The format of /proc/cpuinfo on other arches is different from the
format on x86. Modify the print_cpu_info function to handle arm and
ppc.
This change also eliminates Perl warnings caused by non-existent
elements in the %cpu hash:
Use of uninitialized value in concatenation (.) or string at
./prog/detect/sensors-detect line 3124.
Based on a patch from Changqing Li <changqing.li@windriver.com>,
GitHub PR: https://github.com/lm-sensors/lm-sensors/pull/168
This adds a config file for the MSI X370 SLI Plus. Similiar models may
work with derived configurations.
[olysonek: fixed up 'compute in3' statement, fixed up whitespace,
reformatted description]
Apparently, Gigabyte uses a 8kOhm and a 2.7kOhm resistor for the
voltage divider (both are commercially available SMD resistors). It
matches my BIOS reading.
Applied from PR
https://github.com/lm-sensors/lm-sensors/pull/203
[olysonek: squashed commits, updated description]
To better support cross-compilation, the Makefile should detect the
compiler's target architecture instead of using `uname -m`, which
returns the host architecture. This approach uses the `gcc`/`clang`
compatible `-dumpmachine` option.
Cross-compilation worked but because of this bug, `isadump` and
`isaset` x86 ISA specific tools were built for PCC and ARM.
This run-time issue became a compile-time issue when `glibc` 2.30
removed the `sys/io.h` I/O port functions on ARM.
Also renamed `MACHINE` to `ARCH` to match cross-compiler conventions
for specifying the target architecture. For example,
`make ARCH=arm CC=arm-linux-gnueabi-gcc AR=arm-linux-gnueabi-ar`
Fixes#190
Signed-off-by: Lucas Magasweran <lucas.magasweran@ieee.org>
Reported-and-tested-by: Millhouse <milhouse@libreelec.tv>
This motherboard supports one external temperature sensor. When there isn't
anything connected it shows very low values (e.g. -55C in my case).
Signed-off-by: Karol Grudziński <krlgrudzinski@gmail.com>
Reorder Nuvoton chips in the superio_ids_winbond array to match the
order in the kernel to make lookup easier. No functional change.
Signed-off-by: Ondřej Lysoněk <olysonek@redhat.com>
Since mainline commit 0599682b826ff7bbf, the nct6775 driver uses a
0xfff8 mask when scanning for devices. Update our detection to match
the kernel.
Fixes#188
Signed-off-by: Ondřej Lysoněk <olysonek@redhat.com>
According to:
"Quad-Core Intel® Xeon(R) Processor 5400 Series Electrical Specifications"
(table 2-12), FSB termination voltage of the CPUs (VTT) is 1.1V +/-5%.
Initial configuration set it to 1.2V - fix that.
Signed-off-by: Lech Perczak <lech.perczak@gmail.com>
scale_value is generic function for scaling values. There is a lower
resolution then json format provides for voltage and current.
The patch is calling scale_value() and showing values with higher
resolution.
For example on Xilinx ZynqMP platform:
ina226-i2c-3-41
Adapter: i2c-0-mux (chan_id 0)
in0: 2.00 mV
in1: 848.00 mV
power1: 287.50 mW
curr1: 330.00 mA
Signed-off-by: Michal Simek <monstr@monstr.eu>
Previously, if an error happened when reading the value of a subfeature,
subCnt would get incremented and a stray comma would get printed in the
following iteration, even though nothing was printed in the previous
iteration. This would produce a syntactically incorrect JSON.
This is based on a patch from the su8 user on GitHub.
Signed-off-by: Ondřej Lysoněk <olysonek@redhat.com>
Don't print a newline if no features were printed.
This is based on a patch from the su8 user on GitHub.
Signed-off-by: Ondřej Lysoněk <olysonek@redhat.com>
Add configuration for Supermicro X7DCL motherboard.
Based on slightly modified setup of X7DBE from this repo.
Runtime tested on X7DCL-3 rev. 1.1a
So far there is only support for onboard sensors,
no support for PSU failure yet, nor external PMBus sensors.
Signed-off-by: Lech Perczak <lech.perczak@gmail.com>
Ondřej Lysoněk