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, 128 insertions, 0 deletions

diff --git a/drivers/mcb/mcb-internal.h b/drivers/mcb/mcb-internal.h
new file mode 100644
index 000000000..3602cb3b2
--- /dev/null
+++ b/drivers/mcb/mcb-internal.h
@@ -0,0 +1,128 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __MCB_INTERNAL
+#define __MCB_INTERNAL
+
+#include <linux/types.h>
+
+#define PCI_VENDOR_ID_MEN		0x1a88
+#define PCI_DEVICE_ID_MEN_CHAMELEON	0x4d45
+#define CHAMELEONV2_MAGIC		0xabce
+#define CHAM_HEADER_SIZE		0x200
+
+enum chameleon_descriptor_type {
+	CHAMELEON_DTYPE_GENERAL = 0x0,
+	CHAMELEON_DTYPE_BRIDGE = 0x1,
+	CHAMELEON_DTYPE_CPU = 0x2,
+	CHAMELEON_DTYPE_BAR = 0x3,
+	CHAMELEON_DTYPE_END = 0xf,
+};
+
+enum chameleon_bus_type {
+	CHAMELEON_BUS_WISHBONE,
+	CHAMELEON_BUS_AVALON,
+	CHAMELEON_BUS_LPC,
+	CHAMELEON_BUS_ISA,
+};
+
+/**
+ * struct chameleon_fpga_header
+ *
+ * @revision:	Revison of Chameleon table in FPGA
+ * @model:	Chameleon table model ASCII char
+ * @minor:	Revision minor
+ * @bus_type:	Bus type (usually %CHAMELEON_BUS_WISHBONE)
+ * @magic:	Chameleon header magic number (0xabce for version 2)
+ * @reserved:	Reserved
+ * @filename:	Filename of FPGA bitstream
+ */
+struct chameleon_fpga_header {
+	u8 revision;
+	char model;
+	u8 minor;
+	u8 bus_type;
+	u16 magic;
+	u16 reserved;
+	/* This one has no '\0' at the end!!! */
+	char filename[CHAMELEON_FILENAME_LEN];
+} __packed;
+#define HEADER_MAGIC_OFFSET 0x4
+
+/**
+ * struct chameleon_gdd - Chameleon General Device Descriptor
+ *
+ * @irq:	the position in the FPGA's IRQ controller vector
+ * @rev:	the revision of the variant's implementation
+ * @var:	the variant of the IP core
+ * @dev:	the device  the IP core is
+ * @dtype:	device descriptor type
+ * @bar:	BAR offset that must be added to module offset
+ * @inst:	the instance number of the device, 0 is first instance
+ * @group:	the group the device belongs to (0 = no group)
+ * @reserved:	reserved
+ * @offset:	beginning of the address window of desired module
+ * @size:	size of the module's address window
+ */
+struct chameleon_gdd {
+	__le32 reg1;
+	__le32 reg2;
+	__le32 offset;
+	__le32 size;
+
+} __packed;
+
+/* GDD Register 1 fields */
+#define GDD_IRQ(x) ((x) & 0x1f)
+#define GDD_REV(x) (((x) >> 5) & 0x3f)
+#define GDD_VAR(x) (((x) >> 11) & 0x3f)
+#define GDD_DEV(x) (((x) >> 18) & 0x3ff)
+#define GDD_DTY(x) (((x) >> 28) & 0xf)
+
+/* GDD Register 2 fields */
+#define GDD_BAR(x) ((x) & 0x7)
+#define GDD_INS(x) (((x) >> 3) & 0x3f)
+#define GDD_GRP(x) (((x) >> 9) & 0x3f)
+
+/**
+ * struct chameleon_bdd - Chameleon Bridge Device Descriptor
+ *
+ * @irq:	the position in the FPGA's IRQ controller vector
+ * @rev:	the revision of the variant's implementation
+ * @var:	the variant of the IP core
+ * @dev:	the device  the IP core is
+ * @dtype:	device descriptor type
+ * @bar:	BAR offset that must be added to module offset
+ * @inst:	the instance number of the device, 0 is first instance
+ * @dbar:	destination bar from the bus _behind_ the bridge
+ * @chamoff:	offset within the BAR of the source bus
+ * @offset:
+ * @size:
+ */
+struct chameleon_bdd {
+	unsigned int irq:6;
+	unsigned int rev:6;
+	unsigned int var:6;
+	unsigned int dev:10;
+	unsigned int dtype:4;
+	unsigned int bar:3;
+	unsigned int inst:6;
+	unsigned int dbar:3;
+	unsigned int group:6;
+	unsigned int reserved:14;
+	u32 chamoff;
+	u32 offset;
+	u32 size;
+} __packed;
+
+struct chameleon_bar {
+	u32 addr;
+	u32 size;
+};
+
+#define BAR_CNT(x) ((x) & 0x07)
+#define CHAMELEON_BAR_MAX	6
+#define BAR_DESC_SIZE(x)	((x) * sizeof(struct chameleon_bar) + sizeof(__le32))
+
+int chameleon_parse_cells(struct mcb_bus *bus, phys_addr_t mapbase,
+			  void __iomem *base);
+
+#endif