Re: [PATCH] sky2: Use deferrable timer for watchdog

Parag Warudkar wrote:

On Dec 20, 2007 2:22 PM, Kok, Auke [off-list ref] wrote:

quoted

ok, that's just bad and if there's no user-defineable limit to the deferral I
definately don't like this change.

Can I safely assume that any irq will cause all deferred timers to run?

I think even other causes for wakeup like process related ones will
cause the CPU to go busy and run the timers.
This, coupled with the fact that no one is yet able to reach 0 wakeups
per second makes it pretty unlikely that deferrable timers will be
deferred indefinitely.

0.8 is easy on single core today.
multicore just increases how idle you can be for a given core.

quoted

If this is the case then for e1000 this patch is still OK since the watchdog needs
to run (1) after a link up/down interrupt or (2) to update statistics. Those
statistics won't increase if there is no traffic of course...

I think it is reasonable for Network driver watchdogs to use a
deferrable timer - if the machine is 100% IDLE there is no one needing
the network to be up. If there is something running even on the other
CPU - that is going to cause an IPI, reschedule, TLB invalidation etc.
which will make it very likely in practice that each CPU will be
interrupted in reasonable amount of time.

this is not correct; many machines are idle waiting for network data. Think of webservers...

Of course there are theoretical cases where we could land into a
situation where a CPU in a multiprocessor machine is IDLE infinitely
and that causes the watchdog that happens to be bound to run on the
same CPU to not run. To take care of these unlikely cases I think the
timer mechanism should have a reasonable limit on how long a CPU can
go IDLE if there are deferrable timers.

how about something else instead: a timer mechanism that takes a range instead..
that at least has defined semantics; the deferrable semantics really are "indefinite".
Lets keep at least the semantics clear and clean.