From patchwork Wed Jul 10 08:28:03 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Sergiy Kibrik X-Patchwork-Id: 13728998 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from lists.xenproject.org (lists.xenproject.org [192.237.175.120]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 37E90C3DA42 for ; Wed, 10 Jul 2024 08:28:31 +0000 (UTC) Received: from list by lists.xenproject.org with outflank-mailman.756612.1165274 (Exim 4.92) (envelope-from ) id 1sRSgX-0007tO-7F; Wed, 10 Jul 2024 08:28:21 +0000 X-Outflank-Mailman: Message body and most headers restored to incoming version Received: by outflank-mailman (output) from mailman id 756612.1165274; Wed, 10 Jul 2024 08:28:21 +0000 Received: from localhost ([127.0.0.1] helo=lists.xenproject.org) by lists.xenproject.org with esmtp (Exim 4.92) (envelope-from ) id 1sRSgX-0007tH-4a; Wed, 10 Jul 2024 08:28:21 +0000 Received: by outflank-mailman (input) for mailman id 756612; Wed, 10 Jul 2024 08:28:19 +0000 Received: from se1-gles-sth1-in.inumbo.com ([159.253.27.254] helo=se1-gles-sth1.inumbo.com) by lists.xenproject.org with esmtp (Exim 4.92) (envelope-from ) id 1sRSgV-0007t9-K5 for xen-devel@lists.xenproject.org; Wed, 10 Jul 2024 08:28:19 +0000 Received: from pb-smtp20.pobox.com (pb-smtp20.pobox.com [173.228.157.52]) by se1-gles-sth1.inumbo.com (Halon) with ESMTPS id 5941cc69-3e96-11ef-bbfb-fd08da9f4363; Wed, 10 Jul 2024 10:28:16 +0200 (CEST) Received: from pb-smtp20.pobox.com (unknown [127.0.0.1]) by pb-smtp20.pobox.com (Postfix) with ESMTP id F1B891842A; Wed, 10 Jul 2024 04:28:12 -0400 (EDT) (envelope-from sakib@darkstar.site) Received: from pb-smtp20.sea.icgroup.com (unknown [127.0.0.1]) by pb-smtp20.pobox.com (Postfix) with ESMTP id E9C9018429; Wed, 10 Jul 2024 04:28:12 -0400 (EDT) (envelope-from sakib@darkstar.site) Received: from localhost (unknown [185.130.54.75]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by pb-smtp20.pobox.com (Postfix) with ESMTPSA id DB51D18428; Wed, 10 Jul 2024 04:28:07 -0400 (EDT) (envelope-from sakib@darkstar.site) X-BeenThere: xen-devel@lists.xenproject.org List-Id: Xen developer discussion List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: xen-devel-bounces@lists.xenproject.org Precedence: list Sender: "Xen-devel" X-Inumbo-ID: 5941cc69-3e96-11ef-bbfb-fd08da9f4363 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed; d=pobox.com; h=from:to:cc :subject:date:message-id:in-reply-to:references:mime-version :content-transfer-encoding; s=sasl; bh=CC1ePQ9OVnZduDauSGZBGNh2o v6yI4vWznWUL0c1qAw=; b=D391T4UVsO9uPS2TPUW3fbQCuioSJGnlLv8hBIKWt MjNzTJkCC8rlzKPOiVz25OTjIbo/FpZaowVju2mqef6EOoRCF9ietuz4EDQz5vZd U2u2SJA7nqMe9G/BBY/z/EsEs5Lh0dKVA8lAuJY6Vx7BCeIulZZCC0iEltxQxhYC gQ= From: Sergiy Kibrik To: xen-devel@lists.xenproject.org Cc: Sergiy Kibrik , Jan Beulich , Andrew Cooper , =?utf-8?q?Roger_Pau_Monn=C3=A9?= , Stefano Stabellini Subject: [XEN PATCH v3 1/2] x86/cpufreq: move ACPI cpufreq driver into separate file Date: Wed, 10 Jul 2024 11:28:03 +0300 Message-Id: X-Mailer: git-send-email 2.25.1 In-Reply-To: References: MIME-Version: 1.0 X-Pobox-Relay-ID: 55E95F96-3E96-11EF-BA7E-C38742FD603B-90055647!pb-smtp20.pobox.com Separate ACPI driver from generic initialization cpufreq code. This way acpi-cpufreq can become optional in the future and be disabled from non-Intel builds. no changes to code were introduced, except: acpi_cpufreq_register() helper added clean up a list of included headers license transformed into an SPDX line Signed-off-by: Sergiy Kibrik Reviewed-by: Jan Beulich --- changes in v3: - do not move generic get_measured_perf() & related code - license transformed into an SPDX line --- xen/arch/x86/acpi/cpufreq/Makefile | 1 + xen/arch/x86/acpi/cpufreq/acpi.c | 518 ++++++++++++++++++++++++++++ xen/arch/x86/acpi/cpufreq/cpufreq.c | 506 +-------------------------- xen/include/acpi/cpufreq/cpufreq.h | 2 + 4 files changed, 523 insertions(+), 504 deletions(-) create mode 100644 xen/arch/x86/acpi/cpufreq/acpi.c diff --git a/xen/arch/x86/acpi/cpufreq/Makefile b/xen/arch/x86/acpi/cpufreq/Makefile index db83aa6b14..44d4c0b497 100644 --- a/xen/arch/x86/acpi/cpufreq/Makefile +++ b/xen/arch/x86/acpi/cpufreq/Makefile @@ -1,3 +1,4 @@ +obj-y += acpi.o obj-y += cpufreq.o obj-y += hwp.o obj-y += powernow.o diff --git a/xen/arch/x86/acpi/cpufreq/acpi.c b/xen/arch/x86/acpi/cpufreq/acpi.c new file mode 100644 index 0000000000..512ada3ead --- /dev/null +++ b/xen/arch/x86/acpi/cpufreq/acpi.c @@ -0,0 +1,518 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * cpufreq.c - ACPI Processor P-States Driver ($Revision: 1.4 $) + * + * Copyright (C) 2001, 2002 Andy Grover + * Copyright (C) 2001, 2002 Paul Diefenbaugh + * Copyright (C) 2002 - 2004 Dominik Brodowski + * Copyright (C) 2006 Denis Sadykov + * + * Feb 2008 - Liu Jinsong + * porting acpi-cpufreq.c from Linux 2.6.23 to Xen hypervisor + */ + +#include +#include +#include +#include +#include +#include + + +enum { + UNDEFINED_CAPABLE = 0, + SYSTEM_INTEL_MSR_CAPABLE, + SYSTEM_IO_CAPABLE, +}; + +#define INTEL_MSR_RANGE (0xffffull) + +static bool __read_mostly acpi_pstate_strict; +boolean_param("acpi_pstate_strict", acpi_pstate_strict); + +static unsigned extract_io(u32 value, struct acpi_cpufreq_data *data) +{ + struct processor_performance *perf; + int i; + + perf = data->acpi_data; + + for (i=0; istate_count; i++) { + if (value == perf->states[i].status) + return data->freq_table[i].frequency; + } + return 0; +} + +static unsigned extract_msr(u32 msr, struct acpi_cpufreq_data *data) +{ + int i; + struct processor_performance *perf; + + msr &= INTEL_MSR_RANGE; + perf = data->acpi_data; + + for (i=0; data->freq_table[i].frequency != CPUFREQ_TABLE_END; i++) { + if (msr == perf->states[data->freq_table[i].index].status) + return data->freq_table[i].frequency; + } + return data->freq_table[0].frequency; +} + +static unsigned extract_freq(u32 val, struct acpi_cpufreq_data *data) +{ + switch (data->arch_cpu_flags) { + case SYSTEM_INTEL_MSR_CAPABLE: + return extract_msr(val, data); + case SYSTEM_IO_CAPABLE: + return extract_io(val, data); + default: + return 0; + } +} + +struct msr_addr { + u32 reg; +}; + +struct io_addr { + u16 port; + u8 bit_width; +}; + +typedef union { + struct msr_addr msr; + struct io_addr io; +} drv_addr_union; + +struct drv_cmd { + unsigned int type; + const cpumask_t *mask; + drv_addr_union addr; + u32 val; +}; + +static void cf_check do_drv_read(void *drvcmd) +{ + struct drv_cmd *cmd; + + cmd = (struct drv_cmd *)drvcmd; + + switch (cmd->type) { + case SYSTEM_INTEL_MSR_CAPABLE: + rdmsrl(cmd->addr.msr.reg, cmd->val); + break; + case SYSTEM_IO_CAPABLE: + acpi_os_read_port((acpi_io_address)cmd->addr.io.port, + &cmd->val, (u32)cmd->addr.io.bit_width); + break; + default: + break; + } +} + +static void cf_check do_drv_write(void *drvcmd) +{ + struct drv_cmd *cmd; + uint64_t msr_content; + + cmd = (struct drv_cmd *)drvcmd; + + switch (cmd->type) { + case SYSTEM_INTEL_MSR_CAPABLE: + rdmsrl(cmd->addr.msr.reg, msr_content); + msr_content = (msr_content & ~INTEL_MSR_RANGE) + | (cmd->val & INTEL_MSR_RANGE); + wrmsrl(cmd->addr.msr.reg, msr_content); + break; + case SYSTEM_IO_CAPABLE: + acpi_os_write_port((acpi_io_address)cmd->addr.io.port, + cmd->val, (u32)cmd->addr.io.bit_width); + break; + default: + break; + } +} + +static void drv_read(struct drv_cmd *cmd) +{ + cmd->val = 0; + + ASSERT(cpumask_weight(cmd->mask) == 1); + + /* to reduce IPI for the sake of performance */ + if (likely(cpumask_test_cpu(smp_processor_id(), cmd->mask))) + do_drv_read((void *)cmd); + else + on_selected_cpus(cmd->mask, do_drv_read, cmd, 1); +} + +static void drv_write(struct drv_cmd *cmd) +{ + if (cpumask_equal(cmd->mask, cpumask_of(smp_processor_id()))) + do_drv_write((void *)cmd); + else + on_selected_cpus(cmd->mask, do_drv_write, cmd, 1); +} + +static u32 get_cur_val(const cpumask_t *mask) +{ + struct cpufreq_policy *policy; + struct processor_performance *perf; + struct drv_cmd cmd; + unsigned int cpu = smp_processor_id(); + + if (unlikely(cpumask_empty(mask))) + return 0; + + if (!cpumask_test_cpu(cpu, mask)) + cpu = cpumask_first(mask); + if (cpu >= nr_cpu_ids || !cpu_online(cpu)) + return 0; + + policy = per_cpu(cpufreq_cpu_policy, cpu); + if (!policy || !cpufreq_drv_data[policy->cpu]) + return 0; + + switch (cpufreq_drv_data[policy->cpu]->arch_cpu_flags) { + case SYSTEM_INTEL_MSR_CAPABLE: + cmd.type = SYSTEM_INTEL_MSR_CAPABLE; + cmd.addr.msr.reg = MSR_IA32_PERF_STATUS; + break; + case SYSTEM_IO_CAPABLE: + cmd.type = SYSTEM_IO_CAPABLE; + perf = cpufreq_drv_data[policy->cpu]->acpi_data; + cmd.addr.io.port = perf->control_register.address; + cmd.addr.io.bit_width = perf->control_register.bit_width; + break; + default: + return 0; + } + + cmd.mask = cpumask_of(cpu); + + drv_read(&cmd); + return cmd.val; +} + +static unsigned int cf_check get_cur_freq_on_cpu(unsigned int cpu) +{ + struct cpufreq_policy *policy; + struct acpi_cpufreq_data *data; + + if (!cpu_online(cpu)) + return 0; + + policy = per_cpu(cpufreq_cpu_policy, cpu); + if (!policy) + return 0; + + data = cpufreq_drv_data[policy->cpu]; + if (unlikely(data == NULL || + data->acpi_data == NULL || data->freq_table == NULL)) + return 0; + + return extract_freq(get_cur_val(cpumask_of(cpu)), data); +} + +void intel_feature_detect(struct cpufreq_policy *policy) +{ + unsigned int eax; + + eax = cpuid_eax(6); + if (eax & 0x2) { + policy->turbo = CPUFREQ_TURBO_ENABLED; + if (cpufreq_verbose) + printk(XENLOG_INFO "CPU%u: Turbo Mode detected and enabled\n", + smp_processor_id()); + } +} + +static void cf_check feature_detect(void *info) +{ + intel_feature_detect(info); +} + +static unsigned int check_freqs(const cpumask_t *mask, unsigned int freq, + struct acpi_cpufreq_data *data) +{ + unsigned int cur_freq; + unsigned int i; + + for (i=0; i<100; i++) { + cur_freq = extract_freq(get_cur_val(mask), data); + if (cur_freq == freq) + return 1; + udelay(10); + } + return 0; +} + +static int cf_check acpi_cpufreq_target( + struct cpufreq_policy *policy, + unsigned int target_freq, unsigned int relation) +{ + struct acpi_cpufreq_data *data = cpufreq_drv_data[policy->cpu]; + struct processor_performance *perf; + struct cpufreq_freqs freqs; + cpumask_t online_policy_cpus; + struct drv_cmd cmd; + unsigned int next_state = 0; /* Index into freq_table */ + unsigned int next_perf_state = 0; /* Index into perf table */ + unsigned int j; + int result = 0; + + if (unlikely(data == NULL || + data->acpi_data == NULL || data->freq_table == NULL)) { + return -ENODEV; + } + + if (policy->turbo == CPUFREQ_TURBO_DISABLED) + if (target_freq > policy->cpuinfo.second_max_freq) + target_freq = policy->cpuinfo.second_max_freq; + + perf = data->acpi_data; + result = cpufreq_frequency_table_target(policy, + data->freq_table, + target_freq, + relation, &next_state); + if (unlikely(result)) + return -ENODEV; + + cpumask_and(&online_policy_cpus, &cpu_online_map, policy->cpus); + + next_perf_state = data->freq_table[next_state].index; + if (perf->state == next_perf_state) { + if (unlikely(policy->resume)) + policy->resume = 0; + else + return 0; + } + + switch (data->arch_cpu_flags) { + case SYSTEM_INTEL_MSR_CAPABLE: + cmd.type = SYSTEM_INTEL_MSR_CAPABLE; + cmd.addr.msr.reg = MSR_IA32_PERF_CTL; + cmd.val = (u32) perf->states[next_perf_state].control; + break; + case SYSTEM_IO_CAPABLE: + cmd.type = SYSTEM_IO_CAPABLE; + cmd.addr.io.port = perf->control_register.address; + cmd.addr.io.bit_width = perf->control_register.bit_width; + cmd.val = (u32) perf->states[next_perf_state].control; + break; + default: + return -ENODEV; + } + + if (policy->shared_type != CPUFREQ_SHARED_TYPE_ANY) + cmd.mask = &online_policy_cpus; + else + cmd.mask = cpumask_of(policy->cpu); + + freqs.old = perf->states[perf->state].core_frequency * 1000; + freqs.new = data->freq_table[next_state].frequency; + + drv_write(&cmd); + + if (acpi_pstate_strict && !check_freqs(cmd.mask, freqs.new, data)) { + printk(KERN_WARNING "Fail transfer to new freq %d\n", freqs.new); + return -EAGAIN; + } + + for_each_cpu(j, &online_policy_cpus) + cpufreq_statistic_update(j, perf->state, next_perf_state); + + perf->state = next_perf_state; + policy->cur = freqs.new; + + return result; +} + +static int cf_check acpi_cpufreq_verify(struct cpufreq_policy *policy) +{ + struct acpi_cpufreq_data *data; + struct processor_performance *perf; + + if (!policy || !(data = cpufreq_drv_data[policy->cpu]) || + !processor_pminfo[policy->cpu]) + return -EINVAL; + + perf = &processor_pminfo[policy->cpu]->perf; + + cpufreq_verify_within_limits(policy, 0, + perf->states[perf->platform_limit].core_frequency * 1000); + + return cpufreq_frequency_table_verify(policy, data->freq_table); +} + +static unsigned long +acpi_cpufreq_guess_freq(struct acpi_cpufreq_data *data, unsigned int cpu) +{ + struct processor_performance *perf = data->acpi_data; + + if (cpu_khz) { + /* search the closest match to cpu_khz */ + unsigned int i; + unsigned long freq; + unsigned long freqn = perf->states[0].core_frequency * 1000; + + for (i=0; i<(perf->state_count-1); i++) { + freq = freqn; + freqn = perf->states[i+1].core_frequency * 1000; + if ((2 * cpu_khz) > (freqn + freq)) { + perf->state = i; + return freq; + } + } + perf->state = perf->state_count-1; + return freqn; + } else { + /* assume CPU is at P0... */ + perf->state = 0; + return perf->states[0].core_frequency * 1000; + } +} + +static int cf_check acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) +{ + unsigned int i; + unsigned int valid_states = 0; + unsigned int cpu = policy->cpu; + struct acpi_cpufreq_data *data; + unsigned int result = 0; + struct cpuinfo_x86 *c = &cpu_data[policy->cpu]; + struct processor_performance *perf; + + data = xzalloc(struct acpi_cpufreq_data); + if (!data) + return -ENOMEM; + + cpufreq_drv_data[cpu] = data; + + data->acpi_data = &processor_pminfo[cpu]->perf; + + perf = data->acpi_data; + policy->shared_type = perf->shared_type; + + switch (perf->control_register.space_id) { + case ACPI_ADR_SPACE_SYSTEM_IO: + if (cpufreq_verbose) + printk("xen_pminfo: @acpi_cpufreq_cpu_init," + "SYSTEM IO addr space\n"); + data->arch_cpu_flags = SYSTEM_IO_CAPABLE; + break; + case ACPI_ADR_SPACE_FIXED_HARDWARE: + if (cpufreq_verbose) + printk("xen_pminfo: @acpi_cpufreq_cpu_init," + "HARDWARE addr space\n"); + if (!cpu_has(c, X86_FEATURE_EIST)) { + result = -ENODEV; + goto err_unreg; + } + data->arch_cpu_flags = SYSTEM_INTEL_MSR_CAPABLE; + break; + default: + result = -ENODEV; + goto err_unreg; + } + + data->freq_table = xmalloc_array(struct cpufreq_frequency_table, + (perf->state_count+1)); + if (!data->freq_table) { + result = -ENOMEM; + goto err_unreg; + } + + /* detect transition latency */ + policy->cpuinfo.transition_latency = 0; + for (i=0; istate_count; i++) { + if ((perf->states[i].transition_latency * 1000) > + policy->cpuinfo.transition_latency) + policy->cpuinfo.transition_latency = + perf->states[i].transition_latency * 1000; + } + + policy->governor = cpufreq_opt_governor ? : CPUFREQ_DEFAULT_GOVERNOR; + + /* table init */ + for (i=0; istate_count; i++) { + if (i>0 && perf->states[i].core_frequency >= + data->freq_table[valid_states-1].frequency / 1000) + continue; + + data->freq_table[valid_states].index = i; + data->freq_table[valid_states].frequency = + perf->states[i].core_frequency * 1000; + valid_states++; + } + data->freq_table[valid_states].frequency = CPUFREQ_TABLE_END; + perf->state = 0; + + result = cpufreq_frequency_table_cpuinfo(policy, data->freq_table); + if (result) + goto err_freqfree; + + switch (perf->control_register.space_id) { + case ACPI_ADR_SPACE_SYSTEM_IO: + /* Current speed is unknown and not detectable by IO port */ + policy->cur = acpi_cpufreq_guess_freq(data, policy->cpu); + break; + case ACPI_ADR_SPACE_FIXED_HARDWARE: + cpufreq_driver.get = get_cur_freq_on_cpu; + policy->cur = get_cur_freq_on_cpu(cpu); + break; + default: + break; + } + + /* Check for APERF/MPERF support in hardware + * also check for boost support */ + if (c->x86_vendor == X86_VENDOR_INTEL && c->cpuid_level >= 6) + on_selected_cpus(cpumask_of(cpu), feature_detect, policy, 1); + + /* + * the first call to ->target() should result in us actually + * writing something to the appropriate registers. + */ + policy->resume = 1; + + return result; + +err_freqfree: + xfree(data->freq_table); +err_unreg: + xfree(data); + cpufreq_drv_data[cpu] = NULL; + + return result; +} + +static int cf_check acpi_cpufreq_cpu_exit(struct cpufreq_policy *policy) +{ + struct acpi_cpufreq_data *data = cpufreq_drv_data[policy->cpu]; + + if (data) { + cpufreq_drv_data[policy->cpu] = NULL; + xfree(data->freq_table); + xfree(data); + } + + return 0; +} + +static const struct cpufreq_driver __initconst_cf_clobber +acpi_cpufreq_driver = { + .name = "acpi-cpufreq", + .verify = acpi_cpufreq_verify, + .target = acpi_cpufreq_target, + .init = acpi_cpufreq_cpu_init, + .exit = acpi_cpufreq_cpu_exit, + .get = get_cur_freq_on_cpu, +}; + + +int __init acpi_cpufreq_register(void) +{ + return cpufreq_register_driver(&acpi_cpufreq_driver); +} diff --git a/xen/arch/x86/acpi/cpufreq/cpufreq.c b/xen/arch/x86/acpi/cpufreq/cpufreq.c index a341f2f020..6244d29496 100644 --- a/xen/arch/x86/acpi/cpufreq/cpufreq.c +++ b/xen/arch/x86/acpi/cpufreq/cpufreq.c @@ -26,200 +26,14 @@ * * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ */ - #include #include -#include -#include #include #include -#include -#include -#include -#include -#include -#include -#include #include -enum { - UNDEFINED_CAPABLE = 0, - SYSTEM_INTEL_MSR_CAPABLE, - SYSTEM_IO_CAPABLE, -}; - -#define INTEL_MSR_RANGE (0xffffull) - struct acpi_cpufreq_data *cpufreq_drv_data[NR_CPUS]; -static bool __read_mostly acpi_pstate_strict; -boolean_param("acpi_pstate_strict", acpi_pstate_strict); - -static unsigned extract_io(u32 value, struct acpi_cpufreq_data *data) -{ - struct processor_performance *perf; - int i; - - perf = data->acpi_data; - - for (i=0; istate_count; i++) { - if (value == perf->states[i].status) - return data->freq_table[i].frequency; - } - return 0; -} - -static unsigned extract_msr(u32 msr, struct acpi_cpufreq_data *data) -{ - int i; - struct processor_performance *perf; - - msr &= INTEL_MSR_RANGE; - perf = data->acpi_data; - - for (i=0; data->freq_table[i].frequency != CPUFREQ_TABLE_END; i++) { - if (msr == perf->states[data->freq_table[i].index].status) - return data->freq_table[i].frequency; - } - return data->freq_table[0].frequency; -} - -static unsigned extract_freq(u32 val, struct acpi_cpufreq_data *data) -{ - switch (data->arch_cpu_flags) { - case SYSTEM_INTEL_MSR_CAPABLE: - return extract_msr(val, data); - case SYSTEM_IO_CAPABLE: - return extract_io(val, data); - default: - return 0; - } -} - -struct msr_addr { - u32 reg; -}; - -struct io_addr { - u16 port; - u8 bit_width; -}; - -typedef union { - struct msr_addr msr; - struct io_addr io; -} drv_addr_union; - -struct drv_cmd { - unsigned int type; - const cpumask_t *mask; - drv_addr_union addr; - u32 val; -}; - -static void cf_check do_drv_read(void *drvcmd) -{ - struct drv_cmd *cmd; - - cmd = (struct drv_cmd *)drvcmd; - - switch (cmd->type) { - case SYSTEM_INTEL_MSR_CAPABLE: - rdmsrl(cmd->addr.msr.reg, cmd->val); - break; - case SYSTEM_IO_CAPABLE: - acpi_os_read_port((acpi_io_address)cmd->addr.io.port, - &cmd->val, (u32)cmd->addr.io.bit_width); - break; - default: - break; - } -} - -static void cf_check do_drv_write(void *drvcmd) -{ - struct drv_cmd *cmd; - uint64_t msr_content; - - cmd = (struct drv_cmd *)drvcmd; - - switch (cmd->type) { - case SYSTEM_INTEL_MSR_CAPABLE: - rdmsrl(cmd->addr.msr.reg, msr_content); - msr_content = (msr_content & ~INTEL_MSR_RANGE) - | (cmd->val & INTEL_MSR_RANGE); - wrmsrl(cmd->addr.msr.reg, msr_content); - break; - case SYSTEM_IO_CAPABLE: - acpi_os_write_port((acpi_io_address)cmd->addr.io.port, - cmd->val, (u32)cmd->addr.io.bit_width); - break; - default: - break; - } -} - -static void drv_read(struct drv_cmd *cmd) -{ - cmd->val = 0; - - ASSERT(cpumask_weight(cmd->mask) == 1); - - /* to reduce IPI for the sake of performance */ - if (likely(cpumask_test_cpu(smp_processor_id(), cmd->mask))) - do_drv_read((void *)cmd); - else - on_selected_cpus(cmd->mask, do_drv_read, cmd, 1); -} - -static void drv_write(struct drv_cmd *cmd) -{ - if (cpumask_equal(cmd->mask, cpumask_of(smp_processor_id()))) - do_drv_write((void *)cmd); - else - on_selected_cpus(cmd->mask, do_drv_write, cmd, 1); -} - -static u32 get_cur_val(const cpumask_t *mask) -{ - struct cpufreq_policy *policy; - struct processor_performance *perf; - struct drv_cmd cmd; - unsigned int cpu = smp_processor_id(); - - if (unlikely(cpumask_empty(mask))) - return 0; - - if (!cpumask_test_cpu(cpu, mask)) - cpu = cpumask_first(mask); - if (cpu >= nr_cpu_ids || !cpu_online(cpu)) - return 0; - - policy = per_cpu(cpufreq_cpu_policy, cpu); - if (!policy || !cpufreq_drv_data[policy->cpu]) - return 0; - - switch (cpufreq_drv_data[policy->cpu]->arch_cpu_flags) { - case SYSTEM_INTEL_MSR_CAPABLE: - cmd.type = SYSTEM_INTEL_MSR_CAPABLE; - cmd.addr.msr.reg = MSR_IA32_PERF_STATUS; - break; - case SYSTEM_IO_CAPABLE: - cmd.type = SYSTEM_IO_CAPABLE; - perf = cpufreq_drv_data[policy->cpu]->acpi_data; - cmd.addr.io.port = perf->control_register.address; - cmd.addr.io.bit_width = perf->control_register.bit_width; - break; - default: - return 0; - } - - cmd.mask = cpumask_of(cpu); - - drv_read(&cmd); - return cmd.val; -} - struct perf_pair { union { struct { @@ -255,7 +69,7 @@ static void cf_check read_measured_perf_ctrs(void *_readin) */ unsigned int get_measured_perf(unsigned int cpu, unsigned int flag) { - struct cpufreq_policy *policy; + struct cpufreq_policy *policy; struct perf_pair readin, cur, *saved; unsigned int perf_percent; @@ -308,322 +122,6 @@ unsigned int get_measured_perf(unsigned int cpu, unsigned int flag) return policy->cpuinfo.perf_freq * perf_percent / 100; } -static unsigned int cf_check get_cur_freq_on_cpu(unsigned int cpu) -{ - struct cpufreq_policy *policy; - struct acpi_cpufreq_data *data; - - if (!cpu_online(cpu)) - return 0; - - policy = per_cpu(cpufreq_cpu_policy, cpu); - if (!policy) - return 0; - - data = cpufreq_drv_data[policy->cpu]; - if (unlikely(data == NULL || - data->acpi_data == NULL || data->freq_table == NULL)) - return 0; - - return extract_freq(get_cur_val(cpumask_of(cpu)), data); -} - -void intel_feature_detect(struct cpufreq_policy *policy) -{ - unsigned int eax; - - eax = cpuid_eax(6); - if (eax & 0x2) { - policy->turbo = CPUFREQ_TURBO_ENABLED; - if (cpufreq_verbose) - printk(XENLOG_INFO "CPU%u: Turbo Mode detected and enabled\n", - smp_processor_id()); - } -} - -static void cf_check feature_detect(void *info) -{ - intel_feature_detect(info); -} - -static unsigned int check_freqs(const cpumask_t *mask, unsigned int freq, - struct acpi_cpufreq_data *data) -{ - unsigned int cur_freq; - unsigned int i; - - for (i=0; i<100; i++) { - cur_freq = extract_freq(get_cur_val(mask), data); - if (cur_freq == freq) - return 1; - udelay(10); - } - return 0; -} - -static int cf_check acpi_cpufreq_target( - struct cpufreq_policy *policy, - unsigned int target_freq, unsigned int relation) -{ - struct acpi_cpufreq_data *data = cpufreq_drv_data[policy->cpu]; - struct processor_performance *perf; - struct cpufreq_freqs freqs; - cpumask_t online_policy_cpus; - struct drv_cmd cmd; - unsigned int next_state = 0; /* Index into freq_table */ - unsigned int next_perf_state = 0; /* Index into perf table */ - unsigned int j; - int result = 0; - - if (unlikely(data == NULL || - data->acpi_data == NULL || data->freq_table == NULL)) { - return -ENODEV; - } - - if (policy->turbo == CPUFREQ_TURBO_DISABLED) - if (target_freq > policy->cpuinfo.second_max_freq) - target_freq = policy->cpuinfo.second_max_freq; - - perf = data->acpi_data; - result = cpufreq_frequency_table_target(policy, - data->freq_table, - target_freq, - relation, &next_state); - if (unlikely(result)) - return -ENODEV; - - cpumask_and(&online_policy_cpus, &cpu_online_map, policy->cpus); - - next_perf_state = data->freq_table[next_state].index; - if (perf->state == next_perf_state) { - if (unlikely(policy->resume)) - policy->resume = 0; - else - return 0; - } - - switch (data->arch_cpu_flags) { - case SYSTEM_INTEL_MSR_CAPABLE: - cmd.type = SYSTEM_INTEL_MSR_CAPABLE; - cmd.addr.msr.reg = MSR_IA32_PERF_CTL; - cmd.val = (u32) perf->states[next_perf_state].control; - break; - case SYSTEM_IO_CAPABLE: - cmd.type = SYSTEM_IO_CAPABLE; - cmd.addr.io.port = perf->control_register.address; - cmd.addr.io.bit_width = perf->control_register.bit_width; - cmd.val = (u32) perf->states[next_perf_state].control; - break; - default: - return -ENODEV; - } - - if (policy->shared_type != CPUFREQ_SHARED_TYPE_ANY) - cmd.mask = &online_policy_cpus; - else - cmd.mask = cpumask_of(policy->cpu); - - freqs.old = perf->states[perf->state].core_frequency * 1000; - freqs.new = data->freq_table[next_state].frequency; - - drv_write(&cmd); - - if (acpi_pstate_strict && !check_freqs(cmd.mask, freqs.new, data)) { - printk(KERN_WARNING "Fail transfer to new freq %d\n", freqs.new); - return -EAGAIN; - } - - for_each_cpu(j, &online_policy_cpus) - cpufreq_statistic_update(j, perf->state, next_perf_state); - - perf->state = next_perf_state; - policy->cur = freqs.new; - - return result; -} - -static int cf_check acpi_cpufreq_verify(struct cpufreq_policy *policy) -{ - struct acpi_cpufreq_data *data; - struct processor_performance *perf; - - if (!policy || !(data = cpufreq_drv_data[policy->cpu]) || - !processor_pminfo[policy->cpu]) - return -EINVAL; - - perf = &processor_pminfo[policy->cpu]->perf; - - cpufreq_verify_within_limits(policy, 0, - perf->states[perf->platform_limit].core_frequency * 1000); - - return cpufreq_frequency_table_verify(policy, data->freq_table); -} - -static unsigned long -acpi_cpufreq_guess_freq(struct acpi_cpufreq_data *data, unsigned int cpu) -{ - struct processor_performance *perf = data->acpi_data; - - if (cpu_khz) { - /* search the closest match to cpu_khz */ - unsigned int i; - unsigned long freq; - unsigned long freqn = perf->states[0].core_frequency * 1000; - - for (i=0; i<(perf->state_count-1); i++) { - freq = freqn; - freqn = perf->states[i+1].core_frequency * 1000; - if ((2 * cpu_khz) > (freqn + freq)) { - perf->state = i; - return freq; - } - } - perf->state = perf->state_count-1; - return freqn; - } else { - /* assume CPU is at P0... */ - perf->state = 0; - return perf->states[0].core_frequency * 1000; - } -} - -static int cf_check acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) -{ - unsigned int i; - unsigned int valid_states = 0; - unsigned int cpu = policy->cpu; - struct acpi_cpufreq_data *data; - unsigned int result = 0; - struct cpuinfo_x86 *c = &cpu_data[policy->cpu]; - struct processor_performance *perf; - - data = xzalloc(struct acpi_cpufreq_data); - if (!data) - return -ENOMEM; - - cpufreq_drv_data[cpu] = data; - - data->acpi_data = &processor_pminfo[cpu]->perf; - - perf = data->acpi_data; - policy->shared_type = perf->shared_type; - - switch (perf->control_register.space_id) { - case ACPI_ADR_SPACE_SYSTEM_IO: - if (cpufreq_verbose) - printk("xen_pminfo: @acpi_cpufreq_cpu_init," - "SYSTEM IO addr space\n"); - data->arch_cpu_flags = SYSTEM_IO_CAPABLE; - break; - case ACPI_ADR_SPACE_FIXED_HARDWARE: - if (cpufreq_verbose) - printk("xen_pminfo: @acpi_cpufreq_cpu_init," - "HARDWARE addr space\n"); - if (!cpu_has(c, X86_FEATURE_EIST)) { - result = -ENODEV; - goto err_unreg; - } - data->arch_cpu_flags = SYSTEM_INTEL_MSR_CAPABLE; - break; - default: - result = -ENODEV; - goto err_unreg; - } - - data->freq_table = xmalloc_array(struct cpufreq_frequency_table, - (perf->state_count+1)); - if (!data->freq_table) { - result = -ENOMEM; - goto err_unreg; - } - - /* detect transition latency */ - policy->cpuinfo.transition_latency = 0; - for (i=0; istate_count; i++) { - if ((perf->states[i].transition_latency * 1000) > - policy->cpuinfo.transition_latency) - policy->cpuinfo.transition_latency = - perf->states[i].transition_latency * 1000; - } - - policy->governor = cpufreq_opt_governor ? : CPUFREQ_DEFAULT_GOVERNOR; - - /* table init */ - for (i=0; istate_count; i++) { - if (i>0 && perf->states[i].core_frequency >= - data->freq_table[valid_states-1].frequency / 1000) - continue; - - data->freq_table[valid_states].index = i; - data->freq_table[valid_states].frequency = - perf->states[i].core_frequency * 1000; - valid_states++; - } - data->freq_table[valid_states].frequency = CPUFREQ_TABLE_END; - perf->state = 0; - - result = cpufreq_frequency_table_cpuinfo(policy, data->freq_table); - if (result) - goto err_freqfree; - - switch (perf->control_register.space_id) { - case ACPI_ADR_SPACE_SYSTEM_IO: - /* Current speed is unknown and not detectable by IO port */ - policy->cur = acpi_cpufreq_guess_freq(data, policy->cpu); - break; - case ACPI_ADR_SPACE_FIXED_HARDWARE: - cpufreq_driver.get = get_cur_freq_on_cpu; - policy->cur = get_cur_freq_on_cpu(cpu); - break; - default: - break; - } - - /* Check for APERF/MPERF support in hardware - * also check for boost support */ - if (c->x86_vendor == X86_VENDOR_INTEL && c->cpuid_level >= 6) - on_selected_cpus(cpumask_of(cpu), feature_detect, policy, 1); - - /* - * the first call to ->target() should result in us actually - * writing something to the appropriate registers. - */ - policy->resume = 1; - - return result; - -err_freqfree: - xfree(data->freq_table); -err_unreg: - xfree(data); - cpufreq_drv_data[cpu] = NULL; - - return result; -} - -static int cf_check acpi_cpufreq_cpu_exit(struct cpufreq_policy *policy) -{ - struct acpi_cpufreq_data *data = cpufreq_drv_data[policy->cpu]; - - if (data) { - cpufreq_drv_data[policy->cpu] = NULL; - xfree(data->freq_table); - xfree(data); - } - - return 0; -} - -static const struct cpufreq_driver __initconst_cf_clobber -acpi_cpufreq_driver = { - .name = "acpi-cpufreq", - .verify = acpi_cpufreq_verify, - .target = acpi_cpufreq_target, - .init = acpi_cpufreq_cpu_init, - .exit = acpi_cpufreq_cpu_exit, - .get = get_cur_freq_on_cpu, -}; - static int __init cf_check cpufreq_driver_init(void) { int ret = 0; @@ -640,7 +138,7 @@ static int __init cf_check cpufreq_driver_init(void) switch ( cpufreq_xen_opts[i] ) { case CPUFREQ_xen: - ret = cpufreq_register_driver(&acpi_cpufreq_driver); + ret = acpi_cpufreq_register(); break; case CPUFREQ_hwp: ret = hwp_register_driver(); diff --git a/xen/include/acpi/cpufreq/cpufreq.h b/xen/include/acpi/cpufreq/cpufreq.h index 443427153b..ec7e139000 100644 --- a/xen/include/acpi/cpufreq/cpufreq.h +++ b/xen/include/acpi/cpufreq/cpufreq.h @@ -260,4 +260,6 @@ int get_hwp_para(unsigned int cpu, int set_hwp_para(struct cpufreq_policy *policy, struct xen_set_cppc_para *set_cppc); +int acpi_cpufreq_register(void); + #endif /* __XEN_CPUFREQ_PM_H__ */ From patchwork Wed Jul 10 08:30:11 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Sergiy Kibrik X-Patchwork-Id: 13728999 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from lists.xenproject.org (lists.xenproject.org [192.237.175.120]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id D1631C3DA42 for ; Wed, 10 Jul 2024 08:30:34 +0000 (UTC) Received: from list by lists.xenproject.org with outflank-mailman.756618.1165284 (Exim 4.92) (envelope-from ) id 1sRSiW-0000wT-N2; Wed, 10 Jul 2024 08:30:24 +0000 X-Outflank-Mailman: Message body and most headers restored to incoming version Received: by outflank-mailman (output) from mailman id 756618.1165284; Wed, 10 Jul 2024 08:30:24 +0000 Received: from localhost ([127.0.0.1] helo=lists.xenproject.org) by lists.xenproject.org with esmtp (Exim 4.92) (envelope-from ) id 1sRSiW-0000wM-K6; Wed, 10 Jul 2024 08:30:24 +0000 Received: by outflank-mailman (input) for mailman id 756618; Wed, 10 Jul 2024 08:30:23 +0000 Received: from se1-gles-flk1-in.inumbo.com ([94.247.172.50] helo=se1-gles-flk1.inumbo.com) by lists.xenproject.org with esmtp (Exim 4.92) (envelope-from ) id 1sRSiV-0000wE-4Y for xen-devel@lists.xenproject.org; Wed, 10 Jul 2024 08:30:23 +0000 Received: from pb-smtp20.pobox.com (pb-smtp20.pobox.com [173.228.157.52]) by se1-gles-flk1.inumbo.com (Halon) with ESMTPS id a463428b-3e96-11ef-8776-851b0ebba9a2; Wed, 10 Jul 2024 10:30:20 +0200 (CEST) Received: from pb-smtp20.pobox.com (unknown [127.0.0.1]) by pb-smtp20.pobox.com (Postfix) with ESMTP id 15BF7184AB; Wed, 10 Jul 2024 04:30:19 -0400 (EDT) (envelope-from sakib@darkstar.site) Received: from pb-smtp20.sea.icgroup.com (unknown [127.0.0.1]) by pb-smtp20.pobox.com (Postfix) with ESMTP id 0CE2B184AA; Wed, 10 Jul 2024 04:30:19 -0400 (EDT) (envelope-from sakib@darkstar.site) Received: from localhost (unknown [185.130.54.75]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by pb-smtp20.pobox.com (Postfix) with ESMTPSA id 7A47F184A8; Wed, 10 Jul 2024 04:30:15 -0400 (EDT) (envelope-from sakib@darkstar.site) X-BeenThere: xen-devel@lists.xenproject.org List-Id: Xen developer discussion List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: xen-devel-bounces@lists.xenproject.org Precedence: list Sender: "Xen-devel" X-Inumbo-ID: a463428b-3e96-11ef-8776-851b0ebba9a2 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed; d=pobox.com; h=from:to:cc :subject:date:message-id:in-reply-to:references:mime-version :content-transfer-encoding; s=sasl; bh=DMyPQcp7MVx9wCHWMI7ShggWQ RVoCrJB3hO/uCOqu1U=; b=lNBDLzpOPwS+C46MRgnyrwTYglnSYE1EDn4tGzaWa 5VCs5+wPGTS9MQwUW1ZRnEUcrrn59AU14ohx22Nko4LKarDuE2RT2uMIlWiBzZqr GEUMI0YXUeIpTzUw5qlZuZ3EpLsK8/KiiiHrynjL83yppUD66ASzNsYTPDU5s0eZ j0= From: Sergiy Kibrik To: xen-devel@lists.xenproject.org Cc: Sergiy Kibrik , Andrew Cooper , =?utf-8?q?Roger_Pau_Monn=C3=A9?= , Stefano Stabellini , Jason Andryuk , Jan Beulich Subject: [XEN PATCH v3 2/2] x86/cpufreq: separate powernow/hwp/acpi cpufreq code Date: Wed, 10 Jul 2024 11:30:11 +0300 Message-Id: <56f61a1e22ba77fb352d7a18203935c26c815646.1720596402.git.Sergiy_Kibrik@epam.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: References: MIME-Version: 1.0 X-Pobox-Relay-ID: A1F7A6F4-3E96-11EF-A1CE-C38742FD603B-90055647!pb-smtp20.pobox.com Build AMD Architectural P-state driver when CONFIG_AMD is on, and Intel Hardware P-States driver together with ACPI Processor P-States driver when CONFIG_INTEL is on respectively, allowing for a platform-specific build. Signed-off-by: Sergiy Kibrik CC: Jason Andryuk CC: Jan Beulich Reviewed-by: Jason Andryuk Acked-by: Jan Beulich --- changes in v3: - more clear wrapping of lines in conditionals + put IS_ENABLED() in front - drop unneeded change to utility.c changes in v2: - disable acpi-cpufreq driver as well when !INTEL - leave a stub for hwp_active() only when !INTEL - updated patch description --- xen/arch/x86/acpi/cpufreq/Makefile | 6 +++--- xen/arch/x86/acpi/cpufreq/cpufreq.c | 8 +++++--- xen/drivers/acpi/pmstat.c | 5 +++-- xen/drivers/cpufreq/cpufreq.c | 3 ++- xen/include/acpi/cpufreq/cpufreq.h | 8 ++++++++ 5 files changed, 21 insertions(+), 9 deletions(-) diff --git a/xen/arch/x86/acpi/cpufreq/Makefile b/xen/arch/x86/acpi/cpufreq/Makefile index 44d4c0b497..e7dbe434a8 100644 --- a/xen/arch/x86/acpi/cpufreq/Makefile +++ b/xen/arch/x86/acpi/cpufreq/Makefile @@ -1,4 +1,4 @@ -obj-y += acpi.o +obj-$(CONFIG_INTEL) += acpi.o obj-y += cpufreq.o -obj-y += hwp.o -obj-y += powernow.o +obj-$(CONFIG_INTEL) += hwp.o +obj-$(CONFIG_AMD) += powernow.o diff --git a/xen/arch/x86/acpi/cpufreq/cpufreq.c b/xen/arch/x86/acpi/cpufreq/cpufreq.c index 6244d29496..61e98b67bd 100644 --- a/xen/arch/x86/acpi/cpufreq/cpufreq.c +++ b/xen/arch/x86/acpi/cpufreq/cpufreq.c @@ -138,10 +138,12 @@ static int __init cf_check cpufreq_driver_init(void) switch ( cpufreq_xen_opts[i] ) { case CPUFREQ_xen: - ret = acpi_cpufreq_register(); + ret = IS_ENABLED(CONFIG_INTEL) ? + acpi_cpufreq_register() : -ENODEV; break; case CPUFREQ_hwp: - ret = hwp_register_driver(); + ret = IS_ENABLED(CONFIG_INTEL) ? + hwp_register_driver() : -ENODEV; break; case CPUFREQ_none: ret = 0; @@ -155,7 +157,7 @@ static int __init cf_check cpufreq_driver_init(void) case X86_VENDOR_AMD: case X86_VENDOR_HYGON: - ret = powernow_register_driver(); + ret = IS_ENABLED(CONFIG_AMD) ? powernow_register_driver() : -ENODEV; break; } } diff --git a/xen/drivers/acpi/pmstat.c b/xen/drivers/acpi/pmstat.c index 998d2e3c65..df309e27b4 100644 --- a/xen/drivers/acpi/pmstat.c +++ b/xen/drivers/acpi/pmstat.c @@ -254,8 +254,9 @@ static int get_cpufreq_para(struct xen_sysctl_pm_op *op) else strlcpy(op->u.get_para.scaling_driver, "Unknown", CPUFREQ_NAME_LEN); - if ( !strncmp(op->u.get_para.scaling_driver, XEN_HWP_DRIVER_NAME, - CPUFREQ_NAME_LEN) ) + if ( IS_ENABLED(CONFIG_INTEL) && + !strncmp(op->u.get_para.scaling_driver, XEN_HWP_DRIVER_NAME, + CPUFREQ_NAME_LEN) ) ret = get_hwp_para(policy->cpu, &op->u.get_para.u.cppc_para); else { diff --git a/xen/drivers/cpufreq/cpufreq.c b/xen/drivers/cpufreq/cpufreq.c index 8659ad3aee..4a103c6de9 100644 --- a/xen/drivers/cpufreq/cpufreq.c +++ b/xen/drivers/cpufreq/cpufreq.c @@ -120,7 +120,8 @@ static int __init cf_check setup_cpufreq_option(const char *str) if ( arg[0] && arg[1] ) ret = cpufreq_cmdline_parse(arg + 1, end); } - else if ( choice < 0 && !cmdline_strcmp(str, "hwp") ) + else if ( IS_ENABLED(CONFIG_INTEL) && choice < 0 && + !cmdline_strcmp(str, "hwp") ) { xen_processor_pmbits |= XEN_PROCESSOR_PM_PX; cpufreq_controller = FREQCTL_xen; diff --git a/xen/include/acpi/cpufreq/cpufreq.h b/xen/include/acpi/cpufreq/cpufreq.h index ec7e139000..08ee2af0f8 100644 --- a/xen/include/acpi/cpufreq/cpufreq.h +++ b/xen/include/acpi/cpufreq/cpufreq.h @@ -254,7 +254,15 @@ void intel_feature_detect(struct cpufreq_policy *policy); int hwp_cmdline_parse(const char *s, const char *e); int hwp_register_driver(void); +#ifdef CONFIG_INTEL bool hwp_active(void); +#else +static inline bool hwp_active(void) +{ + return false; +} +#endif + int get_hwp_para(unsigned int cpu, struct xen_cppc_para *cppc_para); int set_hwp_para(struct cpufreq_policy *policy,