If we don't have --gc-sections support, linking kmod fails:
libkmod/.libs/libkmod-util.a(libkmod-util.o): In function 'underscores':
libkmod/libkmod-util.c:117: undefined reference to 'kmod_log'

This is because libkmod-util.la uses kmod_log(), that is in libkmod.la.
Move the function so we don't have a dependency loop while building the
libraries and it works with compilers with no support for --gc-sections.

The test uses the ext4 module dependencies as the testcase, checking
both the number and the names of the returned modules.

Add a modprobe.conf with some blacklist entries in a test rootfs, and
then ensure our blacklist function actually cuts out the two listed
entries (and doesn't cut out the others).

We can't use the rootfs directory because it breaks out-of-tree build
and in future we want to make modifications to the fake filesystem such
as adding and removing files.

We need to call "chmod -R +w" in the resulting directory because when we
distribute the source with make dist all files will be readonly.

Fix 'make distcheck'

Fix 'make dist'

The current configuration is dumb in any number of ways:
1) If the rationale was for space savings, it works the opposite- the
   git repo gets more bloated because we are adding binary compressed
   blobs that share little in common with their parent, and anyone that
   wants to run the test suite has to unzip it anyway.
2) It is a pain in the butt to add new tests, and not accidentally lose
   any new rootfs you built in the directory.
3) `git status` won't help you if you are tweaking files in the rootfs
   and don't know they have been changed, or if some test did that and
   you couldn't detect it.
4) `git log` won't help you find out what is changing in the rootfs test
   directory itself when changes are made to the binary blob, such as
   new files being added or even existing files being tweaked.
5) The files just aren't that big anyway- 2.7MB unzipped.

There's no reason kmod_log should be exported, remove it from linker
script. This doesn't break the API/ABI because we are luck: since the
function had visibility=hidden it was not getting exported as a global
symbol.

This reverts commit 88a170dbd63965b8403b6b40b12ded4f82d89ccd.

There's no reason for users of the API to call this method, it's just
wrong to export it.

The bug that this patch fixed needs to be fixed another way, not
exporting this function.

Previously, depmod would relegate failures of kmod_module_get_symbols()
to debug output, assuming the "error" was simply a lack of symbols.
Leave the ENOENT return to debug output, but report anything else as a
real error.

zlib won't necessarily set the system errno, and this is particularly
evident on corrupted data (which results in a double free). Use zlib's
gzerror to detect the failure, returning a generic EINVAL when zlib
doesn't provide us with an errno.