linux-stable/drivers/cpufreq/vexpress-spc-cpufreq.c

579 lines
14 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/*
* Versatile Express SPC CPUFreq Interface driver
*
* Copyright (C) 2013 - 2019 ARM Ltd.
* Sudeep Holla <sudeep.holla@arm.com>
*
* Copyright (C) 2013 Linaro.
* Viresh Kumar <viresh.kumar@linaro.org>
*/
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/clk.h>
#include <linux/cpu.h>
#include <linux/cpufreq.h>
#include <linux/cpumask.h>
#include <linux/device.h>
#include <linux/module.h>
#include <linux/mutex.h>
#include <linux/platform_device.h>
#include <linux/pm_opp.h>
#include <linux/slab.h>
#include <linux/topology.h>
#include <linux/types.h>
/* Currently we support only two clusters */
#define A15_CLUSTER 0
#define A7_CLUSTER 1
#define MAX_CLUSTERS 2
#ifdef CONFIG_BL_SWITCHER
#include <asm/bL_switcher.h>
static bool bL_switching_enabled;
#define is_bL_switching_enabled() bL_switching_enabled
#define set_switching_enabled(x) (bL_switching_enabled = (x))
#else
#define is_bL_switching_enabled() false
#define set_switching_enabled(x) do { } while (0)
#define bL_switch_request(...) do { } while (0)
#define bL_switcher_put_enabled() do { } while (0)
#define bL_switcher_get_enabled() do { } while (0)
#endif
#define ACTUAL_FREQ(cluster, freq) ((cluster == A7_CLUSTER) ? freq << 1 : freq)
#define VIRT_FREQ(cluster, freq) ((cluster == A7_CLUSTER) ? freq >> 1 : freq)
static struct clk *clk[MAX_CLUSTERS];
static struct cpufreq_frequency_table *freq_table[MAX_CLUSTERS + 1];
static atomic_t cluster_usage[MAX_CLUSTERS + 1];
static unsigned int clk_big_min; /* (Big) clock frequencies */
static unsigned int clk_little_max; /* Maximum clock frequency (Little) */
static DEFINE_PER_CPU(unsigned int, physical_cluster);
static DEFINE_PER_CPU(unsigned int, cpu_last_req_freq);
static struct mutex cluster_lock[MAX_CLUSTERS];
static inline int raw_cpu_to_cluster(int cpu)
{
return topology_physical_package_id(cpu);
}
static inline int cpu_to_cluster(int cpu)
{
return is_bL_switching_enabled() ?
MAX_CLUSTERS : raw_cpu_to_cluster(cpu);
}
static unsigned int find_cluster_maxfreq(int cluster)
{
int j;
u32 max_freq = 0, cpu_freq;
for_each_online_cpu(j) {
cpu_freq = per_cpu(cpu_last_req_freq, j);
if (cluster == per_cpu(physical_cluster, j) &&
max_freq < cpu_freq)
max_freq = cpu_freq;
}
return max_freq;
}
static unsigned int clk_get_cpu_rate(unsigned int cpu)
{
u32 cur_cluster = per_cpu(physical_cluster, cpu);
u32 rate = clk_get_rate(clk[cur_cluster]) / 1000;
/* For switcher we use virtual A7 clock rates */
if (is_bL_switching_enabled())
rate = VIRT_FREQ(cur_cluster, rate);
return rate;
}
static unsigned int ve_spc_cpufreq_get_rate(unsigned int cpu)
{
if (is_bL_switching_enabled())
return per_cpu(cpu_last_req_freq, cpu);
else
return clk_get_cpu_rate(cpu);
}
static unsigned int
ve_spc_cpufreq_set_rate(u32 cpu, u32 old_cluster, u32 new_cluster, u32 rate)
{
u32 new_rate, prev_rate;
int ret;
bool bLs = is_bL_switching_enabled();
mutex_lock(&cluster_lock[new_cluster]);
if (bLs) {
prev_rate = per_cpu(cpu_last_req_freq, cpu);
per_cpu(cpu_last_req_freq, cpu) = rate;
per_cpu(physical_cluster, cpu) = new_cluster;
new_rate = find_cluster_maxfreq(new_cluster);
new_rate = ACTUAL_FREQ(new_cluster, new_rate);
} else {
new_rate = rate;
}
ret = clk_set_rate(clk[new_cluster], new_rate * 1000);
if (!ret) {
/*
* FIXME: clk_set_rate hasn't returned an error here however it
* may be that clk_change_rate failed due to hardware or
* firmware issues and wasn't able to report that due to the
* current design of the clk core layer. To work around this
* problem we will read back the clock rate and check it is
* correct. This needs to be removed once clk core is fixed.
*/
if (clk_get_rate(clk[new_cluster]) != new_rate * 1000)
ret = -EIO;
}
if (WARN_ON(ret)) {
if (bLs) {
per_cpu(cpu_last_req_freq, cpu) = prev_rate;
per_cpu(physical_cluster, cpu) = old_cluster;
}
mutex_unlock(&cluster_lock[new_cluster]);
return ret;
}
mutex_unlock(&cluster_lock[new_cluster]);
/* Recalc freq for old cluster when switching clusters */
if (old_cluster != new_cluster) {
/* Switch cluster */
bL_switch_request(cpu, new_cluster);
mutex_lock(&cluster_lock[old_cluster]);
/* Set freq of old cluster if there are cpus left on it */
new_rate = find_cluster_maxfreq(old_cluster);
new_rate = ACTUAL_FREQ(old_cluster, new_rate);
if (new_rate &&
clk_set_rate(clk[old_cluster], new_rate * 1000)) {
pr_err("%s: clk_set_rate failed: %d, old cluster: %d\n",
__func__, ret, old_cluster);
}
mutex_unlock(&cluster_lock[old_cluster]);
}
return 0;
}
/* Set clock frequency */
static int ve_spc_cpufreq_set_target(struct cpufreq_policy *policy,
unsigned int index)
{
u32 cpu = policy->cpu, cur_cluster, new_cluster, actual_cluster;
unsigned int freqs_new;
cur_cluster = cpu_to_cluster(cpu);
new_cluster = actual_cluster = per_cpu(physical_cluster, cpu);
freqs_new = freq_table[cur_cluster][index].frequency;
if (is_bL_switching_enabled()) {
if (actual_cluster == A15_CLUSTER && freqs_new < clk_big_min)
new_cluster = A7_CLUSTER;
else if (actual_cluster == A7_CLUSTER &&
freqs_new > clk_little_max)
new_cluster = A15_CLUSTER;
}
cpufreq: move invariance setter calls in cpufreq core To properly scale its per-entity load-tracking signals, the task scheduler needs to be given a frequency scale factor, i.e. some image of the current frequency the CPU is running at. Currently, this scale can be computed either by using counters (APERF/MPERF on x86, AMU on arm64), or by piggy-backing on the frequency selection done by cpufreq. For the latter, drivers have to explicitly set the scale factor themselves, despite it being purely boiler-plate code: the required information depends entirely on the kind of frequency switch callback implemented by the driver, i.e. either of: target_index(), target(), fast_switch() and setpolicy(). The fitness of those callbacks with regard to driving the Frequency Invariance Engine (FIE) is studied below: target_index() ============== Documentation states that the chosen frequency "must be determined by freq_table[index].frequency". It isn't clear if it *has* to be that frequency, or if it can use that frequency value to do some computation that ultimately leads to a different frequency selection. All drivers go for the former, while the vexpress-spc-cpufreq has an atypical implementation which is handled separately. Therefore, the hook works on the assumption the core can use freq_table[index].frequency. target() ======= This has been flagged as deprecated since: commit 9c0ebcf78fde ("cpufreq: Implement light weight ->target_index() routine") It also doesn't have that many users: gx-suspmod.c:439: .target = cpufreq_gx_target, s3c24xx-cpufreq.c:428: .target = s3c_cpufreq_target, intel_pstate.c:2528: .target = intel_cpufreq_target, cppc_cpufreq.c:401: .target = cppc_cpufreq_set_target, cpufreq-nforce2.c:371: .target = nforce2_target, sh-cpufreq.c:163: .target = sh_cpufreq_target, pcc-cpufreq.c:573: .target = pcc_cpufreq_target, Similarly to the path taken for target_index() calls in the cpufreq core during a frequency change, all of the drivers above will mark the end of a frequency change by a call to cpufreq_freq_transition_end(). Therefore, cpufreq_freq_transition_end() can be used as the location for the arch_set_freq_scale() call to potentially inform the scheduler of the frequency change. This change maintains the previous functionality for the drivers that implement the target_index() callback, while also adding support for the few drivers that implement the deprecated target() callback. fast_switch() ============= This callback *has* to return the frequency that was selected. setpolicy() =========== This callback does not have any designated way of informing what was the end choice. But there are only two drivers using setpolicy(), and none of them have current FIE support: drivers/cpufreq/longrun.c:281: .setpolicy = longrun_set_policy, drivers/cpufreq/intel_pstate.c:2215: .setpolicy = intel_pstate_set_policy, The intel_pstate is known to use counter-driven frequency invariance. Conclusion ========== Given that the significant majority of current FIE enabled drivers use callbacks that lend themselves to triggering the setting of the FIE scale factor in a generic way, move the invariance setter calls to cpufreq core. As a result of setting the frequency scale factor in cpufreq core, after callbacks that lend themselves to trigger it, remove this functionality from the driver side. To be noted that despite marking a successful frequency change, many cpufreq drivers will consider the new frequency as the requested frequency, although this is might not be the one granted by the hardware. Therefore, the call to arch_set_freq_scale() is a "best effort" one, and it is up to the architecture if the new frequency is used in the new frequency scale factor setting (determined by the implementation of arch_set_freq_scale()) or eventually used by the scheduler (determined by the implementation of arch_scale_freq_capacity()). The architecture is in a better position to decide if it has better methods to obtain more accurate information regarding the current frequency and use that information instead (for example, the use of counters). Also, the implementation to arch_set_freq_scale() will now have to handle error conditions (current frequency == 0) in order to prevent the overhead in cpufreq core when the default arch_set_freq_scale() implementation is used. Signed-off-by: Ionela Voinescu <ionela.voinescu@arm.com> Suggested-by: Valentin Schneider <valentin.schneider@arm.com> Acked-by: Viresh Kumar <viresh.kumar@linaro.org> Acked-by: Sudeep Holla <sudeep.holla@arm.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2020-09-01 20:55:46 +00:00
return ve_spc_cpufreq_set_rate(cpu, actual_cluster, new_cluster,
freqs_new);
}
static inline u32 get_table_count(struct cpufreq_frequency_table *table)
{
int count;
for (count = 0; table[count].frequency != CPUFREQ_TABLE_END; count++)
;
return count;
}
/* get the minimum frequency in the cpufreq_frequency_table */
static inline u32 get_table_min(struct cpufreq_frequency_table *table)
{
struct cpufreq_frequency_table *pos;
u32 min_freq = ~0;
cpufreq_for_each_entry(pos, table)
if (pos->frequency < min_freq)
min_freq = pos->frequency;
return min_freq;
}
/* get the maximum frequency in the cpufreq_frequency_table */
static inline u32 get_table_max(struct cpufreq_frequency_table *table)
{
struct cpufreq_frequency_table *pos;
u32 max_freq = 0;
cpufreq_for_each_entry(pos, table)
if (pos->frequency > max_freq)
max_freq = pos->frequency;
return max_freq;
}
static bool search_frequency(struct cpufreq_frequency_table *table, int size,
unsigned int freq)
{
int count;
for (count = 0; count < size; count++) {
if (table[count].frequency == freq)
return true;
}
return false;
}
static int merge_cluster_tables(void)
{
int i, j, k = 0, count = 1;
struct cpufreq_frequency_table *table;
for (i = 0; i < MAX_CLUSTERS; i++)
count += get_table_count(freq_table[i]);
table = kcalloc(count, sizeof(*table), GFP_KERNEL);
if (!table)
return -ENOMEM;
freq_table[MAX_CLUSTERS] = table;
/* Add in reverse order to get freqs in increasing order */
for (i = MAX_CLUSTERS - 1; i >= 0; i--, count = k) {
for (j = 0; freq_table[i][j].frequency != CPUFREQ_TABLE_END;
j++) {
if (i == A15_CLUSTER &&
search_frequency(table, count, freq_table[i][j].frequency))
continue; /* skip duplicates */
table[k++].frequency =
VIRT_FREQ(i, freq_table[i][j].frequency);
}
}
table[k].driver_data = k;
table[k].frequency = CPUFREQ_TABLE_END;
return 0;
}
static void _put_cluster_clk_and_freq_table(struct device *cpu_dev,
const struct cpumask *cpumask)
{
u32 cluster = raw_cpu_to_cluster(cpu_dev->id);
if (!freq_table[cluster])
return;
clk_put(clk[cluster]);
dev_pm_opp_free_cpufreq_table(cpu_dev, &freq_table[cluster]);
}
static void put_cluster_clk_and_freq_table(struct device *cpu_dev,
const struct cpumask *cpumask)
{
u32 cluster = cpu_to_cluster(cpu_dev->id);
int i;
if (atomic_dec_return(&cluster_usage[cluster]))
return;
if (cluster < MAX_CLUSTERS)
return _put_cluster_clk_and_freq_table(cpu_dev, cpumask);
for_each_present_cpu(i) {
struct device *cdev = get_cpu_device(i);
if (!cdev)
return;
_put_cluster_clk_and_freq_table(cdev, cpumask);
}
/* free virtual table */
kfree(freq_table[cluster]);
}
static int _get_cluster_clk_and_freq_table(struct device *cpu_dev,
const struct cpumask *cpumask)
{
u32 cluster = raw_cpu_to_cluster(cpu_dev->id);
int ret;
if (freq_table[cluster])
return 0;
/*
* platform specific SPC code must initialise the opp table
* so just check if the OPP count is non-zero
*/
ret = dev_pm_opp_get_opp_count(cpu_dev) <= 0;
if (ret)
goto out;
ret = dev_pm_opp_init_cpufreq_table(cpu_dev, &freq_table[cluster]);
if (ret)
goto out;
clk[cluster] = clk_get(cpu_dev, NULL);
if (!IS_ERR(clk[cluster]))
return 0;
dev_err(cpu_dev, "%s: Failed to get clk for cpu: %d, cluster: %d\n",
__func__, cpu_dev->id, cluster);
ret = PTR_ERR(clk[cluster]);
dev_pm_opp_free_cpufreq_table(cpu_dev, &freq_table[cluster]);
out:
dev_err(cpu_dev, "%s: Failed to get data for cluster: %d\n", __func__,
cluster);
return ret;
}
static int get_cluster_clk_and_freq_table(struct device *cpu_dev,
const struct cpumask *cpumask)
{
u32 cluster = cpu_to_cluster(cpu_dev->id);
int i, ret;
if (atomic_inc_return(&cluster_usage[cluster]) != 1)
return 0;
if (cluster < MAX_CLUSTERS) {
ret = _get_cluster_clk_and_freq_table(cpu_dev, cpumask);
if (ret)
atomic_dec(&cluster_usage[cluster]);
return ret;
}
/*
* Get data for all clusters and fill virtual cluster with a merge of
* both
*/
for_each_present_cpu(i) {
struct device *cdev = get_cpu_device(i);
if (!cdev)
return -ENODEV;
ret = _get_cluster_clk_and_freq_table(cdev, cpumask);
if (ret)
goto put_clusters;
}
ret = merge_cluster_tables();
if (ret)
goto put_clusters;
/* Assuming 2 cluster, set clk_big_min and clk_little_max */
clk_big_min = get_table_min(freq_table[A15_CLUSTER]);
clk_little_max = VIRT_FREQ(A7_CLUSTER,
get_table_max(freq_table[A7_CLUSTER]));
return 0;
put_clusters:
for_each_present_cpu(i) {
struct device *cdev = get_cpu_device(i);
if (!cdev)
return -ENODEV;
_put_cluster_clk_and_freq_table(cdev, cpumask);
}
atomic_dec(&cluster_usage[cluster]);
return ret;
}
/* Per-CPU initialization */
static int ve_spc_cpufreq_init(struct cpufreq_policy *policy)
{
u32 cur_cluster = cpu_to_cluster(policy->cpu);
struct device *cpu_dev;
int ret;
cpu_dev = get_cpu_device(policy->cpu);
if (!cpu_dev) {
pr_err("%s: failed to get cpu%d device\n", __func__,
policy->cpu);
return -ENODEV;
}
if (cur_cluster < MAX_CLUSTERS) {
int cpu;
dev_pm_opp_get_sharing_cpus(cpu_dev, policy->cpus);
for_each_cpu(cpu, policy->cpus)
per_cpu(physical_cluster, cpu) = cur_cluster;
} else {
/* Assumption: during init, we are always running on A15 */
per_cpu(physical_cluster, policy->cpu) = A15_CLUSTER;
}
ret = get_cluster_clk_and_freq_table(cpu_dev, policy->cpus);
if (ret)
return ret;
policy->freq_table = freq_table[cur_cluster];
policy->cpuinfo.transition_latency = 1000000; /* 1 ms */
if (is_bL_switching_enabled())
per_cpu(cpu_last_req_freq, policy->cpu) =
clk_get_cpu_rate(policy->cpu);
dev_info(cpu_dev, "%s: CPU %d initialized\n", __func__, policy->cpu);
return 0;
}
static int ve_spc_cpufreq_exit(struct cpufreq_policy *policy)
{
struct device *cpu_dev;
cpu_dev = get_cpu_device(policy->cpu);
if (!cpu_dev) {
pr_err("%s: failed to get cpu%d device\n", __func__,
policy->cpu);
return -ENODEV;
}
put_cluster_clk_and_freq_table(cpu_dev, policy->related_cpus);
return 0;
}
static struct cpufreq_driver ve_spc_cpufreq_driver = {
.name = "vexpress-spc",
cpufreq: Remove CPUFREQ_STICKY flag During cpufreq driver's registration, if the ->init() callback for all the CPUs fail then there is not much point in keeping the driver around as it will only account for more of unnecessary noise, for example cpufreq core will try to suspend/resume the driver which never got registered properly. The removal of such a driver is avoided if the driver carries the CPUFREQ_STICKY flag. This was added way back [1] in 2004 and perhaps no one should ever need it now. A lot of drivers do set this flag, probably because they just copied it from other drivers. This was added earlier for some platforms [2] because their cpufreq drivers were getting registered before the CPUs were registered with subsys framework. And hence they used to fail. The same isn't true anymore though. The current code flow in the kernel is: start_kernel() -> kernel_init() -> kernel_init_freeable() -> do_basic_setup() -> driver_init() -> cpu_dev_init() -> subsys_system_register() //For CPUs -> do_initcalls() -> cpufreq_register_driver() Clearly, the CPUs will always get registered with subsys framework before any cpufreq driver can get probed. Remove the flag and update the relevant drivers. Link: https://git.kernel.org/pub/scm/linux/kernel/git/tglx/history.git/commit/include/linux/cpufreq.h?id=7cc9f0d9a1ab04cedc60d64fd8dcf7df224a3b4d # [1] Link: https://git.kernel.org/pub/scm/linux/kernel/git/tglx/history.git/commit/arch/arm/mach-sa1100/cpu-sa1100.c?id=f59d3bbe35f6268d729f51be82af8325d62f20f5 # [2] Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2021-02-02 04:55:11 +00:00
.flags = CPUFREQ_HAVE_GOVERNOR_PER_POLICY |
CPUFREQ_NEED_INITIAL_FREQ_CHECK,
.verify = cpufreq_generic_frequency_table_verify,
.target_index = ve_spc_cpufreq_set_target,
.get = ve_spc_cpufreq_get_rate,
.init = ve_spc_cpufreq_init,
.exit = ve_spc_cpufreq_exit,
.register_em = cpufreq_register_em_with_opp,
.attr = cpufreq_generic_attr,
};
#ifdef CONFIG_BL_SWITCHER
static int bL_cpufreq_switcher_notifier(struct notifier_block *nfb,
unsigned long action, void *_arg)
{
pr_debug("%s: action: %ld\n", __func__, action);
switch (action) {
case BL_NOTIFY_PRE_ENABLE:
case BL_NOTIFY_PRE_DISABLE:
cpufreq_unregister_driver(&ve_spc_cpufreq_driver);
break;
case BL_NOTIFY_POST_ENABLE:
set_switching_enabled(true);
cpufreq_register_driver(&ve_spc_cpufreq_driver);
break;
case BL_NOTIFY_POST_DISABLE:
set_switching_enabled(false);
cpufreq_register_driver(&ve_spc_cpufreq_driver);
break;
default:
return NOTIFY_DONE;
}
return NOTIFY_OK;
}
static struct notifier_block bL_switcher_notifier = {
.notifier_call = bL_cpufreq_switcher_notifier,
};
static int __bLs_register_notifier(void)
{
return bL_switcher_register_notifier(&bL_switcher_notifier);
}
static int __bLs_unregister_notifier(void)
{
return bL_switcher_unregister_notifier(&bL_switcher_notifier);
}
#else
static int __bLs_register_notifier(void) { return 0; }
static int __bLs_unregister_notifier(void) { return 0; }
#endif
static int ve_spc_cpufreq_probe(struct platform_device *pdev)
{
int ret, i;
set_switching_enabled(bL_switcher_get_enabled());
for (i = 0; i < MAX_CLUSTERS; i++)
mutex_init(&cluster_lock[i]);
if (!is_bL_switching_enabled())
ve_spc_cpufreq_driver.flags |= CPUFREQ_IS_COOLING_DEV;
ret = cpufreq_register_driver(&ve_spc_cpufreq_driver);
if (ret) {
pr_info("%s: Failed registering platform driver: %s, err: %d\n",
__func__, ve_spc_cpufreq_driver.name, ret);
} else {
ret = __bLs_register_notifier();
if (ret)
cpufreq_unregister_driver(&ve_spc_cpufreq_driver);
else
pr_info("%s: Registered platform driver: %s\n",
__func__, ve_spc_cpufreq_driver.name);
}
bL_switcher_put_enabled();
return ret;
}
static int ve_spc_cpufreq_remove(struct platform_device *pdev)
{
bL_switcher_get_enabled();
__bLs_unregister_notifier();
cpufreq_unregister_driver(&ve_spc_cpufreq_driver);
bL_switcher_put_enabled();
pr_info("%s: Un-registered platform driver: %s\n", __func__,
ve_spc_cpufreq_driver.name);
return 0;
}
static struct platform_driver ve_spc_cpufreq_platdrv = {
.driver = {
.name = "vexpress-spc-cpufreq",
},
.probe = ve_spc_cpufreq_probe,
.remove = ve_spc_cpufreq_remove,
};
module_platform_driver(ve_spc_cpufreq_platdrv);
MODULE_ALIAS("platform:vexpress-spc-cpufreq");
MODULE_AUTHOR("Viresh Kumar <viresh.kumar@linaro.org>");
MODULE_AUTHOR("Sudeep Holla <sudeep.holla@arm.com>");
MODULE_DESCRIPTION("Vexpress SPC ARM big LITTLE cpufreq driver");
MODULE_LICENSE("GPL v2");