On 11.10.2008 07:18, Peter Stuge wrote: > Carl-Daniel Hailfinger wrote: > >> That XIP stuff needs a good explanation anyway. We may want to have >> it cover the boot block only or the bootblock and initram. Either >> way, reading the whole ROM to cache it may be less than useful and >> a severe performance killer early on. >> > > XIP could never perform better than fetching from cache except if the > cache is broken. I would like XIP on/off to be an expert option in > Kconfig unless we can reliably detect cache failure at runtime and > decide on XIP or not from that. >
We have to differentiate between two issues here: Marking the ROM as cached (good) and preemptively reading all of it to have it in cache (debatable). > On the other hand, too much logic this early on is not appealing. > I try to keep the complexity low. More patches will be forthcoming in due time. Regards, Carl-Daniel -- http://www.hailfinger.org/ -- coreboot mailing list: [email protected] http://www.coreboot.org/mailman/listinfo/coreboot
