On a system with course grain resolution of energy unit (milli J) the
accumulation thread can be executed less frequently than on the system
with fine grain resolution(micro J).
This patch sets the accumulation thread interval to an optimum value
calculated based on the (energy unit) resolution supported by the
hardware (assuming a peak wattage of 240W).
Signed-off-by: Naveen Krishna Chatradhi <nchatrad@amd.com>
Link: https://lore.kernel.org/r/20200929105322.8919-3-nchatrad@amd.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
0-day rightfully complains about a sometimes uninitialized variable
in pmbus_get_boolean().
drivers/hwmon/pmbus/pmbus_core.c:903:13: warning:
variable 'ret' is used uninitialized whenever 'if' condition is true
} else if (!s1 || !s2) {
While that is technically true, it won't be hit in the field since the
condition indicates a programming error. Move the check of that condition
into the code generating the attribute entry, and refuse generating the
attribute if the condition is true. Swap the condition check in
pmbus_get_boolean() to ensure that static analyzers don't get a hiccup
(because we check if s1 and s2 are NULL, static analyzers may believe
that they can be NULL independently of each other).
Reported-by: kernel test robot <lkp@intel.com>
Cc: Alex Qiu <xqiu@google.com>
Reviewed-by: Alex Qiu <xqiu@google.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Enable runtime debug control of whether the PEC byte is exchanged with
the PMBus device.
Some manufacturers have asked for the PEC to be disabled as part of
debugging driver communication issues with devices.
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Link: https://lore.kernel.org/r/20200910021106.2958382-1-andrew@aj.id.au
[groeck: Replace %1llu with %llu]
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Caching register values can be very expensive for PMBus chips. Some
modern chips may have 10 or more pages, with several sensors supported
per page. For example, MAX16601 creates more than 90 sysfs attributes.
Register caching for such chips is time consuming, especially if only a
few attributes are read on a regular basis. For MAX16601, it was observed
that it can take up to two seconds to read all attributes on a slow I2C
bus. In this situation, register caching results in the opposite of its
intention: It increases the number of I2C operations, in some cases
substantially, and it results in large latency when trying to access
individual sensor data.
Drop all register caching to solve the problem. Since it is no longer
necessary, drop status register mapping as part of the change, and specify
status registers directly.
Cc: Alex Qiu <xqiu@google.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Alex Qiu <xqiu@google.com>
Tested-by: Alex Qiu <xqiu@google.com>
Link: https://lore.kernel.org/r/20200904163314.259087-1-linux@roeck-us.net
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
The debugfs code was intended to aid figuring out functionality
of undocumented registers. Turns out that wasn't very helpful,
since register locations change too much between AMD chip revisions,
and the data isn't really valuable for chips where it isn't already
supported. On top of that, its existence has been used as argument
for providing pseudo-API debugfs functions in other drivers.
So let's just take it out.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Use the nvmem kernel api to expose the black box
chip functionality to userspace.
Using this feature, the device is capable of recording
to nonvolatile flash memory the vital data about the
system status that caused the system to perform a
black box write.
A blackbox is 64 bytes of data containing all the
status registers, last two states of the sequencer,
timestamp and counters. The mapping of this data is
described in the adm1266 datasheet.
On power-up the driver sets the unix time to
the adm1266 using the SET_RTC command. This value
is incremented by an internal clock and it is used
as timestamp for the black box feature.
Signed-off-by: Alexandru Tachici <alexandru.tachici@analog.com>
Link: https://lore.kernel.org/r/20200812142055.9213-6-alexandru.tachici@analog.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
As part of the ongoing i2c transition to the simple probe
("probe_new"), this patch uses i2c_match_id to retrieve the
driver_data for the probed device. The id parameter is thus no longer
necessary and the simple probe can be used instead.
Signed-off-by: Stephen Kitt <steve@sk2.org>
Link: https://lore.kernel.org/r/20200821160035.590142-1-steve@sk2.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
As part of the ongoing i2c transition to the simple probe
("probe_new"), this patch uses i2c_match_id to retrieve the
driver_data for the probed device. The id parameter is thus no longer
necessary and the simple probe can be used instead.
Signed-off-by: Stephen Kitt <steve@sk2.org>
Link: https://lore.kernel.org/r/20200821160159.591293-1-steve@sk2.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
As part of the ongoing i2c transition to the simple probe
("probe_new"), this patch uses i2c_match_id to retrieve the
driver_data for the probed device. The id parameter is thus no longer
necessary and the simple probe can be used instead.
Signed-off-by: Stephen Kitt <steve@sk2.org>
Link: https://lore.kernel.org/r/20200821160231.592571-1-steve@sk2.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Many hwmon drivers don't use the id information provided by the old
i2c probe function, and the remainder can easily be adapted to the new
form ("probe_new") by calling i2c_match_id explicitly.
This avoids scanning the identifier tables during probes.
Drivers which didn't use the id are converted as-is; drivers which did
are modified to call i2c_match_id() with the same level of
error-handling (if any) as before.
This patch wraps up the transition for hwmon, with four stragglers not
included in the previous large patch.
Signed-off-by: Stephen Kitt <steve@sk2.org>
Link: https://lore.kernel.org/r/20200821160354.594715-1-steve@sk2.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Many hwmon drivers don't use the id information provided by the old
i2c probe function, and the remainder can easily be adapted to the new
form ("probe_new") by calling i2c_match_id explicitly.
This avoids scanning the identifier tables during probes.
Drivers which didn't use the id are converted as-is; drivers which did
are modified as follows:
* if the information in i2c_client is sufficient, that's used instead
(client->name);
* anything else is handled by calling i2c_match_id() with the same
level of error-handling (if any) as before.
A few drivers aren't included in this patch because they have a
different set of maintainers. They will be covered by other patches.
Signed-off-by: Stephen Kitt <steve@sk2.org>
Link: https://lore.kernel.org/r/20200813160222.1503401-1-steve@sk2.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
pmbus_do_probe doesn't use the id information provided in its second
argument, so this can be removed, which then allows using the
single-parameter i2c probe function ("probe_new") for probes.
This avoids scanning the identifier tables during probes.
Drivers which didn't use the id are converted as-is; drivers which did
are modified as follows:
* if the information in i2c_client is sufficient, that's used instead
(client->name);
* configured v. probed comparisons are performed by comparing the
configured name to the detected name, instead of the ids; this
involves strcmp but is still cheaper than comparing all the device
names when scanning the tables;
* anything else is handled by calling i2c_match_id() with the same
level of error-handling (if any) as before.
Additionally, the mismatch message in the ltc2978 driver is adjusted
so that it no longer assumes that the driver_data is an index into
ltc2978_id.
Signed-off-by: Stephen Kitt <steve@sk2.org>
Acked-by: Wolfram Sang <wsa@kernel.org>
Link: https://lore.kernel.org/r/20200808210004.30880-1-steve@sk2.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
On some drives, reading the drive temperature resets the drive's
spin-down timer. If the drive temperature is read too often, affected
drives will never spin down. Add this information as usage note to
the driver documentation.
Reported-by: Peter Sulyok <peter@sulyok.net>
Cc: Peter Sulyok <peter@sulyok.net>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Adding implementation for new attributes (rated_min/rated_max) for
currentX, inX, powerX, tempX and humidityX.
Tested with OpenBMC stack and simple hwmon driver using
rated_min/rated_max for the following types of sensors: hwmon_temp,
hwmon_in, hwmon_curr, hwmon_power, hwmon_humidity.
For each sensor rated attributes were available and returned
expected values.
Signed-off-by: Zbigniew Lukwinski <zbigniew.lukwinski@linux.intel.com>
Link: https://lore.kernel.org/r/1596224237-32280-3-git-send-email-zbigniew.lukwinski@linux.intel.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Smatch has a new check for resource leaks which found a bug in probe:
drivers/hwmon/w83627ehf.c:2417 w83627ehf_probe()
warn: 'res->start' not released on lines: 2412.
We need to clean up if devm_hwmon_device_register_with_info() fails.
Fixes: 266cd58359 ("hwmon: (w83627ehf) convert to with_info interface")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Dr. David Alan Gilbert <linux@treblig.org>
Link: https://lore.kernel.org/r/20200921125212.GA1128194@mwanda
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Starting with MAX34451, the chips of this series support STATUS_IOUT and
STATUS_TEMPERATURE commands, and no longer report over-current and
over-temperature status with STATUS_MFR_SPECIFIC.
Fixes: 7a001dbab4 ("hwmon: (pmbus/max34440) Add support for MAX34451.")
Fixes: 50115ac9b6 ("hwmon: (pmbus/max34440) Add support for MAX34460 and MAX34461")
Reported-by: Steve Foreman <foremans@google.com>
Cc: Steve Foreman <foremans@google.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
If the temperature is read before the internal calibration is
completed, the driver returns -EIO. Instead it should return -EAGAIN
to encourage repeating the operation.
Note (groeck): Returning -EAGAIN to userspace may result in hard loops;
some userspace code interprets -EAGAIN as request to retry immediately.
I would prefer -ENODATA, but it turns out that the thermal subsystem only
handles -EAGAIN silently, so we'll have to stick with that.
Signed-off-by: Lars Povlsen <lars.povlsen@microchip.com>
Link: https://lore.kernel.org/r/20200903134704.8949-1-lars.povlsen@microchip.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Pull syscall tracing fix from Borislav Petkov:
"Fix the seccomp syscall rewriting so that trace and audit see the
rewritten syscall number, from Kees Cook"
* tag 'core_urgent_for_v5.9_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
core/entry: Report syscall correctly for trace and audit
Pull locking fixes from Borislav Petkov:
"Two fixes from the locking/urgent pile:
- Fix lockdep's detection of "USED" <- "IN-NMI" inversions (Peter
Zijlstra)
- Make percpu-rwsem operations on the semaphore's ->read_count
IRQ-safe because it can be used in an IRQ context (Hou Tao)"
* tag 'locking_urgent_for_v5.9_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
locking/percpu-rwsem: Use this_cpu_{inc,dec}() for read_count
locking/lockdep: Fix "USED" <- "IN-NMI" inversions
Pull EFI fix from Borislav Petkov:
"Ensure that the EFI bootloader control module only probes successfully
on systems that support the EFI SetVariable runtime service"
[ Tag and commit from Ard Biesheuvel, forwarded by Borislav ]
* tag 'efi-urgent-for-v5.9-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
efi: efibc: check for efivars write capability
