On 19.08.2009 13:52, Stefan Reinauer wrote:
On 8/17/09 5:57 PM, Carl-Daniel Hailfinger wrote:
If FT2232H SPI is not enabled, it should be compiled out completely. We can't remove ft2232_spi.o from unconditional OBJS yet due to our makefile structure (make features), but this patch adds #ifdefs around all FT2232H code, so the net effect is the same.
This is 0.9.1 material.
Signed-off-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net
Please address the one issue below.
I worked on that thing for hours and didn't get it to work reliably. Detailed explanation below.
Acked-by: Stefan Reinauer stepan@coresystems.de
Thanks, r691.
Once I know how to solve the makefile issue, I'll send a followup patch addressing your comments.
For serprog.c the mechanism is moved to the Makefile
ifeq ($(CONFIG_SERPROG), yes) FEATURE_CFLAGS += -D'SERPROG_SUPPORT=1' OBJS += serprog.o endif
Maybe the two cases should be unified?
CONFIG_SERPROG is not determined during make features, so we can use it in ifeq easily. FT2232_SPI support is determined during make features which is after any ifeq statements have been parsed, and thus the ifeq solution will always give the same result regardless of whether FT2232_SPI was actually selected or not.
There are two ways out: - Run make twice automatically (which also means running make features and make compiler and make pciutils twice). I once had such a patch in flashrom, but I was asked to convert this back to the run-once model so I did that. - Move make config to a separate configure.sh
Neither of these ways sounds great.
Any advice is appreciated.
Regards, Carl-Daniel