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, 140 insertions, 0 deletions

diff --git a/drivers/bus/stm32_etzpc.c b/drivers/bus/stm32_etzpc.c
new file mode 100644
index 000000000..4918a14e5
--- /dev/null
+++ b/drivers/bus/stm32_etzpc.c
@@ -0,0 +1,140 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2023, STMicroelectronics - All Rights Reserved
+ */
+
+#include <linux/bitfield.h>
+#include <linux/bits.h>
+#include <linux/bus/stm32_firewall.h>
+#include <linux/device.h>
+#include <linux/err.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/platform_device.h>
+#include <linux/types.h>
+
+/*
+ * ETZPC registers
+ */
+#define ETZPC_DECPROT			0x10
+#define ETZPC_HWCFGR			0x3F0
+
+/*
+ * HWCFGR register
+ */
+#define ETZPC_HWCFGR_NUM_TZMA		GENMASK(7, 0)
+#define ETZPC_HWCFGR_NUM_PER_SEC	GENMASK(15, 8)
+#define ETZPC_HWCFGR_NUM_AHB_SEC	GENMASK(23, 16)
+#define ETZPC_HWCFGR_CHUNKS1N4		GENMASK(31, 24)
+
+/*
+ * ETZPC miscellaneous
+ */
+#define ETZPC_PROT_MASK			GENMASK(1, 0)
+#define ETZPC_PROT_A7NS			0x3
+#define ETZPC_DECPROT_SHIFT		1
+
+#define IDS_PER_DECPROT_REGS		16
+
+static int stm32_etzpc_grant_access(struct stm32_firewall_controller *ctrl, u32 firewall_id)
+{
+	u32 offset, reg_offset, sec_val;
+
+	if (firewall_id >= ctrl->max_entries) {
+		dev_err(ctrl->dev, "Invalid sys bus ID %u", firewall_id);
+		return -EINVAL;
+	}
+
+	/* Check access configuration, 16 peripherals per register */
+	reg_offset = ETZPC_DECPROT + 0x4 * (firewall_id / IDS_PER_DECPROT_REGS);
+	offset = (firewall_id % IDS_PER_DECPROT_REGS) << ETZPC_DECPROT_SHIFT;
+
+	/* Verify peripheral is non-secure and attributed to cortex A7 */
+	sec_val = (readl(ctrl->mmio + reg_offset) >> offset) & ETZPC_PROT_MASK;
+	if (sec_val != ETZPC_PROT_A7NS) {
+		dev_dbg(ctrl->dev, "Invalid bus configuration: reg_offset %#x, value %d\n",
+			reg_offset, sec_val);
+		return -EACCES;
+	}
+
+	return 0;
+}
+
+static void stm32_etzpc_release_access(struct stm32_firewall_controller *ctrl __maybe_unused,
+				       u32 firewall_id __maybe_unused)
+{
+}
+
+static int stm32_etzpc_probe(struct platform_device *pdev)
+{
+	struct stm32_firewall_controller *etzpc_controller;
+	struct device_node *np = pdev->dev.of_node;
+	u32 nb_per, nb_master;
+	struct resource *res;
+	void __iomem *mmio;
+	int rc;
+
+	etzpc_controller = devm_kzalloc(&pdev->dev, sizeof(*etzpc_controller), GFP_KERNEL);
+	if (!etzpc_controller)
+		return -ENOMEM;
+
+	mmio = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
+	if (IS_ERR(mmio))
+		return PTR_ERR(mmio);
+
+	etzpc_controller->dev = &pdev->dev;
+	etzpc_controller->mmio = mmio;
+	etzpc_controller->name = dev_driver_string(etzpc_controller->dev);
+	etzpc_controller->type = STM32_PERIPHERAL_FIREWALL | STM32_MEMORY_FIREWALL;
+	etzpc_controller->grant_access = stm32_etzpc_grant_access;
+	etzpc_controller->release_access = stm32_etzpc_release_access;
+
+	/* Get number of etzpc entries*/
+	nb_per = FIELD_GET(ETZPC_HWCFGR_NUM_PER_SEC,
+			   readl(etzpc_controller->mmio + ETZPC_HWCFGR));
+	nb_master = FIELD_GET(ETZPC_HWCFGR_NUM_AHB_SEC,
+			      readl(etzpc_controller->mmio + ETZPC_HWCFGR));
+	etzpc_controller->max_entries = nb_per + nb_master;
+
+	platform_set_drvdata(pdev, etzpc_controller);
+
+	rc = stm32_firewall_controller_register(etzpc_controller);
+	if (rc) {
+		dev_err(etzpc_controller->dev, "Couldn't register as a firewall controller: %d",
+			rc);
+		return rc;
+	}
+
+	rc = stm32_firewall_populate_bus(etzpc_controller);
+	if (rc) {
+		dev_err(etzpc_controller->dev, "Couldn't populate ETZPC bus: %d",
+			rc);
+		return rc;
+	}
+
+	/* Populate all allowed nodes */
+	return of_platform_populate(np, NULL, NULL, &pdev->dev);
+}
+
+static const struct of_device_id stm32_etzpc_of_match[] = {
+	{ .compatible = "st,stm32-etzpc" },
+	{}
+};
+MODULE_DEVICE_TABLE(of, stm32_etzpc_of_match);
+
+static struct platform_driver stm32_etzpc_driver = {
+	.probe  = stm32_etzpc_probe,
+	.driver = {
+		.name = "stm32-etzpc",
+		.of_match_table = stm32_etzpc_of_match,
+	},
+};
+module_platform_driver(stm32_etzpc_driver);
+
+MODULE_AUTHOR("Gatien Chevallier <gatien.chevallier@foss.st.com>");
+MODULE_DESCRIPTION("STMicroelectronics ETZPC driver");
+MODULE_LICENSE("GPL");