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, 230 insertions, 0 deletions

diff --git a/drivers/firmware/imx/sm-misc.c b/drivers/firmware/imx/sm-misc.c
new file mode 100644
index 000000000..fb8d7bdb5
--- /dev/null
+++ b/drivers/firmware/imx/sm-misc.c
@@ -0,0 +1,230 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2024 NXP
+ */
+
+#include <linux/debugfs.h>
+#include <linux/device/devres.h>
+#include <linux/firmware/imx/sm.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/scmi_protocol.h>
+#include <linux/scmi_imx_protocol.h>
+#include <linux/seq_file.h>
+#include <linux/sizes.h>
+
+static const struct scmi_imx_misc_proto_ops *imx_misc_ctrl_ops;
+static struct scmi_protocol_handle *ph;
+static struct notifier_block scmi_imx_misc_ctrl_nb;
+
+static const char * const rst_imx95[] = {
+	"cm33_lockup", "cm33_swreq", "cm7_lockup", "cm7_swreq", "fccu",
+	"jtag_sw", "ele", "tempsense", "wdog1", "wdog2", "wdog3", "wdog4",
+	"wdog5", "jtag", "cm33_exc", "bbm", "sw", "sm_err", "fusa_sreco",
+	"pmic", "unused", "unused", "unused", "unused", "unused", "unused",
+	"unused", "unused", "unused", "unused", "unused", "por",
+};
+
+static const char * const rst_imx94[] = {
+	"cm33_lockup", "cm33_swreq", "cm70_lockup", "cm70_swreq", "fccu",
+	"jtag_sw", "ele", "tempsense", "wdog1", "wdog2", "wdog3", "wdog4",
+	"wdog5", "jtag", "wdog6", "wdog7", "wdog8", "wo_netc", "cm33s_lockup",
+	"cm33s_swreq", "cm71_lockup", "cm71_swreq", "cm33_exc", "bbm", "sw",
+	"sm_err", "fusa_sreco", "pmic", "unused", "unused", "unused", "por",
+};
+
+static const struct of_device_id allowlist[] = {
+	{ .compatible = "fsl,imx952", .data = rst_imx95 },
+	{ .compatible = "fsl,imx95", .data = rst_imx95 },
+	{ .compatible = "fsl,imx94", .data = rst_imx94 },
+	{ /* Sentinel */ }
+};
+
+int scmi_imx_misc_ctrl_set(u32 id, u32 val)
+{
+	if (!ph)
+		return -EPROBE_DEFER;
+
+	return imx_misc_ctrl_ops->misc_ctrl_set(ph, id, 1, &val);
+};
+EXPORT_SYMBOL(scmi_imx_misc_ctrl_set);
+
+int scmi_imx_misc_ctrl_get(u32 id, u32 *num, u32 *val)
+{
+	if (!ph)
+		return -EPROBE_DEFER;
+
+	return imx_misc_ctrl_ops->misc_ctrl_get(ph, id, num, val);
+}
+EXPORT_SYMBOL(scmi_imx_misc_ctrl_get);
+
+static int scmi_imx_misc_ctrl_notifier(struct notifier_block *nb,
+				       unsigned long event, void *data)
+{
+	/*
+	 * notifier_chain_register requires a valid notifier_block and
+	 * valid notifier_call. SCMI_EVENT_IMX_MISC_CONTROL is needed
+	 * to let SCMI firmware enable control events, but the hook here
+	 * is just a dummy function to avoid kernel panic as of now.
+	 */
+	return 0;
+}
+
+static int syslog_show(struct seq_file *file, void *priv)
+{
+	/* 4KB is large enough for syslog */
+	void *syslog __free(kfree) = kmalloc(SZ_4K, GFP_KERNEL);
+	/* syslog API use num words, not num bytes */
+	u16 size = SZ_4K / 4;
+	int ret;
+
+	if (!syslog)
+		return -ENOMEM;
+
+	if (!ph)
+		return -ENODEV;
+
+	ret = imx_misc_ctrl_ops->misc_syslog(ph, &size, syslog);
+	if (ret)
+		return ret;
+
+	seq_hex_dump(file, " ", DUMP_PREFIX_NONE, 16, sizeof(u32), syslog, size * 4, false);
+	seq_putc(file, '\n');
+
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(syslog);
+
+static void scmi_imx_misc_put(void *p)
+{
+	debugfs_remove((struct dentry *)p);
+}
+
+static int scmi_imx_misc_get_reason(struct scmi_device *sdev)
+{
+	struct scmi_imx_misc_reset_reason boot, shutdown;
+	const char **rst;
+	bool system = true;
+	int ret;
+
+	if (!of_machine_device_match(allowlist))
+		return 0;
+
+	rst = (const char **)of_machine_get_match_data(allowlist);
+
+	ret = imx_misc_ctrl_ops->misc_reset_reason(ph, system, &boot, &shutdown, NULL);
+	if (!ret) {
+		if (boot.valid)
+			dev_info(&sdev->dev, "%s Boot reason: %s, origin: %d, errid: %d\n",
+				 system ? "SYS" : "LM", rst[boot.reason],
+				 boot.orig_valid ? boot.origin : -1,
+				 boot.err_valid ? boot.errid : -1);
+		if (shutdown.valid)
+			dev_info(&sdev->dev, "%s shutdown reason: %s, origin: %d, errid: %d\n",
+				 system ? "SYS" : "LM", rst[shutdown.reason],
+				 shutdown.orig_valid ? shutdown.origin : -1,
+				 shutdown.err_valid ? shutdown.errid : -1);
+	} else {
+		dev_err(&sdev->dev, "Failed to get system reset reason: %d\n", ret);
+	}
+
+	system = false;
+	ret = imx_misc_ctrl_ops->misc_reset_reason(ph, system, &boot, &shutdown, NULL);
+	if (!ret) {
+		if (boot.valid)
+			dev_info(&sdev->dev, "%s Boot reason: %s, origin: %d, errid: %d\n",
+				 system ? "SYS" : "LM", rst[boot.reason],
+				 boot.orig_valid ? boot.origin : -1,
+				 boot.err_valid ? boot.errid : -1);
+		if (shutdown.valid)
+			dev_info(&sdev->dev, "%s shutdown reason: %s, origin: %d, errid: %d\n",
+				 system ? "SYS" : "LM", rst[shutdown.reason],
+				 shutdown.orig_valid ? shutdown.origin : -1,
+				 shutdown.err_valid ? shutdown.errid : -1);
+	} else {
+		dev_err(&sdev->dev, "Failed to get lm reset reason: %d\n", ret);
+	}
+
+	return 0;
+}
+
+static int scmi_imx_misc_ctrl_probe(struct scmi_device *sdev)
+{
+	const struct scmi_handle *handle = sdev->handle;
+	struct device_node *np = sdev->dev.of_node;
+	struct dentry *scmi_imx_dentry;
+	u32 src_id, flags;
+	int ret, i, num;
+
+	if (!handle)
+		return -ENODEV;
+
+	if (imx_misc_ctrl_ops) {
+		dev_err(&sdev->dev, "misc ctrl already initialized\n");
+		return -EEXIST;
+	}
+
+	imx_misc_ctrl_ops = handle->devm_protocol_get(sdev, SCMI_PROTOCOL_IMX_MISC, &ph);
+	if (IS_ERR(imx_misc_ctrl_ops))
+		return PTR_ERR(imx_misc_ctrl_ops);
+
+	num = of_property_count_u32_elems(np, "nxp,ctrl-ids");
+	if (num % 2) {
+		dev_err(&sdev->dev, "Invalid wakeup-sources\n");
+		return -EINVAL;
+	}
+
+	scmi_imx_misc_ctrl_nb.notifier_call = &scmi_imx_misc_ctrl_notifier;
+	for (i = 0; i < num; i += 2) {
+		ret = of_property_read_u32_index(np, "nxp,ctrl-ids", i, &src_id);
+		if (ret) {
+			dev_err(&sdev->dev, "Failed to read ctrl-id: %i\n", i);
+			continue;
+		}
+
+		ret = of_property_read_u32_index(np, "nxp,ctrl-ids", i + 1, &flags);
+		if (ret) {
+			dev_err(&sdev->dev, "Failed to read ctrl-id value: %d\n", i + 1);
+			continue;
+		}
+
+		ret = handle->notify_ops->devm_event_notifier_register(sdev, SCMI_PROTOCOL_IMX_MISC,
+								       SCMI_EVENT_IMX_MISC_CONTROL,
+								       &src_id,
+								       &scmi_imx_misc_ctrl_nb);
+		if (ret) {
+			dev_err(&sdev->dev, "Failed to register scmi misc event: %d\n", src_id);
+		} else {
+			ret = imx_misc_ctrl_ops->misc_ctrl_req_notify(ph, src_id,
+								      SCMI_EVENT_IMX_MISC_CONTROL,
+								      flags);
+			if (ret)
+				dev_err(&sdev->dev, "Failed to req notify: %d\n", src_id);
+		}
+	}
+
+	scmi_imx_dentry = debugfs_create_dir("scmi_imx", NULL);
+	debugfs_create_file("syslog", 0444, scmi_imx_dentry, &sdev->dev, &syslog_fops);
+
+	scmi_imx_misc_get_reason(sdev);
+
+	return devm_add_action_or_reset(&sdev->dev, scmi_imx_misc_put, scmi_imx_dentry);
+}
+
+static const struct scmi_device_id scmi_id_table[] = {
+	{ SCMI_PROTOCOL_IMX_MISC, "imx-misc-ctrl" },
+	{ },
+};
+MODULE_DEVICE_TABLE(scmi, scmi_id_table);
+
+static struct scmi_driver scmi_imx_misc_ctrl_driver = {
+	.name = "scmi-imx-misc-ctrl",
+	.probe = scmi_imx_misc_ctrl_probe,
+	.id_table = scmi_id_table,
+};
+module_scmi_driver(scmi_imx_misc_ctrl_driver);
+
+MODULE_AUTHOR("Peng Fan <peng.fan@nxp.com>");
+MODULE_DESCRIPTION("IMX SM MISC driver");
+MODULE_LICENSE("GPL");