svn commit: r216134 - in head: share/man/man9 sys/amd64/include sys/arm/include sys/i386/include sys/ia64/include sys/mips/include sys/pc98/include sys/powerpc/include sys/sparc64/include sys/sun4v...

[Bruce Evans]

On Fri, 3 Dec 2010, Bruce Cran wrote:

> On Fri, 3 Dec 2010 20:45:12 +1100 (EST)
> Bruce Evans <brde at optusnet.com.au> wrote:
>
>> KASSERT() in little inline functions gives a lot of bloat for such an
>> unlikely error.  Stupid callers can still pass any garbage count
>> except 0.
>
> Yes, this catches a specific case that hps raised a few years ago:
> sending zero-length packets/frames would fail by causing the system to
> hang. Should we just document the restriction in the man page and not
> try and prevent it at runtime?

That is enough for me, and hps should be the last person to write this
bug :-).  If zero lengths can be generated at runtime then they should
be checked for in callers and not handled by panicing.

Bruce