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

3 files changed, 150 insertions, 0 deletions

diff --git a/drivers/soc/litex/Kconfig b/drivers/soc/litex/Kconfig
new file mode 100644
index 000000000..f3f869639
--- /dev/null
+++ b/drivers/soc/litex/Kconfig
@@ -0,0 +1,20 @@
+# SPDX-License_Identifier: GPL-2.0
+
+menu "Enable LiteX SoC Builder specific drivers"
+
+config LITEX
+	bool
+
+config LITEX_SOC_CONTROLLER
+	tristate "Enable LiteX SoC Controller driver"
+	depends on OF
+	depends on HAS_IOMEM
+	select LITEX
+	help
+	  This option enables the SoC Controller Driver which verifies
+	  LiteX CSR access and provides common litex_[read|write]*
+	  accessors.
+	  All drivers that use functions from litex.h must depend on
+	  LITEX.
+
+endmenu
diff --git a/drivers/soc/litex/Makefile b/drivers/soc/litex/Makefile
new file mode 100644
index 000000000..98ff7325b
--- /dev/null
+++ b/drivers/soc/litex/Makefile
@@ -0,0 +1,3 @@
+# SPDX-License_Identifier: GPL-2.0
+
+obj-$(CONFIG_LITEX_SOC_CONTROLLER)	+= litex_soc_ctrl.o
diff --git a/drivers/soc/litex/litex_soc_ctrl.c b/drivers/soc/litex/litex_soc_ctrl.c
new file mode 100644
index 000000000..104a5f9bf
--- /dev/null
+++ b/drivers/soc/litex/litex_soc_ctrl.c
@@ -0,0 +1,127 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * LiteX SoC Controller Driver
+ *
+ * Copyright (C) 2020 Antmicro <www.antmicro.com>
+ *
+ */
+
+#include <linux/litex.h>
+#include <linux/device.h>
+#include <linux/errno.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/printk.h>
+#include <linux/module.h>
+#include <linux/io.h>
+#include <linux/reboot.h>
+
+/* reset register located at the base address */
+#define RESET_REG_OFF           0x00
+#define RESET_REG_VALUE         0x00000001
+
+#define SCRATCH_REG_OFF         0x04
+#define SCRATCH_REG_VALUE       0x12345678
+#define SCRATCH_TEST_VALUE      0xdeadbeef
+
+/*
+ * Check LiteX CSR read/write access
+ *
+ * This function reads and writes a scratch register in order to verify if CSR
+ * access works.
+ *
+ * In case any problems are detected, the driver should panic.
+ *
+ * Access to the LiteX CSR is, by design, done in CPU native endianness.
+ * The driver should not dynamically configure access functions when
+ * the endianness mismatch is detected. Such situation indicates problems in
+ * the soft SoC design and should be solved at the LiteX generator level,
+ * not in the software.
+ */
+static int litex_check_csr_access(void __iomem *reg_addr)
+{
+	unsigned long reg;
+
+	reg = litex_read32(reg_addr + SCRATCH_REG_OFF);
+
+	if (reg != SCRATCH_REG_VALUE) {
+		panic("Scratch register read error - the system is probably broken! Expected: 0x%x but got: 0x%lx",
+			SCRATCH_REG_VALUE, reg);
+		return -EINVAL;
+	}
+
+	litex_write32(reg_addr + SCRATCH_REG_OFF, SCRATCH_TEST_VALUE);
+	reg = litex_read32(reg_addr + SCRATCH_REG_OFF);
+
+	if (reg != SCRATCH_TEST_VALUE) {
+		panic("Scratch register write error - the system is probably broken! Expected: 0x%x but got: 0x%lx",
+			SCRATCH_TEST_VALUE, reg);
+		return -EINVAL;
+	}
+
+	/* restore original value of the SCRATCH register */
+	litex_write32(reg_addr + SCRATCH_REG_OFF, SCRATCH_REG_VALUE);
+
+	pr_info("LiteX SoC Controller driver initialized");
+
+	return 0;
+}
+
+struct litex_soc_ctrl_device {
+	void __iomem *base;
+};
+
+static int litex_reset_handler(struct sys_off_data *data)
+{
+	struct litex_soc_ctrl_device *soc_ctrl_dev = data->cb_data;
+
+	litex_write32(soc_ctrl_dev->base + RESET_REG_OFF, RESET_REG_VALUE);
+	return NOTIFY_DONE;
+}
+
+static const struct of_device_id litex_soc_ctrl_of_match[] = {
+	{.compatible = "litex,soc-controller"},
+	{},
+};
+MODULE_DEVICE_TABLE(of, litex_soc_ctrl_of_match);
+
+static int litex_soc_ctrl_probe(struct platform_device *pdev)
+{
+	struct litex_soc_ctrl_device *soc_ctrl_dev;
+	int error;
+
+	soc_ctrl_dev = devm_kzalloc(&pdev->dev, sizeof(*soc_ctrl_dev), GFP_KERNEL);
+	if (!soc_ctrl_dev)
+		return -ENOMEM;
+
+	soc_ctrl_dev->base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(soc_ctrl_dev->base))
+		return PTR_ERR(soc_ctrl_dev->base);
+
+	error = litex_check_csr_access(soc_ctrl_dev->base);
+	if (error)
+		return error;
+
+	error = devm_register_restart_handler(&pdev->dev,
+					      litex_reset_handler,
+					      soc_ctrl_dev);
+	if (error) {
+		dev_warn(&pdev->dev, "cannot register restart handler: %d\n",
+			 error);
+	}
+
+	return 0;
+}
+
+static struct platform_driver litex_soc_ctrl_driver = {
+	.driver = {
+		.name = "litex-soc-controller",
+		.of_match_table = litex_soc_ctrl_of_match,
+	},
+	.probe = litex_soc_ctrl_probe,
+};
+
+module_platform_driver(litex_soc_ctrl_driver);
+MODULE_DESCRIPTION("LiteX SoC Controller driver");
+MODULE_AUTHOR("Antmicro <www.antmicro.com>");
+MODULE_LICENSE("GPL v2");