Patrick Rudolph has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/32682 )
Change subject: mb/samsung/lumpy: Fix MRC raminit ......................................................................
Patch Set 2:
Patch Set 2:
What was broken? Is this change based on hardware testing or review of source code?
You need to remember that somewhere along the way, MRC.bin _and_ version of pei_data changed, that was related to 0xf0 removal from SPD map.
It's not that long ago (maybe 4 months) when I last did fresh build of samsung/lumpy using non-native raminit (I was looking into the ill-located MRC heap) and I am pretty sure I did not have any trouble related to amount of detected memory (2 GiB on-board + 4 GiB SO-DIMM).
It's not broken (yet). I'm migrating PEI data to devicetree: https://review.coreboot.org/q/topic:%22mrc_to_devicetree%22+(status:open%20O...)
https://review.coreboot.org/c/coreboot/+/32069 will set dimm_channel0_disabled/dimm_channel1_disabled by looking at spd_addresses and spd_data and as spd_data[2] isn't set, it will always disable channel1.
Can you test this on real hardware with mrc.bin and confirm that spd_data[2] works, too?