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, 185 insertions, 0 deletions

diff --git a/arch/arm/mach-versatile/versatile.c b/arch/arm/mach-versatile/versatile.c
new file mode 100644
index 000000000..581c97dc4
--- /dev/null
+++ b/arch/arm/mach-versatile/versatile.c
@@ -0,0 +1,185 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Versatile board support using the device tree
+ *
+ *  Copyright (C) 2010 Secret Lab Technologies Ltd.
+ *  Copyright (C) 2009 Jeremy Kerr <jeremy.kerr@canonical.com>
+ *  Copyright (C) 2004 ARM Limited
+ *  Copyright (C) 2000 Deep Blue Solutions Ltd
+ */
+
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/of_irq.h>
+#include <linux/of_platform.h>
+#include <linux/slab.h>
+#include <linux/amba/bus.h>
+#include <linux/amba/mmci.h>
+#include <asm/mach-types.h>
+#include <asm/mach/arch.h>
+#include <asm/mach/map.h>
+
+/* macro to get at MMIO space when running virtually */
+#define IO_ADDRESS(x)		(((x) & 0x0fffffff) + (((x) >> 4) & 0x0f000000) + 0xf0000000)
+#define __io_address(n)		((void __iomem __force *)IO_ADDRESS(n))
+
+/*
+ * ------------------------------------------------------------------------
+ *  Versatile Registers
+ * ------------------------------------------------------------------------
+ */
+#define VERSATILE_SYS_PCICTL_OFFSET           0x44
+#define VERSATILE_SYS_MCI_OFFSET              0x48
+
+/*
+ * VERSATILE peripheral addresses
+ */
+#define VERSATILE_MMCI0_BASE           0x10005000	/* MMC interface */
+#define VERSATILE_MMCI1_BASE           0x1000B000	/* MMC Interface */
+#define VERSATILE_SCTL_BASE            0x101E0000	/* System controller */
+
+/*
+ * System controller bit assignment
+ */
+#define VERSATILE_REFCLK	0
+#define VERSATILE_TIMCLK	1
+
+#define VERSATILE_TIMER1_EnSel	15
+#define VERSATILE_TIMER2_EnSel	17
+#define VERSATILE_TIMER3_EnSel	19
+#define VERSATILE_TIMER4_EnSel	21
+
+static void __iomem *versatile_sys_base;
+
+static unsigned int mmc_status(struct device *dev)
+{
+	struct amba_device *adev = container_of(dev, struct amba_device, dev);
+	u32 mask;
+
+	if (adev->res.start == VERSATILE_MMCI0_BASE)
+		mask = 1;
+	else
+		mask = 2;
+
+	return readl(versatile_sys_base + VERSATILE_SYS_MCI_OFFSET) & mask;
+}
+
+static struct mmci_platform_data mmc0_plat_data = {
+	.ocr_mask	= MMC_VDD_32_33|MMC_VDD_33_34,
+	.status		= mmc_status,
+};
+
+static struct mmci_platform_data mmc1_plat_data = {
+	.ocr_mask	= MMC_VDD_32_33|MMC_VDD_33_34,
+	.status		= mmc_status,
+};
+
+/*
+ * Lookup table for attaching a specific name and platform_data pointer to
+ * devices as they get created by of_platform_populate().  Ideally this table
+ * would not exist, but the current clock implementation depends on some devices
+ * having a specific name.
+ */
+struct of_dev_auxdata versatile_auxdata_lookup[] __initdata = {
+	OF_DEV_AUXDATA("arm,primecell", VERSATILE_MMCI0_BASE, "fpga:05", &mmc0_plat_data),
+	OF_DEV_AUXDATA("arm,primecell", VERSATILE_MMCI1_BASE, "fpga:0b", &mmc1_plat_data),
+	{}
+};
+
+static struct map_desc versatile_io_desc[] __initdata __maybe_unused = {
+	{
+		.virtual	=  IO_ADDRESS(VERSATILE_SCTL_BASE),
+		.pfn		= __phys_to_pfn(VERSATILE_SCTL_BASE),
+		.length		= SZ_4K * 9,
+		.type		= MT_DEVICE
+	}
+};
+
+static void __init versatile_map_io(void)
+{
+	debug_ll_io_init();
+	iotable_init(versatile_io_desc, ARRAY_SIZE(versatile_io_desc));
+}
+
+static void __init versatile_init_early(void)
+{
+	u32 val;
+
+	/*
+	 * set clock frequency:
+	 *	VERSATILE_REFCLK is 32KHz
+	 *	VERSATILE_TIMCLK is 1MHz
+	 */
+	val = readl(__io_address(VERSATILE_SCTL_BASE));
+	writel((VERSATILE_TIMCLK << VERSATILE_TIMER1_EnSel) |
+	       (VERSATILE_TIMCLK << VERSATILE_TIMER2_EnSel) |
+	       (VERSATILE_TIMCLK << VERSATILE_TIMER3_EnSel) |
+	       (VERSATILE_TIMCLK << VERSATILE_TIMER4_EnSel) | val,
+	       __io_address(VERSATILE_SCTL_BASE));
+}
+
+static void __init versatile_dt_pci_init(void)
+{
+	u32 val;
+	struct device_node *np;
+	struct property *newprop;
+
+	np = of_find_compatible_node(NULL, NULL, "arm,versatile-pci");
+	if (!np)
+		return;
+
+	/* Check if PCI backplane is detected */
+	val = readl(versatile_sys_base + VERSATILE_SYS_PCICTL_OFFSET);
+	if (val & 1) {
+		/*
+		 * Enable PCI accesses. Note that the documentation is
+		 * inconsistent whether or not this is needed, but the old
+		 * driver had it so we will keep it.
+		 */
+		writel(1, versatile_sys_base + VERSATILE_SYS_PCICTL_OFFSET);
+		goto out_put_node;
+	}
+
+	newprop = kzalloc_obj(*newprop);
+	if (!newprop)
+		goto out_put_node;
+
+	newprop->name = kstrdup("status", GFP_KERNEL);
+	newprop->value = kstrdup("disabled", GFP_KERNEL);
+	newprop->length = sizeof("disabled");
+	of_update_property(np, newprop);
+
+	pr_info("Not plugged into PCI backplane!\n");
+
+out_put_node:
+	of_node_put(np);
+}
+
+static void __init versatile_dt_init(void)
+{
+	struct device_node *np;
+
+	np = of_find_compatible_node(NULL, NULL, "arm,core-module-versatile");
+	if (np)
+		versatile_sys_base = of_iomap(np, 0);
+	WARN_ON(!versatile_sys_base);
+
+	versatile_dt_pci_init();
+
+	of_platform_default_populate(NULL, versatile_auxdata_lookup, NULL);
+}
+
+static const char *const versatile_dt_match[] __initconst = {
+	"arm,versatile-ab",
+	"arm,versatile-pb",
+	NULL,
+};
+
+DT_MACHINE_START(VERSATILE_PB, "ARM-Versatile (Device Tree Support)")
+	.map_io		= versatile_map_io,
+	.init_early	= versatile_init_early,
+	.init_machine	= versatile_dt_init,
+	.dt_compat	= versatile_dt_match,
+MACHINE_END