hwdb is broken

[ryao]

udev_hwdb_get_properties_list_entry() is invoked during udevd initialization, but hwdb is not initialized. This does not appear to be used at all at the moment, so f74b773 was introduced to workaround it. However, we have two things that need to be done:

1. We need to fix/delete this code. Since it does not block development with the patch applied, we can leave it in the tree for now.
2. We need to log a warning when udev_hwdb_get_properties_list_entry() is invoked without an initialized hwdb, provided that we elect to fix this instead of deleting it.

[ryao]

It appears that hwdb was commited in a broken state to systemd at the time that we forked. There have been several patches in systemd to fix it. When we evaluate how we want to handle this, we should consider pulling them into eudev provided that we want to make this code work internally as the systemd developers intended.

[ssuominengentoo]

Just pointing out here too that sys-fs/udisks:0 relies on the internal hwdb support being initialized after applying this patch to sys-fs/udisks:0 in gentoo-x86 with a revision bump,

https://projects.archlinux.org/svntogit/packages.git/plain/trunk/drop-pci-db.patch?h=packages/udisks

And the problem is NOT introduced by this change, the problem was already existing with sys-fs/udisks:2 already relying upon it, we just didn't take notice

(I know you already know after our IRC conversation, but wanted to make this note more sticky)

[axs-gentoo]

The hwdb functions were missing from the symbol version script; they were added as of commit c18029a

Also, all functional differences between systemd-udev-196 and eudev relating to hwdb were merged as of commit 5f1d6a9 (note this reverts f74b773 in favor of what upstream did)