Re: tsleep/wakeup patch #2 (Re: tsleep/wakeup)

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

:How does/will tsleep integrate with LWKT? Will it end up being included
:in only for compatiblity with older designs, or will we eventually have
:something like tsleep_async that will let it participate in port
:servicing threads?
:
:Thanks,
:-Eric

    tsleep/wakeup is actually quite a nice design for SMP, even if 
    the original planners didn't have that in mind.  Since we pass a
    cookie to tsleep/wakeup we can simply distribute that to candidate 
    cpus and not have to worry about structural pointers getting ripped out
    from under us. 

    I doubt we will ever have an async tsleep.  I actually put one into
    FreeBSD several years ago, intended to deal with certain types of
    interlock races, but never found a good use for it.

					-Matt
					Matthew Dillon 
					<dillon@xxxxxxxxxxxxx>