Point out that 2.4 kernel headers should probably be used even if a 2.2 kernel is going to be used

[markh@…]

Point out that 2.4 kernel headers should probably be used even if a 2.2 kernel is going to be used.

I've just been speaking to mjt_ on #lfs (well, a private conversation) and he ran into problems with util-linux because 2.2 doesn't seem to define RAW_MAJOR in linux/major.h

I think we should point out that it's best to build everything against the newest kernel headers even if you intend to use an older kernel. (For the record, I advised him to wipe the chapter 6 stuff and re-do it using 2.4 headers; there should've been no need to redo ch5 however)

This may need to go to lfs-dev to check that I'm correct first but I think it's a non-obvious point which may need adding (and not just for 'newbies' - as I know we're taking a lot of handholding out of the book).

[scot@…]

OK tell me if I am wrong, but won't this open up other problems. I.e. software trying to utilize kernel functions that may not exist? I have forgotten how, but there are two ways to deal with the RAW devices. One is to disable it in util-linux, the other is to use one of the patches that adds it to 2.2 kernels.

IMO, since using older kernel is a deviation from the book, I don't think we should be adjusting the book. Otherwise we may be looking at the book "supporting" other deviations as well.

I wouldn't post this here but in lfs.book/dev, but I think this isn't a good idea for both reasons.

Scot

[scot@…]

Sorry, I meant "I would have posted this comment on lfs.dev/book, but I thouught it important enough to post it here.

[gerard@…]

I agree with Scot on this one. Best to just assume people will use a recent kernel. I figure if they have a need to use an old kernel, they should also be prepared to deal with how to install an older kernel and make it work properly.