On Aug 17, 2007, at 6:53 PM, Paul Mackerras wrote: > Kumar Gala writes: > >> Its a regression in that the ULI M1575 doesn't really work properly >> on some of these boards in certain situations. > > Well, "it doesn't work" isn't of itself a regression, a regression is > "it used to work and now it doesn't".
Interesting, maybe I was grouping both of these together as things that could go in after the -rc1. I was under the impression that bug fixes could also go in post -rc1. > In any case the lateness x largeness product (LxL) of that patch is a > bit too high. Is there a simpler way to fix it that doesn't involve > moving all that code around? I'm not really sure how to accomplish that. I could possibly duplicate the 'fixes' in each board port but that seems pointless and I doubt would reduce the code modifications amount any. Most of this has to due with the kernel / firmware / device tree interface. If we have a new firmware that modifies the memory map and requires changes to the device tree and kernel is that a fix? I understand the concern related to the LxL. I'd also say the set of users this effects is very small since its just the two boards. I leave it up to you if you want to grab this patch or not for 2.6.23. If not please pull all these into your for-2.6.24 branch. - k _______________________________________________ Linuxppc-dev mailing list [email protected] https://ozlabs.org/mailman/listinfo/linuxppc-dev
