On Nov 21, 2007, at 9:15 AM, Jerry Van Baren wrote:
Kumar Gala wrote:
Modify libfdt/Makefile to conditionally compile the *.c files based on the board config. Signed-off-by: Kumar Gala galak@kernel.crashing.org
This is against u-boot-testing libfdt/Makefile | 2 +- libfdt/fdt.c | 5 ----- libfdt/fdt_ro.c | 5 ----- libfdt/fdt_rw.c | 5 ----- libfdt/fdt_strerror.c | 5 ----- libfdt/fdt_sw.c | 5 ----- libfdt/fdt_wip.c | 5 ----- 7 files changed, 1 insertions(+), 31 deletions(-)
Joy!
It looks like the side effect of Grant's improvement just removed all of the #ifdefs I hacked into the libfdt primary .c files.
Side note: I was playing last night with the "what if" of deleting (git rm libfdt/*.[ch]) all the "standard" libfdt code, then re- importing it. Effectively rebaselining in a kerchunk rather than a patch-forward that Kumar has been doing.
The advantages would be:
- It would be clearer that u-boot/libfdt version 2007-11-X == dtc/
libfdt version 2007-11-X and then we could more easily track changes forward.
- The removal patch is less than 40K (email list limit), but the re-
import patch is around 70K (IIRC) so that doesn't help.
My current conclusion is that it isn't worth the effort.
- The re-import still has a size vs. email list problem
- Since some libfdt routines changed, other (client) routines must
be changed at the same time as the re-import (Kumar's patch rolls libfdt forward _and_ fixes the handful of calls, so it doesn't have a "broken spot" in between patches).
I've just created a libfdt_testing branch that is based on this patch + u-boot-testing.
If you do a diff between dtc/libfdt and u-boot/libfdt we now see no differences.
- k