sched_ext: Maintain an online cid mask in the scheduler arenagrafted

Schedulers on the default cid mapping treat [0, nr_online_cids) as the
online set and restart on hotplug. Schedulers that install their own mapping
with scx_bpf_cid_override() have no way to learn which cids are online: the
count no longer identifies members and the CPU-form cpumask is unusable from
cid programs. This is an obvious hole in the cid API.

Add scx_bpf_online_cmask(), a kernel-maintained cmask in the scheduler's
arena, allocated alongside the per-CPU scratch masks and populated after the
cid mapping is finalized and before ops.init(), for child schedulers too.
The pointer stays valid through ops.exit() with no reference to take. It is
the arena offset as a void pointer, the same form struct_ops arena arguments
arrive in. The verifier types the void return as a scalar for the program's
arena cast.

The mask follows the SCX hotplug notifications: seeded from cpu_active_mask
and updated before ops.cid_online/offline() runs, so it lags cpu_online_mask
only inside a hotplug transition. Updates walk the scheduler list under the
lock that also serializes unlinking. Reads are live, not atomic snapshots.
Root initialization excludes hotplug.

v2: Reworded the getter kerneldoc (Andrea Righi).

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>

1 files changed, 415 insertions, 0 deletions

diff --git a/drivers/clocksource/timer-tegra.c b/drivers/clocksource/timer-tegra.c
new file mode 100644
index 000000000..35b6ce9de
--- /dev/null
+++ b/drivers/clocksource/timer-tegra.c
@@ -0,0 +1,415 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2010 Google, Inc.
+ *
+ * Author:
+ *	Colin Cross <ccross@google.com>
+ */
+
+#define pr_fmt(fmt)	"tegra-timer: " fmt
+
+#include <linux/clk.h>
+#include <linux/clockchips.h>
+#include <linux/cpu.h>
+#include <linux/cpumask.h>
+#include <linux/delay.h>
+#include <linux/err.h>
+#include <linux/interrupt.h>
+#include <linux/of_address.h>
+#include <linux/of_irq.h>
+#include <linux/percpu.h>
+#include <linux/sched_clock.h>
+#include <linux/time.h>
+
+#include "timer-of.h"
+
+#define RTC_SECONDS		0x08
+#define RTC_SHADOW_SECONDS	0x0c
+#define RTC_MILLISECONDS	0x10
+
+#define TIMERUS_CNTR_1US	0x10
+#define TIMERUS_USEC_CFG	0x14
+#define TIMERUS_CNTR_FREEZE	0x4c
+
+#define TIMER_PTV		0x0
+#define TIMER_PTV_EN		BIT(31)
+#define TIMER_PTV_PER		BIT(30)
+#define TIMER_PCR		0x4
+#define TIMER_PCR_INTR_CLR	BIT(30)
+
+#define TIMER1_BASE		0x00
+#define TIMER2_BASE		0x08
+#define TIMER3_BASE		0x50
+#define TIMER4_BASE		0x58
+#define TIMER10_BASE		0x90
+
+#define TIMER1_IRQ_IDX		0
+#define TIMER10_IRQ_IDX		10
+
+#define TIMER_1MHz		1000000
+
+static u32 usec_config;
+static void __iomem *timer_reg_base;
+
+static int tegra_timer_set_next_event(unsigned long cycles,
+				      struct clock_event_device *evt)
+{
+	void __iomem *reg_base = timer_of_base(to_timer_of(evt));
+
+	/*
+	 * Tegra's timer uses n+1 scheme for the counter, i.e. timer will
+	 * fire after one tick if 0 is loaded.
+	 *
+	 * The minimum and maximum numbers of oneshot ticks are defined
+	 * by clockevents_config_and_register(1, 0x1fffffff + 1) invocation
+	 * below in the code. Hence the cycles (ticks) can't be outside of
+	 * a range supportable by hardware.
+	 */
+	writel_relaxed(TIMER_PTV_EN | (cycles - 1), reg_base + TIMER_PTV);
+
+	return 0;
+}
+
+static int tegra_timer_shutdown(struct clock_event_device *evt)
+{
+	void __iomem *reg_base = timer_of_base(to_timer_of(evt));
+
+	writel_relaxed(0, reg_base + TIMER_PTV);
+
+	return 0;
+}
+
+static int tegra_timer_set_periodic(struct clock_event_device *evt)
+{
+	void __iomem *reg_base = timer_of_base(to_timer_of(evt));
+	unsigned long period = timer_of_period(to_timer_of(evt));
+
+	writel_relaxed(TIMER_PTV_EN | TIMER_PTV_PER | (period - 1),
+		       reg_base + TIMER_PTV);
+
+	return 0;
+}
+
+static irqreturn_t tegra_timer_isr(int irq, void *dev_id)
+{
+	struct clock_event_device *evt = dev_id;
+	void __iomem *reg_base = timer_of_base(to_timer_of(evt));
+
+	writel_relaxed(TIMER_PCR_INTR_CLR, reg_base + TIMER_PCR);
+	evt->event_handler(evt);
+
+	return IRQ_HANDLED;
+}
+
+static void tegra_timer_suspend(struct clock_event_device *evt)
+{
+	void __iomem *reg_base = timer_of_base(to_timer_of(evt));
+
+	writel_relaxed(TIMER_PCR_INTR_CLR, reg_base + TIMER_PCR);
+}
+
+static void tegra_timer_resume(struct clock_event_device *evt)
+{
+	writel_relaxed(usec_config, timer_reg_base + TIMERUS_USEC_CFG);
+}
+
+static DEFINE_PER_CPU(struct timer_of, tegra_to) = {
+	.flags = TIMER_OF_CLOCK | TIMER_OF_BASE,
+
+	.clkevt = {
+		.name = "tegra_timer",
+		.features = CLOCK_EVT_FEAT_ONESHOT | CLOCK_EVT_FEAT_PERIODIC,
+		.set_next_event = tegra_timer_set_next_event,
+		.set_state_shutdown = tegra_timer_shutdown,
+		.set_state_periodic = tegra_timer_set_periodic,
+		.set_state_oneshot = tegra_timer_shutdown,
+		.tick_resume = tegra_timer_shutdown,
+		.suspend = tegra_timer_suspend,
+		.resume = tegra_timer_resume,
+	},
+};
+
+static int tegra_timer_setup(unsigned int cpu)
+{
+	struct timer_of *to = per_cpu_ptr(&tegra_to, cpu);
+
+	writel_relaxed(0, timer_of_base(to) + TIMER_PTV);
+	writel_relaxed(TIMER_PCR_INTR_CLR, timer_of_base(to) + TIMER_PCR);
+
+	irq_force_affinity(to->clkevt.irq, cpumask_of(cpu));
+	enable_irq(to->clkevt.irq);
+
+	/*
+	 * Tegra's timer uses n+1 scheme for the counter, i.e. timer will
+	 * fire after one tick if 0 is loaded and thus minimum number of
+	 * ticks is 1. In result both of the clocksource's tick limits are
+	 * higher than a minimum and maximum that hardware register can
+	 * take by 1, this is then taken into account by set_next_event
+	 * callback.
+	 */
+	clockevents_config_and_register(&to->clkevt, timer_of_rate(to),
+					1, /* min */
+					0x1fffffff + 1); /* max 29 bits + 1 */
+
+	return 0;
+}
+
+static int tegra_timer_stop(unsigned int cpu)
+{
+	struct timer_of *to = per_cpu_ptr(&tegra_to, cpu);
+
+	disable_irq_nosync(to->clkevt.irq);
+
+	return 0;
+}
+
+static u64 notrace tegra_read_sched_clock(void)
+{
+	return readl_relaxed(timer_reg_base + TIMERUS_CNTR_1US);
+}
+
+#ifdef CONFIG_ARM
+static unsigned long tegra_delay_timer_read_counter_long(void)
+{
+	return readl_relaxed(timer_reg_base + TIMERUS_CNTR_1US);
+}
+
+static struct delay_timer tegra_delay_timer = {
+	.read_current_timer = tegra_delay_timer_read_counter_long,
+	.freq = TIMER_1MHz,
+};
+#endif
+
+static struct timer_of suspend_rtc_to = {
+	.flags = TIMER_OF_BASE | TIMER_OF_CLOCK,
+};
+
+/*
+ * tegra_rtc_read - Reads the Tegra RTC registers
+ * Care must be taken that this function is not called while the
+ * tegra_rtc driver could be executing to avoid race conditions
+ * on the RTC shadow register
+ */
+static u64 tegra_rtc_read_ms(struct clocksource *cs)
+{
+	void __iomem *reg_base = timer_of_base(&suspend_rtc_to);
+
+	u32 ms = readl_relaxed(reg_base + RTC_MILLISECONDS);
+	u32 s = readl_relaxed(reg_base + RTC_SHADOW_SECONDS);
+
+	return (u64)s * MSEC_PER_SEC + ms;
+}
+
+static struct clocksource suspend_rtc_clocksource = {
+	.name	= "tegra_suspend_timer",
+	.rating	= 200,
+	.read	= tegra_rtc_read_ms,
+	.mask	= CLOCKSOURCE_MASK(32),
+	.flags	= CLOCK_SOURCE_IS_CONTINUOUS | CLOCK_SOURCE_SUSPEND_NONSTOP,
+};
+
+static inline unsigned int tegra_base_for_cpu(int cpu, bool tegra20)
+{
+	if (tegra20) {
+		switch (cpu) {
+		case 0:
+			return TIMER1_BASE;
+		case 1:
+			return TIMER2_BASE;
+		case 2:
+			return TIMER3_BASE;
+		default:
+			return TIMER4_BASE;
+		}
+	}
+
+	return TIMER10_BASE + cpu * 8;
+}
+
+static inline unsigned int tegra_irq_idx_for_cpu(int cpu, bool tegra20)
+{
+	if (tegra20)
+		return TIMER1_IRQ_IDX + cpu;
+
+	return TIMER10_IRQ_IDX + cpu;
+}
+
+static inline unsigned long tegra_rate_for_timer(struct timer_of *to,
+						 bool tegra20)
+{
+	/*
+	 * TIMER1-9 are fixed to 1MHz, TIMER10-13 are running off the
+	 * parent clock.
+	 */
+	if (tegra20)
+		return TIMER_1MHz;
+
+	return timer_of_rate(to);
+}
+
+static int __init tegra_init_timer(struct device_node *np, bool tegra20,
+				   int rating)
+{
+	struct timer_of *to;
+	int cpu, ret;
+
+	to = this_cpu_ptr(&tegra_to);
+	ret = timer_of_init(np, to);
+	if (ret)
+		goto out;
+
+	timer_reg_base = timer_of_base(to);
+
+	/*
+	 * Configure microsecond timers to have 1MHz clock
+	 * Config register is 0xqqww, where qq is "dividend", ww is "divisor"
+	 * Uses n+1 scheme
+	 */
+	switch (timer_of_rate(to)) {
+	case 12000000:
+		usec_config = 0x000b; /* (11+1)/(0+1) */
+		break;
+	case 12800000:
+		usec_config = 0x043f; /* (63+1)/(4+1) */
+		break;
+	case 13000000:
+		usec_config = 0x000c; /* (12+1)/(0+1) */
+		break;
+	case 16800000:
+		usec_config = 0x0453; /* (83+1)/(4+1) */
+		break;
+	case 19200000:
+		usec_config = 0x045f; /* (95+1)/(4+1) */
+		break;
+	case 26000000:
+		usec_config = 0x0019; /* (25+1)/(0+1) */
+		break;
+	case 38400000:
+		usec_config = 0x04bf; /* (191+1)/(4+1) */
+		break;
+	case 48000000:
+		usec_config = 0x002f; /* (47+1)/(0+1) */
+		break;
+	default:
+		ret = -EINVAL;
+		goto out;
+	}
+
+	writel_relaxed(usec_config, timer_reg_base + TIMERUS_USEC_CFG);
+
+	for_each_possible_cpu(cpu) {
+		struct timer_of *cpu_to = per_cpu_ptr(&tegra_to, cpu);
+		unsigned long flags = IRQF_TIMER | IRQF_NOBALANCING;
+		unsigned long rate = tegra_rate_for_timer(to, tegra20);
+		unsigned int base = tegra_base_for_cpu(cpu, tegra20);
+		unsigned int idx = tegra_irq_idx_for_cpu(cpu, tegra20);
+		unsigned int irq = irq_of_parse_and_map(np, idx);
+
+		if (!irq) {
+			pr_err("failed to map irq for cpu%d\n", cpu);
+			ret = -EINVAL;
+			goto out_irq;
+		}
+
+		cpu_to->clkevt.irq = irq;
+		cpu_to->clkevt.rating = rating;
+		cpu_to->clkevt.cpumask = cpumask_of(cpu);
+		cpu_to->of_base.base = timer_reg_base + base;
+		cpu_to->of_clk.period = rate / HZ;
+		cpu_to->of_clk.rate = rate;
+
+		irq_set_status_flags(cpu_to->clkevt.irq, IRQ_NOAUTOEN);
+
+		ret = request_irq(cpu_to->clkevt.irq, tegra_timer_isr, flags,
+				  cpu_to->clkevt.name, &cpu_to->clkevt);
+		if (ret) {
+			pr_err("failed to set up irq for cpu%d: %d\n",
+			       cpu, ret);
+			irq_dispose_mapping(cpu_to->clkevt.irq);
+			cpu_to->clkevt.irq = 0;
+			goto out_irq;
+		}
+	}
+
+	sched_clock_register(tegra_read_sched_clock, 32, TIMER_1MHz);
+
+	ret = clocksource_mmio_init(timer_reg_base + TIMERUS_CNTR_1US,
+				    "timer_us", TIMER_1MHz, 300, 32,
+				    clocksource_mmio_readl_up);
+	if (ret)
+		pr_err("failed to register clocksource: %d\n", ret);
+
+#ifdef CONFIG_ARM
+	register_current_timer_delay(&tegra_delay_timer);
+#endif
+
+	ret = cpuhp_setup_state(CPUHP_AP_TEGRA_TIMER_STARTING,
+				"AP_TEGRA_TIMER_STARTING", tegra_timer_setup,
+				tegra_timer_stop);
+	if (ret)
+		pr_err("failed to set up cpu hp state: %d\n", ret);
+
+	return ret;
+
+out_irq:
+	for_each_possible_cpu(cpu) {
+		struct timer_of *cpu_to;
+
+		cpu_to = per_cpu_ptr(&tegra_to, cpu);
+		if (cpu_to->clkevt.irq) {
+			free_irq(cpu_to->clkevt.irq, &cpu_to->clkevt);
+			irq_dispose_mapping(cpu_to->clkevt.irq);
+		}
+	}
+
+	to->of_base.base = timer_reg_base;
+out:
+	timer_of_cleanup(to);
+
+	return ret;
+}
+
+static int __init tegra210_init_timer(struct device_node *np)
+{
+	/*
+	 * Arch-timer can't survive across power cycle of CPU core and
+	 * after CPUPORESET signal due to a system design shortcoming,
+	 * hence tegra-timer is more preferable on Tegra210.
+	 */
+	return tegra_init_timer(np, false, 460);
+}
+TIMER_OF_DECLARE(tegra210_timer, "nvidia,tegra210-timer", tegra210_init_timer);
+
+static int __init tegra20_init_timer(struct device_node *np)
+{
+	int rating;
+
+	/*
+	 * Tegra20 and Tegra30 have Cortex A9 CPU that has a TWD timer,
+	 * that timer runs off the CPU clock and hence is subjected to
+	 * a jitter caused by DVFS clock rate changes. Tegra-timer is
+	 * more preferable for older Tegra's, while later SoC generations
+	 * have arch-timer as a main per-CPU timer and it is not affected
+	 * by DVFS changes.
+	 */
+	if (of_machine_is_compatible("nvidia,tegra20") ||
+	    of_machine_is_compatible("nvidia,tegra30"))
+		rating = 460;
+	else
+		rating = 330;
+
+	return tegra_init_timer(np, true, rating);
+}
+TIMER_OF_DECLARE(tegra20_timer, "nvidia,tegra20-timer", tegra20_init_timer);
+
+static int __init tegra20_init_rtc(struct device_node *np)
+{
+	int ret;
+
+	ret = timer_of_init(np, &suspend_rtc_to);
+	if (ret)
+		return ret;
+
+	return clocksource_register_hz(&suspend_rtc_clocksource, 1000);
+}
+TIMER_OF_DECLARE(tegra20_rtc, "nvidia,tegra20-rtc", tegra20_init_rtc);