[PATCH v2] Improve performance of strstr

[Wilco Dijkstra]

Hi Rich,

> This is a major performance regression and should not be committed.

Exactly which case regresses? I'd love to hear about it since I can't find
any.

> The test case does not cover anything near the worst cases. With a
> limit more like 16 bytes it might be reasonable but 256 is plenty
> large to see the quadratic effects (64k is a big number).

We currently don't have a good strstr benchmark in GLIBC, which is why
I posted a patch to improve this. If you have further suggestions how to
improve it just let me know.

Cheers,
Wilco