svn commit: r294327 - in head/sys: dev/cxgb dev/cxgbe dev/e1000 dev/hyperv/netvsc dev/ixgbe dev/mxge netinet sys

[Pedro Giffuni]

On 02/11/16 11:47, Warner Losh wrote:
>
>
> On Wed, Feb 10, 2016 at 8:08 AM, Pedro Giffuni <pfg at freebsd.org
...

>
>
>     FWIW, our libc qsort() has an additional enhancement:
>
>     http://svnweb.freebsd.org/base?view=revision&revision=279663
>
>     In my measurements qsort(3) was now always faster than mergesort(3).
>
>
> If it is faster, is there any good reason to maintain both qsort and
> mergesort in the kernel then?
>

I don't know, that happened ages ago, way before I was around ;).

The late enhancement is only in the libc qsort and the code is likely to 
be different in other ways (is it recursive? .. haven't looked).

I think I saw some comments somewhere about degenerate cases that
affect our version of qsort, but the performance is known to be
very good[1].

Pedro.

[1] http://calmerthanyouare.org/2013/05/31/qsort-shootout.html