On Tue, May 6, 2008 at 5:15 AM, Carl-Daniel Hailfinger
Indeed. Since we don't use the struct device passed in to ide_init, why don't we give Geode IDE its own struct device_operations and store the IDE config there?
That's a great question. The reason is that the cs5536 is one chip. On v2, we treated it as more than one. For my goal of easing non-experts into the code, I decided to try making it one chip and one dts. This is an experiment that may, in the end, prove wrong.
I still think one dts per chip is the right way to go, but do we need more than one level of dts? I really think this all should be in one file, it got much simpler that way.
I have problems parsing the sentence above: "... I don't think it was ever IDE."
I should not write when that tired.
Let's try again.
Works on 2.6.24. hangs on 2.6.25, I thought for a second it was NOT ide, but it is clear that 2.6.25 is hanging on probing ide1 -- or it sure looks that way. And, again, the post code of 00 is weird; I keep thinking there are vsa issues here. It just doesn't act like Linux hangs I see on other boards.
You know my preference for killing dev_find_pci_device, but that can come later if it is too difficult.
I have no disagreement as long as, in so doing, we reduce code overall and increase comprehensability.
Acked-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net
many thanks!
ron