On Mon, Feb 19, 2018 at 03:50:29PM +0100, Rasmus Villemoes wrote:

Hi,

Please consider adding

dd3f6983b4a4 mmc: sdhci-of-esdhc: disable SD clock for clock value 0 97618aca1440 mmc: sdhci-of-esdhc: fix eMMC couldn't work after kexec f2bc600008bd mmc: sdhci-of-esdhc: fix the mmc error after sleep on ls1046ardb

(present in mainline since 4.15-rc1, 4.16-rc1 and 4.16-rc1, respectively) to 4.14-stable.

The third removes all the code added by the second, but the clearing of the ESDHC_TB_EN bit from esdhc_init() (whose sole user is esdhc_probe) still happens during esdhc_probe() due to

esdhc_probe() -> sdhci_add_host() -> __sdhci_add_host() -> sdhci_init(soft=0) -> sdhci_do_reset(host, mask=SDHCI_RESET_ALL) -> host->ops->reset(host, mask);

Presumably they could/should have been squashed, but I suppose -stable prefers keeping a 1-1 correspondence with mainline when possible.

Yes, 1-1 is best, all now queued up, thanks.

greg k-h