x86/kprobes: Fix crash when probing CS CALL instructionsgrafted

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) <mhiramat@kernel.org>
Signed-off-by: Jinke Han <jinkehan@didiglobal.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Acked-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000

1 files changed, 227 insertions, 0 deletions

diff --git a/drivers/clocksource/timer-pxa.c b/drivers/clocksource/timer-pxa.c
new file mode 100644
index 000000000..7ad0e5adb
--- /dev/null
+++ b/drivers/clocksource/timer-pxa.c
@@ -0,0 +1,227 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * arch/arm/mach-pxa/time.c
+ *
+ * PXA clocksource, clockevents, and OST interrupt handlers.
+ * Copyright (c) 2007 by Bill Gatliff <bgat@billgatliff.com>.
+ *
+ * Derived from Nicolas Pitre's PXA timer handler Copyright (c) 2001
+ * by MontaVista Software, Inc.  (Nico, your code rocks!)
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/clk.h>
+#include <linux/clockchips.h>
+#include <linux/of_address.h>
+#include <linux/of_irq.h>
+#include <linux/sched/clock.h>
+#include <linux/sched_clock.h>
+
+#include <clocksource/pxa.h>
+
+#include <asm/div64.h>
+
+#define OSMR0		0x00	/* OS Timer 0 Match Register */
+#define OSMR1		0x04	/* OS Timer 1 Match Register */
+#define OSMR2		0x08	/* OS Timer 2 Match Register */
+#define OSMR3		0x0C	/* OS Timer 3 Match Register */
+
+#define OSCR		0x10	/* OS Timer Counter Register */
+#define OSSR		0x14	/* OS Timer Status Register */
+#define OWER		0x18	/* OS Timer Watchdog Enable Register */
+#define OIER		0x1C	/* OS Timer Interrupt Enable Register */
+
+#define OSSR_M3		(1 << 3)	/* Match status channel 3 */
+#define OSSR_M2		(1 << 2)	/* Match status channel 2 */
+#define OSSR_M1		(1 << 1)	/* Match status channel 1 */
+#define OSSR_M0		(1 << 0)	/* Match status channel 0 */
+
+#define OIER_E0		(1 << 0)	/* Interrupt enable channel 0 */
+
+/*
+ * This is PXA's sched_clock implementation. This has a resolution
+ * of at least 308 ns and a maximum value of 208 days.
+ *
+ * The return value is guaranteed to be monotonic in that range as
+ * long as there is always less than 582 seconds between successive
+ * calls to sched_clock() which should always be the case in practice.
+ */
+
+#define timer_readl(reg) readl_relaxed(timer_base + (reg))
+#define timer_writel(val, reg) writel_relaxed((val), timer_base + (reg))
+
+static void __iomem *timer_base;
+
+static u64 notrace pxa_read_sched_clock(void)
+{
+	return timer_readl(OSCR);
+}
+
+
+#define MIN_OSCR_DELTA 16
+
+static irqreturn_t
+pxa_ost0_interrupt(int irq, void *dev_id)
+{
+	struct clock_event_device *c = dev_id;
+
+	/* Disarm the compare/match, signal the event. */
+	timer_writel(timer_readl(OIER) & ~OIER_E0, OIER);
+	timer_writel(OSSR_M0, OSSR);
+	c->event_handler(c);
+
+	return IRQ_HANDLED;
+}
+
+static int
+pxa_osmr0_set_next_event(unsigned long delta, struct clock_event_device *dev)
+{
+	unsigned long next, oscr;
+
+	timer_writel(timer_readl(OIER) | OIER_E0, OIER);
+	next = timer_readl(OSCR) + delta;
+	timer_writel(next, OSMR0);
+	oscr = timer_readl(OSCR);
+
+	return (signed)(next - oscr) <= MIN_OSCR_DELTA ? -ETIME : 0;
+}
+
+static int pxa_osmr0_shutdown(struct clock_event_device *evt)
+{
+	/* initializing, released, or preparing for suspend */
+	timer_writel(timer_readl(OIER) & ~OIER_E0, OIER);
+	timer_writel(OSSR_M0, OSSR);
+	return 0;
+}
+
+#ifdef CONFIG_PM
+static unsigned long osmr[4], oier, oscr;
+
+static void pxa_timer_suspend(struct clock_event_device *cedev)
+{
+	osmr[0] = timer_readl(OSMR0);
+	osmr[1] = timer_readl(OSMR1);
+	osmr[2] = timer_readl(OSMR2);
+	osmr[3] = timer_readl(OSMR3);
+	oier = timer_readl(OIER);
+	oscr = timer_readl(OSCR);
+}
+
+static void pxa_timer_resume(struct clock_event_device *cedev)
+{
+	/*
+	 * Ensure that we have at least MIN_OSCR_DELTA between match
+	 * register 0 and the OSCR, to guarantee that we will receive
+	 * the one-shot timer interrupt.  We adjust OSMR0 in preference
+	 * to OSCR to guarantee that OSCR is monotonically incrementing.
+	 */
+	if (osmr[0] - oscr < MIN_OSCR_DELTA)
+		osmr[0] += MIN_OSCR_DELTA;
+
+	timer_writel(osmr[0], OSMR0);
+	timer_writel(osmr[1], OSMR1);
+	timer_writel(osmr[2], OSMR2);
+	timer_writel(osmr[3], OSMR3);
+	timer_writel(oier, OIER);
+	timer_writel(oscr, OSCR);
+}
+#else
+#define pxa_timer_suspend NULL
+#define pxa_timer_resume NULL
+#endif
+
+static struct clock_event_device ckevt_pxa_osmr0 = {
+	.name			= "osmr0",
+	.features		= CLOCK_EVT_FEAT_ONESHOT,
+	.rating			= 200,
+	.set_next_event		= pxa_osmr0_set_next_event,
+	.set_state_shutdown	= pxa_osmr0_shutdown,
+	.set_state_oneshot	= pxa_osmr0_shutdown,
+	.suspend		= pxa_timer_suspend,
+	.resume			= pxa_timer_resume,
+};
+
+static int __init pxa_timer_common_init(int irq, unsigned long clock_tick_rate)
+{
+	int ret;
+
+	timer_writel(0, OIER);
+	timer_writel(OSSR_M0 | OSSR_M1 | OSSR_M2 | OSSR_M3, OSSR);
+
+	sched_clock_register(pxa_read_sched_clock, 32, clock_tick_rate);
+
+	ckevt_pxa_osmr0.cpumask = cpumask_of(0);
+
+	ret = request_irq(irq, pxa_ost0_interrupt, IRQF_TIMER | IRQF_IRQPOLL,
+			  "ost0", &ckevt_pxa_osmr0);
+	if (ret) {
+		pr_err("Failed to setup irq\n");
+		return ret;
+	}
+
+	ret = clocksource_mmio_init(timer_base + OSCR, "oscr0", clock_tick_rate, 200,
+				    32, clocksource_mmio_readl_up);
+	if (ret) {
+		pr_err("Failed to init clocksource\n");
+		return ret;
+	}
+
+	clockevents_config_and_register(&ckevt_pxa_osmr0, clock_tick_rate,
+					MIN_OSCR_DELTA * 2, 0x7fffffff);
+
+	return 0;
+}
+
+static int __init pxa_timer_dt_init(struct device_node *np)
+{
+	struct clk *clk;
+	int irq, ret;
+
+	/* timer registers are shared with watchdog timer */
+	timer_base = of_iomap(np, 0);
+	if (!timer_base) {
+		pr_err("%pOFn: unable to map resource\n", np);
+		return -ENXIO;
+	}
+
+	clk = of_clk_get(np, 0);
+	if (IS_ERR(clk)) {
+		pr_crit("%pOFn: unable to get clk\n", np);
+		return PTR_ERR(clk);
+	}
+
+	ret = clk_prepare_enable(clk);
+	if (ret) {
+		pr_crit("Failed to prepare clock\n");
+		return ret;
+	}
+
+	/* we are only interested in OS-timer0 irq */
+	irq = irq_of_parse_and_map(np, 0);
+	if (irq <= 0) {
+		pr_crit("%pOFn: unable to parse OS-timer0 irq\n", np);
+		return -EINVAL;
+	}
+
+	return pxa_timer_common_init(irq, clk_get_rate(clk));
+}
+TIMER_OF_DECLARE(pxa_timer, "marvell,pxa-timer", pxa_timer_dt_init);
+
+/*
+ * Legacy timer init for non device-tree boards.
+ */
+void __init pxa_timer_nodt_init(int irq, void __iomem *base)
+{
+	struct clk *clk;
+
+	timer_base = base;
+	clk = clk_get(NULL, "OSTIMER0");
+	if (clk && !IS_ERR(clk)) {
+		clk_prepare_enable(clk);
+		pxa_timer_common_init(irq, clk_get_rate(clk));
+	} else {
+		pr_crit("%s: unable to get clk\n", __func__);
+	}
+}