Suspend/resume without bringing down SDIO

Dongas dongas86 at
Mon Sep 21 00:03:33 EDT 2009

2009/9/6, Pierre Ossman <pierre at>:
> On Mon, 03 Aug 2009 17:24:55 -0400
> Chris Ball <cjb at> wrote:
> > Hi Pierre, and libertas-dev,
> >
> > We're using a Marvell SDIO 8686 wifi module in the new OLPC XO-1.5,
> > and need the SDIO card to stay up across suspend so that it can wake
> > us with an SDIO interrupt when a packet arrives while we're suspended.
> > (We do this with the USB 8388 on the XO-1, too.)
> >
> > I see that the SD stack doesn't yet have support for per-driver
> > suspend/resume hooks, though.  We're using a standard sdhci host,
> > which we keep powered during suspend.  Any suggestions on what we
> > should work on, or pointers to patches that accomplish these if
> > they exist already?  It sounds like we're looking for:
> >
> > * a way to tell the core not to remove the card on the way into
> >   suspend; it's always going to be present when we resume.
> The missing key here really is that the MMC layer never grew the
> functionality to keep cards powered during suspend. Once it has that it
> can just poke the card after resume to see if it is still running, and
> only do a reinit if that fails.

Is there any story behind that why all cards must be powered down
during suspend?

There're real cases that some SDIO WiFi cards indeed DO NOT like to be
powered down during suspend, e.g they want to enter deep sleep.



> > * a way to register driver hooks to call suspend/resume functions
> >   in the same way current libertas if_usb.c does.
> That would be sensible once the above is fixed.
> Rgds
> --
>     -- Pierre Ossman
>  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.
> _______________________________________________
> libertas-dev mailing list
> libertas-dev at

More information about the libertas-dev mailing list