On Sat, Jan 14, 2023 at 07:51:00PM +0100, Pali Rohár wrote:
On Friday 13 January 2023 21:00:21 Tom Rini wrote:
On Sat, Jan 14, 2023 at 02:41:32AM +0100, Pali Rohár wrote:
On Friday 13 January 2023 16:38:55 Tony Dinh wrote:
@@ -16,4 +19,9 @@ obj-$(CONFIG_SPL_BUILD) += mv_ddr_build_message.o obj-$(CONFIG_SPL_BUILD) += mv_ddr_common.o obj-$(CONFIG_SPL_BUILD) += mv_ddr_spd.o obj-$(CONFIG_SPL_BUILD) += mv_ddr_topology.o +obj-$(CONFIG_SPL_BUILD) += mv_ddr4_mpr_pda_if.o +obj-$(CONFIG_SPL_BUILD) += mv_ddr4_training.o +obj-$(CONFIG_SPL_BUILD) += mv_ddr4_training_calibration.o +obj-$(CONFIG_SPL_BUILD) += mv_ddr4_training_db.o +obj-$(CONFIG_SPL_BUILD) += mv_ddr4_training_leveling.o obj-$(CONFIG_SPL_BUILD) += xor.o
And all these new files are ddr4 specific, so should be wrapped in makefile section: ifdef CONFIG_DDR4
Looking at the Makefile in question, I think we might want to make the whole thing ifdef CONFIG_SPL_BUILD ... endif and then more finely control building of what objects are built. Perhaps: drivers/Makefile:obj-$(CONFIG_ARMADA_38X) += ddr/marvell/a38x/ should only be for SPL instead, even?
Some cleanup like this can be done. But it is related to DDR4 support and is mostly independent of it. So lets do it after having DDR4 there.
We're going to also want to not build the DDR3 code on the DDR4 platforms, right? A little clean up would make adding the DDR4 code a bit cleaner for both cases. It's not a hard no, if someone really wants to do the clean-up after.