Attention is currently required from: Tim Wawrzynczak, Nick Vaccaro, Zhuohao Lee, Patrick Rudolph. Subrata Banik has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/62294 )
Change subject: mb, soc: Add the SPD_CACHE_ENABLE ......................................................................
Patch Set 3:
(1 comment)
Patchset:
PS2:
Hi Subrata/Tim,
Sorry, i could be misunderstand your meaning. I thought the original request is asking to move the below full training setting to the function read_spd_dimm() https://review.coreboot.org/c/coreboot/+/62294/3/src/soc/intel/common/block/... :
if (dimms_changed) { memupd->FspmArchUpd.NvsBufferPtr = 0; memupd->FspmArchUpd.BootMode = FSP_BOOT_WITH_FULL_CONFIGURATION; }
Thats correct, I expected this to handle inside https://review.coreboot.org/c/coreboot/+/62294/3/src/soc/intel/common/block/... itself.
However, that change will require the FSPM_UPD *memupd be passing to the function mem_populate_channel_data() in order to pass it to the read_spd_dimm(). But the mem_populate_channel_data() is a common api and will require the other intel soc change(eg. tigerlake). https://github.com/coreboot/coreboot/blob/master/src/soc/intel/common/block/...
Do we need to change it or could we just apply the current approach?
True, we need to pass the FSPM_UPD structure rather soc config structure alone. if you have time constraint, please go with current approach and I will make it proper in coming days.