Jon Loeliger wrote:
On Thu, 2008-03-20 at 10:32, Wolfgang Denk wrote:
Um... that would fix this specific situation - but we might still run into the same problem with the next attempt to convert code to the new config style.
Yeah, we just need to pick 'em off one-by-one... :-)
For example, there might be code which might be compiled depending on CONFIG_PREBOOT.
Ah, as I understand that one, I think it gets solved in a slightly different manner. The kernel introduces a "HAS" variant that indicates if the feature is enabled and then uses something like CONFIG_HAS_PREBOOT to indicate and test for its presence. The actual _value_ remains the CONFIG_PREBOOT symbol.
jdl
Hi Jon,
In a recent patch (picked up by Stefan but not in the current ToT) http://article.gmane.org/gmane.comp.boot-loaders.u-boot/38092 I added a config CONFIG_FLASH_SHOW_PROGRESS which I used to provide a countdown value as well as configure code (*not* a .o file). Based on my limited understanding and perusal of the linux Kconfig methodology, this looked like an acceptable thing.
Is this a violation of Good Design[tm], or is the "CONFIG_HAS_*" principle an additional rule that only applies if you have a separately compiled file, necessary so that the "COBJS-y +=" trick works?
Thanks for clarifying, gvb