On Wed, 3 May 2000, Jason R Thorpe wrote:

> On Wed, May 03, 2000 at 08:49:31AM -0700, Eduardo Horvath wrote:
> 
>  > mi_switch() does not do any spl() stuff so it's not blocking
>  > interrupts.  And it lools like you took the interrupt at the very
>  > beginning of mi_switch() so cpu_switch() isn't the culprit either.
> 
> Actually, I'm pretty sure mi_switch() is supposed to be called at
> splstatclock().

Yes, I know.  But it never lowers the IPL so if an interrupt is taken
there it's not because it had interrupts disabled too long and then
dropped the IPL.  Unless there's something strange on the '060 and IPL
transitions don't take effect immediately but are delayed for some number
of instructions.

Eduardo Horvath