From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-alma10-1.taild15c8.ts.net [100.103.45.18]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 0764327467F; Fri, 19 Jun 2026 14:52:12 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=100.103.45.18 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1781880734; cv=none; b=jp/ZWyvqASR6UtWRSHz8WCeOaM6DdccPixARv/6YV0cYrbZ88MZE6kEizPQ4eI3Nf2YWnVaiSarKxPX7b4/+Cr1BAHPQP4vTPyG1AsSw5mFE9f4/860v3OtnIX5gFpkaaDGgK+cLJoEEHludHp3Ghts4P9goQ5FeOhsgfWd2Mck= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1781880734; c=relaxed/simple; bh=1TKHOOgIi2B54S5F5zBztQ//B5Kzm+e8gYrFWEcvawA=; h=From:To:Cc:Subject:Date:Message-ID:MIME-Version:Content-Type; b=PnIOPhlWb2ZXpmVzI8r2YplBciEmjH03yJ+BQXo6mM4Gcet9PQ9cSgviVXr6+mQ0TLv3BPDmkdphT5m2gum/yxSYqUMuV5rl3LAq8Y3so2rsYzQH3g0mdtLyJC6RQQx8QXuFX9/VWgby3JE3SjR9Vc59wYgYc4Cvhcg6BXaFsMA= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=Ztvcb7gK; arc=none smtp.client-ip=100.103.45.18 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="Ztvcb7gK" Received: by smtp.kernel.org (Postfix) with ESMTPSA id CB5411F000E9; Fri, 19 Jun 2026 14:52:10 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kernel.org; s=k20260515; t=1781880732; bh=LUJ5P/oH3jsbOHH+ze6NU0xGCFjKU1qMIrAx9AtT24s=; h=From:To:Cc:Subject:Date; b=Ztvcb7gKIIW0IcaVJUIbmR5h+z+RMZANRpUC4qUnfJN0Ef+oZXe2JxtumV9q9Foe4 pTpj06bDh9iOD1tFrRrqbnTmaKc2fbiz7URwPV1n91Juq+R6QHuKD7drU/TQtH9E1A 5uzRIkeU47oKWxwYUhul84NCIekEdloAdo1cuMgszq2gfHVNXu6q2czwfKGHAHG8EN fKflnke8khfCqP4JMhEsCQIJj3vfURRNx4vSe1uBTFweVSWTB4RPWyYN1VJn0d/6BG UM79FCUzT2/Xwqp8J25PT3qcgJXhArlvNnEvg1UvcHuxu38DrAVhiirra4SlgjlSwY xbEIYdwBXskOA== From: "Rafael J. Wysocki" To: Linux PM Cc: LKML , Viresh Kumar , Srinivas Pandruvada , Christian Loehle , Peter Zijlstra , Vincent Guittot , Zhongqiu Han , Mario Limonciello Subject: [PATCH v2] cpufreq: intel_pstate: Adjust the .adjust_perf() driver callback Date: Fri, 19 Jun 2026 16:52:08 +0200 Message-ID: <6277654.lOV4Wx5bFT@rafael.j.wysocki> Organization: Linux Kernel Development Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: 7Bit Content-Type: text/plain; charset="UTF-8" From: Rafael J. Wysocki In some cases, the processor may not actually stick to the "desired" performance level programmed through the driver's .adjust_perf() callback and may go above it, which may not be desirable (for instance, there may be a UCLAMP_MAX limit set for the task currently running on the given CPU which should be respected). Address that by adjusting the .adjust_perf() callback to take an additional argument, max_perf, representing the maximum allowed performance level of the CPU and update the intel_pstate driver to take that argument into account as appropriate. Accordingly, adjust cpufreq_driver_adjust_perf() and the other existing user of .adjust_perf(), which is the amd-pstate driver (but the behavior of that driver is not changed). While at it, also update the cpufreq_driver_adjust_perf() documentation to reflect this change and some previous code changes that have not been taken into account in it. Signed-off-by: Rafael J. Wysocki Acked-by: Viresh Kumar --- This is an update of https://lore.kernel.org/linux-pm/14060154.uLZWGnKmhe@rafael.j.wysocki/ sent mainly because the v1 did not update the Rust bindings by omission. It also fixes a few typos present in the v1. Thanks! --- drivers/cpufreq/amd-pstate.c | 1 + drivers/cpufreq/cpufreq.c | 14 +++++++++----- drivers/cpufreq/intel_pstate.c | 9 ++++++++- include/linux/cpufreq.h | 2 ++ kernel/sched/cpufreq_schedutil.c | 4 +++- rust/kernel/cpufreq.rs | 6 ++++-- 6 files changed, 27 insertions(+), 9 deletions(-) --- a/drivers/cpufreq/amd-pstate.c +++ b/drivers/cpufreq/amd-pstate.c @@ -781,6 +781,7 @@ static unsigned int amd_pstate_fast_swit static void amd_pstate_adjust_perf(struct cpufreq_policy *policy, unsigned long _min_perf, unsigned long target_perf, + unsigned long _max_perf, unsigned long capacity) { u8 max_perf, min_perf, des_perf, cap_perf; --- a/drivers/cpufreq/cpufreq.c +++ b/drivers/cpufreq/cpufreq.c @@ -2252,14 +2252,17 @@ EXPORT_SYMBOL_GPL(cpufreq_driver_fast_sw * @policy: cpufreq policy object of the target CPU. * @min_perf: Minimum (required) performance level (units of @capacity). * @target_perf: Target (desired) performance level (units of @capacity). + * @max_perf: Maximum (allowed) performance level (units of @capacity). * @capacity: Capacity of the target CPU. * - * Carry out a fast performance level switch of @cpu without sleeping. + * Carry out a fast performance level adjustment for the CPU represented by + * @policy without sleeping. * * The driver's ->adjust_perf() callback invoked by this function must be - * suitable for being called from within RCU-sched read-side critical sections - * and it is expected to select a suitable performance level equal to or above - * @min_perf and preferably equal to or below @target_perf. + * suitable for calling from within RCU-sched read-side critical sections and + * it is expected to program the processor to select suitable performance + * levels between @min_perf and @max_perf inclusive and preferably close to + * @target_perf going forward for the CPU represented by @policy. * * This function must not be called if policy->fast_switch_enabled is unset. * @@ -2271,9 +2274,10 @@ EXPORT_SYMBOL_GPL(cpufreq_driver_fast_sw void cpufreq_driver_adjust_perf(struct cpufreq_policy *policy, unsigned long min_perf, unsigned long target_perf, + unsigned long max_perf, unsigned long capacity) { - cpufreq_driver->adjust_perf(policy, min_perf, target_perf, capacity); + cpufreq_driver->adjust_perf(policy, min_perf, target_perf, max_perf, capacity); } /** --- a/drivers/cpufreq/intel_pstate.c +++ b/drivers/cpufreq/intel_pstate.c @@ -3241,6 +3241,7 @@ static unsigned int intel_cpufreq_fast_s static void intel_cpufreq_adjust_perf(struct cpufreq_policy *policy, unsigned long min_perf, unsigned long target_perf, + unsigned long max_perf, unsigned long capacity) { struct cpudata *cpu = all_cpu_data[policy->cpu]; @@ -3271,7 +3272,13 @@ static void intel_cpufreq_adjust_perf(st if (min_pstate > cpu->max_perf_ratio) min_pstate = cpu->max_perf_ratio; - max_pstate = min(cap_pstate, cpu->max_perf_ratio); + max_pstate = cap_pstate; + if (max_perf < capacity) + max_pstate = DIV_ROUND_UP(cap_pstate * max_perf, capacity); + + if (max_pstate > cpu->max_perf_ratio) + max_pstate = cpu->max_perf_ratio; + if (max_pstate < min_pstate) max_pstate = min_pstate; --- a/include/linux/cpufreq.h +++ b/include/linux/cpufreq.h @@ -379,6 +379,7 @@ struct cpufreq_driver { void (*adjust_perf)(struct cpufreq_policy *policy, unsigned long min_perf, unsigned long target_perf, + unsigned long max_perf, unsigned long capacity); /* @@ -624,6 +625,7 @@ unsigned int cpufreq_driver_fast_switch( void cpufreq_driver_adjust_perf(struct cpufreq_policy *policy, unsigned long min_perf, unsigned long target_perf, + unsigned long max_perf, unsigned long capacity); bool cpufreq_driver_has_adjust_perf(void); int cpufreq_driver_target(struct cpufreq_policy *policy, --- a/kernel/sched/cpufreq_schedutil.c +++ b/kernel/sched/cpufreq_schedutil.c @@ -50,6 +50,7 @@ struct sugov_cpu { unsigned long util; unsigned long bw_min; + unsigned long bw_max; /* The field below is for single-CPU policies only: */ #ifdef CONFIG_NO_HZ_COMMON @@ -232,6 +233,7 @@ static void sugov_get_util(struct sugov_ util = effective_cpu_util(sg_cpu->cpu, util, &min, &max); util = max(util, boost); sg_cpu->bw_min = min; + sg_cpu->bw_max = max; sg_cpu->util = sugov_effective_cpu_perf(sg_cpu->cpu, util, min, max); } @@ -484,7 +486,7 @@ static void sugov_update_single_perf(str sg_cpu->util = prev_util; cpufreq_driver_adjust_perf(sg_policy->policy, sg_cpu->bw_min, - sg_cpu->util, max_cap); + sg_cpu->util, sg_cpu->bw_max, max_cap); sg_policy->last_freq_update_time = time; } --- a/rust/kernel/cpufreq.rs +++ b/rust/kernel/cpufreq.rs @@ -792,7 +792,8 @@ pub trait Driver { } /// Driver's `adjust_perf` callback. - fn adjust_perf(_policy: &mut Policy, _min_perf: usize, _target_perf: usize, _capacity: usize) { + fn adjust_perf(_policy: &mut Policy, _min_perf: usize, _target_perf: usize, + _max_perf: usize, _capacity: usize) { build_error!(VTABLE_DEFAULT_ERROR) } @@ -1262,12 +1263,13 @@ impl Registration { ptr: *mut bindings::cpufreq_policy, min_perf: c_ulong, target_perf: c_ulong, + max_perf: c_ulong, capacity: c_ulong, ) { // SAFETY: The `ptr` is guaranteed to be valid by the contract with the C code for the // lifetime of `policy`. let policy = unsafe { Policy::from_raw_mut(ptr) }; - T::adjust_perf(policy, min_perf, target_perf, capacity); + T::adjust_perf(policy, min_perf, target_perf, max_perf, capacity); } /// Driver's `get_intermediate` callback.