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, 136 insertions, 0 deletions

diff --git a/drivers/net/ipa/reg.h b/drivers/net/ipa/reg.h
new file mode 100644
index 000000000..53c16e594
--- /dev/null
+++ b/drivers/net/ipa/reg.h
@@ -0,0 +1,136 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+/* Copyright (C) 2022-2024 Linaro Ltd. */
+
+#ifndef _REG_H_
+#define _REG_H_
+
+#include <linux/array_size.h>
+#include <linux/bits.h>
+#include <linux/bug.h>
+#include <linux/log2.h>
+#include <linux/types.h>
+
+/**
+ * struct reg - A register descriptor
+ * @offset:	Register offset relative to base of register memory
+ * @stride:	Distance between two instances, if parameterized
+ * @fcount:	Number of entries in the @fmask array
+ * @fmask:	Array of mask values defining position and width of fields
+ * @name:	Upper-case name of the register
+ */
+struct reg {
+	u32 offset;
+	u32 stride;
+	u32 fcount;
+	const u32 *fmask;			/* BIT(nr) or GENMASK(h, l) */
+	const char *name;
+};
+
+/* Helper macro for defining "simple" (non-parameterized) registers */
+#define REG(__NAME, __reg_id, __offset)					\
+	REG_STRIDE(__NAME, __reg_id, __offset, 0)
+
+/* Helper macro for defining parameterized registers, specifying stride */
+#define REG_STRIDE(__NAME, __reg_id, __offset, __stride)		\
+	static const struct reg reg_ ## __reg_id = {			\
+		.name	= #__NAME,					\
+		.offset	= __offset,					\
+		.stride	= __stride,					\
+	}
+
+#define REG_FIELDS(__NAME, __name, __offset)				\
+	REG_STRIDE_FIELDS(__NAME, __name, __offset, 0)
+
+#define REG_STRIDE_FIELDS(__NAME, __name, __offset, __stride)		\
+	static const struct reg reg_ ## __name = {			\
+		.name   = #__NAME,					\
+		.offset = __offset,					\
+		.stride = __stride,					\
+		.fcount = ARRAY_SIZE(reg_ ## __name ## _fmask),		\
+		.fmask  = reg_ ## __name ## _fmask,			\
+	}
+
+/**
+ * struct regs - Description of registers supported by hardware
+ * @reg_count:	Number of registers in the @reg[] array
+ * @reg:	Array of register descriptors
+ */
+struct regs {
+	u32 reg_count;
+	const struct reg **reg;
+};
+
+static inline const struct reg *reg(const struct regs *regs, u32 reg_id)
+{
+	if (WARN(reg_id >= regs->reg_count,
+		 "reg out of range (%u > %u)\n", reg_id, regs->reg_count - 1))
+		return NULL;
+
+	return regs->reg[reg_id];
+}
+
+/* Return the field mask for a field in a register, or 0 on error */
+static inline u32 reg_fmask(const struct reg *reg, u32 field_id)
+{
+	if (!reg || WARN_ON(field_id >= reg->fcount))
+		return 0;
+
+	return reg->fmask[field_id];
+}
+
+/* Return the mask for a single-bit field in a register, or 0 on error  */
+static inline u32 reg_bit(const struct reg *reg, u32 field_id)
+{
+	u32 fmask = reg_fmask(reg, field_id);
+
+	if (WARN_ON(!is_power_of_2(fmask)))
+		return 0;
+
+	return fmask;
+}
+
+/* Return the maximum value representable by the given field; always 2^n - 1 */
+static inline u32 reg_field_max(const struct reg *reg, u32 field_id)
+{
+	u32 fmask = reg_fmask(reg, field_id);
+
+	return fmask ? fmask >> __ffs(fmask) : 0;
+}
+
+/* Encode a value into the given field of a register */
+static inline u32 reg_encode(const struct reg *reg, u32 field_id, u32 val)
+{
+	u32 fmask = reg_fmask(reg, field_id);
+
+	if (!fmask)
+		return 0;
+
+	val <<= __ffs(fmask);
+	if (WARN_ON(val & ~fmask))
+		return 0;
+
+	return val;
+}
+
+/* Given a register value, decode (extract) the value in the given field */
+static inline u32 reg_decode(const struct reg *reg, u32 field_id, u32 val)
+{
+	u32 fmask = reg_fmask(reg, field_id);
+
+	return fmask ? (val & fmask) >> __ffs(fmask) : 0;
+}
+
+/* Returns 0 for NULL reg; warning should have already been issued */
+static inline u32 reg_offset(const struct reg *reg)
+{
+	return reg ? reg->offset : 0;
+}
+
+/* Returns 0 for NULL reg; warning should have already been issued */
+static inline u32 reg_n_offset(const struct reg *reg, u32 n)
+{
+	return reg ? reg->offset + n * reg->stride : 0;
+}
+
+#endif /* _REG_H_ */