[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: Making powerd=YES default
On 26.07.2011 22:10, Jukka Ruohonen wrote:
> On Tue, Jul 26, 2011 at 09:37:30PM +0200, Jean-Yves Migeon wrote:
> Making a mountain out of a mole hill, but nevertheless...
> See, also sysmon_envsys(9) uses this internally. If you want to do the
> dubious and error-prone task of enumerating all powerd(8) users, you have to
> go through all kernel configuration files on all architectures and all (MD
> or MI) device drivers or other components that may use either sysmon_envsys(9)
> or sysmon_pswitch(9). Then you have to maintain it.
> I would say that the kernel does not need this kind of useless cruft.
> Rather, we should make good default choices even if there is m68k.
>> Granted, the current registering operation is a NOP; but I think these
>> functions were introduced for one good reason, and here it is: if you
>> don't have any sort of event registered when powerd(8) starts, you
>> shouldn't expect it to handle some in a "stateless" manner later.
> I think the actual reason is: sysmon(4) was never really finished.
>> I can change the semantic of these operations and force them to
>> reference a power backend. In that case, that will force all sysmon
>> callers to:
>> - first register a power backend
>> - then register different sort of events (like pswitch).
> Please don't.
Then this leaves the sysinst route.
I can foretell a good amount of questions for NetBSD-6 and up, where
people upgrading straight from already installed systems will come in
and ask why xm/xl is dead frozen in dom0 for a save (because powerd(8)
wasn't started beforehand and only newly installed systems will have it
enabled by default).
Note that once you issue a save operation, you cannot really undo it at
this step. The domain will be marked in a "migrating-<dom-name>" state.
So if you issue a save and forgot to start powerd(8), you will have to
run powerd/scripts/sleep_button manually in domU.
Main Index |
Thread Index |