Upgraded today to ~amd64 gentoo-sources-2.6.30, attempted to rebuild openafs-kernel-1.4.9, failed. Reproducible: Always Steps to Reproduce: 1. Build and boot gentoo-sources-2.6.30 (probably same with vanilla, too) 2. Attempt to rebuild openafs-kernel-1.4.9 3. Compilation fails. Actual Results: Compilation fails : CC [M] /var/tmp/portage/net-fs/openafs-kernel-1.4.9/work/openafs-1.4.9/src/libafs/MODLOAD-2.6.30-gentoo-MP/osi_module.o In file included from /var/tmp/portage/net-fs/openafs-kernel-1.4.9/work/openafs-1.4.9/src/libafs/MODLOAD-2.6.30-gentoo-MP/osi_module.c:42: include/linux/seq_file.h:48: warning: 'printk' is an unrecognized format function type /var/tmp/portage/net-fs/openafs-kernel-1.4.9/work/openafs-1.4.9/src/libafs/MODLOAD-2.6.30-gentoo-MP/osi_module.c: In function 'afsproc_init': /var/tmp/portage/net-fs/openafs-kernel-1.4.9/work/openafs-1.4.9/src/libafs/MODLOAD-2.6.30-gentoo-MP/osi_module.c:254: error: 'struct proc_dir_entry' has no member named 'owner' make[6]: *** [/var/tmp/portage/net-fs/openafs-kernel-1.4.9/work/openafs-1.4.9/src/libafs/MODLOAD-2.6.30-gentoo-MP/osi_module.o] Error 1 Expected Results: Openafs-kernel should have compiled and installed, /etc/init.d/openafs start should have run, klog, and back on my way with the new kernel. A quick peek with google indicates that openafs-1.4.10 has been tweaked to work with 2.6.30, but there is no Gentoo ebuild yet available for it. I'm out of time at the moment, buy may try and scare this together on my own.
It looks like I forgot to mention this, but net-fs/openafs-1.4.10 has been in the tree for some time now. Builds and seems to work with gentoo-sources-2.6.30-r4. Not marking as fixed until it makes it to the stable tree.
*** Bug 280503 has been marked as a duplicate of this bug. ***
FYI: net-fs/openafs-1.4.10 works for me on amd64 under 2.6.30-gentoo-r4
(In reply to comment #3) > FYI: net-fs/openafs-1.4.10 works for me on amd64 under 2.6.30-gentoo-r4 > Ditto. Do we know why 1.4.10 is still marked unstable? It's worked quite well for me for the last few weeks.
(In reply to comment #4) > (In reply to comment #3) > > FYI: net-fs/openafs-1.4.10 works for me on amd64 under 2.6.30-gentoo-r4 > > > > Ditto. Do we know why 1.4.10 is still marked unstable? It's worked quite well > for me for the last few weeks. > For what it's worth, openafs-1.4.10 has been working well for me, too. BUT... Openafs-1.4.10 won't build with kernel-2.6.31, you need openafs-1.4.11 for that. BUT... The pam_afs.so.1 that comes with openafs-1.4.11 is broken, so you need the old copy from openafs-1.4.10 for that.
I recall a mechanism used in the ivtv ebuilds awhile back (and it may still be the case) that different series of the ebuilds (0.6.x, 0.7.x, etc.) required different kernel series, and the build would error out with a specific warning if you tried to mix'n'match ivtv drivers and kernels. Can we add something similar to this, where: - 1.4.9 is required for kernels <=2.6.29 (and is kept stable) - 1.4.10 is required for kernel 2.6.30 (and is marked stable on x86/amd64) - 1.4.11 is required for kernels >=2.6.31 (and is still marked unstable due to crashes) Would that be a feasible solution? It would require minimal patching to the existing ebuilds, and would ensure users get a nice error message when they mismatch openafs to kernel rather than some obscure build problem. Finally, it would bring openafs-1.4.x stable status up to date with the latest gentoo-sources (as 2.6.30-r4 is stable, but 2.6.31-* is not).
(In reply to comment #4) > Ditto. Do we know why 1.4.10 is still marked unstable? It's worked quite well > for me for the last few weeks. I'd like to refer you to bug 282196 for this.
*** Bug 289582 has been marked as a duplicate of this bug. ***
openafs-kernel-1.4.12.1 should fix things