On Tue, May 30, 2000 at 11:49:12AM -0400, John Kohl wrote:

 > Bill> Hmm.  Can we do a splhigh() before suspend, splx() on resume, just so
 > Bill> an interrupt doesn't sneak in before the power hooks have a chance to
 > Bill> run?
 > 
 > Alas, some APM BIOSes are lame and don't work right unless interrupts
 > are enabled when the BIOS is called.  So splhigh() (== cli) isn't going
 > to work.  We'd need to use whatever splfoo() blocks all interrupt
 > processing, but not interrupt delivery (splstatclock()? splclock()?).

No, splhigh() does NOT cli.  It merely marks all interrupts as `masked'
in a kernel data strcture.

-- 
        -- Jason R. Thorpe <thorpej@zembu.com>