svn commit: r326218 - head/sys/kern

[John Baldwin]

On Monday, November 27, 2017 02:04:59 PM Nathan Whitehorn wrote:
> Unfortunately, it's unfixable on ppc64. Apologies for breaking dtrace! 
> Would you like me to remove the KASSERT() here? I'm happy to do that in 
> a few hours (unless you beat me to it first) -- although I do think that 
> explicitly checking for CPU_ABSENT is a much better behavior in client 
> code than checking the return value of pcpu_find().

It sounds like the dtrace change is actually fine as-is.  At the very least
fix the style of the kassert for now I guess.  In general though I think our
kernel basically depends on NULL faulting and in the past I recall Bruce has
usually advocated for depending on the page fault from NULL rather than
adding explicit assertions just for 'foo != NULL'.

>From some discussion on IRC after this e-mail though it seems that NULL
might not be entirely unfixable on powerpc64?

-- 
John Baldwin