kldunload DIAGNOSTIC idea...

[Poul-Henning Kamp]

In message <20040721.111045.88466992.imp at bsdimp.com>, "M. Warner Losh" writes:

>We still ahve to wait for all threads with references to softc to
>drain out before we unload the module.  I see a small benefit to that,
>but it seems that the requirements for softc are stronger than ifnet
>and already cover it.  The detach routine of the driver could easily
>wait for the ifnet reference count to drop before returning for
>network drivers (heck the if_detach call could wait all references are
>gone, so that it isn't actually in the driver).  Where it is stored
>seems not to matter so much.

I have not studied the ifnet scenario in details, so I won't argue
about it.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.