aufs2-0_p20110120 appears to compile, but fails to insert/modpobe. I am using gentoo-sources, it compiled and loaded before. It's important enough to me that I'd fix it my self even if it meant learning how; unfortunately I would need a pointer to the right resources to learn how. Reproducible: Always Steps to Reproduce: 1. Compile 2.6.37 2. Run 2.6.37 3. Try to compile aufs Actual Results: Right at the end of the compile: WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol fsnotify_put_mark WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol __lookup_one_len WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol file_sb_list_del WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol do_splice_to WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol fsnotify_alloc_group WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol inode_lock WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol fsnotify_init_mark WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol fsnotify_destroy_mark WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol lookup_hash WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol fsnotify_put_group WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol cap_file_mmap WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol do_splice_from WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol do_truncate WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol deny_write_access WARNING: //lib/modules/2.6.37-gentoo/misc/aufs.ko needs unknown symbol fsnotify_add_mark [ ok ] For me aufs is very useful, just as COW style block devices with LVM are useful. However I am unsure if it qualifies as a 'major feature'.
Reassigning to maintainers.
You need to compile and boot the kernel with the aufs patches applied.
My mistake; should I make a feature request based on the answer to this question: My main source of the confusion is that in every other case I've ever seen modules, even out of tree video drivers, can be loaded without recompiling the kernel. If a kernel recompile is necessary then why is this being built as a module?
(In reply to comment #3) > My mistake; should I make a feature request based on the answer to this > question: > > My main source of the confusion is that in every other case I've ever seen > modules, even out of tree video drivers, can be loaded without recompiling the > kernel. If a kernel recompile is necessary then why is this being built as a > module? > because there are some other source files patched, which aren't build into the module, but into other parts of the kernel.
Yes, that much is obvious when you watch the files that are being patched (this scrolls by far too quickly when months ago it had been setup with kernel-patch use flag); the result is a compiled module that seems to fail. This breaks the standard conventions one thinks of when they think modules. There is a bit that is a module, and a part that isn't that module. So why make the module if part /HAS/ to be compiled in to the kernel? Why not instead apply the patch and ask the user to rebuild the kernel. (I saw no changes to the .config or config.gz btw; there is no easy way of testing if this is part of a kernel.)
You are right with this. I will check and include a more verbose warning about the need of a kernel rebuild and reload. Additionally there will be a aufs2 patched gentoo sources in future. But this needs another 2 month until I have time for this.
+ 27 Feb 2011; Justin Lecher <jlec@gentoo.org> aufs2-0_p20100809.ebuild, + aufs2-0_p20101122.ebuild, aufs2-0_p20110120.ebuild, metadata.xml: + Increase Warning level for rebuilding kernel with patches, #356267; Changed + to me as the primary maintainer +