From:   Tianrui Zhao <zhaotianrui@loongson.cn>
To:     Paolo Bonzini <pbonzini@redhat.com>
Cc:     Huacai Chen <chenhuacai@kernel.org>,
        WANG Xuerui <kernel@xen0n.name>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        loongarch@lists.linux.dev, linux-kernel@vger.kernel.org,
        kvm@vger.kernel.org, Jens Axboe <axboe@kernel.dk>,
        Mark Brown <broonie@kernel.org>,
        Alex Deucher <alexander.deucher@amd.com>,
        Oliver Upton <oliver.upton@linux.dev>, maobibo@loongson.cn,
        Xi Ruoyao <xry111@xry111.site>
Subject: [PATCH v4 18/29] LoongArch: KVM: Implement vcpu timer operations
Date:   Tue, 21 Mar 2023 11:56:40 +0800
Message-Id: <20230321035651.598505-19-zhaotianrui@loongson.cn>
In-Reply-To: <20230321035651.598505-1-zhaotianrui@loongson.cn>
References: <20230321035651.598505-1-zhaotianrui@loongson.cn>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

Implement loongarch vcpu timer operations such as init kvm timer,
require kvm timer, save kvm timer and restore kvm timer. When
vcpu exit, we use kvm soft timer to emulate hardware timer. If
timeout happens, the vcpu timer interrupt will be set and it is
going to be handled at vcpu next entrance.

Signed-off-by: Tianrui Zhao <zhaotianrui@loongson.cn>
---
 arch/loongarch/kvm/timer.c | 266 +++++++++++++++++++++++++++++++++++++
 1 file changed, 266 insertions(+)
 create mode 100644 arch/loongarch/kvm/timer.c

diff --git a/arch/loongarch/kvm/timer.c b/arch/loongarch/kvm/timer.c
new file mode 100644
index 000000000..2c7677248
--- /dev/null
+++ b/arch/loongarch/kvm/timer.c
@@ -0,0 +1,266 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2020-2023 Loongson Technology Corporation Limited
+ */
+
+#include <linux/kvm_host.h>
+#include <asm/kvm_csr.h>
+#include <asm/kvm_vcpu.h>
+
+/* low level hrtimer wake routine */
+enum hrtimer_restart kvm_swtimer_wakeup(struct hrtimer *timer)
+{
+	struct kvm_vcpu *vcpu;
+
+	vcpu = container_of(timer, struct kvm_vcpu, arch.swtimer);
+	_kvm_queue_irq(vcpu, LARCH_INT_TIMER);
+	rcuwait_wake_up(&vcpu->wait);
+	return kvm_count_timeout(vcpu);
+}
+
+/*
+ * ktime_to_tick() - Scale ktime_t to a 64-bit stable timer.
+ *
+ * Caches the dynamic nanosecond bias in vcpu->arch.timer_dyn_bias.
+ */
+static unsigned long ktime_to_tick(struct kvm_vcpu *vcpu, ktime_t now)
+{
+	s64 now_ns, periods;
+	unsigned long delta;
+
+	now_ns = ktime_to_ns(now);
+	delta = now_ns + vcpu->arch.timer_dyn_bias;
+
+	if (delta >= vcpu->arch.timer_period) {
+		/* If delta is out of safe range the bias needs adjusting */
+		periods = div64_s64(now_ns, vcpu->arch.timer_period);
+		vcpu->arch.timer_dyn_bias = -periods * vcpu->arch.timer_period;
+		/* Recalculate delta with new bias */
+		delta = now_ns + vcpu->arch.timer_dyn_bias;
+	}
+
+	/*
+	 * We've ensured that:
+	 *   delta < timer_period
+	 */
+	return div_u64(delta * vcpu->arch.timer_mhz, MNSEC_PER_SEC);
+}
+
+/**
+ * kvm_resume_hrtimer() - Resume hrtimer, updating expiry.
+ * @vcpu:	Virtual CPU.
+ * @now:	ktime at point of resume.
+ * @val:	stable timer at point of resume.
+ *
+ * Resumes the timer and updates the timer expiry based on @now and @count.
+ */
+static void kvm_resume_hrtimer(struct kvm_vcpu *vcpu, ktime_t now,
+				unsigned long val)
+{
+	unsigned long delta;
+	ktime_t expire;
+
+	/* Stable timer decreased to zero or
+	 * initialize to zero, set 4 second timer
+	 */
+	delta = div_u64(val * MNSEC_PER_SEC, vcpu->arch.timer_mhz);
+	expire = ktime_add_ns(now, delta);
+
+	/* Update hrtimer to use new timeout */
+	hrtimer_cancel(&vcpu->arch.swtimer);
+	hrtimer_start(&vcpu->arch.swtimer, expire, HRTIMER_MODE_ABS_PINNED);
+}
+
+/**
+ * kvm_init_timer() - Initialise stable timer.
+ * @vcpu:	Virtual CPU.
+ * @timer_hz:	Frequency of timer.
+ *
+ * Initialise the timer to the specified frequency, zero it, and set it going if
+ * it's enabled.
+ */
+void kvm_init_timer(struct kvm_vcpu *vcpu, unsigned long timer_hz)
+{
+	ktime_t now;
+	unsigned long ticks;
+	struct loongarch_csrs *csr = vcpu->arch.csr;
+
+	ticks = (unsigned long)MNSEC_PER_SEC * CSR_TCFG_VAL;
+	vcpu->arch.timer_mhz = timer_hz >> 20;
+	vcpu->arch.timer_period = div_u64(ticks, vcpu->arch.timer_mhz);
+	vcpu->arch.timer_dyn_bias = 0;
+
+	/* Starting at 0 */
+	ticks = 0;
+	now = ktime_get();
+	vcpu->arch.timer_bias = ticks - ktime_to_tick(vcpu, now);
+	vcpu->arch.timer_bias &= CSR_TCFG_VAL;
+	kvm_write_sw_gcsr(csr, LOONGARCH_CSR_TVAL, ticks);
+}
+
+/**
+ * kvm_count_timeout() - Push timer forward on timeout.
+ * @vcpu:	Virtual CPU.
+ *
+ * Handle an hrtimer event by push the hrtimer forward a period.
+ *
+ * Returns:	The hrtimer_restart value to return to the hrtimer subsystem.
+ */
+enum hrtimer_restart kvm_count_timeout(struct kvm_vcpu *vcpu)
+{
+	unsigned long cfg;
+
+	/* Add the Count period to the current expiry time */
+	cfg = kvm_read_sw_gcsr(vcpu->arch.csr, LOONGARCH_CSR_TCFG);
+	if (cfg & CSR_TCFG_PERIOD) {
+		hrtimer_add_expires_ns(&vcpu->arch.swtimer, cfg & CSR_TCFG_VAL);
+		return HRTIMER_RESTART;
+	} else
+		return HRTIMER_NORESTART;
+}
+
+/*
+ * kvm_restore_timer() - Restore timer state.
+ * @vcpu:       Virtual CPU.
+ *
+ * Restore soft timer state from saved context.
+ */
+void kvm_restore_timer(struct kvm_vcpu *vcpu)
+{
+	struct loongarch_csrs *csr = vcpu->arch.csr;
+	ktime_t saved_ktime, now;
+	unsigned long val, new, delta;
+	int expired = 0;
+	unsigned long cfg;
+
+	/*
+	 * Set guest stable timer cfg csr
+	 */
+	cfg = kvm_read_sw_gcsr(csr, LOONGARCH_CSR_TCFG);
+	kvm_restore_hw_gcsr(csr, LOONGARCH_CSR_ESTAT);
+	if (!(cfg & CSR_TCFG_EN)) {
+		kvm_restore_hw_gcsr(csr, LOONGARCH_CSR_TCFG);
+		kvm_restore_hw_gcsr(csr, LOONGARCH_CSR_TVAL);
+		return;
+	}
+
+	now = ktime_get();
+	saved_ktime = vcpu->arch.stable_ktime_saved;
+	val = kvm_read_sw_gcsr(csr, LOONGARCH_CSR_TVAL);
+
+	/*hrtimer not expire */
+	delta = ktime_to_tick(vcpu, ktime_sub(now, saved_ktime));
+	if (delta >= val) {
+		expired = 1;
+		if (cfg & CSR_TCFG_PERIOD)
+			new = (delta - val) % (cfg & CSR_TCFG_VAL);
+		else
+			new = 1;
+	} else
+		new = val - delta;
+
+	new &= CSR_TCFG_VAL;
+	write_gcsr_timercfg(cfg);
+	write_gcsr_timertick(new);
+	if (expired)
+		_kvm_queue_irq(vcpu, LARCH_INT_TIMER);
+}
+
+/*
+ * kvm_acquire_timer() - Switch to hard timer state.
+ * @vcpu:       Virtual CPU.
+ *
+ * Restore hard timer state on top of existing soft timer state if possible.
+ *
+ * Since hard timer won't remain active over preemption, preemption should be
+ * disabled by the caller.
+ */
+void kvm_acquire_timer(struct kvm_vcpu *vcpu)
+{
+	unsigned long flags, guestcfg;
+
+	guestcfg = read_csr_gcfg();
+	if (!(guestcfg & CSR_GCFG_TIT))
+		return;
+
+	/* enable guest access to hard timer */
+	write_csr_gcfg(guestcfg & ~CSR_GCFG_TIT);
+
+	/*
+	 * Freeze the soft-timer and sync the guest stable timer with it. We do
+	 * this with interrupts disabled to avoid latency.
+	 */
+	local_irq_save(flags);
+	hrtimer_cancel(&vcpu->arch.swtimer);
+	local_irq_restore(flags);
+}
+
+
+/*
+ * _kvm_save_timer() - Switch to software emulation of guest timer.
+ * @vcpu:       Virtual CPU.
+ *
+ * Save guest timer state and switch to software emulation of guest
+ * timer. The hard timer must already be in use, so preemption should be
+ * disabled.
+ */
+static ktime_t _kvm_save_timer(struct kvm_vcpu *vcpu, unsigned long *val)
+{
+	unsigned long end_time;
+	ktime_t before_time;
+
+	before_time = ktime_get();
+
+	/*
+	 * Record a final stable timer which we will transfer to the soft-timer.
+	 */
+	end_time = read_gcsr_timertick();
+	*val = end_time;
+
+	kvm_resume_hrtimer(vcpu, before_time, end_time);
+	return before_time;
+}
+
+/*
+ * kvm_save_timer() - Save guest timer state.
+ * @vcpu:       Virtual CPU.
+ *
+ * Save guest timer state and switch to soft guest timer if hard timer was in
+ * use.
+ */
+void kvm_save_timer(struct kvm_vcpu *vcpu)
+{
+	struct loongarch_csrs *csr = vcpu->arch.csr;
+	unsigned long guestcfg, val;
+	ktime_t save_ktime;
+
+	preempt_disable();
+	guestcfg = read_csr_gcfg();
+	if (!(guestcfg & CSR_GCFG_TIT)) {
+		/* disable guest use of hard timer */
+		write_csr_gcfg(guestcfg | CSR_GCFG_TIT);
+
+		/* save hard timer state */
+		kvm_save_hw_gcsr(csr, LOONGARCH_CSR_TCFG);
+		if (kvm_read_sw_gcsr(csr, LOONGARCH_CSR_TCFG) & CSR_TCFG_EN) {
+			save_ktime = _kvm_save_timer(vcpu, &val);
+			kvm_write_sw_gcsr(csr, LOONGARCH_CSR_TVAL, val);
+			vcpu->arch.stable_ktime_saved = save_ktime;
+			if (val == CSR_TCFG_VAL)
+				_kvm_queue_irq(vcpu, LARCH_INT_TIMER);
+		} else {
+			kvm_save_hw_gcsr(csr, LOONGARCH_CSR_TVAL);
+		}
+	}
+
+	/* save timer-related state to VCPU context */
+	kvm_save_hw_gcsr(csr, LOONGARCH_CSR_ESTAT);
+	preempt_enable();
+}
+
+void kvm_reset_timer(struct kvm_vcpu *vcpu)
+{
+	write_gcsr_timercfg(0);
+	kvm_write_sw_gcsr(vcpu->arch.csr, LOONGARCH_CSR_TCFG, 0);
+	hrtimer_cancel(&vcpu->arch.swtimer);
+}
-- 
2.31.1