[PATCH] pcmcia: don't sleep when you aren't tired

[Dominik Brodowski]

Hi Russell,

On Sun, Jul 06, 2003 at 11:42:22PM +0100, Russell King wrote:
> On Fri, Jun 20, 2003 at 11:43:56PM +0200, Dominik Brodowski wrote:
> > There seems to be no reason why ds.c should wait for 100ms before it
> > unregisters the drivers etc. Also, check whether it's safe to sleep
> > here...
> 
> I'm not so sure we want to make this change just yet.  Eventually, yes.
> Net devices are currently undergoing something of a overhaul, sorting
> out the lifetime of the "net_device" object.  I think this work needs
> to be completed before we think about removing this timer.

I agree completely. At first I thought it's because this migh_sleep(), but
in fact unloading of devices is quite borken..

> This is all part of a crude attempt at sequencing events to make stuff
> occur in the right order.  ie:
> 
> - unplug device
> - start 50ms timer to remove processing in card drivers
>   (crude attempt to wait for card driver interrupt handlers to return.)
> - start 100ms timer to delay notification to cardmgr
> - card driver removes all references
> - (50ms wait caused by difference in timers)
> - card removal notification hits user space
> 
> Other than that, its something we should do.

Yes, but as you pointed out, not at this time.

	Dominik