| Message ID | 1558969089-13204-3-git-send-email-mikelley@microsoft.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Drivers: hv: Move Hyper-V clock/timer code to separate clocksource driver | expand |
Michael Kelley <mikelley@microsoft.com> writes: > Code for the Hyper-V specific clocksources is currently mixed > in with other Hyper-V code. Move the code to the Hyper-V specific > driver in the "clocksource" directory, while separating out > ISA dependencies so that the clocksource driver remains ISA > independent. Update the Hyper-V initialization code to call > initialization and cleanup routines since the Hyper-V synthetic > timers are not independently enumerated in ACPI. Update Hyper-V > clocksource users KVM and VDSO to get definitions from a new > include file. > > No behavior is changed and no new functionality is added. > > Suggested-by: Marc Zyngier <marc.zyngier@arm.com> > Signed-off-by: Michael Kelley <mikelley@microsoft.com> > --- > arch/x86/entry/vdso/vclock_gettime.c | 1 + > arch/x86/entry/vdso/vma.c | 2 +- > arch/x86/hyperv/hv_init.c | 91 +----------------------- > arch/x86/include/asm/mshyperv.h | 81 +++------------------- > arch/x86/kvm/x86.c | 1 + > drivers/clocksource/hyperv_timer.c | 130 +++++++++++++++++++++++++++++++++++ > drivers/hv/hv_util.c | 1 + > include/clocksource/hyperv_timer.h | 78 +++++++++++++++++++++ > 8 files changed, 226 insertions(+), 159 deletions(-) > > diff --git a/arch/x86/entry/vdso/vclock_gettime.c b/arch/x86/entry/vdso/vclock_gettime.c > index 98c7d12..7983ca2 100644 > --- a/arch/x86/entry/vdso/vclock_gettime.c > +++ b/arch/x86/entry/vdso/vclock_gettime.c > @@ -21,6 +21,7 @@ > #include <linux/math64.h> > #include <linux/time.h> > #include <linux/kernel.h> > +#include <clocksource/hyperv_timer.h> > > #define gtod (&VVAR(vsyscall_gtod_data)) > > diff --git a/arch/x86/entry/vdso/vma.c b/arch/x86/entry/vdso/vma.c > index babc4e7..99b38e9 100644 > --- a/arch/x86/entry/vdso/vma.c > +++ b/arch/x86/entry/vdso/vma.c > @@ -22,7 +22,7 @@ > #include <asm/page.h> > #include <asm/desc.h> > #include <asm/cpufeature.h> > -#include <asm/mshyperv.h> > +#include <clocksource/hyperv_timer.h> > > #if defined(CONFIG_X86_64) > unsigned int __read_mostly vdso64_enabled = 1; > diff --git a/arch/x86/hyperv/hv_init.c b/arch/x86/hyperv/hv_init.c > index e4ba467..79f626a 100644 > --- a/arch/x86/hyperv/hv_init.c > +++ b/arch/x86/hyperv/hv_init.c > @@ -27,64 +27,13 @@ > #include <linux/version.h> > #include <linux/vmalloc.h> > #include <linux/mm.h> > -#include <linux/clockchips.h> > #include <linux/hyperv.h> > #include <linux/slab.h> > #include <linux/cpuhotplug.h> > - > -#ifdef CONFIG_HYPERV_TSCPAGE > - > -static struct ms_hyperv_tsc_page *tsc_pg; > - > -struct ms_hyperv_tsc_page *hv_get_tsc_page(void) > -{ > - return tsc_pg; > -} > -EXPORT_SYMBOL_GPL(hv_get_tsc_page); > - > -static u64 read_hv_clock_tsc(struct clocksource *arg) > -{ > - u64 current_tick = hv_read_tsc_page(tsc_pg); > - > - if (current_tick == U64_MAX) > - rdmsrl(HV_X64_MSR_TIME_REF_COUNT, current_tick); > - > - return current_tick; > -} > - > -static struct clocksource hyperv_cs_tsc = { > - .name = "hyperv_clocksource_tsc_page", > - .rating = 400, > - .read = read_hv_clock_tsc, > - .mask = CLOCKSOURCE_MASK(64), > - .flags = CLOCK_SOURCE_IS_CONTINUOUS, > -}; > -#endif > - > -static u64 read_hv_clock_msr(struct clocksource *arg) > -{ > - u64 current_tick; > - /* > - * Read the partition counter to get the current tick count. This count > - * is set to 0 when the partition is created and is incremented in > - * 100 nanosecond units. > - */ > - rdmsrl(HV_X64_MSR_TIME_REF_COUNT, current_tick); > - return current_tick; > -} > - > -static struct clocksource hyperv_cs_msr = { > - .name = "hyperv_clocksource_msr", > - .rating = 400, > - .read = read_hv_clock_msr, > - .mask = CLOCKSOURCE_MASK(64), > - .flags = CLOCK_SOURCE_IS_CONTINUOUS, > -}; > +#include <clocksource/hyperv_timer.h> > > void *hv_hypercall_pg; > EXPORT_SYMBOL_GPL(hv_hypercall_pg); > -struct clocksource *hyperv_cs; > -EXPORT_SYMBOL_GPL(hyperv_cs); > > u32 *hv_vp_index; > EXPORT_SYMBOL_GPL(hv_vp_index); > @@ -353,42 +302,8 @@ void __init hyperv_init(void) > > x86_init.pci.arch_init = hv_pci_init; > > - /* > - * Register Hyper-V specific clocksource. > - */ > -#ifdef CONFIG_HYPERV_TSCPAGE > - if (ms_hyperv.features & HV_MSR_REFERENCE_TSC_AVAILABLE) { > - union hv_x64_msr_hypercall_contents tsc_msr; > - > - tsc_pg = __vmalloc(PAGE_SIZE, GFP_KERNEL, PAGE_KERNEL); > - if (!tsc_pg) > - goto register_msr_cs; > - > - hyperv_cs = &hyperv_cs_tsc; > - > - rdmsrl(HV_X64_MSR_REFERENCE_TSC, tsc_msr.as_uint64); > - > - tsc_msr.enable = 1; > - tsc_msr.guest_physical_address = vmalloc_to_pfn(tsc_pg); > - > - wrmsrl(HV_X64_MSR_REFERENCE_TSC, tsc_msr.as_uint64); > - > - hyperv_cs_tsc.archdata.vclock_mode = VCLOCK_HVCLOCK; > - > - clocksource_register_hz(&hyperv_cs_tsc, NSEC_PER_SEC/100); > - return; > - } > -register_msr_cs: > -#endif > - /* > - * For 32 bit guests just use the MSR based mechanism for reading > - * the partition counter. > - */ > - > - hyperv_cs = &hyperv_cs_msr; > - if (ms_hyperv.features & HV_MSR_TIME_REF_COUNT_AVAILABLE) > - clocksource_register_hz(&hyperv_cs_msr, NSEC_PER_SEC/100); > - > + /* Register Hyper-V specific clocksource */ > + hv_init_clocksource(); > return; > > remove_cpuhp_state: > diff --git a/arch/x86/include/asm/mshyperv.h b/arch/x86/include/asm/mshyperv.h > index cc60e61..f4fa8a9 100644 > --- a/arch/x86/include/asm/mshyperv.h > +++ b/arch/x86/include/asm/mshyperv.h > @@ -105,6 +105,17 @@ static inline void vmbus_signal_eom(struct hv_message *msg, u32 old_msg_type) > #define hv_get_crash_ctl(val) \ > rdmsrl(HV_X64_MSR_CRASH_CTL, val) > > +#define hv_get_time_ref_count(val) \ > + rdmsrl(HV_X64_MSR_TIME_REF_COUNT, val) > + > +#define hv_get_reference_tsc(val) \ > + rdmsrl(HV_X64_MSR_REFERENCE_TSC, val) > +#define hv_set_reference_tsc(val) \ > + wrmsrl(HV_X64_MSR_REFERENCE_TSC, val) > +#define hv_set_clocksource_vdso(val) \ > + ((val).archdata.vclock_mode = VCLOCK_HVCLOCK) > +#define hv_get_raw_timer() rdtsc_ordered() > + > void hyperv_callback_vector(void); > void hyperv_reenlightenment_vector(void); > #ifdef CONFIG_TRACING > @@ -133,7 +144,6 @@ static inline void hv_disable_stimer0_percpu_irq(int irq) {} > > > #if IS_ENABLED(CONFIG_HYPERV) > -extern struct clocksource *hyperv_cs; > extern void *hv_hypercall_pg; > extern void __percpu **hyperv_pcpu_input_arg; > > @@ -387,73 +397,4 @@ static inline int hyperv_flush_guest_mapping_range(u64 as, > } > #endif /* CONFIG_HYPERV */ > > -#ifdef CONFIG_HYPERV_TSCPAGE > -struct ms_hyperv_tsc_page *hv_get_tsc_page(void); > -static inline u64 hv_read_tsc_page_tsc(const struct ms_hyperv_tsc_page *tsc_pg, > - u64 *cur_tsc) > -{ > - u64 scale, offset; > - u32 sequence; > - > - /* > - * The protocol for reading Hyper-V TSC page is specified in Hypervisor > - * Top-Level Functional Specification ver. 3.0 and above. To get the > - * reference time we must do the following: > - * - READ ReferenceTscSequence > - * A special '0' value indicates the time source is unreliable and we > - * need to use something else. The currently published specification > - * versions (up to 4.0b) contain a mistake and wrongly claim '-1' > - * instead of '0' as the special value, see commit c35b82ef0294. > - * - ReferenceTime = > - * ((RDTSC() * ReferenceTscScale) >> 64) + ReferenceTscOffset > - * - READ ReferenceTscSequence again. In case its value has changed > - * since our first reading we need to discard ReferenceTime and repeat > - * the whole sequence as the hypervisor was updating the page in > - * between. > - */ > - do { > - sequence = READ_ONCE(tsc_pg->tsc_sequence); > - if (!sequence) > - return U64_MAX; > - /* > - * Make sure we read sequence before we read other values from > - * TSC page. > - */ > - smp_rmb(); > - > - scale = READ_ONCE(tsc_pg->tsc_scale); > - offset = READ_ONCE(tsc_pg->tsc_offset); > - *cur_tsc = rdtsc_ordered(); > - > - /* > - * Make sure we read sequence after we read all other values > - * from TSC page. > - */ > - smp_rmb(); > - > - } while (READ_ONCE(tsc_pg->tsc_sequence) != sequence); > - > - return mul_u64_u64_shr(*cur_tsc, scale, 64) + offset; > -} > - > -static inline u64 hv_read_tsc_page(const struct ms_hyperv_tsc_page *tsc_pg) > -{ > - u64 cur_tsc; > - > - return hv_read_tsc_page_tsc(tsc_pg, &cur_tsc); > -} > - > -#else > -static inline struct ms_hyperv_tsc_page *hv_get_tsc_page(void) > -{ > - return NULL; > -} > - > -static inline u64 hv_read_tsc_page_tsc(const struct ms_hyperv_tsc_page *tsc_pg, > - u64 *cur_tsc) > -{ > - BUG(); > - return U64_MAX; > -} > -#endif > #endif > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > index 536b78c..3fbaac0 100644 > --- a/arch/x86/kvm/x86.c > +++ b/arch/x86/kvm/x86.c > @@ -70,6 +70,7 @@ > #include <asm/mshyperv.h> > #include <asm/hypervisor.h> > #include <asm/intel_pt.h> > +#include <clocksource/hyperv_timer.h> > > #define CREATE_TRACE_POINTS > #include "trace.h" > diff --git a/drivers/clocksource/hyperv_timer.c b/drivers/clocksource/hyperv_timer.c > index 30615f3..274614d 100644 > --- a/drivers/clocksource/hyperv_timer.c > +++ b/drivers/clocksource/hyperv_timer.c > @@ -14,6 +14,8 @@ > #include <linux/percpu.h> > #include <linux/cpumask.h> > #include <linux/clockchips.h> > +#include <linux/clocksource.h> > +#include <linux/sched_clock.h> > #include <linux/mm.h> > #include <clocksource/hyperv_timer.h> > #include <asm/hyperv-tlfs.h> > @@ -189,3 +191,131 @@ void hv_stimer_global_cleanup(void) > hv_stimer_free(); > } > EXPORT_SYMBOL_GPL(hv_stimer_global_cleanup); > + > +/* > + * Code and definitions for the Hyper-V clocksources. Two > + * clocksources are defined: one that reads the Hyper-V defined MSR, and > + * the other that uses the TSC reference page feature as defined in the > + * TLFS. The MSR version is for compatibility with old versions of > + * Hyper-V and 32-bit x86. The TSC reference page version is preferred. > + */ > + > +struct clocksource *hyperv_cs; > +EXPORT_SYMBOL_GPL(hyperv_cs); > + > +#ifdef CONFIG_HYPERV_TSCPAGE > + > +static struct ms_hyperv_tsc_page *tsc_pg; > + > +struct ms_hyperv_tsc_page *hv_get_tsc_page(void) > +{ > + return tsc_pg; > +} > +EXPORT_SYMBOL_GPL(hv_get_tsc_page); > + > +static u64 read_hv_sched_clock_tsc(void) > +{ > + u64 current_tick = hv_read_tsc_page(tsc_pg); > + > + if (current_tick == U64_MAX) > + hv_get_time_ref_count(current_tick); > + > + return current_tick; > +} > + > +static u64 read_hv_clock_tsc(struct clocksource *arg) > +{ > + return read_hv_sched_clock_tsc(); > +} > + > +static struct clocksource hyperv_cs_tsc = { > + .name = "hyperv_clocksource_tsc_page", > + .rating = 400, > + .read = read_hv_clock_tsc, > + .mask = CLOCKSOURCE_MASK(64), > + .flags = CLOCK_SOURCE_IS_CONTINUOUS, > +}; > +#endif > + > +static u64 read_hv_sched_clock_msr(void) > +{ > + u64 current_tick; > + /* > + * Read the partition counter to get the current tick count. This count > + * is set to 0 when the partition is created and is incremented in > + * 100 nanosecond units. > + */ > + hv_get_time_ref_count(current_tick); > + return current_tick; > +} > + > +static u64 read_hv_clock_msr(struct clocksource *arg) > +{ > + return read_hv_sched_clock_msr(); > +} > + > +static struct clocksource hyperv_cs_msr = { > + .name = "hyperv_clocksource_msr", > + .rating = 400, > + .read = read_hv_clock_msr, > + .mask = CLOCKSOURCE_MASK(64), > + .flags = CLOCK_SOURCE_IS_CONTINUOUS, > +}; > + > +void __init hv_init_clocksource(void) > +{ > +#ifdef CONFIG_HYPERV_TSCPAGE > + if (ms_hyperv.features & HV_MSR_REFERENCE_TSC_AVAILABLE) { > + > + u64 tsc_msr; > + phys_addr_t phys_addr; > + > + tsc_pg = vmalloc(PAGE_SIZE); > + if (!tsc_pg) > + goto register_msr_cs; > + > + hyperv_cs = &hyperv_cs_tsc; > + phys_addr = page_to_phys(vmalloc_to_page(tsc_pg)); > + > + /* The Hyper-V TLFS specifies to preserve the value of > + * reserved bits in registers. So read the existing > + * value, preserve the low order 12 bits, and add in > + * the guest physical address (which already has at > + * least the low 12 bits set to zero since it is page > + * aligned). Also set the "enable" bit, which is bit 0. > + */ > + hv_get_reference_tsc(tsc_msr); > + tsc_msr &= GENMASK_ULL(11, 0); > + tsc_msr = tsc_msr | 0x1 | (u64)phys_addr; > + hv_set_reference_tsc(tsc_msr); > + > + hv_set_clocksource_vdso(hyperv_cs_tsc); > + clocksource_register_hz(&hyperv_cs_tsc, NSEC_PER_SEC/100); > + > + /* > + * sched_clock_register is needed on ARM64 but > + * is a no-op on x86 > + */ > + sched_clock_register(read_hv_sched_clock_tsc, > + 64, HV_CLOCK_HZ); > + return; > + } > +register_msr_cs: > +#endif > + /* > + * For 32 bit guests just use the MSR based mechanism for reading > + * the partition counter. > + */ > + hyperv_cs = &hyperv_cs_msr; > + if (ms_hyperv.features & HV_MSR_TIME_REF_COUNT_AVAILABLE) { > + clocksource_register_hz(&hyperv_cs_msr, NSEC_PER_SEC/100); > + > + /* > + * sched_clock_register is needed on ARM64 but > + * is a no-op on x86 > + */ > + sched_clock_register(read_hv_sched_clock_msr, > + 64, HV_CLOCK_HZ); I'm not sure about ARM, but MSR-based clocksource would be a really bad choice for sched clock on x86, this will slow things down significantly. Luckily, as you're validly stating above, sched_clock_register() is a no-op on x86 as we don't define CONFIG_GENERIC_SCHED_CLOCK. Can we actually *not* do sched_clock_register() in case TSC page is unavailable (and revert to counting jiffies or whatever)? > + } > +} > +EXPORT_SYMBOL_GPL(hv_init_clocksource); > diff --git a/drivers/hv/hv_util.c b/drivers/hv/hv_util.c > index f10eeb1..9eff85e 100644 > --- a/drivers/hv/hv_util.c > +++ b/drivers/hv/hv_util.c > @@ -29,6 +29,7 @@ > #include <linux/hyperv.h> > #include <linux/clockchips.h> > #include <linux/ptp_clock_kernel.h> > +#include <clocksource/hyperv_timer.h> > #include <asm/mshyperv.h> > > #include "hyperv_vmbus.h" > diff --git a/include/clocksource/hyperv_timer.h b/include/clocksource/hyperv_timer.h > index ba5dc17..e17e8a2 100644 > --- a/include/clocksource/hyperv_timer.h > +++ b/include/clocksource/hyperv_timer.h > @@ -13,6 +13,10 @@ > #ifndef __CLKSOURCE_HYPERV_TIMER_H > #define __CLKSOURCE_HYPERV_TIMER_H > > +#include <linux/clocksource.h> > +#include <linux/math64.h> > +#include <asm/mshyperv.h> > + > #define HV_MAX_MAX_DELTA_TICKS 0xffffffff > #define HV_MIN_DELTA_TICKS 1 > > @@ -24,4 +28,78 @@ > extern void hv_stimer_global_cleanup(void); > extern void hv_stimer0_isr(void); > > +#if IS_ENABLED(CONFIG_HYPERV) > +extern struct clocksource *hyperv_cs; > +extern void hv_init_clocksource(void); > +#endif /* CONFIG_HYPERV */ > + > +#ifdef CONFIG_HYPERV_TSCPAGE > +extern struct ms_hyperv_tsc_page *hv_get_tsc_page(void); > +static inline u64 hv_read_tsc_page_tsc(const struct ms_hyperv_tsc_page *tsc_pg, > + u64 *cur_tsc) > +{ > + u64 scale, offset; > + u32 sequence; > + > + /* > + * The protocol for reading Hyper-V TSC page is specified in Hypervisor > + * Top-Level Functional Specification ver. 3.0 and above. To get the > + * reference time we must do the following: > + * - READ ReferenceTscSequence > + * A special '0' value indicates the time source is unreliable and we > + * need to use something else. The currently published specification > + * versions (up to 4.0b) contain a mistake and wrongly claim '-1' > + * instead of '0' as the special value, see commit c35b82ef0294. > + * - ReferenceTime = > + * ((RDTSC() * ReferenceTscScale) >> 64) + ReferenceTscOffset > + * - READ ReferenceTscSequence again. In case its value has changed > + * since our first reading we need to discard ReferenceTime and repeat > + * the whole sequence as the hypervisor was updating the page in > + * between. > + */ > + do { > + sequence = READ_ONCE(tsc_pg->tsc_sequence); > + if (!sequence) > + return U64_MAX; > + /* > + * Make sure we read sequence before we read other values from > + * TSC page. > + */ > + smp_rmb(); > + > + scale = READ_ONCE(tsc_pg->tsc_scale); > + offset = READ_ONCE(tsc_pg->tsc_offset); > + *cur_tsc = hv_get_raw_timer(); > + > + /* > + * Make sure we read sequence after we read all other values > + * from TSC page. > + */ > + smp_rmb(); > + > + } while (READ_ONCE(tsc_pg->tsc_sequence) != sequence); > + > + return mul_u64_u64_shr(*cur_tsc, scale, 64) + offset; > +} > + > +static inline u64 hv_read_tsc_page(const struct ms_hyperv_tsc_page *tsc_pg) > +{ > + u64 cur_tsc; > + > + return hv_read_tsc_page_tsc(tsc_pg, &cur_tsc); > +} > + > +#else /* CONFIG_HYPERV_TSC_PAGE */ > +static inline struct ms_hyperv_tsc_page *hv_get_tsc_page(void) > +{ > + return NULL; > +} > + > +static inline u64 hv_read_tsc_page_tsc(const struct ms_hyperv_tsc_page *tsc_pg, > + u64 *cur_tsc) > +{ > + return U64_MAX; > +} > +#endif /* CONFIG_HYPERV_TSCPAGE */ > + > #endif With the (theoretical) question above answered, Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
From: Vitaly Kuznetsov <vkuznets@redhat.com> Sent: Thursday, May 30, 2019 2:48 AM > > + /* > > + * sched_clock_register is needed on ARM64 but > > + * is a no-op on x86 > > + */ > > + sched_clock_register(read_hv_sched_clock_msr, > > + 64, HV_CLOCK_HZ); > > I'm not sure about ARM, but MSR-based clocksource would be a really bad > choice for sched clock on x86, this will slow things down > significantly. Luckily, as you're validly stating above, > sched_clock_register() is a no-op on x86 as we don't define > CONFIG_GENERIC_SCHED_CLOCK. > > Can we actually *not* do sched_clock_register() in case > TSC page is unavailable (and revert to counting jiffies or whatever)? > We can't skip the sched_clock_register() on ARM64 because it does define CONFIG_GENERIC_SCHED_CLOCK. However, Hyper-V should always provide REFERENCE_TSC_AVAILALBE on ARM64, so we should never end up in the MSR-based code on ARM64. Arguably that means the call to sched_clock_register() could be removed since it's a no-op on x86. But I'd like to keep it for symmetry and in case there's a testing/debugging situation on ARM64 where we want to clear REFERENCE_TSC_AVAILABLE and go down the MSR-based code path. Michael
Michael Kelley <mikelley@microsoft.com> writes: > From: Vitaly Kuznetsov <vkuznets@redhat.com> Sent: Thursday, May 30, 2019 2:48 AM >> > + /* >> > + * sched_clock_register is needed on ARM64 but >> > + * is a no-op on x86 >> > + */ >> > + sched_clock_register(read_hv_sched_clock_msr, >> > + 64, HV_CLOCK_HZ); >> >> I'm not sure about ARM, but MSR-based clocksource would be a really bad >> choice for sched clock on x86, this will slow things down >> significantly. Luckily, as you're validly stating above, >> sched_clock_register() is a no-op on x86 as we don't define >> CONFIG_GENERIC_SCHED_CLOCK. >> >> Can we actually *not* do sched_clock_register() in case >> TSC page is unavailable (and revert to counting jiffies or whatever)? >> > > We can't skip the sched_clock_register() on ARM64 because it > does define CONFIG_GENERIC_SCHED_CLOCK. However, Hyper-V > should always provide REFERENCE_TSC_AVAILALBE on ARM64, > so we should never end up in the MSR-based code on ARM64. > Arguably that means the call to sched_clock_register() could be > removed since it's a no-op on x86. But I'd like to keep it for symmetry > and in case there's a testing/debugging situation on ARM64 where > we want to clear REFERENCE_TSC_AVAILABLE and go down the > MSR-based code path. Ok, so it is just a fall-back and not going to be actively used. Thanks!
diff --git a/arch/x86/entry/vdso/vclock_gettime.c b/arch/x86/entry/vdso/vclock_gettime.c index 98c7d12..7983ca2 100644 --- a/arch/x86/entry/vdso/vclock_gettime.c +++ b/arch/x86/entry/vdso/vclock_gettime.c @@ -21,6 +21,7 @@ #include <linux/math64.h> #include <linux/time.h> #include <linux/kernel.h> +#include <clocksource/hyperv_timer.h> #define gtod (&VVAR(vsyscall_gtod_data)) diff --git a/arch/x86/entry/vdso/vma.c b/arch/x86/entry/vdso/vma.c index babc4e7..99b38e9 100644 --- a/arch/x86/entry/vdso/vma.c +++ b/arch/x86/entry/vdso/vma.c @@ -22,7 +22,7 @@ #include <asm/page.h> #include <asm/desc.h> #include <asm/cpufeature.h> -#include <asm/mshyperv.h> +#include <clocksource/hyperv_timer.h> #if defined(CONFIG_X86_64) unsigned int __read_mostly vdso64_enabled = 1; diff --git a/arch/x86/hyperv/hv_init.c b/arch/x86/hyperv/hv_init.c index e4ba467..79f626a 100644 --- a/arch/x86/hyperv/hv_init.c +++ b/arch/x86/hyperv/hv_init.c @@ -27,64 +27,13 @@ #include <linux/version.h> #include <linux/vmalloc.h> #include <linux/mm.h> -#include <linux/clockchips.h> #include <linux/hyperv.h> #include <linux/slab.h> #include <linux/cpuhotplug.h> - -#ifdef CONFIG_HYPERV_TSCPAGE - -static struct ms_hyperv_tsc_page *tsc_pg; - -struct ms_hyperv_tsc_page *hv_get_tsc_page(void) -{ - return tsc_pg; -} -EXPORT_SYMBOL_GPL(hv_get_tsc_page); - -static u64 read_hv_clock_tsc(struct clocksource *arg) -{ - u64 current_tick = hv_read_tsc_page(tsc_pg); - - if (current_tick == U64_MAX) - rdmsrl(HV_X64_MSR_TIME_REF_COUNT, current_tick); - - return current_tick; -} - -static struct clocksource hyperv_cs_tsc = { - .name = "hyperv_clocksource_tsc_page", - .rating = 400, - .read = read_hv_clock_tsc, - .mask = CLOCKSOURCE_MASK(64), - .flags = CLOCK_SOURCE_IS_CONTINUOUS, -}; -#endif - -static u64 read_hv_clock_msr(struct clocksource *arg) -{ - u64 current_tick; - /* - * Read the partition counter to get the current tick count. This count - * is set to 0 when the partition is created and is incremented in - * 100 nanosecond units. - */ - rdmsrl(HV_X64_MSR_TIME_REF_COUNT, current_tick); - return current_tick; -} - -static struct clocksource hyperv_cs_msr = { - .name = "hyperv_clocksource_msr", - .rating = 400, - .read = read_hv_clock_msr, - .mask = CLOCKSOURCE_MASK(64), - .flags = CLOCK_SOURCE_IS_CONTINUOUS, -}; +#include <clocksource/hyperv_timer.h> void *hv_hypercall_pg; EXPORT_SYMBOL_GPL(hv_hypercall_pg); -struct clocksource *hyperv_cs; -EXPORT_SYMBOL_GPL(hyperv_cs); u32 *hv_vp_index; EXPORT_SYMBOL_GPL(hv_vp_index); @@ -353,42 +302,8 @@ void __init hyperv_init(void) x86_init.pci.arch_init = hv_pci_init; - /* - * Register Hyper-V specific clocksource. - */ -#ifdef CONFIG_HYPERV_TSCPAGE - if (ms_hyperv.features & HV_MSR_REFERENCE_TSC_AVAILABLE) { - union hv_x64_msr_hypercall_contents tsc_msr; - - tsc_pg = __vmalloc(PAGE_SIZE, GFP_KERNEL, PAGE_KERNEL); - if (!tsc_pg) - goto register_msr_cs; - - hyperv_cs = &hyperv_cs_tsc; - - rdmsrl(HV_X64_MSR_REFERENCE_TSC, tsc_msr.as_uint64); - - tsc_msr.enable = 1; - tsc_msr.guest_physical_address = vmalloc_to_pfn(tsc_pg); - - wrmsrl(HV_X64_MSR_REFERENCE_TSC, tsc_msr.as_uint64); - - hyperv_cs_tsc.archdata.vclock_mode = VCLOCK_HVCLOCK; - - clocksource_register_hz(&hyperv_cs_tsc, NSEC_PER_SEC/100); - return; - } -register_msr_cs: -#endif - /* - * For 32 bit guests just use the MSR based mechanism for reading - * the partition counter. - */ - - hyperv_cs = &hyperv_cs_msr; - if (ms_hyperv.features & HV_MSR_TIME_REF_COUNT_AVAILABLE) - clocksource_register_hz(&hyperv_cs_msr, NSEC_PER_SEC/100); - + /* Register Hyper-V specific clocksource */ + hv_init_clocksource(); return; remove_cpuhp_state: diff --git a/arch/x86/include/asm/mshyperv.h b/arch/x86/include/asm/mshyperv.h index cc60e61..f4fa8a9 100644 --- a/arch/x86/include/asm/mshyperv.h +++ b/arch/x86/include/asm/mshyperv.h @@ -105,6 +105,17 @@ static inline void vmbus_signal_eom(struct hv_message *msg, u32 old_msg_type) #define hv_get_crash_ctl(val) \ rdmsrl(HV_X64_MSR_CRASH_CTL, val) +#define hv_get_time_ref_count(val) \ + rdmsrl(HV_X64_MSR_TIME_REF_COUNT, val) + +#define hv_get_reference_tsc(val) \ + rdmsrl(HV_X64_MSR_REFERENCE_TSC, val) +#define hv_set_reference_tsc(val) \ + wrmsrl(HV_X64_MSR_REFERENCE_TSC, val) +#define hv_set_clocksource_vdso(val) \ + ((val).archdata.vclock_mode = VCLOCK_HVCLOCK) +#define hv_get_raw_timer() rdtsc_ordered() + void hyperv_callback_vector(void); void hyperv_reenlightenment_vector(void); #ifdef CONFIG_TRACING @@ -133,7 +144,6 @@ static inline void hv_disable_stimer0_percpu_irq(int irq) {} #if IS_ENABLED(CONFIG_HYPERV) -extern struct clocksource *hyperv_cs; extern void *hv_hypercall_pg; extern void __percpu **hyperv_pcpu_input_arg; @@ -387,73 +397,4 @@ static inline int hyperv_flush_guest_mapping_range(u64 as, } #endif /* CONFIG_HYPERV */ -#ifdef CONFIG_HYPERV_TSCPAGE -struct ms_hyperv_tsc_page *hv_get_tsc_page(void); -static inline u64 hv_read_tsc_page_tsc(const struct ms_hyperv_tsc_page *tsc_pg, - u64 *cur_tsc) -{ - u64 scale, offset; - u32 sequence; - - /* - * The protocol for reading Hyper-V TSC page is specified in Hypervisor - * Top-Level Functional Specification ver. 3.0 and above. To get the - * reference time we must do the following: - * - READ ReferenceTscSequence - * A special '0' value indicates the time source is unreliable and we - * need to use something else. The currently published specification - * versions (up to 4.0b) contain a mistake and wrongly claim '-1' - * instead of '0' as the special value, see commit c35b82ef0294. - * - ReferenceTime = - * ((RDTSC() * ReferenceTscScale) >> 64) + ReferenceTscOffset - * - READ ReferenceTscSequence again. In case its value has changed - * since our first reading we need to discard ReferenceTime and repeat - * the whole sequence as the hypervisor was updating the page in - * between. - */ - do { - sequence = READ_ONCE(tsc_pg->tsc_sequence); - if (!sequence) - return U64_MAX; - /* - * Make sure we read sequence before we read other values from - * TSC page. - */ - smp_rmb(); - - scale = READ_ONCE(tsc_pg->tsc_scale); - offset = READ_ONCE(tsc_pg->tsc_offset); - *cur_tsc = rdtsc_ordered(); - - /* - * Make sure we read sequence after we read all other values - * from TSC page. - */ - smp_rmb(); - - } while (READ_ONCE(tsc_pg->tsc_sequence) != sequence); - - return mul_u64_u64_shr(*cur_tsc, scale, 64) + offset; -} - -static inline u64 hv_read_tsc_page(const struct ms_hyperv_tsc_page *tsc_pg) -{ - u64 cur_tsc; - - return hv_read_tsc_page_tsc(tsc_pg, &cur_tsc); -} - -#else -static inline struct ms_hyperv_tsc_page *hv_get_tsc_page(void) -{ - return NULL; -} - -static inline u64 hv_read_tsc_page_tsc(const struct ms_hyperv_tsc_page *tsc_pg, - u64 *cur_tsc) -{ - BUG(); - return U64_MAX; -} -#endif #endif diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 536b78c..3fbaac0 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -70,6 +70,7 @@ #include <asm/mshyperv.h> #include <asm/hypervisor.h> #include <asm/intel_pt.h> +#include <clocksource/hyperv_timer.h> #define CREATE_TRACE_POINTS #include "trace.h" diff --git a/drivers/clocksource/hyperv_timer.c b/drivers/clocksource/hyperv_timer.c index 30615f3..274614d 100644 --- a/drivers/clocksource/hyperv_timer.c +++ b/drivers/clocksource/hyperv_timer.c @@ -14,6 +14,8 @@ #include <linux/percpu.h> #include <linux/cpumask.h> #include <linux/clockchips.h> +#include <linux/clocksource.h> +#include <linux/sched_clock.h> #include <linux/mm.h> #include <clocksource/hyperv_timer.h> #include <asm/hyperv-tlfs.h> @@ -189,3 +191,131 @@ void hv_stimer_global_cleanup(void) hv_stimer_free(); } EXPORT_SYMBOL_GPL(hv_stimer_global_cleanup); + +/* + * Code and definitions for the Hyper-V clocksources. Two + * clocksources are defined: one that reads the Hyper-V defined MSR, and + * the other that uses the TSC reference page feature as defined in the + * TLFS. The MSR version is for compatibility with old versions of + * Hyper-V and 32-bit x86. The TSC reference page version is preferred. + */ + +struct clocksource *hyperv_cs; +EXPORT_SYMBOL_GPL(hyperv_cs); + +#ifdef CONFIG_HYPERV_TSCPAGE + +static struct ms_hyperv_tsc_page *tsc_pg; + +struct ms_hyperv_tsc_page *hv_get_tsc_page(void) +{ + return tsc_pg; +} +EXPORT_SYMBOL_GPL(hv_get_tsc_page); + +static u64 read_hv_sched_clock_tsc(void) +{ + u64 current_tick = hv_read_tsc_page(tsc_pg); + + if (current_tick == U64_MAX) + hv_get_time_ref_count(current_tick); + + return current_tick; +} + +static u64 read_hv_clock_tsc(struct clocksource *arg) +{ + return read_hv_sched_clock_tsc(); +} + +static struct clocksource hyperv_cs_tsc = { + .name = "hyperv_clocksource_tsc_page", + .rating = 400, + .read = read_hv_clock_tsc, + .mask = CLOCKSOURCE_MASK(64), + .flags = CLOCK_SOURCE_IS_CONTINUOUS, +}; +#endif + +static u64 read_hv_sched_clock_msr(void) +{ + u64 current_tick; + /* + * Read the partition counter to get the current tick count. This count + * is set to 0 when the partition is created and is incremented in + * 100 nanosecond units. + */ + hv_get_time_ref_count(current_tick); + return current_tick; +} + +static u64 read_hv_clock_msr(struct clocksource *arg) +{ + return read_hv_sched_clock_msr(); +} + +static struct clocksource hyperv_cs_msr = { + .name = "hyperv_clocksource_msr", + .rating = 400, + .read = read_hv_clock_msr, + .mask = CLOCKSOURCE_MASK(64), + .flags = CLOCK_SOURCE_IS_CONTINUOUS, +}; + +void __init hv_init_clocksource(void) +{ +#ifdef CONFIG_HYPERV_TSCPAGE + if (ms_hyperv.features & HV_MSR_REFERENCE_TSC_AVAILABLE) { + + u64 tsc_msr; + phys_addr_t phys_addr; + + tsc_pg = vmalloc(PAGE_SIZE); + if (!tsc_pg) + goto register_msr_cs; + + hyperv_cs = &hyperv_cs_tsc; + phys_addr = page_to_phys(vmalloc_to_page(tsc_pg)); + + /* The Hyper-V TLFS specifies to preserve the value of + * reserved bits in registers. So read the existing + * value, preserve the low order 12 bits, and add in + * the guest physical address (which already has at + * least the low 12 bits set to zero since it is page + * aligned). Also set the "enable" bit, which is bit 0. + */ + hv_get_reference_tsc(tsc_msr); + tsc_msr &= GENMASK_ULL(11, 0); + tsc_msr = tsc_msr | 0x1 | (u64)phys_addr; + hv_set_reference_tsc(tsc_msr); + + hv_set_clocksource_vdso(hyperv_cs_tsc); + clocksource_register_hz(&hyperv_cs_tsc, NSEC_PER_SEC/100); + + /* + * sched_clock_register is needed on ARM64 but + * is a no-op on x86 + */ + sched_clock_register(read_hv_sched_clock_tsc, + 64, HV_CLOCK_HZ); + return; + } +register_msr_cs: +#endif + /* + * For 32 bit guests just use the MSR based mechanism for reading + * the partition counter. + */ + hyperv_cs = &hyperv_cs_msr; + if (ms_hyperv.features & HV_MSR_TIME_REF_COUNT_AVAILABLE) { + clocksource_register_hz(&hyperv_cs_msr, NSEC_PER_SEC/100); + + /* + * sched_clock_register is needed on ARM64 but + * is a no-op on x86 + */ + sched_clock_register(read_hv_sched_clock_msr, + 64, HV_CLOCK_HZ); + } +} +EXPORT_SYMBOL_GPL(hv_init_clocksource); diff --git a/drivers/hv/hv_util.c b/drivers/hv/hv_util.c index f10eeb1..9eff85e 100644 --- a/drivers/hv/hv_util.c +++ b/drivers/hv/hv_util.c @@ -29,6 +29,7 @@ #include <linux/hyperv.h> #include <linux/clockchips.h> #include <linux/ptp_clock_kernel.h> +#include <clocksource/hyperv_timer.h> #include <asm/mshyperv.h> #include "hyperv_vmbus.h" diff --git a/include/clocksource/hyperv_timer.h b/include/clocksource/hyperv_timer.h index ba5dc17..e17e8a2 100644 --- a/include/clocksource/hyperv_timer.h +++ b/include/clocksource/hyperv_timer.h @@ -13,6 +13,10 @@ #ifndef __CLKSOURCE_HYPERV_TIMER_H #define __CLKSOURCE_HYPERV_TIMER_H +#include <linux/clocksource.h> +#include <linux/math64.h> +#include <asm/mshyperv.h> + #define HV_MAX_MAX_DELTA_TICKS 0xffffffff #define HV_MIN_DELTA_TICKS 1 @@ -24,4 +28,78 @@ extern void hv_stimer_global_cleanup(void); extern void hv_stimer0_isr(void); +#if IS_ENABLED(CONFIG_HYPERV) +extern struct clocksource *hyperv_cs; +extern void hv_init_clocksource(void); +#endif /* CONFIG_HYPERV */ + +#ifdef CONFIG_HYPERV_TSCPAGE +extern struct ms_hyperv_tsc_page *hv_get_tsc_page(void); +static inline u64 hv_read_tsc_page_tsc(const struct ms_hyperv_tsc_page *tsc_pg, + u64 *cur_tsc) +{ + u64 scale, offset; + u32 sequence; + + /* + * The protocol for reading Hyper-V TSC page is specified in Hypervisor + * Top-Level Functional Specification ver. 3.0 and above. To get the + * reference time we must do the following: + * - READ ReferenceTscSequence + * A special '0' value indicates the time source is unreliable and we + * need to use something else. The currently published specification + * versions (up to 4.0b) contain a mistake and wrongly claim '-1' + * instead of '0' as the special value, see commit c35b82ef0294. + * - ReferenceTime = + * ((RDTSC() * ReferenceTscScale) >> 64) + ReferenceTscOffset + * - READ ReferenceTscSequence again. In case its value has changed + * since our first reading we need to discard ReferenceTime and repeat + * the whole sequence as the hypervisor was updating the page in + * between. + */ + do { + sequence = READ_ONCE(tsc_pg->tsc_sequence); + if (!sequence) + return U64_MAX; + /* + * Make sure we read sequence before we read other values from + * TSC page. + */ + smp_rmb(); + + scale = READ_ONCE(tsc_pg->tsc_scale); + offset = READ_ONCE(tsc_pg->tsc_offset); + *cur_tsc = hv_get_raw_timer(); + + /* + * Make sure we read sequence after we read all other values + * from TSC page. + */ + smp_rmb(); + + } while (READ_ONCE(tsc_pg->tsc_sequence) != sequence); + + return mul_u64_u64_shr(*cur_tsc, scale, 64) + offset; +} + +static inline u64 hv_read_tsc_page(const struct ms_hyperv_tsc_page *tsc_pg) +{ + u64 cur_tsc; + + return hv_read_tsc_page_tsc(tsc_pg, &cur_tsc); +} + +#else /* CONFIG_HYPERV_TSC_PAGE */ +static inline struct ms_hyperv_tsc_page *hv_get_tsc_page(void) +{ + return NULL; +} + +static inline u64 hv_read_tsc_page_tsc(const struct ms_hyperv_tsc_page *tsc_pg, + u64 *cur_tsc) +{ + return U64_MAX; +} +#endif /* CONFIG_HYPERV_TSCPAGE */ + #endif
Code for the Hyper-V specific clocksources is currently mixed in with other Hyper-V code. Move the code to the Hyper-V specific driver in the "clocksource" directory, while separating out ISA dependencies so that the clocksource driver remains ISA independent. Update the Hyper-V initialization code to call initialization and cleanup routines since the Hyper-V synthetic timers are not independently enumerated in ACPI. Update Hyper-V clocksource users KVM and VDSO to get definitions from a new include file. No behavior is changed and no new functionality is added. Suggested-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Michael Kelley <mikelley@microsoft.com> --- arch/x86/entry/vdso/vclock_gettime.c | 1 + arch/x86/entry/vdso/vma.c | 2 +- arch/x86/hyperv/hv_init.c | 91 +----------------------- arch/x86/include/asm/mshyperv.h | 81 +++------------------- arch/x86/kvm/x86.c | 1 + drivers/clocksource/hyperv_timer.c | 130 +++++++++++++++++++++++++++++++++++ drivers/hv/hv_util.c | 1 + include/clocksource/hyperv_timer.h | 78 +++++++++++++++++++++ 8 files changed, 226 insertions(+), 159 deletions(-)