For families where it makes sense (ITE, Winbond, Fintek), when we have
an unknown Super-I/O, we scan all logical devices in search of one
with the typical address for hardware monitoring (0x290). If we find
it, we let the user know there are probably sensors in the chip.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4144 7894878c-1315-0410-8ee3-d5d059ff63e0
The detection is based of the PCI device class (SMBus is 0x0c05).
This will let us spot new adapters. Of course we can't load or even
suggest a driver for an unknown adapter, but at least the user will
know there's something to be investigated, and this should help us
when users report such configurations.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4143 7894878c-1315-0410-8ee3-d5d059ff63e0
(Linux 2.6). Still fallback to /proc (Linux 2.4). sysfs provides the PCI
device domain and class, which were hard to get from /proc.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4142 7894878c-1315-0410-8ee3-d5d059ff63e0
with the Super-I/O detection, and less reliable. Our w83627ehf driver
only supports the Super-I/O detection anyway.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4141 7894878c-1315-0410-8ee3-d5d059ff63e0
There are too many of these, the list was not up-to-date. And there
are different drivers which can give access to their I2C busses,
we can't guess which one the user wants. It was somewhat confusing
to suggest a driver which is not in mainline anyway.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4140 7894878c-1315-0410-8ee3-d5d059ff63e0
* Fix project homepage.
* Init scripts are in /etc/init.d on modern systems, rather than
/etc/rc.d/init.d. That way it works on Fedora, Suse and Gentoo, at
least.
* Don't tell the user to copy the init script if it's already there.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4131 7894878c-1315-0410-8ee3-d5d059ff63e0
* Give a name to Super-I/O chip families.
* Merge the two Winbond families.
* Be smarter when testing Super-I/O chips, read the device ID only once.
* When an unknown chip is found, print the ID only once.
* Reformat the warning message when a module isn't found.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4129 7894878c-1315-0410-8ee3-d5d059ff63e0
* Shorten the date part in revision string.
* Make the chip names more consistent.
* Align the result of ISA and Super-I/O probes.
* Replace "Failed!" with a less needlessly alarming "No".
* Make the ISA and Super-I/O probe results more compact.
* Adjust some explanations.
* Don't ask again before probing secondary Super-I/O address.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4128 7894878c-1315-0410-8ee3-d5d059ff63e0
* Kill the annoying blank space in front of half of the messages.
* Stop frightening the user with problems which might have happened some
years ago, but are very rare nowadays.
* Skip some messages on 2.6 kernels where they do not apply.
* Don't even mention i2c-dev if it is already loaded or build into the
kernel.
* Reword some messages.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4120 7894878c-1315-0410-8ee3-d5d059ff63e0
sense to try and continue as a non-root user, as we won't be able to load
modules, nor to probe i2c busses, nor to access the ISA ports. It's way
less confusing for the user to clearly state he/she needs to be root,
and stop.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4119 7894878c-1315-0410-8ee3-d5d059ff63e0
we don't use SMBus transactions the adapter doesn't support. This could
be extended to the transactions used during detection itself.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4086 7894878c-1315-0410-8ee3-d5d059ff63e0
IT8712F so there's not much to do. The most significant difference is that
fan speeds are now reported as 16-bit values, without fan clock dividers.
For this reason, I made the fan_divs optional in sensors. I also made the
fans themselves optional, as I noticed that a number of motherboards don't
use (and don't enable) fan3.
I've made a separate section (chip "it8716-*") in sensors.conf.eg, because
new motherboards tend to have different wirings (e.g. negative voltage
lines are not monitored) so this will make for saner defaults.
Thanks to Stian Oksavik for testing.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4083 7894878c-1315-0410-8ee3-d5d059ff63e0
less I2C addresses by default than the script itself does. The feature
was never used, and I don't think it's particuarly useful, so removing
it for simpler code makes sense.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@4082 7894878c-1315-0410-8ee3-d5d059ff63e0
