Andy Green wrote:
> +     /* we are going to do more commands to override this in
> +      * mmc_suspend_host(), so we need to change sd_idleclk for the
> +      * duration as well
> +      */

Hmm, isn't this a race between the card's internal controller doing
whatever it does and the glamo's PLLs being shut down, i.e., when the
clock is being stopped for real ?

I think it would be good if someone could test this by turning off
TCLK right after mmc_suspend_host returns. If there is still
corruption, add a suitable usleep or msleep. Then measure the time
the system spends in mmc_suspend_host and add that to the sleep.

It's also not so nice that there's per-device state in a single
global variable. But okay, one Glamo per system is bad enough, who
would want to have more anyway ? ;-)

- Werner

Reply via email to