constify struct cfattach

uninitialised value caused some lmenv(4) fans to never appear sometimes;
the bug was reported long time ago by Simon Slaytor with adm9240;
the eyes were cleared today by LLVM/Clang Static Analyser; ok deraadt

rework sensor tasks to use the kernels generic workq rather than a special
kernel thread of its own. the api has changed (which will be fixed in the
manpage shortly) so all the users of sensor tasks that i can find have
been fixed too.

noone tested, so its going in to force people to run with it.
"put it in" deraadt@

split userland & kernel struct sensor/sensordev so that the addition
of new fields in the future is less disruptive. This is done similar
to how struct proc is handled for ps(1). ok jmc (man page changes)
tested fkr simon, and more suggestions from millert

adapt to new two-level sensor api; Constantine A. Murenin

slightly nicer sensor names

Replace SENSOR_ADD() macro with a pair of functions
sensor_add()/sensor_del() so that sensors can be attached
and detached dynamicaly.

ok kettenis@ deraadt@ dlg@

spacing

interrupt clearing applies to lm87 class too (not just lm81 class)

if the device is busy trying to deliver an interrupt, it will stall on
scanning. at startup, that is dumb. clear the interrupt bit. now i get
40 sensors on the alpha... crazy

on lm81 and clones, reflag external sensor as invalid every time through

match about 15 more clone chips

if we write to the START bit, splat out a message

if start is already set, avoid setting it again

Fix typo.

read the channel register, hopefully initialized by some rom (Yes, Apple
does in their rom, but not in their kernel code). This tells us if the
2 AIN/FAN pins are wired as AIN or FAN (G4 xserve fans, G5 xserve more
voltage readings..); ok kettenis

temperature readings of 0x80 mean "diode not connected", i think; kettenis ok

Start monitoring operations such that values actually change.
ok deraadt@

avoid integer overflow in fan speed calculations, from kettenis
also, if things appear majorly wonky, mark sensors invalid

instead of passing OF nodes down to the drivers, pass name/compat string
pointers. This lets their match() functions actually make real decisions.
OF-capable machines will pass name/compat pointers, but other machines
will not. grudging ok kettenis

move lm87 to dev/i2c; ok kettenis

uninitialised value caused some lmenv(4) fans to never appear sometimes;
the bug was reported long time ago by Simon Slaytor with adm9240;
the eyes were cleared today by LLVM/Clang Static Analyser; ok deraadt

rework sensor tasks to use the kernels generic workq rather than a special
kernel thread of its own. the api has changed (which will be fixed in the
manpage shortly) so all the users of sensor tasks that i can find have
been fixed too.

noone tested, so its going in to force people to run with it.
"put it in" deraadt@

split userland & kernel struct sensor/sensordev so that the addition
of new fields in the future is less disruptive. This is done similar
to how struct proc is handled for ps(1). ok jmc (man page changes)
tested fkr simon, and more suggestions from millert

adapt to new two-level sensor api; Constantine A. Murenin

slightly nicer sensor names

Replace SENSOR_ADD() macro with a pair of functions
sensor_add()/sensor_del() so that sensors can be attached
and detached dynamicaly.

ok kettenis@ deraadt@ dlg@

spacing

interrupt clearing applies to lm87 class too (not just lm81 class)

if the device is busy trying to deliver an interrupt, it will stall on
scanning. at startup, that is dumb. clear the interrupt bit. now i get
40 sensors on the alpha... crazy

on lm81 and clones, reflag external sensor as invalid every time through

match about 15 more clone chips

if we write to the START bit, splat out a message

if start is already set, avoid setting it again

Fix typo.

read the channel register, hopefully initialized by some rom (Yes, Apple
does in their rom, but not in their kernel code). This tells us if the
2 AIN/FAN pins are wired as AIN or FAN (G4 xserve fans, G5 xserve more
voltage readings..); ok kettenis

temperature readings of 0x80 mean "diode not connected", i think; kettenis ok

Start monitoring operations such that values actually change.
ok deraadt@

avoid integer overflow in fan speed calculations, from kettenis
also, if things appear majorly wonky, mark sensors invalid

instead of passing OF nodes down to the drivers, pass name/compat string
pointers. This lets their match() functions actually make real decisions.
OF-capable machines will pass name/compat pointers, but other machines
will not. grudging ok kettenis

move lm87 to dev/i2c; ok kettenis