Are you suggesting that each subsystem define its own default log level?
That sounds truly horrible. It would mean that when reading code or
adding new messages you'd always have to check what the default loglevel
is for a particular subsystem.
I really like the idea of being able to easily increase the verbosity for
a particular subsystem, but please make sure that by default all messages
with the same level are treated the same, irrespective of which subsystem
they belong to.
So IMO
printk(KERN_WARNING "APIC calibration not consistent "
and
printk(KERN_WARNING KERN_APIC "APIC calibration not consistent "
should do exactly the same thing if no 'loglevel=apic:X' is used and that
should be guaranteed for all subsystems.
Cheers,
FJP
--