mmaped fdi cache

[Richard Hughes]

strings /var/cache/hald/fdi-cache | grep power_management.quirk.vbe_post | wc -l
141

du -h /home/hughsie/.root/var/cache/hald/fdi-cache
120K    /home/hughsie/.root/var/cache/hald/fdi-cache

Now I'm guessing that we can reduce the size of that cache by 90% just
by compiling a string hash and then just saving the previous offset of
the same string. This has big savings for embedded at no decoding
speed cost.

Anyone want to have a go at this?

Richard.