On Sat, Oct 2, 2010 at 12:27 PM, Warren Turkal wt@penguintechs.org wrote:
I think that making arguments against this code on grounds that it doesn't feel like other common code is a little shortsighted. When bugs are fixed in one bit of code, we should be able to take advantage of those in other parts of code. For instance, I just had to copy/paste some CPP logic to fix the via/vt8454c CAR init due to a tiny bootblock. That would have already been fixed had the via and amd logic been unified since they are basically identical bits of code.
well, you have to be careful. We've had more than one problem with "common" code that was not so common. CAR is so tricky, and one platform changes may break another platform in ways we don't expect. CAR, in fact, scares me. It's fragile. I'm particularly concerned about the idea of unifying via and amd code. The changes some well-intentioned people made years ago to some via northbridge code to "clean it up" -- it broke many things -- are still fresh in my mind.
I contribute very little code at this point, really zero in the last year, so I don't believe my opinions should be taken that seriously :-) I'm really more than anything recommending you all to be very careful. Anything you do to CAR needs to be extensively tested.
So let's call it my .01, not my .02 :-)
ron