We move the following functions into a new file drivers/acpi/pm-op.c, as
they are all more fitting in performance controling and only called by
do_pm_op():
- get_cpufreq_para()
- set_cpufreq_para()
- set_cpufreq_gov()
- set_cpufreq_cppc()
- cpufreq_driver_getavg()
- cpufreq_update_turbo()
- cpufreq_get_turbo_status()
We introduce a new Kconfig CONFIG_PM_OP to wrap the new file.
Also, although the following helpers are only called by do_pm_op(), they have
dependency on local variable, we wrap them with CONFIG_PM_OP in place:
- write_userspace_scaling_setspeed()
- write_ondemand_sampling_rate()
- write_ondemand_up_threshold()
- get_cpufreq_ondemand_para()
- cpufreq_driver.update()
- get_hwp_para()
Various style corrections shall be applied at the same time while moving these
functions, including:
- add extra space before and after bracket of if() and switch()
- fix indentation
- drop all the unnecessary inner figure braces
Signed-off-by: Penny Zheng <Penny.Zheng@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Acked-by: Jan Beulich <jbeulich@suse.com>
---
v2 -> v3
- new commit
---
v3 -> v4:
- rename the file to pm-op.c
- drop all the unnecessary inner figure braces
- be consistent with the comment on the #endif
---
xen/arch/x86/acpi/cpufreq/hwp.c | 6 +
xen/arch/x86/acpi/cpufreq/powernow.c | 4 +
xen/common/Kconfig | 7 +
xen/common/sysctl.c | 2 +
xen/drivers/acpi/Makefile | 1 +
xen/drivers/acpi/pm-op.c | 397 +++++++++++++++++++
xen/drivers/acpi/pmstat.c | 357 -----------------
xen/drivers/cpufreq/cpufreq_misc_governors.c | 2 +
xen/drivers/cpufreq/cpufreq_ondemand.c | 2 +
xen/drivers/cpufreq/utility.c | 41 --
xen/include/acpi/cpufreq/cpufreq.h | 3 -
11 files changed, 421 insertions(+), 401 deletions(-)
create mode 100644 xen/drivers/acpi/pm-op.c
diff --git a/xen/arch/x86/acpi/cpufreq/hwp.c b/xen/arch/x86/acpi/cpufreq/hwp.c
index d5fa3d47ca..e4c09244ab 100644
--- a/xen/arch/x86/acpi/cpufreq/hwp.c
+++ b/xen/arch/x86/acpi/cpufreq/hwp.c
@@ -466,6 +466,7 @@ static int cf_check hwp_cpufreq_cpu_exit(struct cpufreq_policy *policy)
return 0;
}
+#ifdef CONFIG_PM_OP
/*
* The SDM reads like turbo should be disabled with MSR_IA32_PERF_CTL and
* PERF_CTL_TURBO_DISENGAGE, but that does not seem to actually work, at least
@@ -508,6 +509,7 @@ static int cf_check hwp_cpufreq_update(unsigned int cpu, struct cpufreq_policy *
return per_cpu(hwp_drv_data, cpu)->ret;
}
+#endif /* CONFIG_PM_OP */
static const struct cpufreq_driver __initconst_cf_clobber
hwp_cpufreq_driver = {
@@ -516,9 +518,12 @@ hwp_cpufreq_driver = {
.target = hwp_cpufreq_target,
.init = hwp_cpufreq_cpu_init,
.exit = hwp_cpufreq_cpu_exit,
+#ifdef CONFIG_PM_OP
.update = hwp_cpufreq_update,
+#endif
};
+#ifdef CONFIG_PM_OP
int get_hwp_para(unsigned int cpu,
struct xen_cppc_para *cppc_para)
{
@@ -639,6 +644,7 @@ int set_hwp_para(struct cpufreq_policy *policy,
return hwp_cpufreq_target(policy, 0, 0);
}
+#endif /* CONFIG_PM_OP */
int __init hwp_register_driver(void)
{
diff --git a/xen/arch/x86/acpi/cpufreq/powernow.c b/xen/arch/x86/acpi/cpufreq/powernow.c
index 69364e1855..12fca45b45 100644
--- a/xen/arch/x86/acpi/cpufreq/powernow.c
+++ b/xen/arch/x86/acpi/cpufreq/powernow.c
@@ -49,6 +49,7 @@ static void cf_check transition_pstate(void *pstate)
wrmsrl(MSR_PSTATE_CTRL, *(unsigned int *)pstate);
}
+#ifdef CONFIG_PM_OP
static void cf_check update_cpb(void *data)
{
struct cpufreq_policy *policy = data;
@@ -77,6 +78,7 @@ static int cf_check powernow_cpufreq_update(
return 0;
}
+#endif /* CONFIG_PM_OP */
static int cf_check powernow_cpufreq_target(
struct cpufreq_policy *policy,
@@ -324,7 +326,9 @@ powernow_cpufreq_driver = {
.target = powernow_cpufreq_target,
.init = powernow_cpufreq_cpu_init,
.exit = powernow_cpufreq_cpu_exit,
+#ifdef CONFIG_PM_OP
.update = powernow_cpufreq_update
+#endif
};
unsigned int __init powernow_register_driver(void)
diff --git a/xen/common/Kconfig b/xen/common/Kconfig
index d84906df24..712c29b1a0 100644
--- a/xen/common/Kconfig
+++ b/xen/common/Kconfig
@@ -591,4 +591,11 @@ config SYSCTL
to reduce Xen footprint.
endmenu
+config PM_OP
+ bool "Enable Performance Management Operation"
+ depends on ACPI && HAS_CPUFREQ && SYSCTL
+ default y
+ help
+ This option shall enable userspace performance management control
+ to do power/performance analyzing and tuning.
endmenu
diff --git a/xen/common/sysctl.c b/xen/common/sysctl.c
index c2d99ae12e..daf57fbe56 100644
--- a/xen/common/sysctl.c
+++ b/xen/common/sysctl.c
@@ -174,7 +174,9 @@ long do_sysctl(XEN_GUEST_HANDLE_PARAM(xen_sysctl_t) u_sysctl)
case XEN_SYSCTL_get_pmstat:
ret = do_get_pm_info(&op->u.get_pmstat);
break;
+#endif
+#ifdef CONFIG_PM_OP
case XEN_SYSCTL_pm_op:
ret = do_pm_op(&op->u.pm_op);
if ( ret == -EAGAIN )
diff --git a/xen/drivers/acpi/Makefile b/xen/drivers/acpi/Makefile
index 2fc5230253..1d811a51a7 100644
--- a/xen/drivers/acpi/Makefile
+++ b/xen/drivers/acpi/Makefile
@@ -6,6 +6,7 @@ obj-bin-y += tables.init.o
obj-$(CONFIG_ACPI_NUMA) += numa.o
obj-y += osl.o
obj-$(CONFIG_HAS_CPUFREQ) += pmstat.o
+obj-$(CONFIG_PM_OP) += pm-op.o
obj-$(CONFIG_X86) += hwregs.o
obj-$(CONFIG_X86) += reboot.o
diff --git a/xen/drivers/acpi/pm-op.c b/xen/drivers/acpi/pm-op.c
new file mode 100644
index 0000000000..b6c590967e
--- /dev/null
+++ b/xen/drivers/acpi/pm-op.c
@@ -0,0 +1,397 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+
+#include <xen/acpi.h>
+#include <xen/domain.h>
+#include <xen/errno.h>
+#include <xen/guest_access.h>
+#include <xen/lib.h>
+#include <xen/sched.h>
+
+#include <acpi/cpufreq/cpufreq.h>
+#include <public/platform.h>
+#include <public/sysctl.h>
+
+/*
+ * 1. Get PM parameter
+ * 2. Provide user PM control
+ */
+static int cpufreq_update_turbo(unsigned int cpu, int new_state)
+{
+ struct cpufreq_policy *policy;
+ int curr_state;
+ int ret = 0;
+
+ if ( new_state != CPUFREQ_TURBO_ENABLED &&
+ new_state != CPUFREQ_TURBO_DISABLED )
+ return -EINVAL;
+
+ policy = per_cpu(cpufreq_cpu_policy, cpu);
+ if ( !policy )
+ return -EACCES;
+
+ if ( policy->turbo == CPUFREQ_TURBO_UNSUPPORTED )
+ return -EOPNOTSUPP;
+
+ curr_state = policy->turbo;
+ if ( curr_state == new_state )
+ return 0;
+
+ policy->turbo = new_state;
+ if ( cpufreq_driver.update )
+ {
+ ret = alternative_call(cpufreq_driver.update, cpu, policy);
+ if ( ret )
+ policy->turbo = curr_state;
+ }
+
+ return ret;
+}
+
+static int cpufreq_get_turbo_status(unsigned int cpu)
+{
+ struct cpufreq_policy *policy;
+
+ policy = per_cpu(cpufreq_cpu_policy, cpu);
+ return policy && policy->turbo == CPUFREQ_TURBO_ENABLED;
+}
+
+static int read_scaling_available_governors(char *scaling_available_governors,
+ unsigned int size)
+{
+ unsigned int i = 0;
+ struct cpufreq_governor *t;
+
+ if ( !scaling_available_governors )
+ return -EINVAL;
+
+ list_for_each_entry(t, &cpufreq_governor_list, governor_list)
+ {
+ i += scnprintf(&scaling_available_governors[i],
+ CPUFREQ_NAME_LEN, "%s ", t->name);
+ if ( i > size )
+ return -EINVAL;
+ }
+ scaling_available_governors[i-1] = '\0';
+
+ return 0;
+}
+
+static int get_cpufreq_para(struct xen_sysctl_pm_op *op)
+{
+ uint32_t ret = 0;
+ const struct processor_pminfo *pmpt;
+ struct cpufreq_policy *policy;
+ uint32_t gov_num = 0;
+ uint32_t *data;
+ char *scaling_available_governors;
+ struct list_head *pos;
+ unsigned int cpu, i = 0;
+
+ pmpt = processor_pminfo[op->cpuid];
+ policy = per_cpu(cpufreq_cpu_policy, op->cpuid);
+
+ if ( !pmpt || !pmpt->perf.states ||
+ !policy || !policy->governor )
+ return -EINVAL;
+
+ list_for_each(pos, &cpufreq_governor_list)
+ gov_num++;
+
+ if ( (op->u.get_para.cpu_num != cpumask_weight(policy->cpus)) ||
+ (op->u.get_para.freq_num != pmpt->perf.state_count) ||
+ (op->u.get_para.gov_num != gov_num) )
+ {
+ op->u.get_para.cpu_num = cpumask_weight(policy->cpus);
+ op->u.get_para.freq_num = pmpt->perf.state_count;
+ op->u.get_para.gov_num = gov_num;
+ return -EAGAIN;
+ }
+
+ if ( !(data = xzalloc_array(uint32_t,
+ max(op->u.get_para.cpu_num,
+ op->u.get_para.freq_num))) )
+ return -ENOMEM;
+
+ for_each_cpu(cpu, policy->cpus)
+ data[i++] = cpu;
+ ret = copy_to_guest(op->u.get_para.affected_cpus,
+ data, op->u.get_para.cpu_num);
+
+ for ( i = 0; i < op->u.get_para.freq_num; i++ )
+ data[i] = pmpt->perf.states[i].core_frequency * 1000;
+ ret += copy_to_guest(op->u.get_para.scaling_available_frequencies,
+ data, op->u.get_para.freq_num);
+
+ xfree(data);
+ if ( ret )
+ return -EFAULT;
+
+ op->u.get_para.cpuinfo_cur_freq =
+ cpufreq_driver.get ? alternative_call(cpufreq_driver.get, op->cpuid)
+ : policy->cur;
+ op->u.get_para.cpuinfo_max_freq = policy->cpuinfo.max_freq;
+ op->u.get_para.cpuinfo_min_freq = policy->cpuinfo.min_freq;
+ op->u.get_para.turbo_enabled = cpufreq_get_turbo_status(op->cpuid);
+
+ if ( cpufreq_driver.name[0] )
+ strlcpy(op->u.get_para.scaling_driver,
+ cpufreq_driver.name, CPUFREQ_NAME_LEN);
+ else
+ strlcpy(op->u.get_para.scaling_driver, "Unknown", 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
+ {
+ if ( !(scaling_available_governors =
+ xzalloc_array(char, gov_num * CPUFREQ_NAME_LEN)) )
+ return -ENOMEM;
+ if ( (ret = read_scaling_available_governors(
+ scaling_available_governors,
+ (gov_num * CPUFREQ_NAME_LEN *
+ sizeof(*scaling_available_governors)))) )
+ {
+ xfree(scaling_available_governors);
+ return ret;
+ }
+ ret = copy_to_guest(op->u.get_para.scaling_available_governors,
+ scaling_available_governors,
+ gov_num * CPUFREQ_NAME_LEN);
+ xfree(scaling_available_governors);
+ if ( ret )
+ return -EFAULT;
+
+ op->u.get_para.u.s.scaling_cur_freq = policy->cur;
+ op->u.get_para.u.s.scaling_max_freq = policy->max;
+ op->u.get_para.u.s.scaling_min_freq = policy->min;
+
+ if ( policy->governor->name[0] )
+ strlcpy(op->u.get_para.u.s.scaling_governor,
+ policy->governor->name, CPUFREQ_NAME_LEN);
+ else
+ strlcpy(op->u.get_para.u.s.scaling_governor, "Unknown",
+ CPUFREQ_NAME_LEN);
+
+ /* governor specific para */
+ if ( !strncasecmp(op->u.get_para.u.s.scaling_governor,
+ "userspace", CPUFREQ_NAME_LEN) )
+ op->u.get_para.u.s.u.userspace.scaling_setspeed = policy->cur;
+
+ if ( !strncasecmp(op->u.get_para.u.s.scaling_governor,
+ "ondemand", CPUFREQ_NAME_LEN) )
+ ret = get_cpufreq_ondemand_para(
+ &op->u.get_para.u.s.u.ondemand.sampling_rate_max,
+ &op->u.get_para.u.s.u.ondemand.sampling_rate_min,
+ &op->u.get_para.u.s.u.ondemand.sampling_rate,
+ &op->u.get_para.u.s.u.ondemand.up_threshold);
+ }
+
+ return ret;
+}
+
+static int set_cpufreq_gov(struct xen_sysctl_pm_op *op)
+{
+ struct cpufreq_policy new_policy, *old_policy;
+
+ old_policy = per_cpu(cpufreq_cpu_policy, op->cpuid);
+ if ( !old_policy )
+ return -EINVAL;
+
+ memcpy(&new_policy, old_policy, sizeof(struct cpufreq_policy));
+
+ new_policy.governor = __find_governor(op->u.set_gov.scaling_governor);
+ if ( new_policy.governor == NULL )
+ return -EINVAL;
+
+ return __cpufreq_set_policy(old_policy, &new_policy);
+}
+
+static int set_cpufreq_para(struct xen_sysctl_pm_op *op)
+{
+ int ret = 0;
+ struct cpufreq_policy *policy;
+
+ policy = per_cpu(cpufreq_cpu_policy, op->cpuid);
+
+ if ( !policy || !policy->governor )
+ return -EINVAL;
+
+ if ( hwp_active() )
+ return -EOPNOTSUPP;
+
+ switch( op->u.set_para.ctrl_type )
+ {
+ case SCALING_MAX_FREQ:
+ {
+ struct cpufreq_policy new_policy;
+
+ memcpy(&new_policy, policy, sizeof(struct cpufreq_policy));
+ new_policy.max = op->u.set_para.ctrl_value;
+ ret = __cpufreq_set_policy(policy, &new_policy);
+
+ break;
+ }
+
+ case SCALING_MIN_FREQ:
+ {
+ struct cpufreq_policy new_policy;
+
+ memcpy(&new_policy, policy, sizeof(struct cpufreq_policy));
+ new_policy.min = op->u.set_para.ctrl_value;
+ ret = __cpufreq_set_policy(policy, &new_policy);
+
+ break;
+ }
+
+ case SCALING_SETSPEED:
+ {
+ unsigned int freq =op->u.set_para.ctrl_value;
+
+ if ( !strncasecmp(policy->governor->name,
+ "userspace", CPUFREQ_NAME_LEN) )
+ ret = write_userspace_scaling_setspeed(op->cpuid, freq);
+ else
+ ret = -EINVAL;
+
+ break;
+ }
+
+ case SAMPLING_RATE:
+ {
+ unsigned int sampling_rate = op->u.set_para.ctrl_value;
+
+ if ( !strncasecmp(policy->governor->name,
+ "ondemand", CPUFREQ_NAME_LEN) )
+ ret = write_ondemand_sampling_rate(sampling_rate);
+ else
+ ret = -EINVAL;
+
+ break;
+ }
+
+ case UP_THRESHOLD:
+ {
+ unsigned int up_threshold = op->u.set_para.ctrl_value;
+
+ if ( !strncasecmp(policy->governor->name,
+ "ondemand", CPUFREQ_NAME_LEN) )
+ ret = write_ondemand_up_threshold(up_threshold);
+ else
+ ret = -EINVAL;
+
+ break;
+ }
+
+ default:
+ ret = -EINVAL;
+ break;
+ }
+
+ return ret;
+}
+
+static int set_cpufreq_cppc(struct xen_sysctl_pm_op *op)
+{
+ struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_policy, op->cpuid);
+
+ if ( !policy || !policy->governor )
+ return -ENOENT;
+
+ if ( !hwp_active() )
+ return -EOPNOTSUPP;
+
+ return set_hwp_para(policy, &op->u.set_cppc);
+}
+
+int do_pm_op(struct xen_sysctl_pm_op *op)
+{
+ int ret = 0;
+ const struct processor_pminfo *pmpt;
+
+ switch ( op->cmd )
+ {
+ case XEN_SYSCTL_pm_op_set_sched_opt_smt:
+ {
+ uint32_t saved_value = sched_smt_power_savings;
+
+ if ( op->cpuid != 0 )
+ return -EINVAL;
+ sched_smt_power_savings = !!op->u.set_sched_opt_smt;
+ op->u.set_sched_opt_smt = saved_value;
+ return 0;
+ }
+
+ case XEN_SYSCTL_pm_op_get_max_cstate:
+ BUILD_BUG_ON(XEN_SYSCTL_CX_UNLIMITED != UINT_MAX);
+ if ( op->cpuid == 0 )
+ op->u.get_max_cstate = acpi_get_cstate_limit();
+ else if ( op->cpuid == 1 )
+ op->u.get_max_cstate = acpi_get_csubstate_limit();
+ else
+ ret = -EINVAL;
+ return ret;
+
+ case XEN_SYSCTL_pm_op_set_max_cstate:
+ if ( op->cpuid == 0 )
+ acpi_set_cstate_limit(op->u.set_max_cstate);
+ else if ( op->cpuid == 1 )
+ acpi_set_csubstate_limit(op->u.set_max_cstate);
+ else
+ ret = -EINVAL;
+ return ret;
+ }
+
+ if ( op->cpuid >= nr_cpu_ids || !cpu_online(op->cpuid) )
+ return -EINVAL;
+ pmpt = processor_pminfo[op->cpuid];
+
+ switch ( op->cmd & PM_PARA_CATEGORY_MASK )
+ {
+ case CPUFREQ_PARA:
+ if ( !(xen_processor_pmbits & XEN_PROCESSOR_PM_PX) )
+ return -ENODEV;
+ if ( !pmpt || !(pmpt->perf.init & XEN_PX_INIT) )
+ return -EINVAL;
+ break;
+ }
+
+ switch ( op->cmd )
+ {
+ case GET_CPUFREQ_PARA:
+ ret = get_cpufreq_para(op);
+ break;
+
+ case SET_CPUFREQ_GOV:
+ ret = set_cpufreq_gov(op);
+ break;
+
+ case SET_CPUFREQ_PARA:
+ ret = set_cpufreq_para(op);
+ break;
+
+ case SET_CPUFREQ_CPPC:
+ ret = set_cpufreq_cppc(op);
+ break;
+
+ case GET_CPUFREQ_AVGFREQ:
+ op->u.get_avgfreq = cpufreq_driver_getavg(op->cpuid, USR_GETAVG);
+ break;
+
+ case XEN_SYSCTL_pm_op_enable_turbo:
+ ret = cpufreq_update_turbo(op->cpuid, CPUFREQ_TURBO_ENABLED);
+ break;
+
+ case XEN_SYSCTL_pm_op_disable_turbo:
+ ret = cpufreq_update_turbo(op->cpuid, CPUFREQ_TURBO_DISABLED);
+ break;
+
+ default:
+ printk("not defined sub-hypercall @ do_pm_op\n");
+ ret = -ENOSYS;
+ break;
+ }
+
+ return ret;
+}
diff --git a/xen/drivers/acpi/pmstat.c b/xen/drivers/acpi/pmstat.c
index abfdc45cc2..61b60e59a2 100644
--- a/xen/drivers/acpi/pmstat.c
+++ b/xen/drivers/acpi/pmstat.c
@@ -330,360 +330,3 @@ int do_get_pm_info(struct xen_sysctl_get_pmstat *op)
return ret;
}
-
-/*
- * 1. Get PM parameter
- * 2. Provide user PM control
- */
-static int read_scaling_available_governors(char *scaling_available_governors,
- unsigned int size)
-{
- unsigned int i = 0;
- struct cpufreq_governor *t;
-
- if ( !scaling_available_governors )
- return -EINVAL;
-
- list_for_each_entry(t, &cpufreq_governor_list, governor_list)
- {
- i += scnprintf(&scaling_available_governors[i],
- CPUFREQ_NAME_LEN, "%s ", t->name);
- if ( i > size )
- return -EINVAL;
- }
- scaling_available_governors[i-1] = '\0';
-
- return 0;
-}
-
-static int get_cpufreq_para(struct xen_sysctl_pm_op *op)
-{
- uint32_t ret = 0;
- const struct processor_pminfo *pmpt;
- struct cpufreq_policy *policy;
- uint32_t gov_num = 0;
- uint32_t *data;
- char *scaling_available_governors;
- struct list_head *pos;
- unsigned int cpu, i = 0;
-
- pmpt = processor_pminfo[op->cpuid];
- policy = per_cpu(cpufreq_cpu_policy, op->cpuid);
-
- if ( !pmpt || !pmpt->perf.states ||
- !policy || !policy->governor )
- return -EINVAL;
-
- list_for_each(pos, &cpufreq_governor_list)
- gov_num++;
-
- if ( (op->u.get_para.cpu_num != cpumask_weight(policy->cpus)) ||
- (op->u.get_para.freq_num != pmpt->perf.state_count) ||
- (op->u.get_para.gov_num != gov_num) )
- {
- op->u.get_para.cpu_num = cpumask_weight(policy->cpus);
- op->u.get_para.freq_num = pmpt->perf.state_count;
- op->u.get_para.gov_num = gov_num;
- return -EAGAIN;
- }
-
- if ( !(data = xzalloc_array(uint32_t,
- max(op->u.get_para.cpu_num,
- op->u.get_para.freq_num))) )
- return -ENOMEM;
-
- for_each_cpu(cpu, policy->cpus)
- data[i++] = cpu;
- ret = copy_to_guest(op->u.get_para.affected_cpus,
- data, op->u.get_para.cpu_num);
-
- for ( i = 0; i < op->u.get_para.freq_num; i++ )
- data[i] = pmpt->perf.states[i].core_frequency * 1000;
- ret += copy_to_guest(op->u.get_para.scaling_available_frequencies,
- data, op->u.get_para.freq_num);
-
- xfree(data);
- if ( ret )
- return -EFAULT;
-
- op->u.get_para.cpuinfo_cur_freq =
- cpufreq_driver.get ? alternative_call(cpufreq_driver.get, op->cpuid)
- : policy->cur;
- op->u.get_para.cpuinfo_max_freq = policy->cpuinfo.max_freq;
- op->u.get_para.cpuinfo_min_freq = policy->cpuinfo.min_freq;
- op->u.get_para.turbo_enabled = cpufreq_get_turbo_status(op->cpuid);
-
- if ( cpufreq_driver.name[0] )
- strlcpy(op->u.get_para.scaling_driver,
- cpufreq_driver.name, CPUFREQ_NAME_LEN);
- else
- strlcpy(op->u.get_para.scaling_driver, "Unknown", 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
- {
- if ( !(scaling_available_governors =
- xzalloc_array(char, gov_num * CPUFREQ_NAME_LEN)) )
- return -ENOMEM;
- if ( (ret = read_scaling_available_governors(
- scaling_available_governors,
- (gov_num * CPUFREQ_NAME_LEN *
- sizeof(*scaling_available_governors)))) )
- {
- xfree(scaling_available_governors);
- return ret;
- }
- ret = copy_to_guest(op->u.get_para.scaling_available_governors,
- scaling_available_governors,
- gov_num * CPUFREQ_NAME_LEN);
- xfree(scaling_available_governors);
- if ( ret )
- return -EFAULT;
-
- op->u.get_para.u.s.scaling_cur_freq = policy->cur;
- op->u.get_para.u.s.scaling_max_freq = policy->max;
- op->u.get_para.u.s.scaling_min_freq = policy->min;
-
- if ( policy->governor->name[0] )
- strlcpy(op->u.get_para.u.s.scaling_governor,
- policy->governor->name, CPUFREQ_NAME_LEN);
- else
- strlcpy(op->u.get_para.u.s.scaling_governor, "Unknown",
- CPUFREQ_NAME_LEN);
-
- /* governor specific para */
- if ( !strncasecmp(op->u.get_para.u.s.scaling_governor,
- "userspace", CPUFREQ_NAME_LEN) )
- op->u.get_para.u.s.u.userspace.scaling_setspeed = policy->cur;
-
- if ( !strncasecmp(op->u.get_para.u.s.scaling_governor,
- "ondemand", CPUFREQ_NAME_LEN) )
- ret = get_cpufreq_ondemand_para(
- &op->u.get_para.u.s.u.ondemand.sampling_rate_max,
- &op->u.get_para.u.s.u.ondemand.sampling_rate_min,
- &op->u.get_para.u.s.u.ondemand.sampling_rate,
- &op->u.get_para.u.s.u.ondemand.up_threshold);
- }
-
- return ret;
-}
-
-static int set_cpufreq_gov(struct xen_sysctl_pm_op *op)
-{
- struct cpufreq_policy new_policy, *old_policy;
-
- old_policy = per_cpu(cpufreq_cpu_policy, op->cpuid);
- if ( !old_policy )
- return -EINVAL;
-
- memcpy(&new_policy, old_policy, sizeof(struct cpufreq_policy));
-
- new_policy.governor = __find_governor(op->u.set_gov.scaling_governor);
- if (new_policy.governor == NULL)
- return -EINVAL;
-
- return __cpufreq_set_policy(old_policy, &new_policy);
-}
-
-static int set_cpufreq_para(struct xen_sysctl_pm_op *op)
-{
- int ret = 0;
- struct cpufreq_policy *policy;
-
- policy = per_cpu(cpufreq_cpu_policy, op->cpuid);
-
- if ( !policy || !policy->governor )
- return -EINVAL;
-
- if ( hwp_active() )
- return -EOPNOTSUPP;
-
- switch(op->u.set_para.ctrl_type)
- {
- case SCALING_MAX_FREQ:
- {
- struct cpufreq_policy new_policy;
-
- memcpy(&new_policy, policy, sizeof(struct cpufreq_policy));
- new_policy.max = op->u.set_para.ctrl_value;
- ret = __cpufreq_set_policy(policy, &new_policy);
-
- break;
- }
-
- case SCALING_MIN_FREQ:
- {
- struct cpufreq_policy new_policy;
-
- memcpy(&new_policy, policy, sizeof(struct cpufreq_policy));
- new_policy.min = op->u.set_para.ctrl_value;
- ret = __cpufreq_set_policy(policy, &new_policy);
-
- break;
- }
-
- case SCALING_SETSPEED:
- {
- unsigned int freq =op->u.set_para.ctrl_value;
-
- if ( !strncasecmp(policy->governor->name,
- "userspace", CPUFREQ_NAME_LEN) )
- ret = write_userspace_scaling_setspeed(op->cpuid, freq);
- else
- ret = -EINVAL;
-
- break;
- }
-
- case SAMPLING_RATE:
- {
- unsigned int sampling_rate = op->u.set_para.ctrl_value;
-
- if ( !strncasecmp(policy->governor->name,
- "ondemand", CPUFREQ_NAME_LEN) )
- ret = write_ondemand_sampling_rate(sampling_rate);
- else
- ret = -EINVAL;
-
- break;
- }
-
- case UP_THRESHOLD:
- {
- unsigned int up_threshold = op->u.set_para.ctrl_value;
-
- if ( !strncasecmp(policy->governor->name,
- "ondemand", CPUFREQ_NAME_LEN) )
- ret = write_ondemand_up_threshold(up_threshold);
- else
- ret = -EINVAL;
-
- break;
- }
-
- default:
- ret = -EINVAL;
- break;
- }
-
- return ret;
-}
-
-static int set_cpufreq_cppc(struct xen_sysctl_pm_op *op)
-{
- struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_policy, op->cpuid);
-
- if ( !policy || !policy->governor )
- return -ENOENT;
-
- if ( !hwp_active() )
- return -EOPNOTSUPP;
-
- return set_hwp_para(policy, &op->u.set_cppc);
-}
-
-int do_pm_op(struct xen_sysctl_pm_op *op)
-{
- int ret = 0;
- const struct processor_pminfo *pmpt;
-
- switch ( op->cmd )
- {
- case XEN_SYSCTL_pm_op_set_sched_opt_smt:
- {
- uint32_t saved_value = sched_smt_power_savings;
-
- if ( op->cpuid != 0 )
- return -EINVAL;
- sched_smt_power_savings = !!op->u.set_sched_opt_smt;
- op->u.set_sched_opt_smt = saved_value;
- return 0;
- }
-
- case XEN_SYSCTL_pm_op_get_max_cstate:
- BUILD_BUG_ON(XEN_SYSCTL_CX_UNLIMITED != UINT_MAX);
- if ( op->cpuid == 0 )
- op->u.get_max_cstate = acpi_get_cstate_limit();
- else if ( op->cpuid == 1 )
- op->u.get_max_cstate = acpi_get_csubstate_limit();
- else
- ret = -EINVAL;
- return ret;
-
- case XEN_SYSCTL_pm_op_set_max_cstate:
- if ( op->cpuid == 0 )
- acpi_set_cstate_limit(op->u.set_max_cstate);
- else if ( op->cpuid == 1 )
- acpi_set_csubstate_limit(op->u.set_max_cstate);
- else
- ret = -EINVAL;
- return ret;
- }
-
- if ( op->cpuid >= nr_cpu_ids || !cpu_online(op->cpuid) )
- return -EINVAL;
- pmpt = processor_pminfo[op->cpuid];
-
- switch ( op->cmd & PM_PARA_CATEGORY_MASK )
- {
- case CPUFREQ_PARA:
- if ( !(xen_processor_pmbits & XEN_PROCESSOR_PM_PX) )
- return -ENODEV;
- if ( !pmpt || !(pmpt->perf.init & XEN_PX_INIT) )
- return -EINVAL;
- break;
- }
-
- switch ( op->cmd )
- {
- case GET_CPUFREQ_PARA:
- {
- ret = get_cpufreq_para(op);
- break;
- }
-
- case SET_CPUFREQ_GOV:
- {
- ret = set_cpufreq_gov(op);
- break;
- }
-
- case SET_CPUFREQ_PARA:
- {
- ret = set_cpufreq_para(op);
- break;
- }
-
- case SET_CPUFREQ_CPPC:
- ret = set_cpufreq_cppc(op);
- break;
-
- case GET_CPUFREQ_AVGFREQ:
- {
- op->u.get_avgfreq = cpufreq_driver_getavg(op->cpuid, USR_GETAVG);
- break;
- }
-
- case XEN_SYSCTL_pm_op_enable_turbo:
- {
- ret = cpufreq_update_turbo(op->cpuid, CPUFREQ_TURBO_ENABLED);
- break;
- }
-
- case XEN_SYSCTL_pm_op_disable_turbo:
- {
- ret = cpufreq_update_turbo(op->cpuid, CPUFREQ_TURBO_DISABLED);
- break;
- }
-
- default:
- printk("not defined sub-hypercall @ do_pm_op\n");
- ret = -ENOSYS;
- break;
- }
-
- return ret;
-}
diff --git a/xen/drivers/cpufreq/cpufreq_misc_governors.c b/xen/drivers/cpufreq/cpufreq_misc_governors.c
index 0327fad23b..e5cb9ab02f 100644
--- a/xen/drivers/cpufreq/cpufreq_misc_governors.c
+++ b/xen/drivers/cpufreq/cpufreq_misc_governors.c
@@ -64,6 +64,7 @@ static int cf_check cpufreq_governor_userspace(
return ret;
}
+#ifdef CONFIG_PM_OP
int write_userspace_scaling_setspeed(unsigned int cpu, unsigned int freq)
{
struct cpufreq_policy *policy;
@@ -80,6 +81,7 @@ int write_userspace_scaling_setspeed(unsigned int cpu, unsigned int freq)
return __cpufreq_driver_target(policy, freq, CPUFREQ_RELATION_L);
}
+#endif /* CONFIG_PM_OP */
static bool __init cf_check
cpufreq_userspace_handle_option(const char *name, const char *val)
diff --git a/xen/drivers/cpufreq/cpufreq_ondemand.c b/xen/drivers/cpufreq/cpufreq_ondemand.c
index 06cfc88d30..0126a3f5d9 100644
--- a/xen/drivers/cpufreq/cpufreq_ondemand.c
+++ b/xen/drivers/cpufreq/cpufreq_ondemand.c
@@ -57,6 +57,7 @@ static struct dbs_tuners {
static DEFINE_PER_CPU(struct timer, dbs_timer);
+#ifdef CONFIG_PM_OP
int write_ondemand_sampling_rate(unsigned int sampling_rate)
{
if ( (sampling_rate > MAX_SAMPLING_RATE / MICROSECS(1)) ||
@@ -93,6 +94,7 @@ int get_cpufreq_ondemand_para(uint32_t *sampling_rate_max,
return 0;
}
+#endif /* CONFIG_PM_OP */
static void dbs_check_cpu(struct cpu_dbs_info_s *this_dbs_info)
{
diff --git a/xen/drivers/cpufreq/utility.c b/xen/drivers/cpufreq/utility.c
index 723045b240..987c3b5929 100644
--- a/xen/drivers/cpufreq/utility.c
+++ b/xen/drivers/cpufreq/utility.c
@@ -224,47 +224,6 @@ int cpufreq_driver_getavg(unsigned int cpu, unsigned int flag)
return policy->cur;
}
-int cpufreq_update_turbo(unsigned int cpu, int new_state)
-{
- struct cpufreq_policy *policy;
- int curr_state;
- int ret = 0;
-
- if (new_state != CPUFREQ_TURBO_ENABLED &&
- new_state != CPUFREQ_TURBO_DISABLED)
- return -EINVAL;
-
- policy = per_cpu(cpufreq_cpu_policy, cpu);
- if (!policy)
- return -EACCES;
-
- if (policy->turbo == CPUFREQ_TURBO_UNSUPPORTED)
- return -EOPNOTSUPP;
-
- curr_state = policy->turbo;
- if (curr_state == new_state)
- return 0;
-
- policy->turbo = new_state;
- if (cpufreq_driver.update)
- {
- ret = alternative_call(cpufreq_driver.update, cpu, policy);
- if (ret)
- policy->turbo = curr_state;
- }
-
- return ret;
-}
-
-
-int cpufreq_get_turbo_status(unsigned int cpu)
-{
- struct cpufreq_policy *policy;
-
- policy = per_cpu(cpufreq_cpu_policy, cpu);
- return policy && policy->turbo == CPUFREQ_TURBO_ENABLED;
-}
-
/*********************************************************************
* POLICY *
*********************************************************************/
diff --git a/xen/include/acpi/cpufreq/cpufreq.h b/xen/include/acpi/cpufreq/cpufreq.h
index 241117a9af..0742aa9f44 100644
--- a/xen/include/acpi/cpufreq/cpufreq.h
+++ b/xen/include/acpi/cpufreq/cpufreq.h
@@ -143,9 +143,6 @@ extern int cpufreq_driver_getavg(unsigned int cpu, unsigned int flag);
#define CPUFREQ_TURBO_UNSUPPORTED 0
#define CPUFREQ_TURBO_ENABLED 1
-int cpufreq_update_turbo(unsigned int cpu, int new_state);
-int cpufreq_get_turbo_status(unsigned int cpu);
-
static inline int
__cpufreq_governor(struct cpufreq_policy *policy, unsigned int event)
{
--
2.34.1On 28.05.2025 11:16, Penny Zheng wrote: > --- a/xen/common/Kconfig > +++ b/xen/common/Kconfig > @@ -591,4 +591,11 @@ config SYSCTL > to reduce Xen footprint. > endmenu > > +config PM_OP > + bool "Enable Performance Management Operation" > + depends on ACPI && HAS_CPUFREQ && SYSCTL > + default y > + help > + This option shall enable userspace performance management control > + to do power/performance analyzing and tuning. > endmenu Why is this being added outside of the "Supported hypercall interfaces" menu? (No matter where it's added, please make sure there's a separating blank line ahdead of the subsequent "endmenu".) Jan
[Public]
Hi,
> -----Original Message-----
> From: Jan Beulich <jbeulich@suse.com>
> Sent: Tuesday, June 10, 2025 9:41 PM
> To: Penny, Zheng <penny.zheng@amd.com>
> Cc: Huang, Ray <Ray.Huang@amd.com>; Andrew Cooper
> <andrew.cooper3@citrix.com>; Roger Pau Monné <roger.pau@citrix.com>;
> Anthony PERARD <anthony.perard@vates.tech>; Orzel, Michal
> <Michal.Orzel@amd.com>; Julien Grall <julien@xen.org>; Stefano Stabellini
> <sstabellini@kernel.org>; xen-devel@lists.xenproject.org
> Subject: Re: [PATCH v4 11/20] xen/pmstat: introduce CONFIG_PM_OP
>
> On 28.05.2025 11:16, Penny Zheng wrote:
> > --- a/xen/common/Kconfig
> > +++ b/xen/common/Kconfig
> > @@ -591,4 +591,11 @@ config SYSCTL
> > to reduce Xen footprint.
> > endmenu
> >
> > +config PM_OP
> > + bool "Enable Performance Management Operation"
> > + depends on ACPI && HAS_CPUFREQ && SYSCTL
> > + default y
> > + help
> > + This option shall enable userspace performance management control
> > + to do power/performance analyzing and tuning.
> > endmenu
>
> Why is this being added outside of the "Supported hypercall interfaces"
> menu? (No matter where it's added, please make sure there's a separating blank
> line ahdead of the subsequent "endmenu".)
Because, I intended to have the only five catogory under "Supported hypercall interfaces":
- sysctl
- domctl
- hvm
- physdev
- platform
Which will be introduced in a patch serie and may contain multiple different features.
PM_OP is more or less a feature to me. And it is only dependent on SYSCTL, like LIVEPATCH, etc
>
> Jan
On 12.06.2025 09:00, Penny, Zheng wrote: > [Public] > > Hi, > >> -----Original Message----- >> From: Jan Beulich <jbeulich@suse.com> >> Sent: Tuesday, June 10, 2025 9:41 PM >> To: Penny, Zheng <penny.zheng@amd.com> >> Cc: Huang, Ray <Ray.Huang@amd.com>; Andrew Cooper >> <andrew.cooper3@citrix.com>; Roger Pau Monné <roger.pau@citrix.com>; >> Anthony PERARD <anthony.perard@vates.tech>; Orzel, Michal >> <Michal.Orzel@amd.com>; Julien Grall <julien@xen.org>; Stefano Stabellini >> <sstabellini@kernel.org>; xen-devel@lists.xenproject.org >> Subject: Re: [PATCH v4 11/20] xen/pmstat: introduce CONFIG_PM_OP >> >> On 28.05.2025 11:16, Penny Zheng wrote: >>> --- a/xen/common/Kconfig >>> +++ b/xen/common/Kconfig >>> @@ -591,4 +591,11 @@ config SYSCTL >>> to reduce Xen footprint. >>> endmenu >>> >>> +config PM_OP >>> + bool "Enable Performance Management Operation" >>> + depends on ACPI && HAS_CPUFREQ && SYSCTL >>> + default y >>> + help >>> + This option shall enable userspace performance management control >>> + to do power/performance analyzing and tuning. >>> endmenu >> >> Why is this being added outside of the "Supported hypercall interfaces" >> menu? (No matter where it's added, please make sure there's a separating blank >> line ahdead of the subsequent "endmenu".) > > Because, I intended to have the only five catogory under "Supported hypercall interfaces": > - sysctl > - domctl > - hvm > - physdev > - platform > Which will be introduced in a patch serie and may contain multiple different features. > PM_OP is more or less a feature to me. And it is only dependent on SYSCTL, like LIVEPATCH, etc Hmm, well, okay. Especially your reference to LIVEPATCH here suggests to me that some re-ordering would then be quite desirable in the file. It is generally pretty desirable for "depends on <feat>" to appear later than "<feat>". While for the graphical UI interfaces of kconfig the difference may be marginal, for the command line one you otherwise get to select the setting for the dependent feature first, just for that setting potentially being undone by the later selection for <feat>. Then again the entire file could do with some rearrangements. Yet I guess to avoid interfering with your work, I'd better postpone any tidying work there. Jan
© 2016 - 2025 Red Hat, Inc.