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, 155 insertions, 0 deletions

diff --git a/drivers/clocksource/timer-gx6605s.c b/drivers/clocksource/timer-gx6605s.c
new file mode 100644
index 000000000..8d386adbe
--- /dev/null
+++ b/drivers/clocksource/timer-gx6605s.c
@@ -0,0 +1,155 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (C) 2018 Hangzhou C-SKY Microsystems co.,ltd.
+
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/sched_clock.h>
+
+#include "timer-of.h"
+
+#define CLKSRC_OFFSET	0x40
+
+#define TIMER_STATUS	0x00
+#define TIMER_VALUE	0x04
+#define TIMER_CONTRL	0x10
+#define TIMER_CONFIG	0x20
+#define TIMER_DIV	0x24
+#define TIMER_INI	0x28
+
+#define GX6605S_STATUS_CLR	BIT(0)
+#define GX6605S_CONTRL_RST	BIT(0)
+#define GX6605S_CONTRL_START	BIT(1)
+#define GX6605S_CONFIG_EN	BIT(0)
+#define GX6605S_CONFIG_IRQ_EN	BIT(1)
+
+static irqreturn_t gx6605s_timer_interrupt(int irq, void *dev)
+{
+	struct clock_event_device *ce = dev;
+	void __iomem *base = timer_of_base(to_timer_of(ce));
+
+	writel_relaxed(GX6605S_STATUS_CLR, base + TIMER_STATUS);
+	writel_relaxed(0, base + TIMER_INI);
+
+	ce->event_handler(ce);
+
+	return IRQ_HANDLED;
+}
+
+static int gx6605s_timer_set_oneshot(struct clock_event_device *ce)
+{
+	void __iomem *base = timer_of_base(to_timer_of(ce));
+
+	/* reset and stop counter */
+	writel_relaxed(GX6605S_CONTRL_RST, base + TIMER_CONTRL);
+
+	/* enable with irq and start */
+	writel_relaxed(GX6605S_CONFIG_EN | GX6605S_CONFIG_IRQ_EN,
+		       base + TIMER_CONFIG);
+
+	return 0;
+}
+
+static int gx6605s_timer_set_next_event(unsigned long delta,
+					struct clock_event_device *ce)
+{
+	void __iomem *base = timer_of_base(to_timer_of(ce));
+
+	/* use reset to pause timer */
+	writel_relaxed(GX6605S_CONTRL_RST, base + TIMER_CONTRL);
+
+	/* config next timeout value */
+	writel_relaxed(ULONG_MAX - delta, base + TIMER_INI);
+	writel_relaxed(GX6605S_CONTRL_START, base + TIMER_CONTRL);
+
+	return 0;
+}
+
+static int gx6605s_timer_shutdown(struct clock_event_device *ce)
+{
+	void __iomem *base = timer_of_base(to_timer_of(ce));
+
+	writel_relaxed(0, base + TIMER_CONTRL);
+	writel_relaxed(0, base + TIMER_CONFIG);
+
+	return 0;
+}
+
+static struct timer_of to = {
+	.flags = TIMER_OF_IRQ | TIMER_OF_BASE | TIMER_OF_CLOCK,
+	.clkevt = {
+		.rating			= 300,
+		.features		= CLOCK_EVT_FEAT_DYNIRQ |
+					  CLOCK_EVT_FEAT_ONESHOT,
+		.set_state_shutdown	= gx6605s_timer_shutdown,
+		.set_state_oneshot	= gx6605s_timer_set_oneshot,
+		.set_next_event		= gx6605s_timer_set_next_event,
+		.cpumask		= cpu_possible_mask,
+	},
+	.of_irq = {
+		.handler		= gx6605s_timer_interrupt,
+		.flags			= IRQF_TIMER | IRQF_IRQPOLL,
+	},
+};
+
+static u64 notrace gx6605s_sched_clock_read(void)
+{
+	void __iomem *base;
+
+	base = timer_of_base(&to) + CLKSRC_OFFSET;
+
+	return (u64)readl_relaxed(base + TIMER_VALUE);
+}
+
+static void gx6605s_clkevt_init(void __iomem *base)
+{
+	writel_relaxed(0, base + TIMER_DIV);
+	writel_relaxed(0, base + TIMER_CONFIG);
+
+	clockevents_config_and_register(&to.clkevt, timer_of_rate(&to), 2,
+					ULONG_MAX);
+}
+
+static int gx6605s_clksrc_init(void __iomem *base)
+{
+	writel_relaxed(0, base + TIMER_DIV);
+	writel_relaxed(0, base + TIMER_INI);
+
+	writel_relaxed(GX6605S_CONTRL_RST, base + TIMER_CONTRL);
+
+	writel_relaxed(GX6605S_CONFIG_EN, base + TIMER_CONFIG);
+
+	writel_relaxed(GX6605S_CONTRL_START, base + TIMER_CONTRL);
+
+	sched_clock_register(gx6605s_sched_clock_read, 32, timer_of_rate(&to));
+
+	return clocksource_mmio_init(base + TIMER_VALUE, "gx6605s",
+			timer_of_rate(&to), 200, 32, clocksource_mmio_readl_up);
+}
+
+static int __init gx6605s_timer_init(struct device_node *np)
+{
+	int ret;
+
+	/*
+	 * The timer driver is for nationalchip gx6605s SOC and there are two
+	 * same timer in gx6605s. We use one for clkevt and another for clksrc.
+	 *
+	 * The timer is mmio map to access, so we need give mmio address in dts.
+	 *
+	 * It provides a 32bit countup timer and interrupt will be caused by
+	 * count-overflow.
+	 * So we need set-next-event by ULONG_MAX - delta in TIMER_INI reg.
+	 *
+	 * The counter at 0x0  offset is clock event.
+	 * The counter at 0x40 offset is clock source.
+	 * They are the same in hardware, just different used by driver.
+	 */
+	ret = timer_of_init(np, &to);
+	if (ret)
+		return ret;
+
+	gx6605s_clkevt_init(timer_of_base(&to));
+
+	return gx6605s_clksrc_init(timer_of_base(&to) + CLKSRC_OFFSET);
+}
+TIMER_OF_DECLARE(csky_gx6605s_timer, "csky,gx6605s-timer", gx6605s_timer_init);