linux/drivers/opp, branch v6.2

OPP: decouple dt properties in opp_parse_supplies()

2022-11-04T05:29:07Z

The opp-microwatt property was added with the intention of providing platforms a way to specify a precise value for the power consumption of a device at a given OPP to enable better energy-aware scheduling decisions by informing the kernel of the total static and dynamic power of a device at a given OPP, removing the reliance on the EM subsystem's often flawed estimations. This property is parsed by opp_parse_supplies(), which creates a hard dependency on the opp-microvolt property. Some platforms, such as Apple Silicon, do not describe their device's voltage regulators in the DT as they cannot be controlled by the kernel and/or rely on opaque firmware algorithms to control their voltage and current characteristics at runtime. We can, however, experimentally determine the power consumption of a given device at a given OPP, taking advantage of opp-microwatt to provide EAS on such devices as was initially intended. Allow platforms to specify and consume any subset of opp-microvolt, opp-microamp, or opp-microwatt without a hard dependency on opp-microvolt to enable this functionality on such platforms. Tested-by: James Calligeros Signed-off-by: James Calligeros Co-developed-by: Viresh Kumar Signed-off-by: Viresh Kumar

OPP: Simplify opp_parse_supplies() by restructuring it

2022-11-04T05:29:05Z

opp_parse_supplies() has grown into too big of a routine (~190 lines) and it is not straight-forward to understand it anymore. Break it into smaller routines and reduce code redundancy a bit by using the same code to parse properties. This shouldn't result in any logical changes. Tested-by: James Calligeros Signed-off-by: Viresh Kumar

OPP: Parse named opp-microwatt property too

2022-11-04T05:29:03Z

We missed parsing the named opp-microwatt- property, fix that. Tested-by: James Calligeros Signed-off-by: Viresh Kumar

OPP: Fix an un-initialized variable usage

2022-08-16T05:18:08Z

smatch complains that 'ret' may be returned un-initialized. Explicitly return 0 if we reach the end of the function (should 'opp_table->clk_count' be 0). Fixes: 8174a3a613af ("OPP: Provide a simple implementation to configure multiple clocks") Signed-off-by: Christophe JAILLET Signed-off-by: Viresh Kumar

Merge tag 'opp-updates-5.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vireshk/pm

2022-08-03T15:49:38Z

Pull operating performance points (OPP) updates for 5.20-rc1 from Viresh Kumar: "- Make dev_pm_opp_set_regulators() accept NULL terminated list (Viresh Kumar). - Add dev_pm_opp_set_config() and friends and migrate other users/helpers to using them (Viresh Kumar). - Add support for multiple clocks for a device (Viresh Kumar and Krzysztof Kozlowski). - Configure resources before adding OPP table for Venus (Stanimir Varbanov). - Keep reference count up for opp->np and opp_table->np while they are still in use (Liang He). - Minor cleanups (Viresh Kumar and Yang Li)." * tag 'opp-updates-5.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vireshk/pm: (43 commits) venus: pm_helpers: Fix warning in OPP during probe OPP: Don't drop opp->np reference while it is still in use OPP: Don't drop opp_table->np reference while it is still in use OPP: Remove dev{m}_pm_opp_of_add_table_noclk() PM / devfreq: tegra30: Register config_clks helper OPP: Allow config_clks helper for single clk case OPP: Provide a simple implementation to configure multiple clocks OPP: Assert clk_count == 1 for single clk helpers OPP: Add key specific assert() method to key finding helpers OPP: Compare bandwidths for all paths in _opp_compare_key() OPP: Allow multiple clocks for a device dt-bindings: opp: accept array of frequencies OPP: Make dev_pm_opp_set_opp() independent of frequency OPP: Reuse _opp_compare_key() in _opp_add_static_v2() OPP: Remove rate_not_available parameter to _opp_add() OPP: Use consistent names for OPP table instances OPP: Use generic key finding helpers for bandwidth key OPP: Use generic key finding helpers for level key OPP: Add generic key finding helpers and use them for freq APIs OPP: Remove dev_pm_opp_find_freq_ceil_by_volt() ...

OPP: Don't drop opp->np reference while it is still in use

2022-07-19T05:37:02Z

The struct dev_pm_opp contains a reference of the DT node, opp->np, throughout its lifetime. We should increase the refcount for the same from _opp_add_static_v2(), and drop it while removing the OPP finally. Signed-off-by: Liang He [ Viresh: Updated subject / commit log, create _of_clear_opp() and drop reference from it] Signed-off-by: Viresh Kumar

OPP: Don't drop opp_table->np reference while it is still in use

2022-07-19T05:05:18Z

The OPP table contains a reference of the DT node, opp_table->np, throughout its lifetime. We shouldn't drop the refcount for the same from _of_init_opp_table(), but do that while removing the OPP table finally. Signed-off-by: Liang He [ Viresh: Updated subject / commit log and drop reference from _of_clear_opp_table() ] Signed-off-by: Viresh Kumar

PM: EM: convert power field to micro-Watts precision and align drivers

2022-07-15T17:17:30Z

The milli-Watts precision causes rounding errors while calculating efficiency cost for each OPP. This is especially visible in the 'simple' Energy Model (EM), where the power for each OPP is provided from OPP framework. This can cause some OPPs to be marked inefficient, while using micro-Watts precision that might not happen. Update all EM users which access 'power' field and assume the value is in milli-Watts. Solve also an issue with potential overflow in calculation of energy estimation on 32bit machine. It's needed now since the power value (thus the 'cost' as well) are higher. Example calculation which shows the rounding error and impact: power = 'dyn-power-coeff' * volt_mV * volt_mV * freq_MHz power_a_uW = (100 * 600mW * 600mW * 500MHz) / 10^6 = 18000 power_a_mW = (100 * 600mW * 600mW * 500MHz) / 10^9 = 18 power_b_uW = (100 * 605mW * 605mW * 600MHz) / 10^6 = 21961 power_b_mW = (100 * 605mW * 605mW * 600MHz) / 10^9 = 21 max_freq = 2000MHz cost_a_mW = 18 * 2000MHz/500MHz = 72 cost_a_uW = 18000 * 2000MHz/500MHz = 72000 cost_b_mW = 21 * 2000MHz/600MHz = 70 // <- artificially better cost_b_uW = 21961 * 2000MHz/600MHz = 73203 The 'cost_b_mW' (which is based on old milli-Watts) is misleadingly better that the 'cost_b_uW' (this patch uses micro-Watts) and such would have impact on the 'inefficient OPPs' information in the Cpufreq framework. This patch set removes the rounding issue. Signed-off-by: Lukasz Luba Acked-by: Daniel Lezcano Acked-by: Viresh Kumar Signed-off-by: Rafael J. Wysocki

OPP: Remove dev{m}_pm_opp_of_add_table_noclk()

2022-07-12T15:05:21Z

Remove the now unused variants and the now unnecessary "getclk" parameter from few routines. Signed-off-by: Viresh Kumar

OPP: Allow config_clks helper for single clk case

2022-07-12T15:05:21Z

There is a corner case with Tegra30, where we want to skip clk configuration via dev_pm_opp_set_opp(), but still want the OPP core to read the "opp-hz" property so we can find the right OPP via freq finding helpers. This is the easiest of the ways to make it work, without any special hacks in the OPP core. Allow config_clks to be passed for single clk case. Tested-by: Dmitry Osipenko Signed-off-by: Viresh Kumar