08:31 <jbrunet> One thing we have been failing to do for years when it comes to MMC, and more specifically eMMC, is properly managing pull-up. In early, we should be pulling up CMD and DAT0. But AFAIU, we should not be pulling anything out of that. DAT1-7 should be pulled by eMMC device itself and I don't think we should be pulling CMD0 or DAT0 in 4 or 8 bits mode. Yet, we are pulling regardless of the mode ... No pull was tried in the

08:31 <jbrunet> past and had problem (both u-boot and Linux) but putting pulls on CMD0 and DAT0 in early modes only, then disabling it has not been tried. I would be more aligned with the JEDEC spec

08:32 <jbrunet> If you want to debug MMC, I'd suggest starting there, fixing the current pinconf setting and adding another one to switch for higher speed modes

13:49 <minute> i'm now circumventing bpi cm4's wifi issue by using an MT7612U based m.2 card that plugs into the wwan (!) slot. works perfectly

16:29 <f_> Huh.

16:29 <f_> Looks like the eMMC gives out EXT_CSD with a HC_WP_GRP_SIZE SPL doesn't like

16:31 <f_> Actually, hold on.

16:32 <f_> It's the other read-only fields it doesn't like

17:16 <f_> EXT_CSD_PARTITIONING_SUPPORT, EXT_CSD_REV, and EXT_CSD_HC_ERASE_GRP_SIZE

17:17 <f_> https://elixir.bootlin.com/u-boot/v2024.07/source/drivers/mmc/mmc.c#L1900