[systemd-devel] Using the hwdb in external programs

[Kay Sievers]

On Wed, Aug 14, 2013 at 1:06 PM, Richard Hughes <hughsient at gmail.com> wrote:
> On 14 August 2013 11:41, Kay Sievers <kay at vrfy.org> wrote:
>> An example is here:
>>   http://cgit.freedesktop.org/systemd/systemd/tree/src/test/test-libudev.c#n432
>
> Great! Thanks. So I know what to use where, what's the cost of those
> functions?

It's all in a binary on-disk patricia trie, and all strings are
de-duplicated. The lookup cost is almost "free", you should not notice
it.

> For instance, is udev_hwdb_new() going to use globs of
> memory until udev_hwdb_unref() is called,

The entire binary database file is mmap()ed into the address space as
long as the context exists, no interesting amount of memory is
allocated in the calling process or during lookup.

> or are any of the functions
> especially CPU intensive?

The time is spends searching is mostly dependent on the length of the
lookup string, and the amount of globs in the database that have
common prefixes. None of that should really matter, you can ignore any
such issues unless you plan to call it 1000s of time in an inner loop.

> I can profile if required, although I'm
> hoping the answer is "very little; very little" :)

Unless something is wrong, it should be even less than little. :)

Kay