svn commit: r230408 - head/sys/net80211

[Adrian Chadd]

On 20 January 2012 18:41, Rui Paulo <rpaulo at freebsd.org> wrote:


> This commit doesn't help much and keeps us more toolchain dependent like
> Alexander pointed out in IRC. Since ether_sprintf() wasn't causing any
> trouble, please consider backing this out.
>
> This commit also introduces unnecessary changes which make reviewing
> harder.
>

Hi,

I'd like to eventually move away from doing it this way. The trouble is
using ether_sprintf() multiple times (and even from multiple threads?) in
the same debug statement results in non-useful behaviour.

What I'm likely going to do is introduce an ether_ssprintf() which takes a
char/len buffer to write into. That way the debug statements can just use
that (with appropriate wrappers around them to only call ether_ssprintf()
if the debug condition is met, otherwise unused sprintf()s will occur.

I'm open to other suggestions though!

Adrian