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, 73 insertions, 0 deletions

diff --git a/include/uapi/asm-generic/stat.h b/include/uapi/asm-generic/stat.h
new file mode 100644
index 000000000..0d962ecd1
--- /dev/null
+++ b/include/uapi/asm-generic/stat.h
@@ -0,0 +1,73 @@
+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+#ifndef __ASM_GENERIC_STAT_H
+#define __ASM_GENERIC_STAT_H
+
+/*
+ * Everybody gets this wrong and has to stick with it for all
+ * eternity. Hopefully, this version gets used by new architectures
+ * so they don't fall into the same traps.
+ *
+ * stat64 is copied from powerpc64, with explicit padding added.
+ * stat is the same structure layout on 64-bit, without the 'long long'
+ * types.
+ *
+ * By convention, 64 bit architectures use the stat interface, while
+ * 32 bit architectures use the stat64 interface. Note that we don't
+ * provide an __old_kernel_stat here, which new architecture should
+ * not have to start with.
+ */
+
+#include <asm/bitsperlong.h>
+
+#define STAT_HAVE_NSEC 1
+
+struct stat {
+	unsigned long	st_dev;		/* Device.  */
+	unsigned long	st_ino;		/* File serial number.  */
+	unsigned int	st_mode;	/* File mode.  */
+	unsigned int	st_nlink;	/* Link count.  */
+	unsigned int	st_uid;		/* User ID of the file's owner.  */
+	unsigned int	st_gid;		/* Group ID of the file's group. */
+	unsigned long	st_rdev;	/* Device number, if device.  */
+	unsigned long	__pad1;
+	long		st_size;	/* Size of file, in bytes.  */
+	int		st_blksize;	/* Optimal block size for I/O.  */
+	int		__pad2;
+	long		st_blocks;	/* Number 512-byte blocks allocated. */
+	long		st_atime;	/* Time of last access.  */
+	unsigned long	st_atime_nsec;
+	long		st_mtime;	/* Time of last modification.  */
+	unsigned long	st_mtime_nsec;
+	long		st_ctime;	/* Time of last status change.  */
+	unsigned long	st_ctime_nsec;
+	unsigned int	__unused4;
+	unsigned int	__unused5;
+};
+
+/* This matches struct stat64 in glibc2.1. Only used for 32 bit. */
+#if __BITS_PER_LONG != 64 || defined(__ARCH_WANT_STAT64)
+struct stat64 {
+	unsigned long long st_dev;	/* Device.  */
+	unsigned long long st_ino;	/* File serial number.  */
+	unsigned int	st_mode;	/* File mode.  */
+	unsigned int	st_nlink;	/* Link count.  */
+	unsigned int	st_uid;		/* User ID of the file's owner.  */
+	unsigned int	st_gid;		/* Group ID of the file's group. */
+	unsigned long long st_rdev;	/* Device number, if device.  */
+	unsigned long long __pad1;
+	long long	st_size;	/* Size of file, in bytes.  */
+	int		st_blksize;	/* Optimal block size for I/O.  */
+	int		__pad2;
+	long long	st_blocks;	/* Number 512-byte blocks allocated. */
+	int		st_atime;	/* Time of last access.  */
+	unsigned int	st_atime_nsec;
+	int		st_mtime;	/* Time of last modification.  */
+	unsigned int	st_mtime_nsec;
+	int		st_ctime;	/* Time of last status change.  */
+	unsigned int	st_ctime_nsec;
+	unsigned int	__unused4;
+	unsigned int	__unused5;
+};
+#endif
+
+#endif /* __ASM_GENERIC_STAT_H */