Re: nano: On Oldworld PowerPC, nano freezes machine when help is being displayed
> There is but it's not a real NMI, it is masked by cli().
That qualifies as 'not a NMI' then :-) Can we make cli() mask all but that
int (is there a concept like IPL in the PowerPC programming model)?
> I've been thinking about implementing a soft-disable mecanism based
> on the OpenPIC interrupt priority mecanism, but it's tricky and
> would slow things down significantly. Can still be a useful debugging
> mecanism though as it would allow breaking via NMI on cli() lockups
> and spin deadlocks.
Please do; it's enough to have that a compile or kernel startup option.