500
> Ken.Fuchs at bench.com wrote:
> 
> > 
> > The manual for the AT91SAM926x processors clearly states that the
> > MCI controller supports MMC 2.2.  I asked Atmel whether 
> this controller
> > could support 4 bits to an MMC 4.x chip and they said _no_. 
>   The answer
> > is supposedly from their engineering group in France.
> > 
> 
> What they answered was if Atmel would support such a solution, not if
> the hardware would.
> 
> > 
> > How does one program the MCI controller to send 4 bit 
> (parallel) data
> > an MMC 4.x chip, with or without telling it that it is communicating
> > to a 4-bit SD chip?  The only way to get 4 bits of data out 
> of the MCI
> > is to tell it that it is connected to a SD (1.0) chip.  If 
> you program
> > the MCI to communicate with an MMC chip, it will send data 
> out only on
> > the low order bit.
> 
> You tell it to use all four bits. That's it. The setting isn't MMC/SD,
> it's 1-bit/4-bit.
> 
> > Both ends of the communication link must considered.  It may not be
> > sufficient that the MMC chip is MMC 4.x; The fact that the MCI
> > controller is only MMC 2.2 or SD 1.0 compliant, may or may 
> not impact
> > this special handling.
> 
> It does not in the slightest. Remember that these controllers are
> extremely dumb. The low level behaviour of the protocol hasn't changed
> since the very first specs, so only software needs to be changed to
> support even the newest features.
> 
> Rgds
> -- 
>      -- Pierre Ossman
> 
>   Linux kernel, MMC maintainer        http://www.kernel.org
>   PulseAudio, core developer          http://pulseaudio.org
>   rdesktop, core developer          http://www.rdesktop.org
> 
> --------------------------------------------------------------
> -----------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
> Don't miss this year's exciting event. There's still time to 
> save $100. 
> Use priority code J8TL2D2. 
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java
.sun.com/javao
ne
_______________________________________________
U-Boot-Users mailing list
U-Boot-Users at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/u-boot-users