Hi Tom,
On 13 September 2016 at 07:16, Tom Rini trini@konsulko.com wrote:
On Mon, Sep 12, 2016 at 11:20:00PM -0600, Simon Glass wrote:
Despite the availability of Kconfig, the number of ad-hoc CONFIG options in U-Boot is still at over 8000.
In February 2015 (commit 741e58e0) there were around 8387 ad-hoc CONFIGs. As of this patch there are about 8336, only a little fewer.
One problem is that new ones are still being added, admittedly at a slower rate. This series adds a Makefile check to detect that and produce a build error. This provides immediate feedback that new CONFIG options should go in Kconfig.
Changes in v2:
- Rebase to mainline
- Fix scripts so that there are no errors
- Add the 'build-whitelist.sh' script to the tree
Simon Glass (2): Kconfig: Add a whitelist of ad-hoc CONFIG options Makefile: Give a build error if ad-hoc CONFIG options are added
OK, this is still a little fragile. When I do: ./tools/buildman/buildman -b master --force-build --step 0 \ -SCdvel 'avr32|nds32|bfin|m68k|microblaze|or32|sh4|sparc|mips|mips64|xtensa'
vs kconfig-working2 I'm seeing a bunch of failures: +(M5475GFE) make[1]: *** [no_new_adhoc_configs_check] Error 1 +(M5475GFE) make: *** [sub-make] Error 2
Yes it looks like I was testing it incorrectly, relying on running on from the SPL series. One problem is that it misses configs with lower case which are defined in Kconfig and then used. I've updated u-boot-dm/kconfig-working2 to deal with that.
I'll take another look at this and have an update out today. But we can decouple this from the painful SPL series if you like, since in principle we can run the generation script to create the first of the two patches.
Regards, Simon