Hi Wolfgang,
On Monday 16 May 2011 09:18 PM, Wolfgang Denk wrote:
Dear Aneesh V,
In message4DD134DD.6040802@ti.com you wrote:
create mode 100644 spl/board/ti/omap4.mk create mode 100644 spl/board/ti/rules-omap.mk
We do you introduce these files? Please omit them, and use a plain Makefile instead.
I introduced them because 'spl/board/ti/sdp4430/Makefile' and 'spl/board /ti/panda/Makefile' were exactly same and wanted to re-use the make rules.
But in the result, the Makefiles are still the same, aren't they?
Yes, but the contents are not duplicated in two files. Easier for maintenance.
I indeed wanted to abstract it out even further in future so that the make rules can be shared between OMAP3 and OMAP4 too. This way the changes needed to support a new board will be lesser and maintenance will be easier while adding new features.
For instance, if I have to add NAND support today I just need to update omap4.mk instead of updating the Makefile of both panda and SDP.
This re-use will be evident when you see a later patch that replicates the same Makefile for panda.
Can you please rather try and move this to a common directory level, then?
The top-level make rule being the following, we will need a Makefile in the board directory, right?
+SPL :$(TIMESTAMP_FILE) $(VERSION_FILE) depend tools + $(MAKE) -C spl/board/$(BOARDDIR) all
I have a bad feeling about this. If the Makefiles are the same, then the code is probably more or less the same, also. Eventually larger parts of the code should be moved to a common directory, too?
In fact, code is already common. All OMAP code specific to SPL comes from spl/board/ti/spl-omap.c . All OMAP code shared with OMAP U-Boot comes from the respective directory within arch/arm/cpu/armv7/*
In the interest of simplicity I deferred all board specific initialization of OMAP4(mainly mux) to U-Boot. So, SPL is board independent today for OMAP4.
best regards, Aneesh