Re: Linux 2.6.27-git3: no SD card reader

[Pierre Ossman]

On Tue, 14 Oct 2008 23:27:59 +0200
"J.A. MagallÃn" <jamagallon@xxxxxxx> wrote:

> 
> Err, sorry if this is stupid. I can see the card reader even if I don't
> load the jmb38x_ms.ko module,just sdhci_pci. I understand that it is needed only if
> I want to read memstick type cards. Or not ?

That is correct.

> I suppose you refer to 'settings for JMicron hardware in sdhci_pci module'.

Kind of. The JMicron controller supports 50 MHz SD cards, but it
restricted to 25 MHz in the name of higher compatibility.

> > Have you tested more than one card?
> > 
> 
> I have two (both unknown brands for me, but I write full names fwiw):
> - One "MBA Executive HiSpeed SD", 2Gb
> - One "takeMS SDHC card Class6",  8Gb
> 
> Both give about 10Mb/s on read with hdparm under 2.6.27.
> 

And both work fine in one of the slots, but not the other?

> > What machine is this? 
> 
> It's an Acer Aspire One. From what I have seen, there are two SD devices.
> Both are capable to read SDHC cards, I have switched cards and they work
> (or at the moment both are working as backwards-compatible in plain SD
> mode, if that even exists...).

The SD/SDHC distinction is all software, so the hardware doesn't really
care.

> 
> I call this half-working because I have to boot with a card inside a slot
> to have it detected. If i boot without the card-in, it even dissapears from
> lspci. This is just with the left SD inserted:
> 

No idea why this occurs. I'd guess some ACPI voodoo. You're going to
have to check with the PCI and/or ACPI guys to figure that part of the
mystery out.

> 
> And now with -git4:
> 
> ...
> mmc0: error -84 whilst initialising SD card
> ...
> mmc1: error -84 whilst initialising SD card
> 

Hmm... I thought you said that it was just the left reader that had
problems? Not both?

> 
> Mmm, looking at the logs I have seen that in 2.6.27 shpchp is loaded
> way before sdhci, and in -git4 it loads after it. Can this have
> any effect ?
> 

Perhaps, but I doubt it. If you rmmod sdhci-pci and modprobe it again
you should have the same behaviour as in .27.

> I solved this in modprobe.conf:
> 
> options sdhci debug_quirks=1
> 
> If I uderstood the source, that touches the controller clock on each reset for
> the operation to work.

Actually, I think you're just seeing the side effect that the
controller will revert back to the 25 MHz restriction.

Rgds
-- 
     -- Pierre Ossman

  Linux kernel, MMC maintainer        http://www.kernel.org
  rdesktop, core developer          http://www.rdesktop.org

  WARNING: This correspondence is being monitored by the
  Swedish government. Make sure your server uses encryption
  for SMTP traffic and consider using PGP for end-to-end
  encryption.

Attachment: signature.asc
Description: PGP signature