Option PARAVIRT_SPINLOCKS is added on LoongArch system, and pv_lock_ops template is added here. If option PARAVIRT_SPINLOCKS is enabled, the native ops works on host machine. Two functions kvm_wait() and kvm_kick_cpu() are added specicial for VM, if VM detects hypervisor supports pv spinlock. With kvm_wait() vCPU thread will exit to hypervisor and give up scheduleing on pCPU, and with function kvm_kick_cpu() one hypercall function is used to notify hypervisor to wakeup previously waited vCPU. Signed-off-by: Bibo Mao <maobibo@xxxxxxxxxxx> --- arch/loongarch/Kconfig | 14 +++ arch/loongarch/include/asm/Kbuild | 1 - arch/loongarch/include/asm/paravirt.h | 47 ++++++++++ arch/loongarch/include/asm/qspinlock.h | 39 ++++++++ .../include/asm/qspinlock_paravirt.h | 6 ++ arch/loongarch/kernel/paravirt.c | 88 +++++++++++++++++++ arch/loongarch/kernel/smp.c | 4 +- 7 files changed, 197 insertions(+), 2 deletions(-) create mode 100644 arch/loongarch/include/asm/qspinlock.h create mode 100644 arch/loongarch/include/asm/qspinlock_paravirt.h diff --git a/arch/loongarch/Kconfig b/arch/loongarch/Kconfig index b81d0eba5c7e..7ad63db2fafd 100644 --- a/arch/loongarch/Kconfig +++ b/arch/loongarch/Kconfig @@ -660,6 +660,20 @@ config PARAVIRT_TIME_ACCOUNTING If in doubt, say N here. +config PARAVIRT_SPINLOCKS + bool "Paravirtual queued spinlocks" + select PARAVIRT + depends on SMP + help + Paravirtualized spinlocks allow a pvops backend to replace the + spinlock implementation with something virtualization-friendly + (for example, block the virtual CPU rather than spinning). + + It has a minimal impact on native kernels and gives a nice performance + benefit on paravirtualized kernels. + + If you are unsure how to answer this question, answer Y. + endmenu config ARCH_SELECT_MEMORY_MODEL diff --git a/arch/loongarch/include/asm/Kbuild b/arch/loongarch/include/asm/Kbuild index 2bb3676429c0..4635b755b2b4 100644 --- a/arch/loongarch/include/asm/Kbuild +++ b/arch/loongarch/include/asm/Kbuild @@ -6,7 +6,6 @@ generic-y += mcs_spinlock.h generic-y += parport.h generic-y += early_ioremap.h generic-y += qrwlock.h -generic-y += qspinlock.h generic-y += user.h generic-y += ioctl.h generic-y += statfs.h diff --git a/arch/loongarch/include/asm/paravirt.h b/arch/loongarch/include/asm/paravirt.h index dddec49671ae..2617d635171b 100644 --- a/arch/loongarch/include/asm/paravirt.h +++ b/arch/loongarch/include/asm/paravirt.h @@ -20,6 +20,47 @@ static inline u64 paravirt_steal_clock(int cpu) int __init pv_ipi_init(void); int __init pv_time_init(void); +#if defined(CONFIG_PARAVIRT_SPINLOCKS) +struct qspinlock; +struct pv_lock_ops { + void (*queued_spin_lock_slowpath)(struct qspinlock *lock, u32 val); + void (*queued_spin_unlock)(struct qspinlock *lock); + void (*wait)(u8 *ptr, u8 val); + void (*kick)(int cpu); + bool (*vcpu_is_preempted)(int cpu); +}; + +extern struct pv_lock_ops pv_lock_ops; + +void __init kvm_spinlock_init(void); +bool pv_is_native_spin_unlock(void); + +static __always_inline void pv_queued_spin_lock_slowpath(struct qspinlock *lock, + u32 val) +{ + pv_lock_ops.queued_spin_lock_slowpath(lock, val); +} + +static __always_inline void pv_queued_spin_unlock(struct qspinlock *lock) +{ + pv_lock_ops.queued_spin_unlock(lock); +} + +static __always_inline void pv_wait(u8 *ptr, u8 val) +{ + pv_lock_ops.wait(ptr, val); +} + +static __always_inline void pv_kick(int cpu) +{ + pv_lock_ops.kick(cpu); +} + +static __always_inline bool pv_vcpu_is_preempted(long cpu) +{ + return pv_lock_ops.vcpu_is_preempted(cpu); +} +#endif /* PARAVIRT_SPINLOCKS */ #else static inline int pv_ipi_init(void) @@ -32,4 +73,10 @@ static inline int pv_time_init(void) return 0; } #endif // CONFIG_PARAVIRT + +#ifndef CONFIG_PARAVIRT_SPINLOCKS +static inline void kvm_spinlock_init(void) +{ +} +#endif #endif diff --git a/arch/loongarch/include/asm/qspinlock.h b/arch/loongarch/include/asm/qspinlock.h new file mode 100644 index 000000000000..8e1b14c9e906 --- /dev/null +++ b/arch/loongarch/include/asm/qspinlock.h @@ -0,0 +1,39 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef _ASM_LOONGARCH_QSPINLOCK_H +#define _ASM_LOONGARCH_QSPINLOCK_H + +#include <asm/paravirt.h> + +#define _Q_PENDING_LOOPS (1 << 9) + +#ifdef CONFIG_PARAVIRT_SPINLOCKS +/* How long a lock should spin before we consider blocking */ +#define SPIN_THRESHOLD (1 << 15) + +extern void native_queued_spin_lock_slowpath(struct qspinlock *lock, u32 val); +extern void __pv_init_lock_hash(void); +extern void __pv_queued_spin_lock_slowpath(struct qspinlock *lock, u32 val); +extern void __pv_queued_spin_unlock(struct qspinlock *lock); +extern bool nopvspin; + +static inline void queued_spin_lock_slowpath(struct qspinlock *lock, u32 val) +{ + pv_queued_spin_lock_slowpath(lock, val); +} + +#define queued_spin_unlock queued_spin_unlock +static inline void queued_spin_unlock(struct qspinlock *lock) +{ + pv_queued_spin_unlock(lock); +} + +#define vcpu_is_preempted vcpu_is_preempted +static inline bool vcpu_is_preempted(long cpu) +{ + return pv_vcpu_is_preempted(cpu); +} +#endif + +#include <asm-generic/qspinlock.h> + +#endif // _ASM_LOONGARCH_QSPINLOCK_H diff --git a/arch/loongarch/include/asm/qspinlock_paravirt.h b/arch/loongarch/include/asm/qspinlock_paravirt.h new file mode 100644 index 000000000000..d6d7f487daea --- /dev/null +++ b/arch/loongarch/include/asm/qspinlock_paravirt.h @@ -0,0 +1,6 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef __ASM_QSPINLOCK_PARAVIRT_H +#define __ASM_QSPINLOCK_PARAVIRT_H + +void __lockfunc __pv_queued_spin_unlock_slowpath(struct qspinlock *lock, u8 locked); +#endif diff --git a/arch/loongarch/kernel/paravirt.c b/arch/loongarch/kernel/paravirt.c index aee44610007d..758039eabdde 100644 --- a/arch/loongarch/kernel/paravirt.c +++ b/arch/loongarch/kernel/paravirt.c @@ -298,3 +298,91 @@ int __init pv_time_init(void) return 0; } + +#ifdef CONFIG_PARAVIRT_SPINLOCKS +static bool native_vcpu_is_preempted(int cpu) +{ + return false; +} + +/** + * queued_spin_unlock - release a queued spinlock + * @lock : Pointer to queued spinlock structure + */ +static void native_queued_spin_unlock(struct qspinlock *lock) +{ + /* + * unlock() needs release semantics: + */ + smp_store_release(&lock->locked, 0); +} + +static void paravirt_nop_kick(int cpu) +{ +} + +static void paravirt_nop_wait(u8 *ptr, u8 val) +{ +} + +static void kvm_wait(u8 *ptr, u8 val) +{ + if (READ_ONCE(*ptr) != val) + return; + + __asm__ __volatile__("idle 0\n\t" : : : "memory"); +} + +/* Kick a cpu. Used to wake up a halted vcpu */ +static void kvm_kick_cpu(int cpu) +{ + kvm_hypercall1(KVM_HCALL_FUNC_KICK, cpu_logical_map(cpu)); +} + +bool pv_is_native_spin_unlock(void) +{ + return pv_lock_ops.queued_spin_unlock == native_queued_spin_unlock; +} + +/* + * Setup pv_lock_ops for guest kernel. + */ +void __init kvm_spinlock_init(void) +{ + int feature; + + /* + * pv_hash()/pv_unhas() need it whatever pv spinlock is + * enabled or not + */ + __pv_init_lock_hash(); + + if (!kvm_para_available()) + return; + + /* Don't use the pvqspinlock code if there is only 1 vCPU. */ + if (num_possible_cpus() == 1) + return; + + feature = kvm_arch_para_features(); + if (!(feature & KVM_FEATURE_PARAVIRT_SPINLOCK)) + return; + + if (nopvspin) + return; + + pr_info("Using paravirt qspinlock\n"); + pv_lock_ops.queued_spin_lock_slowpath = __pv_queued_spin_lock_slowpath; + pv_lock_ops.queued_spin_unlock = __pv_queued_spin_unlock; + pv_lock_ops.wait = kvm_wait; + pv_lock_ops.kick = kvm_kick_cpu; +} + +struct pv_lock_ops pv_lock_ops = { + .queued_spin_lock_slowpath = native_queued_spin_lock_slowpath, + .queued_spin_unlock = native_queued_spin_unlock, + .wait = paravirt_nop_wait, + .kick = paravirt_nop_kick, + .vcpu_is_preempted = native_vcpu_is_preempted, +}; +#endif /* CONFIG_PARAVIRT_SPINLOCKS */ diff --git a/arch/loongarch/kernel/smp.c b/arch/loongarch/kernel/smp.c index 1436d2465939..6bc0b182a2ce 100644 --- a/arch/loongarch/kernel/smp.c +++ b/arch/loongarch/kernel/smp.c @@ -463,7 +463,7 @@ core_initcall(ipi_pm_init); #endif /* Preload SMP state for boot cpu */ -void smp_prepare_boot_cpu(void) +void __init smp_prepare_boot_cpu(void) { unsigned int cpu, node, rr_node; @@ -496,6 +496,8 @@ void smp_prepare_boot_cpu(void) rr_node = next_node_in(rr_node, node_online_map); } } + + kvm_spinlock_init(); } /* called from main before smp_init() */ -- 2.39.3