On Mon, Feb 16, 2009 at 09:24:11AM +0900, YAMAMOTO Takashi wrote: > > I'm inclined to think a big part of it is that the buffer cache tries > > (to support LFS) to be virtually indexed and never physically indexed; > > that is, all buffers belong to vnodes and file offsets rather than > > devices and device offsets. The problem is that this doesn't work for and UDF yes > > I think the buffer cache needs to be restructured so it can be either > > virtually or physically indexed. This is going to be a big hassle. i think the easiest way is to make the buffer cache virtually indexed only since its file granularity that you want to cache on. It does not make sense to cache a large file when its contents is already deleted. A Physical block cache needs to exist in the discs device driver, not in a FS. The transition to virtual indexed cache is almost trivial; FFS stores its indirect blocks as *negative* indexes already f.e. (yuck, but what to do about it now). > actually filesystems can use any kind of numbers as buffer cache index > for their own vnodes. Indeed, see FFS > i don't understand how you think it's related to the bug, tho. I didn't follow the start of the conversation, so no comment here. With regards, Reinoud
Attachment:
pgpWzb2WhXXQW.pgp
Description: PGP signature