On 17/07/17 13:47, Adrian Hunter wrote:
This change seems like a reasonable justification for adding a new
SDHCI quirk, even if we in general wants to avoid that.
Adrian?
If we add a quirk for every register we end up with a very ugly version of
CONFIG_MMC_SDHCI_IO_ACCESSORS. So CONFIG_MMC_SDHCI_IO_ACCESSORS is a better
option. Another possibility is to add a member to struct sdhci_host,
initialized to SDHCI_DEFAULT_BOUNDARY_ARG, and write that. Drivers could
then change the value if necessary.
Second option seems to be better one. I will give it a try and see!
thanks,
srini
