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/cpufreq/davinci-cpufreq.c b/drivers/cpufreq/davinci-cpufreq.c
new file mode 100644
index 000000000..2c277eb37
--- /dev/null
+++ b/drivers/cpufreq/davinci-cpufreq.c
@@ -0,0 +1,155 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * CPU frequency scaling for DaVinci
+ *
+ * Copyright (C) 2009 Texas Instruments Incorporated - https://www.ti.com/
+ *
+ * Based on linux/arch/arm/plat-omap/cpu-omap.c. Original Copyright follows:
+ *
+ *  Copyright (C) 2005 Nokia Corporation
+ *  Written by Tony Lindgren <tony@atomide.com>
+ *
+ *  Based on cpu-sa1110.c, Copyright (C) 2001 Russell King
+ *
+ * Copyright (C) 2007-2008 Texas Instruments, Inc.
+ * Updated to support OMAP3
+ * Rajendra Nayak <rnayak@ti.com>
+ */
+#include <linux/types.h>
+#include <linux/cpufreq.h>
+#include <linux/init.h>
+#include <linux/err.h>
+#include <linux/clk.h>
+#include <linux/platform_data/davinci-cpufreq.h>
+#include <linux/platform_device.h>
+#include <linux/export.h>
+
+struct davinci_cpufreq {
+	struct device *dev;
+	struct clk *armclk;
+	struct clk *asyncclk;
+	unsigned long asyncrate;
+};
+static struct davinci_cpufreq cpufreq;
+
+static int davinci_target(struct cpufreq_policy *policy, unsigned int idx)
+{
+	struct davinci_cpufreq_config *pdata = cpufreq.dev->platform_data;
+	struct clk *armclk = cpufreq.armclk;
+	unsigned int old_freq, new_freq;
+	int ret = 0;
+
+	old_freq = policy->cur;
+	new_freq = pdata->freq_table[idx].frequency;
+
+	/* if moving to higher frequency, up the voltage beforehand */
+	if (pdata->set_voltage && new_freq > old_freq) {
+		ret = pdata->set_voltage(idx);
+		if (ret)
+			return ret;
+	}
+
+	ret = clk_set_rate(armclk, new_freq * 1000);
+	if (ret)
+		return ret;
+
+	if (cpufreq.asyncclk) {
+		ret = clk_set_rate(cpufreq.asyncclk, cpufreq.asyncrate);
+		if (ret)
+			return ret;
+	}
+
+	/* if moving to lower freq, lower the voltage after lowering freq */
+	if (pdata->set_voltage && new_freq < old_freq)
+		pdata->set_voltage(idx);
+
+	return 0;
+}
+
+static int davinci_cpu_init(struct cpufreq_policy *policy)
+{
+	int result = 0;
+	struct davinci_cpufreq_config *pdata = cpufreq.dev->platform_data;
+	struct cpufreq_frequency_table *freq_table = pdata->freq_table;
+
+	if (policy->cpu != 0)
+		return -EINVAL;
+
+	/* Finish platform specific initialization */
+	if (pdata->init) {
+		result = pdata->init();
+		if (result)
+			return result;
+	}
+
+	policy->clk = cpufreq.armclk;
+
+	/*
+	 * Time measurement across the target() function yields ~1500-1800us
+	 * time taken with no drivers on notification list.
+	 * Setting the latency to 2000 us to accommodate addition of drivers
+	 * to pre/post change notification list.
+	 */
+	cpufreq_generic_init(policy, freq_table, 2000 * 1000);
+	return 0;
+}
+
+static struct cpufreq_driver davinci_driver = {
+	.flags		= CPUFREQ_NEED_INITIAL_FREQ_CHECK,
+	.verify		= cpufreq_generic_frequency_table_verify,
+	.target_index	= davinci_target,
+	.get		= cpufreq_generic_get,
+	.init		= davinci_cpu_init,
+	.name		= "davinci",
+};
+
+static int __init davinci_cpufreq_probe(struct platform_device *pdev)
+{
+	struct davinci_cpufreq_config *pdata = pdev->dev.platform_data;
+	struct clk *asyncclk;
+
+	if (!pdata)
+		return -EINVAL;
+	if (!pdata->freq_table)
+		return -EINVAL;
+
+	cpufreq.dev = &pdev->dev;
+
+	cpufreq.armclk = clk_get(NULL, "arm");
+	if (IS_ERR(cpufreq.armclk)) {
+		dev_err(cpufreq.dev, "Unable to get ARM clock\n");
+		return PTR_ERR(cpufreq.armclk);
+	}
+
+	asyncclk = clk_get(cpufreq.dev, "async");
+	if (!IS_ERR(asyncclk)) {
+		cpufreq.asyncclk = asyncclk;
+		cpufreq.asyncrate = clk_get_rate(asyncclk);
+	}
+
+	return cpufreq_register_driver(&davinci_driver);
+}
+
+static void __exit davinci_cpufreq_remove(struct platform_device *pdev)
+{
+	cpufreq_unregister_driver(&davinci_driver);
+
+	clk_put(cpufreq.armclk);
+
+	if (cpufreq.asyncclk)
+		clk_put(cpufreq.asyncclk);
+}
+
+static struct platform_driver davinci_cpufreq_driver = {
+	.driver = {
+		.name	 = "cpufreq-davinci",
+	},
+	.remove = __exit_p(davinci_cpufreq_remove),
+};
+
+int __init davinci_cpufreq_init(void)
+{
+	return platform_driver_probe(&davinci_cpufreq_driver,
+							davinci_cpufreq_probe);
+}
+