linux-stable/arch/x86/kernel/cpu/umwait.c

239 lines
6 KiB
C
Raw Normal View History

x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
// SPDX-License-Identifier: GPL-2.0
#include <linux/syscore_ops.h>
#include <linux/suspend.h>
#include <linux/cpu.h>
#include <asm/msr.h>
#include <asm/mwait.h>
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
#define UMWAIT_C02_ENABLE 0
2019-06-20 01:33:56 +00:00
#define UMWAIT_CTRL_VAL(max_time, c02_disable) \
(((max_time) & MSR_IA32_UMWAIT_CONTROL_TIME_MASK) | \
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
((c02_disable) & MSR_IA32_UMWAIT_CONTROL_C02_DISABLE))
/*
* Cache IA32_UMWAIT_CONTROL MSR. This is a systemwide control. By default,
* umwait max time is 100000 in TSC-quanta and C0.2 is enabled
*/
static u32 umwait_control_cached = UMWAIT_CTRL_VAL(100000, UMWAIT_C02_ENABLE);
x86/umwait: Fix error handling in umwait_init() Currently, failure of cpuhp_setup_state() is ignored and the syscore ops and the control interfaces can still be added even after the failure. But, this error handling will cause a few issues: 1. The CPUs may have different values in the IA32_UMWAIT_CONTROL MSR because there is no way to roll back the control MSR on the CPUs which already set the MSR before the failure. 2. If the sysfs interface is added successfully, there will be a mismatch between the global control value and the control MSR: - The interface shows the default global control value. But, the control MSR is not set to the value because the CPU online function, which is supposed to set the MSR to the value, is not installed. - If the sysadmin changes the global control value through the interface, the control MSR on all current online CPUs is set to the new value. But, the control MSR on newly onlined CPUs after the value change will not be set to the new value due to lack of the CPU online function. 3. On resume from suspend/hibernation, the boot CPU restores the control MSR to the global control value through the syscore ops. But, the control MSR on all APs is not set due to lake of the CPU online function. To solve the issues and enforce consistent behavior on the failure of the CPU hotplug setup, make the following changes: 1. Cache the original control MSR value which is configured by hardware or BIOS before kernel boot. This value is likely to be 0. But it could be a different number as well. Cache the control MSR only once before the MSR is changed. 2. Add the CPU offline function so that the MSR is restored to the original control value on all CPUs on the failure. 3. On the failure, exit from cpumait_init() so that the syscore ops and the control interfaces are not added. Reported-by: Valdis Kletnieks <valdis.kletnieks@vt.edu> Suggested-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Link: https://lkml.kernel.org/r/1565401237-60936-1-git-send-email-fenghua.yu@intel.com
2019-08-10 01:40:37 +00:00
/*
* Cache the original IA32_UMWAIT_CONTROL MSR value which is configured by
* hardware or BIOS before kernel boot.
*/
static u32 orig_umwait_control_cached __ro_after_init;
2019-06-20 01:33:56 +00:00
/*
* Serialize access to umwait_control_cached and IA32_UMWAIT_CONTROL MSR in
* the sysfs write functions.
*/
static DEFINE_MUTEX(umwait_lock);
static void umwait_update_control_msr(void * unused)
{
lockdep_assert_irqs_disabled();
wrmsr(MSR_IA32_UMWAIT_CONTROL, READ_ONCE(umwait_control_cached), 0);
}
/*
* The CPU hotplug callback sets the control MSR to the global control
* value.
*
* Disable interrupts so the read of umwait_control_cached and the WRMSR
* are protected against a concurrent sysfs write. Otherwise the sysfs
* write could update the cached value after it had been read on this CPU
* and issue the IPI before the old value had been written. The IPI would
* interrupt, write the new value and after return from IPI the previous
* value would be written by this CPU.
*
* With interrupts disabled the upcoming CPU either sees the new control
* value or the IPI is updating this CPU to the new control value after
* interrupts have been reenabled.
*/
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
static int umwait_cpu_online(unsigned int cpu)
{
2019-06-20 01:33:56 +00:00
local_irq_disable();
umwait_update_control_msr(NULL);
local_irq_enable();
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
return 0;
}
x86/umwait: Fix error handling in umwait_init() Currently, failure of cpuhp_setup_state() is ignored and the syscore ops and the control interfaces can still be added even after the failure. But, this error handling will cause a few issues: 1. The CPUs may have different values in the IA32_UMWAIT_CONTROL MSR because there is no way to roll back the control MSR on the CPUs which already set the MSR before the failure. 2. If the sysfs interface is added successfully, there will be a mismatch between the global control value and the control MSR: - The interface shows the default global control value. But, the control MSR is not set to the value because the CPU online function, which is supposed to set the MSR to the value, is not installed. - If the sysadmin changes the global control value through the interface, the control MSR on all current online CPUs is set to the new value. But, the control MSR on newly onlined CPUs after the value change will not be set to the new value due to lack of the CPU online function. 3. On resume from suspend/hibernation, the boot CPU restores the control MSR to the global control value through the syscore ops. But, the control MSR on all APs is not set due to lake of the CPU online function. To solve the issues and enforce consistent behavior on the failure of the CPU hotplug setup, make the following changes: 1. Cache the original control MSR value which is configured by hardware or BIOS before kernel boot. This value is likely to be 0. But it could be a different number as well. Cache the control MSR only once before the MSR is changed. 2. Add the CPU offline function so that the MSR is restored to the original control value on all CPUs on the failure. 3. On the failure, exit from cpumait_init() so that the syscore ops and the control interfaces are not added. Reported-by: Valdis Kletnieks <valdis.kletnieks@vt.edu> Suggested-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Link: https://lkml.kernel.org/r/1565401237-60936-1-git-send-email-fenghua.yu@intel.com
2019-08-10 01:40:37 +00:00
/*
* The CPU hotplug callback sets the control MSR to the original control
* value.
*/
static int umwait_cpu_offline(unsigned int cpu)
{
/*
* This code is protected by the CPU hotplug already and
* orig_umwait_control_cached is never changed after it caches
* the original control MSR value in umwait_init(). So there
* is no race condition here.
*/
wrmsr(MSR_IA32_UMWAIT_CONTROL, orig_umwait_control_cached, 0);
return 0;
}
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
/*
* On resume, restore IA32_UMWAIT_CONTROL MSR on the boot processor which
* is the only active CPU at this time. The MSR is set up on the APs via the
* CPU hotplug callback.
*
* This function is invoked on resume from suspend and hibernation. On
* resume from suspend the restore should be not required, but we neither
* trust the firmware nor does it matter if the same value is written
* again.
*/
static void umwait_syscore_resume(void)
{
2019-06-20 01:33:56 +00:00
umwait_update_control_msr(NULL);
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
}
static struct syscore_ops umwait_syscore_ops = {
.resume = umwait_syscore_resume,
};
2019-06-20 01:33:56 +00:00
/* sysfs interface */
/*
* When bit 0 in IA32_UMWAIT_CONTROL MSR is 1, C0.2 is disabled.
* Otherwise, C0.2 is enabled.
*/
static inline bool umwait_ctrl_c02_enabled(u32 ctrl)
{
return !(ctrl & MSR_IA32_UMWAIT_CONTROL_C02_DISABLE);
}
static inline u32 umwait_ctrl_max_time(u32 ctrl)
{
return ctrl & MSR_IA32_UMWAIT_CONTROL_TIME_MASK;
}
static inline void umwait_update_control(u32 maxtime, bool c02_enable)
{
u32 ctrl = maxtime & MSR_IA32_UMWAIT_CONTROL_TIME_MASK;
if (!c02_enable)
ctrl |= MSR_IA32_UMWAIT_CONTROL_C02_DISABLE;
WRITE_ONCE(umwait_control_cached, ctrl);
/* Propagate to all CPUs */
on_each_cpu(umwait_update_control_msr, NULL, 1);
}
static ssize_t
enable_c02_show(struct device *dev, struct device_attribute *attr, char *buf)
{
u32 ctrl = READ_ONCE(umwait_control_cached);
return sprintf(buf, "%d\n", umwait_ctrl_c02_enabled(ctrl));
}
static ssize_t enable_c02_store(struct device *dev,
struct device_attribute *attr,
const char *buf, size_t count)
{
bool c02_enable;
u32 ctrl;
int ret;
ret = kstrtobool(buf, &c02_enable);
if (ret)
return ret;
mutex_lock(&umwait_lock);
ctrl = READ_ONCE(umwait_control_cached);
if (c02_enable != umwait_ctrl_c02_enabled(ctrl))
umwait_update_control(ctrl, c02_enable);
mutex_unlock(&umwait_lock);
return count;
}
static DEVICE_ATTR_RW(enable_c02);
static ssize_t
max_time_show(struct device *kobj, struct device_attribute *attr, char *buf)
{
u32 ctrl = READ_ONCE(umwait_control_cached);
return sprintf(buf, "%u\n", umwait_ctrl_max_time(ctrl));
}
static ssize_t max_time_store(struct device *kobj,
struct device_attribute *attr,
const char *buf, size_t count)
{
u32 max_time, ctrl;
int ret;
ret = kstrtou32(buf, 0, &max_time);
if (ret)
return ret;
/* bits[1:0] must be zero */
if (max_time & ~MSR_IA32_UMWAIT_CONTROL_TIME_MASK)
return -EINVAL;
mutex_lock(&umwait_lock);
ctrl = READ_ONCE(umwait_control_cached);
if (max_time != umwait_ctrl_max_time(ctrl))
umwait_update_control(max_time, umwait_ctrl_c02_enabled(ctrl));
mutex_unlock(&umwait_lock);
return count;
}
static DEVICE_ATTR_RW(max_time);
2019-06-20 01:33:56 +00:00
static struct attribute *umwait_attrs[] = {
&dev_attr_enable_c02.attr,
&dev_attr_max_time.attr,
2019-06-20 01:33:56 +00:00
NULL
};
static struct attribute_group umwait_attr_group = {
.attrs = umwait_attrs,
.name = "umwait_control",
};
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
static int __init umwait_init(void)
{
2019-06-20 01:33:56 +00:00
struct device *dev;
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
int ret;
if (!boot_cpu_has(X86_FEATURE_WAITPKG))
return -ENODEV;
x86/umwait: Fix error handling in umwait_init() Currently, failure of cpuhp_setup_state() is ignored and the syscore ops and the control interfaces can still be added even after the failure. But, this error handling will cause a few issues: 1. The CPUs may have different values in the IA32_UMWAIT_CONTROL MSR because there is no way to roll back the control MSR on the CPUs which already set the MSR before the failure. 2. If the sysfs interface is added successfully, there will be a mismatch between the global control value and the control MSR: - The interface shows the default global control value. But, the control MSR is not set to the value because the CPU online function, which is supposed to set the MSR to the value, is not installed. - If the sysadmin changes the global control value through the interface, the control MSR on all current online CPUs is set to the new value. But, the control MSR on newly onlined CPUs after the value change will not be set to the new value due to lack of the CPU online function. 3. On resume from suspend/hibernation, the boot CPU restores the control MSR to the global control value through the syscore ops. But, the control MSR on all APs is not set due to lake of the CPU online function. To solve the issues and enforce consistent behavior on the failure of the CPU hotplug setup, make the following changes: 1. Cache the original control MSR value which is configured by hardware or BIOS before kernel boot. This value is likely to be 0. But it could be a different number as well. Cache the control MSR only once before the MSR is changed. 2. Add the CPU offline function so that the MSR is restored to the original control value on all CPUs on the failure. 3. On the failure, exit from cpumait_init() so that the syscore ops and the control interfaces are not added. Reported-by: Valdis Kletnieks <valdis.kletnieks@vt.edu> Suggested-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Link: https://lkml.kernel.org/r/1565401237-60936-1-git-send-email-fenghua.yu@intel.com
2019-08-10 01:40:37 +00:00
/*
* Cache the original control MSR value before the control MSR is
* changed. This is the only place where orig_umwait_control_cached
* is modified.
*/
rdmsrl(MSR_IA32_UMWAIT_CONTROL, orig_umwait_control_cached);
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
ret = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "umwait:online",
x86/umwait: Fix error handling in umwait_init() Currently, failure of cpuhp_setup_state() is ignored and the syscore ops and the control interfaces can still be added even after the failure. But, this error handling will cause a few issues: 1. The CPUs may have different values in the IA32_UMWAIT_CONTROL MSR because there is no way to roll back the control MSR on the CPUs which already set the MSR before the failure. 2. If the sysfs interface is added successfully, there will be a mismatch between the global control value and the control MSR: - The interface shows the default global control value. But, the control MSR is not set to the value because the CPU online function, which is supposed to set the MSR to the value, is not installed. - If the sysadmin changes the global control value through the interface, the control MSR on all current online CPUs is set to the new value. But, the control MSR on newly onlined CPUs after the value change will not be set to the new value due to lack of the CPU online function. 3. On resume from suspend/hibernation, the boot CPU restores the control MSR to the global control value through the syscore ops. But, the control MSR on all APs is not set due to lake of the CPU online function. To solve the issues and enforce consistent behavior on the failure of the CPU hotplug setup, make the following changes: 1. Cache the original control MSR value which is configured by hardware or BIOS before kernel boot. This value is likely to be 0. But it could be a different number as well. Cache the control MSR only once before the MSR is changed. 2. Add the CPU offline function so that the MSR is restored to the original control value on all CPUs on the failure. 3. On the failure, exit from cpumait_init() so that the syscore ops and the control interfaces are not added. Reported-by: Valdis Kletnieks <valdis.kletnieks@vt.edu> Suggested-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Link: https://lkml.kernel.org/r/1565401237-60936-1-git-send-email-fenghua.yu@intel.com
2019-08-10 01:40:37 +00:00
umwait_cpu_online, umwait_cpu_offline);
if (ret < 0) {
/*
* On failure, the control MSR on all CPUs has the
* original control value.
*/
return ret;
}
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
register_syscore_ops(&umwait_syscore_ops);
2019-06-20 01:33:56 +00:00
/*
* Add umwait control interface. Ignore failure, so at least the
* default values are set up in case the machine manages to boot.
*/
dev = cpu_subsys.dev_root;
return sysfs_create_group(&dev->kobj, &umwait_attr_group);
x86/umwait: Initialize umwait control values umwait or tpause allows the processor to enter a light-weight power/performance optimized state (C0.1 state) or an improved power/performance optimized state (C0.2 state) for a period specified by the instruction or until the system time limit or until a store to the monitored address range in umwait. IA32_UMWAIT_CONTROL MSR register allows the OS to enable/disable C0.2 on the processor and to set the maximum time the processor can reside in C0.1 or C0.2. By default C0.2 is enabled so the user wait instructions can enter the C0.2 state to save more power with slower wakeup time. Andy Lutomirski proposed to set the maximum umwait time to 100000 cycles by default. A quote from Andy: "What I want to avoid is the case where it works dramatically differently on NO_HZ_FULL systems as compared to everything else. Also, UMWAIT may behave a bit differently if the max timeout is hit, and I'd like that path to get exercised widely by making it happen even on default configs." A sysfs interface to adjust the time and the C0.2 enablement is provided in a follow up change. [ tglx: Renamed MSR_IA32_UMWAIT_CONTROL_MAX_TIME to MSR_IA32_UMWAIT_CONTROL_TIME_MASK because the constant is used as mask throughout the code. Massaged comments and changelog ] Signed-off-by: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Ashok Raj <ashok.raj@intel.com> Reviewed-by: Andy Lutomirski <luto@kernel.org> Cc: "Borislav Petkov" <bp@alien8.de> Cc: "H Peter Anvin" <hpa@zytor.com> Cc: "Peter Zijlstra" <peterz@infradead.org> Cc: "Tony Luck" <tony.luck@intel.com> Cc: "Ravi V Shankar" <ravi.v.shankar@intel.com> Link: https://lkml.kernel.org/r/1560994438-235698-3-git-send-email-fenghua.yu@intel.com
2019-06-20 01:33:55 +00:00
}
device_initcall(umwait_init);