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, 187 insertions, 0 deletions

diff --git a/drivers/xen/arm-device.c b/drivers/xen/arm-device.c
new file mode 100644
index 000000000..7bc0fc400
--- /dev/null
+++ b/drivers/xen/arm-device.c
@@ -0,0 +1,187 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2015, Linaro Limited, Shannon Zhao
+ */
+
+#include <linux/platform_device.h>
+#include <linux/acpi.h>
+#include <xen/xen.h>
+#include <xen/page.h>
+#include <xen/interface/memory.h>
+#include <asm/xen/hypervisor.h>
+#include <asm/xen/hypercall.h>
+
+static int xen_unmap_device_mmio(const struct resource *resources,
+				 unsigned int count)
+{
+	unsigned int i, j, nr;
+	int rc = 0;
+	const struct resource *r;
+	struct xen_remove_from_physmap xrp;
+
+	for (i = 0; i < count; i++) {
+		r = &resources[i];
+		nr = DIV_ROUND_UP(resource_size(r), XEN_PAGE_SIZE);
+		if ((resource_type(r) != IORESOURCE_MEM) || (nr == 0))
+			continue;
+
+		for (j = 0; j < nr; j++) {
+			xrp.domid = DOMID_SELF;
+			xrp.gpfn = XEN_PFN_DOWN(r->start) + j;
+			rc = HYPERVISOR_memory_op(XENMEM_remove_from_physmap,
+						  &xrp);
+			if (rc)
+				return rc;
+		}
+	}
+
+	return rc;
+}
+
+static int xen_map_device_mmio(const struct resource *resources,
+			       unsigned int count)
+{
+	unsigned int i, j, nr;
+	int rc = 0;
+	const struct resource *r;
+	xen_pfn_t *gpfns;
+	xen_ulong_t *idxs;
+	int *errs;
+
+	for (i = 0; i < count; i++) {
+		struct xen_add_to_physmap_range xatp = {
+			.domid = DOMID_SELF,
+			.space = XENMAPSPACE_dev_mmio
+		};
+
+		r = &resources[i];
+		nr = DIV_ROUND_UP(resource_size(r), XEN_PAGE_SIZE);
+		if ((resource_type(r) != IORESOURCE_MEM) || (nr == 0))
+			continue;
+
+		gpfns = kzalloc_objs(xen_pfn_t, nr);
+		idxs = kzalloc_objs(xen_ulong_t, nr);
+		errs = kzalloc_objs(int, nr);
+		if (!gpfns || !idxs || !errs) {
+			kfree(gpfns);
+			kfree(idxs);
+			kfree(errs);
+			rc = -ENOMEM;
+			goto unmap;
+		}
+
+		for (j = 0; j < nr; j++) {
+			/*
+			 * The regions are always mapped 1:1 to DOM0 and this is
+			 * fine because the memory map for DOM0 is the same as
+			 * the host (except for the RAM).
+			 */
+			gpfns[j] = XEN_PFN_DOWN(r->start) + j;
+			idxs[j] = XEN_PFN_DOWN(r->start) + j;
+		}
+
+		xatp.size = nr;
+
+		set_xen_guest_handle(xatp.gpfns, gpfns);
+		set_xen_guest_handle(xatp.idxs, idxs);
+		set_xen_guest_handle(xatp.errs, errs);
+
+		rc = HYPERVISOR_memory_op(XENMEM_add_to_physmap_range, &xatp);
+		kfree(gpfns);
+		kfree(idxs);
+		kfree(errs);
+		if (rc)
+			goto unmap;
+	}
+
+	return rc;
+
+unmap:
+	xen_unmap_device_mmio(resources, i);
+	return rc;
+}
+
+static int xen_platform_notifier(struct notifier_block *nb,
+				 unsigned long action, void *data)
+{
+	struct platform_device *pdev = to_platform_device(data);
+	int r = 0;
+
+	if (pdev->num_resources == 0 || pdev->resource == NULL)
+		return NOTIFY_OK;
+
+	switch (action) {
+	case BUS_NOTIFY_ADD_DEVICE:
+		r = xen_map_device_mmio(pdev->resource, pdev->num_resources);
+		break;
+	case BUS_NOTIFY_DEL_DEVICE:
+		r = xen_unmap_device_mmio(pdev->resource, pdev->num_resources);
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+	if (r)
+		dev_err(&pdev->dev, "Platform: Failed to %s device %s MMIO!\n",
+			action == BUS_NOTIFY_ADD_DEVICE ? "map" :
+			(action == BUS_NOTIFY_DEL_DEVICE ? "unmap" : "?"),
+			pdev->name);
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block platform_device_nb = {
+	.notifier_call = xen_platform_notifier,
+};
+
+static int __init register_xen_platform_notifier(void)
+{
+	if (!xen_initial_domain() || acpi_disabled)
+		return 0;
+
+	return bus_register_notifier(&platform_bus_type, &platform_device_nb);
+}
+
+arch_initcall(register_xen_platform_notifier);
+
+#ifdef CONFIG_ARM_AMBA
+#include <linux/amba/bus.h>
+
+static int xen_amba_notifier(struct notifier_block *nb,
+			     unsigned long action, void *data)
+{
+	struct amba_device *adev = to_amba_device(data);
+	int r = 0;
+
+	switch (action) {
+	case BUS_NOTIFY_ADD_DEVICE:
+		r = xen_map_device_mmio(&adev->res, 1);
+		break;
+	case BUS_NOTIFY_DEL_DEVICE:
+		r = xen_unmap_device_mmio(&adev->res, 1);
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+	if (r)
+		dev_err(&adev->dev, "AMBA: Failed to %s device %s MMIO!\n",
+			action == BUS_NOTIFY_ADD_DEVICE ? "map" :
+			(action == BUS_NOTIFY_DEL_DEVICE ? "unmap" : "?"),
+			adev->dev.init_name);
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block amba_device_nb = {
+	.notifier_call = xen_amba_notifier,
+};
+
+static int __init register_xen_amba_notifier(void)
+{
+	if (!xen_initial_domain() || acpi_disabled)
+		return 0;
+
+	return bus_register_notifier(&amba_bustype, &amba_device_nb);
+}
+
+arch_initcall(register_xen_amba_notifier);
+#endif