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-mp-csky.c | 173 ++++++++++++++++++++++++++++++++++++ 1 file changed, 173 insertions(+) create mode 100644 drivers/clocksource/timer-mp-csky.c (limited to 'drivers/clocksource/timer-mp-csky.c') diff --git a/drivers/clocksource/timer-mp-csky.c b/drivers/clocksource/timer-mp-csky.c new file mode 100644 index 000000000..183a99551 --- /dev/null +++ b/drivers/clocksource/timer-mp-csky.c @@ -0,0 +1,173 @@ +// SPDX-License-Identifier: GPL-2.0 +// Copyright (C) 2018 Hangzhou C-SKY Microsystems co.,ltd. + +#include +#include +#include +#include +#include +#include + +#include "timer-of.h" + +#define PTIM_CCVR "cr<3, 14>" +#define PTIM_CTLR "cr<0, 14>" +#define PTIM_LVR "cr<6, 14>" +#define PTIM_TSR "cr<1, 14>" + +static int csky_mptimer_irq; + +static int csky_mptimer_set_next_event(unsigned long delta, + struct clock_event_device *ce) +{ + mtcr(PTIM_LVR, delta); + + return 0; +} + +static int csky_mptimer_shutdown(struct clock_event_device *ce) +{ + mtcr(PTIM_CTLR, 0); + + return 0; +} + +static int csky_mptimer_oneshot(struct clock_event_device *ce) +{ + mtcr(PTIM_CTLR, 1); + + return 0; +} + +static int csky_mptimer_oneshot_stopped(struct clock_event_device *ce) +{ + mtcr(PTIM_CTLR, 0); + + return 0; +} + +static DEFINE_PER_CPU(struct timer_of, csky_to) = { + .flags = TIMER_OF_CLOCK, + .clkevt = { + .rating = 300, + .features = CLOCK_EVT_FEAT_PERCPU | + CLOCK_EVT_FEAT_ONESHOT, + .set_state_shutdown = csky_mptimer_shutdown, + .set_state_oneshot = csky_mptimer_oneshot, + .set_state_oneshot_stopped = csky_mptimer_oneshot_stopped, + .set_next_event = csky_mptimer_set_next_event, + }, +}; + +static irqreturn_t csky_timer_interrupt(int irq, void *dev) +{ + struct timer_of *to = this_cpu_ptr(&csky_to); + + mtcr(PTIM_TSR, 0); + + to->clkevt.event_handler(&to->clkevt); + + return IRQ_HANDLED; +} + +/* + * clock event for percpu + */ +static int csky_mptimer_starting_cpu(unsigned int cpu) +{ + struct timer_of *to = per_cpu_ptr(&csky_to, cpu); + + to->clkevt.cpumask = cpumask_of(cpu); + + enable_percpu_irq(csky_mptimer_irq, 0); + + clockevents_config_and_register(&to->clkevt, timer_of_rate(to), + 2, ULONG_MAX); + + return 0; +} + +static int csky_mptimer_dying_cpu(unsigned int cpu) +{ + disable_percpu_irq(csky_mptimer_irq); + + return 0; +} + +/* + * clock source + */ +static u64 notrace sched_clock_read(void) +{ + return (u64)mfcr(PTIM_CCVR); +} + +static u64 clksrc_read(struct clocksource *c) +{ + return (u64)mfcr(PTIM_CCVR); +} + +struct clocksource csky_clocksource = { + .name = "csky", + .rating = 400, + .mask = CLOCKSOURCE_MASK(32), + .flags = CLOCK_SOURCE_IS_CONTINUOUS, + .read = clksrc_read, +}; + +static int __init csky_mptimer_init(struct device_node *np) +{ + int ret, cpu, cpu_rollback; + struct timer_of *to = NULL; + + /* + * Csky_mptimer is designed for C-SKY SMP multi-processors and + * every core has it's own private irq and regs for clkevt and + * clksrc. + * + * The regs is accessed by cpu instruction: mfcr/mtcr instead of + * mmio map style. So we needn't mmio-address in dts, but we still + * need to give clk and irq number. + * + * We use private irq for the mptimer and irq number is the same + * for every core. So we use request_percpu_irq() in timer_of_init. + */ + csky_mptimer_irq = irq_of_parse_and_map(np, 0); + if (csky_mptimer_irq <= 0) + return -EINVAL; + + ret = request_percpu_irq(csky_mptimer_irq, csky_timer_interrupt, + "csky_mp_timer", &csky_to); + if (ret) + return -EINVAL; + + for_each_possible_cpu(cpu) { + to = per_cpu_ptr(&csky_to, cpu); + ret = timer_of_init(np, to); + if (ret) + goto rollback; + } + + clocksource_register_hz(&csky_clocksource, timer_of_rate(to)); + sched_clock_register(sched_clock_read, 32, timer_of_rate(to)); + + ret = cpuhp_setup_state(CPUHP_AP_CSKY_TIMER_STARTING, + "clockevents/csky/timer:starting", + csky_mptimer_starting_cpu, + csky_mptimer_dying_cpu); + if (ret) + return -EINVAL; + + return 0; + +rollback: + for_each_possible_cpu(cpu_rollback) { + if (cpu_rollback == cpu) + break; + + to = per_cpu_ptr(&csky_to, cpu_rollback); + timer_of_cleanup(to); + } + return -EINVAL; +} +TIMER_OF_DECLARE(csky_mptimer, "csky,mptimer", csky_mptimer_init); -- cgit v1.3.1