On Tue, Apr 19, 2011 at 2:52 PM, Christopher Li <sparse@xxxxxxxxxxx> wrote:
>
> I am suppressed that you can get 3-4% from just comparing the stream names.
I was too. But the kernel sources have gotten more complicated, and
"input_stream_number" can get very high. Having 200 streams is common,
and 400 streams is not unusual for some more complex kernel files that
include a lot of files.
So what happened was that a single '#include" at the end of such a
situation would compare the pathname against hundreds of longish
strings. So it ends up O(n**2) in number of #includes - each strcmp is
cheap, but there's a _lot_ of them.
I don't remember these kinds of numbers from years ago when I did more
profiling, but I suspect the kernel tree has gotten way more
include-happy too.
Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
