From a5f7a5bb3b7f28ba7e4fa246775b29a0e5537255 Mon Sep 17 00:00:00 2001 From: Jinke Han Date: Tue, 8 Sep 2026 15:37:42 +0800 Subject: x86/kprobes: Fix crash when probing CS CALL instructions When using eBPF to probe CS CALL instructions within a function, a crash can be triggered. The eBPF tool probes offset 257 of the __hrtimer_run_queues() function: <__hrtimer_run_queues+249>: nopl 0x0(%rax,%rax,1) <__hrtimer_run_queues+254>: mov %r14,%rdi <__hrtimer_run_queues+257>: cs call <__x86_indirect_thunk_r12> <__hrtimer_run_queues+263>: mov %eax,%r12d <__hrtimer_run_queues+266>: xchg %ax,%ax <__hrtimer_run_queues+268>: mov %r13,%rdi Which triggers this crash: BUG: unable to handle page fault for address: 00000000000f41c9 #PF: supervisor write access in kernel mode #PF: error_code(0x0002) - not-present page PGD 0 P4D 0 Oops: 0002 [#1] SMP NOPTI CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Tainted: P RIP: 0010:__hrtimer_run_queues+0x106/0x230 Note that __hrtimer_run_queues+0x106 is __hrtimer_run_queues+262, which is at the 6th byte of the above CS CALL instruction. Since the CS CALL instruction occupies 6 bytes, the exception occurred in the middle of that call instruction. The root cause is that when using eBPF tools to probe in the middle of a function, a kprobe with INT3 is used as the underlying implementation. During single-step emulation of the original CALL instruction, int3_emulate_call() assumes that the probed CALL instruction is 5 bytes long. However, the actual CS-prefixed CALL instruction occupies 6 bytes, so it constructs an incorrect exception return address. When the CPU returns from the kprobe handler, the next instruction to be executed is at the address of the last byte of that CS CALL instruction. Coincidentally, starting from that address, the CPU fetches and decodes a completely different instruction, which ultimately triggers a kernel crash. Fix the issue by using the actual instruction length obtained from the instruction decoder when constructing the exception return address, rather than relying on the hardcoded CALL_INSN_SIZE macro. [ mingo: Refined the changelog ] Fixes: 6256e668b7af ("x86/kprobes: Use int3 instead of debug trap for single-step") Suggested-by: Masami Hiramatsu (Google) Signed-off-by: Jinke Han Signed-off-by: Ingo Molnar Reviewed-by: Masami Hiramatsu (Google) Acked-by: Yafang Shao Acked-by: Borislav Petkov Cc: Peter Zijlstra Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000 --- drivers/clocksource/timer-owl.c | 176 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 176 insertions(+) create mode 100644 drivers/clocksource/timer-owl.c (limited to 'drivers/clocksource/timer-owl.c') diff --git a/drivers/clocksource/timer-owl.c b/drivers/clocksource/timer-owl.c new file mode 100644 index 000000000..ac97420bf --- /dev/null +++ b/drivers/clocksource/timer-owl.c @@ -0,0 +1,176 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * Actions Semi Owl timer + * + * Copyright 2012 Actions Semi Inc. + * Author: Actions Semi, Inc. + * + * Copyright (c) 2017 SUSE Linux GmbH + * Author: Andreas Färber + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#define OWL_Tx_CTL 0x0 +#define OWL_Tx_CMP 0x4 +#define OWL_Tx_VAL 0x8 + +#define OWL_Tx_CTL_PD BIT(0) +#define OWL_Tx_CTL_INTEN BIT(1) +#define OWL_Tx_CTL_EN BIT(2) + +static void __iomem *owl_timer_base; +static void __iomem *owl_clksrc_base; +static void __iomem *owl_clkevt_base; + +static inline void owl_timer_reset(void __iomem *base) +{ + writel(0, base + OWL_Tx_CTL); + writel(0, base + OWL_Tx_VAL); + writel(0, base + OWL_Tx_CMP); +} + +static inline void owl_timer_set_enabled(void __iomem *base, bool enabled) +{ + u32 ctl = readl(base + OWL_Tx_CTL); + + /* PD bit is cleared when set */ + ctl &= ~OWL_Tx_CTL_PD; + + if (enabled) + ctl |= OWL_Tx_CTL_EN; + else + ctl &= ~OWL_Tx_CTL_EN; + + writel(ctl, base + OWL_Tx_CTL); +} + +static u64 notrace owl_timer_sched_read(void) +{ + return (u64)readl(owl_clksrc_base + OWL_Tx_VAL); +} + +static int owl_timer_set_state_shutdown(struct clock_event_device *evt) +{ + owl_timer_set_enabled(owl_clkevt_base, false); + + return 0; +} + +static int owl_timer_set_state_oneshot(struct clock_event_device *evt) +{ + owl_timer_reset(owl_clkevt_base); + + return 0; +} + +static int owl_timer_tick_resume(struct clock_event_device *evt) +{ + return 0; +} + +static int owl_timer_set_next_event(unsigned long evt, + struct clock_event_device *ev) +{ + void __iomem *base = owl_clkevt_base; + + owl_timer_set_enabled(base, false); + writel(OWL_Tx_CTL_INTEN, base + OWL_Tx_CTL); + writel(0, base + OWL_Tx_VAL); + writel(evt, base + OWL_Tx_CMP); + owl_timer_set_enabled(base, true); + + return 0; +} + +static struct clock_event_device owl_clockevent = { + .name = "owl_tick", + .rating = 200, + .features = CLOCK_EVT_FEAT_ONESHOT | + CLOCK_EVT_FEAT_DYNIRQ, + .set_state_shutdown = owl_timer_set_state_shutdown, + .set_state_oneshot = owl_timer_set_state_oneshot, + .tick_resume = owl_timer_tick_resume, + .set_next_event = owl_timer_set_next_event, +}; + +static irqreturn_t owl_timer1_interrupt(int irq, void *dev_id) +{ + struct clock_event_device *evt = (struct clock_event_device *)dev_id; + + writel(OWL_Tx_CTL_PD, owl_clkevt_base + OWL_Tx_CTL); + + evt->event_handler(evt); + + return IRQ_HANDLED; +} + +static int __init owl_timer_init(struct device_node *node) +{ + struct clk *clk; + unsigned long rate; + int timer1_irq, ret; + + owl_timer_base = of_io_request_and_map(node, 0, "owl-timer"); + if (IS_ERR(owl_timer_base)) { + pr_err("Can't map timer registers\n"); + return PTR_ERR(owl_timer_base); + } + + owl_clksrc_base = owl_timer_base + 0x08; + owl_clkevt_base = owl_timer_base + 0x14; + + timer1_irq = of_irq_get_byname(node, "timer1"); + if (timer1_irq <= 0) { + pr_err("Can't parse timer1 IRQ\n"); + return -EINVAL; + } + + clk = of_clk_get(node, 0); + if (IS_ERR(clk)) { + ret = PTR_ERR(clk); + pr_err("Failed to get clock for clocksource (%d)\n", ret); + return ret; + } + + rate = clk_get_rate(clk); + + owl_timer_reset(owl_clksrc_base); + owl_timer_set_enabled(owl_clksrc_base, true); + + sched_clock_register(owl_timer_sched_read, 32, rate); + ret = clocksource_mmio_init(owl_clksrc_base + OWL_Tx_VAL, node->name, + rate, 200, 32, clocksource_mmio_readl_up); + if (ret) { + pr_err("Failed to register clocksource (%d)\n", ret); + return ret; + } + + owl_timer_reset(owl_clkevt_base); + + ret = request_irq(timer1_irq, owl_timer1_interrupt, IRQF_TIMER, + "owl-timer", &owl_clockevent); + if (ret) { + pr_err("failed to request irq %d\n", timer1_irq); + return ret; + } + + owl_clockevent.cpumask = cpumask_of(0); + owl_clockevent.irq = timer1_irq; + + clockevents_config_and_register(&owl_clockevent, rate, + 0xf, 0xffffffff); + + return 0; +} +TIMER_OF_DECLARE(owl_s500, "actions,s500-timer", owl_timer_init); +TIMER_OF_DECLARE(owl_s700, "actions,s700-timer", owl_timer_init); +TIMER_OF_DECLARE(owl_s900, "actions,s900-timer", owl_timer_init); -- cgit v1.3.1