recent changes in Giant usage -> panic

[Poul-Henning Kamp]

In message <20040315012321.X1848 at gamplex.bde.org>, Bruce Evans writes:

>vgone() has interesting locking problems.  It seems to be a fundamentally
>broken interface that only works if the kernel is not premptible or
>possibly if everything related to vgone() is locked by Giant.  Apparently
>the lower levels know that Giant locking is needed.

That's a fair way of putting it.  Quite a bit of the brokenness comes
from the revoke systemcall.

-- 
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.