Re: [RFC net-next 0/2] net: Use net_<level>_ratelimit

From: Johannes Berg <redacted>
Date: Tue, 15 May 2012 20:29:08 +0200

On Tue, 2012-05-15 at 14:24 -0400, David Miller wrote:

quoted

quoted

But don't you have to do something in the sources to actually get
dynamic_debug enabled? I'm concerned this will make mac80211 debugging
inconsistent - the normal bits are just printk() still, and the
rate-limited bits need some special enabling? That seems odd.

You can turn them all off or on with a simple flip of a boolean switch
at run-time.  Or, alternatively, you can turn them on or off in a
finer grained manner with other run-time facilities.

There should be a transition away from explicit KERN_DEBUG.

I support that, but I think it's confusing to have things mixed. Also,
we seem to have printk_ratelimit(), so I'd prefer to have that used in
mac80211 instead until everything there moves over to newer facilities.

No, you don't want that, it's different from the networking ratelimit
mechanism.  Joe explained this, and I just explaing it to you once
again in another reply.