From 194c396e8a0d25f06dfb02683f8e33727a58882f Mon Sep 17 00:00:00 2001 From: Atul Kumar Pant Date: Sat, 5 Apr 2025 11:39:09 +0530 Subject: [PATCH 1/7] cpuidle: teo: Fix typos in two comments Fix a typo and correct a spelling mistake in comments. Signed-off-by: Atul Kumar Pant Link: https://patch.msgid.link/20250405060909.2026332-1-atulpant.linux@gmail.com [ rjw: Subject and changelog edits ] Signed-off-by: Rafael J. Wysocki --- drivers/cpuidle/governors/teo.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/cpuidle/governors/teo.c b/drivers/cpuidle/governors/teo.c index 8fe5e1b47ef9..bfa55c1eab5b 100644 --- a/drivers/cpuidle/governors/teo.c +++ b/drivers/cpuidle/governors/teo.c @@ -19,7 +19,7 @@ * * Of course, non-timer wakeup sources are more important in some use cases, * but even then it is generally unnecessary to consider idle duration values - * greater than the time time till the next timer event, referred as the sleep + * greater than the time till the next timer event, referred as the sleep * length in what follows, because the closest timer will ultimately wake up the * CPU anyway unless it is woken up earlier. * @@ -311,7 +311,7 @@ static int teo_select(struct cpuidle_driver *drv, struct cpuidle_device *dev, struct cpuidle_state *s = &drv->states[i]; /* - * Update the sums of idle state mertics for all of the states + * Update the sums of idle state metrics for all of the states * shallower than the current one. */ intercept_sum += prev_bin->intercepts; From d4a7882f93bf2520315d10ab600ea4701e22be69 Mon Sep 17 00:00:00 2001 From: Zhongqiu Han Date: Sat, 5 Apr 2025 21:53:08 +0800 Subject: [PATCH 2/7] cpuidle: menu: Optimize bucket assignment when next_timer_ns equals KTIME_MAX Directly assign the last bucket value instead of calling which_bucket() when next_timer_ns equals KTIME_MAX, the largest possible value that always falls into the last bucket. This avoids unnecessary calculations and enhances performance. Reviewed-by: Christian Loehle Signed-off-by: Zhongqiu Han Link: https://patch.msgid.link/20250405135308.1854342-1-quic_zhonhan@quicinc.com Signed-off-by: Rafael J. Wysocki --- drivers/cpuidle/governors/menu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/cpuidle/governors/menu.c b/drivers/cpuidle/governors/menu.c index 39aa0aea61c6..52d5d26fc7c6 100644 --- a/drivers/cpuidle/governors/menu.c +++ b/drivers/cpuidle/governors/menu.c @@ -255,7 +255,7 @@ static int menu_select(struct cpuidle_driver *drv, struct cpuidle_device *dev, */ data->next_timer_ns = KTIME_MAX; delta_tick = TICK_NSEC / 2; - data->bucket = which_bucket(KTIME_MAX); + data->bucket = BUCKETS - 1; } if (unlikely(drv->state_count <= 1 || latency_req == 0) || From af5376a77e8759084a9a135f50761607fea31743 Mon Sep 17 00:00:00 2001 From: Sudeep Holla Date: Mon, 17 Mar 2025 10:13:13 +0000 Subject: [PATCH 3/7] cpuidle: psci: Transition to the faux device interface The PSCI cpuidle driver does not require the creation of a platform device. Originally, this approach was chosen for simplicity when the driver was first implemented. With the introduction of the lightweight faux device interface, we now have a more appropriate alternative. Migrate the driver to utilize the faux bus, given that the platform device it previously created was not a real one anyway. This will simplify the code, reducing its footprint while maintaining functionality. Signed-off-by: Sudeep Holla Link: https://patch.msgid.link/20250317-plat2faux_dev-v1-1-5fe67c085ad5@arm.com Signed-off-by: Rafael J. Wysocki --- drivers/cpuidle/cpuidle-psci.c | 26 +++++++++----------------- 1 file changed, 9 insertions(+), 17 deletions(-) diff --git a/drivers/cpuidle/cpuidle-psci.c b/drivers/cpuidle/cpuidle-psci.c index b46a83f5ffe4..ee35ac816321 100644 --- a/drivers/cpuidle/cpuidle-psci.c +++ b/drivers/cpuidle/cpuidle-psci.c @@ -16,7 +16,7 @@ #include #include #include -#include +#include #include #include #include @@ -407,14 +407,14 @@ deinit: * to register cpuidle driver then rollback to cancel all CPUs * registration. */ -static int psci_cpuidle_probe(struct platform_device *pdev) +static int psci_cpuidle_probe(struct faux_device *fdev) { int cpu, ret; struct cpuidle_driver *drv; struct cpuidle_device *dev; for_each_present_cpu(cpu) { - ret = psci_idle_init_cpu(&pdev->dev, cpu); + ret = psci_idle_init_cpu(&fdev->dev, cpu); if (ret) goto out_fail; } @@ -434,26 +434,18 @@ out_fail: return ret; } -static struct platform_driver psci_cpuidle_driver = { +static struct faux_device_ops psci_cpuidle_ops = { .probe = psci_cpuidle_probe, - .driver = { - .name = "psci-cpuidle", - }, }; static int __init psci_idle_init(void) { - struct platform_device *pdev; - int ret; + struct faux_device *fdev; - ret = platform_driver_register(&psci_cpuidle_driver); - if (ret) - return ret; - - pdev = platform_device_register_simple("psci-cpuidle", -1, NULL, 0); - if (IS_ERR(pdev)) { - platform_driver_unregister(&psci_cpuidle_driver); - return PTR_ERR(pdev); + fdev = faux_device_create("psci-cpuidle", NULL, &psci_cpuidle_ops); + if (!fdev) { + pr_err("Failed to create psci-cpuidle device\n"); + return -ENODEV; } return 0; From 6138f34515162340520b0415184367e12775f68a Mon Sep 17 00:00:00 2001 From: Artem Bityutskiy Date: Mon, 17 Mar 2025 15:55:39 +0200 Subject: [PATCH 4/7] intel_idle: Add C1 demotion on/off sysfs knob Add a sysfs knob to enable/disable C1 demotion for the following Intel platforms: Sapphire Rapids Xeon, Emerald Rapids Xeon, Granite Rapids Xeon, Sierra Forest Xeon, and Grand Ridge SoC. This sysfs file shows up as "/sys/devices/system/cpu/cpuidle/intel_c1_demotion". The C1 demotion feature involves the platform firmware demoting deep C-state requests from the OS (e.g., C6 requests) to C1. The idea is that firmware monitors CPU wake-up rate, and if it is higher than a platform-specific threshold, the firmware demotes deep C-state requests to C1. For example, Linux requests C6, but firmware noticed too many wake-ups per second, and it keeps the CPU in C1. When the CPU stays in C1 long enough, the platform promotes it back to C6. The default value for C1 demotion is whatever is configured by BIOS. Signed-off-by: Artem Bityutskiy Link: https://patch.msgid.link/20250317135541.1471754-2-dedekind1@gmail.com Signed-off-by: Rafael J. Wysocki --- drivers/idle/intel_idle.c | 102 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 102 insertions(+) diff --git a/drivers/idle/intel_idle.c b/drivers/idle/intel_idle.c index 976f5be54e36..3292bf74e3c2 100644 --- a/drivers/idle/intel_idle.c +++ b/drivers/idle/intel_idle.c @@ -48,9 +48,11 @@ #include #include #include +#include #include #include #include +#include #include #include #include @@ -92,9 +94,15 @@ struct idle_cpu { */ unsigned long auto_demotion_disable_flags; bool disable_promotion_to_c1e; + bool c1_demotion_supported; bool use_acpi; }; +static bool c1_demotion_supported; +static DEFINE_MUTEX(c1_demotion_mutex); + +static struct device *sysfs_root __initdata; + static const struct idle_cpu *icpu __initdata; static struct cpuidle_state *cpuidle_state_table __initdata; @@ -1549,18 +1557,21 @@ static const struct idle_cpu idle_cpu_gmt __initconst = { static const struct idle_cpu idle_cpu_spr __initconst = { .state_table = spr_cstates, .disable_promotion_to_c1e = true, + .c1_demotion_supported = true, .use_acpi = true, }; static const struct idle_cpu idle_cpu_gnr __initconst = { .state_table = gnr_cstates, .disable_promotion_to_c1e = true, + .c1_demotion_supported = true, .use_acpi = true, }; static const struct idle_cpu idle_cpu_gnrd __initconst = { .state_table = gnrd_cstates, .disable_promotion_to_c1e = true, + .c1_demotion_supported = true, .use_acpi = true, }; @@ -1599,12 +1610,14 @@ static const struct idle_cpu idle_cpu_snr __initconst = { static const struct idle_cpu idle_cpu_grr __initconst = { .state_table = grr_cstates, .disable_promotion_to_c1e = true, + .c1_demotion_supported = true, .use_acpi = true, }; static const struct idle_cpu idle_cpu_srf __initconst = { .state_table = srf_cstates, .disable_promotion_to_c1e = true, + .c1_demotion_supported = true, .use_acpi = true, }; @@ -2324,6 +2337,88 @@ static void __init intel_idle_cpuidle_devices_uninit(void) cpuidle_unregister_device(per_cpu_ptr(intel_idle_cpuidle_devices, i)); } +static void intel_c1_demotion_toggle(void *enable) +{ + unsigned long long msr_val; + + rdmsrl(MSR_PKG_CST_CONFIG_CONTROL, msr_val); + /* + * Enable/disable C1 undemotion along with C1 demotion, as this is the + * most sensible configuration in general. + */ + if (enable) + msr_val |= NHM_C1_AUTO_DEMOTE | SNB_C1_AUTO_UNDEMOTE; + else + msr_val &= ~(NHM_C1_AUTO_DEMOTE | SNB_C1_AUTO_UNDEMOTE); + wrmsrl(MSR_PKG_CST_CONFIG_CONTROL, msr_val); +} + +static ssize_t intel_c1_demotion_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + bool enable; + int err; + + err = kstrtobool(buf, &enable); + if (err) + return err; + + mutex_lock(&c1_demotion_mutex); + /* Enable/disable C1 demotion on all CPUs */ + on_each_cpu(intel_c1_demotion_toggle, (void *)enable, 1); + mutex_unlock(&c1_demotion_mutex); + + return count; +} + +static ssize_t intel_c1_demotion_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + unsigned long long msr_val; + + /* + * Read the MSR value for a CPU and assume it is the same for all CPUs. Any other + * configuration would be a BIOS bug. + */ + rdmsrl(MSR_PKG_CST_CONFIG_CONTROL, msr_val); + return sysfs_emit(buf, "%d\n", !!(msr_val & NHM_C1_AUTO_DEMOTE)); +} +static DEVICE_ATTR_RW(intel_c1_demotion); + +static int __init intel_idle_sysfs_init(void) +{ + int err; + + if (!c1_demotion_supported) + return 0; + + sysfs_root = bus_get_dev_root(&cpu_subsys); + if (!sysfs_root) + return 0; + + err = sysfs_add_file_to_group(&sysfs_root->kobj, + &dev_attr_intel_c1_demotion.attr, + "cpuidle"); + if (err) { + put_device(sysfs_root); + return err; + } + + return 0; +} + +static void __init intel_idle_sysfs_uninit(void) +{ + if (!sysfs_root) + return; + + sysfs_remove_file_from_group(&sysfs_root->kobj, + &dev_attr_intel_c1_demotion.attr, + "cpuidle"); + put_device(sysfs_root); +} + static int __init intel_idle_init(void) { const struct x86_cpu_id *id; @@ -2374,6 +2469,8 @@ static int __init intel_idle_init(void) auto_demotion_disable_flags = icpu->auto_demotion_disable_flags; if (icpu->disable_promotion_to_c1e) c1e_promotion = C1E_PROMOTION_DISABLE; + if (icpu->c1_demotion_supported) + c1_demotion_supported = true; if (icpu->use_acpi || force_use_acpi) intel_idle_acpi_cst_extract(); } else if (!intel_idle_acpi_cst_extract()) { @@ -2387,6 +2484,10 @@ static int __init intel_idle_init(void) if (!intel_idle_cpuidle_devices) return -ENOMEM; + retval = intel_idle_sysfs_init(); + if (retval) + pr_warn("failed to initialized sysfs"); + intel_idle_cpuidle_driver_init(&intel_idle_driver); retval = cpuidle_register_driver(&intel_idle_driver); @@ -2411,6 +2512,7 @@ hp_setup_fail: intel_idle_cpuidle_devices_uninit(); cpuidle_unregister_driver(&intel_idle_driver); init_driver_fail: + intel_idle_sysfs_uninit(); free_percpu(intel_idle_cpuidle_devices); return retval; From af3a1b6a18258168f09e4ebc014b368bb0e845b0 Mon Sep 17 00:00:00 2001 From: Artem Bityutskiy Date: Mon, 17 Mar 2025 15:55:40 +0200 Subject: [PATCH 5/7] Documentation: admin-guide: pm: Document intel_idle C1 demotion Document the intel_idle driver sysfs file for enabling/disabling C1 demotion. Signed-off-by: Artem Bityutskiy Link: https://patch.msgid.link/20250317135541.1471754-3-dedekind1@gmail.com Signed-off-by: Rafael J. Wysocki --- Documentation/admin-guide/pm/intel_idle.rst | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) diff --git a/Documentation/admin-guide/pm/intel_idle.rst b/Documentation/admin-guide/pm/intel_idle.rst index 5940528146eb..ed6f055d4b14 100644 --- a/Documentation/admin-guide/pm/intel_idle.rst +++ b/Documentation/admin-guide/pm/intel_idle.rst @@ -38,6 +38,27 @@ instruction at all. only way to pass early-configuration-time parameters to it is via the kernel command line. +Sysfs Interface +=============== + +The ``intel_idle`` driver exposes the following ``sysfs`` attributes in +``/sys/devices/system/cpu/cpuidle/``: + +``intel_c1_demotion`` + Enable or disable C1 demotion for all CPUs in the system. This file is + only exposed on platforms that support the C1 demotion feature and where + it was tested. Value 0 means that C1 demotion is disabled, value 1 means + that it is enabled. Write 0 or 1 to disable or enable C1 demotion for + all CPUs. + + The C1 demotion feature involves the platform firmware demoting deep + C-state requests from the OS (e.g., C6 requests) to C1. The idea is that + firmware monitors CPU wake-up rate, and if it is higher than a + platform-specific threshold, the firmware demotes deep C-state requests + to C1. For example, Linux requests C6, but firmware noticed too many + wake-ups per second, and it keeps the CPU in C1. When the CPU stays in + C1 long enough, the platform promotes it back to C6. This may improve + some workloads' performance, but it may also increase power consumption. .. _intel-idle-enumeration-of-states: From e80e1342eabaac86da02f2c74fffc010e4f68c55 Mon Sep 17 00:00:00 2001 From: Artem Bityutskiy Date: Mon, 17 Mar 2025 15:55:41 +0200 Subject: [PATCH 6/7] Documentation: ABI: testing: document the new cpuidle sysfs file Mention the new 'intel_c1_demotion' sysfs file in the "cpuidle" section and refer to "Documentation/admin-guide/pm/intel_idle.rst" for more information. Signed-off-by: Artem Bityutskiy Link: https://patch.msgid.link/20250317135541.1471754-4-dedekind1@gmail.com Signed-off-by: Rafael J. Wysocki --- Documentation/ABI/testing/sysfs-devices-system-cpu | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/Documentation/ABI/testing/sysfs-devices-system-cpu b/Documentation/ABI/testing/sysfs-devices-system-cpu index 206079d3bd5b..6c4934d3f4db 100644 --- a/Documentation/ABI/testing/sysfs-devices-system-cpu +++ b/Documentation/ABI/testing/sysfs-devices-system-cpu @@ -111,6 +111,7 @@ What: /sys/devices/system/cpu/cpuidle/available_governors /sys/devices/system/cpu/cpuidle/current_driver /sys/devices/system/cpu/cpuidle/current_governor /sys/devices/system/cpu/cpuidle/current_governer_ro + /sys/devices/system/cpu/cpuidle/intel_c1_demotion Date: September 2007 Contact: Linux kernel mailing list Description: Discover cpuidle policy and mechanism @@ -132,7 +133,11 @@ Description: Discover cpuidle policy and mechanism current_governor_ro: (RO) displays current idle policy. - See Documentation/admin-guide/pm/cpuidle.rst and + intel_c1_demotion: (RW) enables/disables the C1 demotion + feature on Intel CPUs. + + See Documentation/admin-guide/pm/cpuidle.rst, + Documentation/admin-guide/pm/intel_idle.rst, and Documentation/driver-api/pm/cpuidle.rst for more information. From 5836ebeb4a2b7e25305fab9170060d5f3de1ff30 Mon Sep 17 00:00:00 2001 From: Sudeep Holla Date: Fri, 2 May 2025 15:01:19 +0100 Subject: [PATCH 7/7] cpuidle: psci: Avoid initializing faux device if no DT idle states are present Commit af5376a77e87 ("cpuidle: psci: Transition to the faux device interface") transitioned the PSCI cpuidle driver from using a platform device to the faux device framework. However, unlike platform devices, the faux device infrastructure logs an error when the probe function fails, even if the failure is intentional or expected. To prevent unnecessary error logs, we can skip creating the faux device entirely if there are no PSCI idle states defined in the device tree. Introduce a check for DT idle states during initialization and avoid setting up the device if none are found. This ensures cleaner logs and avoids misleading probe failure messages when PSCI idle support is intentionally not described in DT. Fixes: af5376a77e87 ("cpuidle: psci: Transition to the faux device interface") Reported-by: Jon Hunter Closes: https://lore.kernel.org/r/cf4e70e4-9fe5-4697-8744-8c12c41b5ff9@nvidia.com Signed-off-by: Sudeep Holla Reviewed-by: Ulf Hansson Link: https://patch.msgid.link/20250502140119.2578909-1-sudeep.holla@arm.com Signed-off-by: Rafael J. Wysocki --- drivers/cpuidle/cpuidle-psci.c | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) diff --git a/drivers/cpuidle/cpuidle-psci.c b/drivers/cpuidle/cpuidle-psci.c index ee35ac816321..40f378c1dc9f 100644 --- a/drivers/cpuidle/cpuidle-psci.c +++ b/drivers/cpuidle/cpuidle-psci.c @@ -438,10 +438,29 @@ static struct faux_device_ops psci_cpuidle_ops = { .probe = psci_cpuidle_probe, }; +static bool __init dt_idle_state_present(void) +{ + struct device_node *cpu_node __free(device_node); + struct device_node *state_node __free(device_node); + + cpu_node = of_cpu_device_node_get(cpumask_first(cpu_possible_mask)); + if (!cpu_node) + return false; + + state_node = of_get_cpu_state_node(cpu_node, 0); + if (!state_node) + return false; + + return !!of_match_node(psci_idle_state_match, state_node); +} + static int __init psci_idle_init(void) { struct faux_device *fdev; + if (!dt_idle_state_present()) + return 0; + fdev = faux_device_create("psci-cpuidle", NULL, &psci_cpuidle_ops); if (!fdev) { pr_err("Failed to create psci-cpuidle device\n");