> I'll try to remove it. Note that sh73a0-kzm9g.dts also uses broken-cd, > without ill effects. This one does have a seperate card-detect-irq (which is not populated for sdhi2) while the other one only has one, generic irq. That's at least a difference. We could try to see if there are unexpected card-detect events by instrumenting mmc_detect_change() in tmio_mmc_pio.c, but my feeling is that the card_busy change you reverted is only making something visible which went wrong before the change, also. So, it might be worth to protect the whole voltage_switch and card_busy functionality with TMIO_MMC_MIN_RCAR2?
Attachment:
signature.asc
Description: PGP signature
