On Fri, Sep 09, 2016 at 10:09:56AM +0900, Masahiro Yamada wrote:
2016-09-09 5:17 GMT+09:00 Tom Rini trini@konsulko.com:
On Thu, Sep 08, 2016 at 02:13:21PM -0400, Tom Rini wrote:
On Thu, Sep 08, 2016 at 09:58:13AM -0600, Stephen Warren wrote:
On 09/07/2016 07:29 PM, Masahiro Yamada wrote:
Hi Stephen
2016-09-08 1:15 GMT+09:00 Stephen Warren swarren@wwwdotorg.org:
Masahiro,
In patch 6e7e9294d321 "usb: add basic USB configs in Kconfig", you added "config USB_STORAGE" to drivers/usb/Kconfig. However, it's still just #defined by many include/configs/*.h rather than being defined in configs/*_defconfig. Is that a problem? It seems to work in practice, but leads people adding new boards to put the definition in configs/*_defconfig which then may be inconsistent with similar existing boards which have it defined in include/configs/*.h.
Once we create an entry in Kconfig, all the defines in include/configs/*.h should be moved.
That's what I imagined. The commit above didn't do that though; are you planning on sending a fix?
Since those are often conflict-happy I'll take care of it. I probably intended to, even.
So, yay for moveconfig.suspicious being a thing now. A number of boards set CONFIG_USB_STORAGE without CONFIG_USB=y in their defconfig file, so they lost CONFIG_USB_STORAGE in migration. Confirming that yes, really, nothing else is added or dropped when we set CONFIG_USB=y in the defconfig files as well (unit testing was good, testing the world now).
I thought so.
A little bit manual intervention is needed for those suspicious boards. (add CONFIGs by your own script and re-sync)
I hope this is not too painful...
Nope, it was rather smooth and got me to notice another problem, that is a little harder to solve "cleanly" until we have Simon's SPL series in.
Thanks for taking care of this!
Thanks for making the tools!