Jim Gifford wrote:
I don't see any of these references in sys/types.h, or am I missing something.
Glibc-2.3.6. This may not be an issue at all, see below. Will let you know tomorrow if I have the time to complete xorg.
----------------------- Greg Schafer wrote:
I'm not sure what you're trying say in this post. It would help if you specified the actual problem you are having.
Sorry for being so vague. I thought the code posted would be sufficient to see the problem (or lack of a problem? see below).
Oh, it's the "Xorg lnx_agp.c fails to compile" issue? If so, then you're off the mark IMHO. Xorg fails to compile due to a Glibc bug with sys/kd.h, now fixed in Glibc-2.4. In other words, Xorg compiles successfully against Glibc-2.4. No seds to lnx_agp.c or other hacks required.
The errors _were_ in the same general area, unfortunately they were different. xorg-server-1.0.2 includes the workaround (include linux/types.h first). I'm probably the only person building against the new headers with glibc-2.3.6. Anyhow, I got redefined fd_set, dev_t and others in sys/{stat,types}.h with previous in linux/types.h. These errors are consistent with the comments in the LLH version of linux/types.h.
I reversed my changes to /usr/include/linux/types.h and I am now unable to reproduce the error in chroot. I must assume user error for now. I thought I had it all figured out so I smoked the build logs. I'll be rebuilding from scratch again very shortly, well now actually because of a problem with xauth in the new setup (I want a completely clean build from scratch before updates go into BLFS). I'll give it one more shot with the new headers. If the errors show again, I'll post the output for furthur review and revert to llh with glibc-2.3.6. Still, even with the little problems I've encountered, the new headers are coming along very nicely.
Sorry for the noise. Thanks for taking a look. Oh, I also forgot about the original problem with sys/kd.h. Thanks for the reminder. Should this be fixed in LFS if continuing with glibc-2.3.6? Xorg is the only known (to me) issue and it's worked around in BLFS.
-- DJ Lucas -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
