On Sun, May 07, 2006 at 11:08:15AM -0700, Jim Gifford wrote:
> Ken, I found the problem it was a typo in the script, try this patch
> before I comm it it. http://ftp.jg555.com/headers/headers-00.60.diff
Jim,
Two problems with this: First, including linux/mod_devicetable.h
was what caused the build to fail.
I looked at it, and didn't see anything that looked like
userspace to me. Mariusz didn't have it, and I think I checked for
other users without finding any, so I pruned it out and compiled.
I must admit, I didn't remove mod_devicetable.h, but I think
anything trying to include it is likely to choke.
Second, linux/pci_regs.h defines all the PCI registers and might be
needed in userspace. Certainly, Mariusz copied it into linux/pci.h
rather than having it as a separate file.
I don't (at the moment) have strong feelings about linux/pci_ids.h
which you are reinstating as available for inclusion, but it seemed
to me that anything likely to want these details (e.g. external
modules) would include the file directly rather than through the
inclusion of pci.h (and again, that matches Mariusz's treatment).
Just in case anybody thinks I'm writing a tribute to Mariusz's
headers, I'll point out that we were using them for some months, and
the only recent problem was with inotify. So, if his old headers
were adequate for what people build after {c,}lfs then taking the
same approach about which files to include is unlikely to harm our
users.
Ken
--
das eine Mal als Tragödie, das andere Mal als Farce
_______________________________________________
Clfs-dev mailing list
[email protected]
http://ninja.linux-phreak.biz/mailman/listinfo/clfs-dev
