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, 150 insertions, 0 deletions

diff --git a/fs/hfs/trans.c b/fs/hfs/trans.c
new file mode 100644
index 000000000..fdb0edb8a
--- /dev/null
+++ b/fs/hfs/trans.c
@@ -0,0 +1,150 @@
+/*
+ *  linux/fs/hfs/trans.c
+ *
+ * Copyright (C) 1995-1997  Paul H. Hargrove
+ * This file may be distributed under the terms of the GNU General Public License.
+ *
+ * This file contains routines for converting between the Macintosh
+ * character set and various other encodings.  This includes dealing
+ * with ':' vs. '/' as the path-element separator.
+ */
+
+#include <linux/types.h>
+#include <linux/nls.h>
+
+#include "hfs_fs.h"
+
+/*================ Global functions ================*/
+
+/*
+ * hfs_mac2asc()
+ *
+ * Given a 'Pascal String' (a string preceded by a length byte) in
+ * the Macintosh character set produce the corresponding filename using
+ * the 'trivial' name-mangling scheme, returning the length of the
+ * mangled filename.  Note that the output string is not NULL
+ * terminated.
+ *
+ * The name-mangling works as follows:
+ * The character '/', which is illegal in Linux filenames is replaced
+ * by ':' which never appears in HFS filenames.	 All other characters
+ * are passed unchanged from input to output.
+ */
+int hfs_mac2asc(struct super_block *sb, char *out, const struct hfs_name *in)
+{
+	struct nls_table *nls_disk = HFS_SB(sb)->nls_disk;
+	struct nls_table *nls_io = HFS_SB(sb)->nls_io;
+	const char *src;
+	char *dst;
+	int srclen, dstlen, size;
+
+	src = in->name;
+	srclen = in->len;
+	if (srclen > HFS_NAMELEN)
+		srclen = HFS_NAMELEN;
+	dst = out;
+	dstlen = HFS_MAX_NAMELEN;
+	if (nls_io) {
+		wchar_t ch;
+
+		while (srclen > 0) {
+			if (nls_disk) {
+				size = nls_disk->char2uni(src, srclen, &ch);
+				if (size <= 0) {
+					ch = '?';
+					size = 1;
+				}
+				src += size;
+				srclen -= size;
+			} else {
+				ch = *src++;
+				srclen--;
+			}
+			if (ch == '/')
+				ch = ':';
+			size = nls_io->uni2char(ch, dst, dstlen);
+			if (size < 0) {
+				if (size == -ENAMETOOLONG)
+					goto out;
+				*dst = '?';
+				size = 1;
+			}
+			dst += size;
+			dstlen -= size;
+		}
+	} else {
+		char ch;
+
+		while (--srclen >= 0)
+			*dst++ = (ch = *src++) == '/' ? ':' : ch;
+	}
+out:
+	return dst - out;
+}
+
+/*
+ * hfs_asc2mac()
+ *
+ * Given an ASCII string (not null-terminated) and its length,
+ * generate the corresponding filename in the Macintosh character set
+ * using the 'trivial' name-mangling scheme, returning the length of
+ * the mangled filename.  Note that the output string is not NULL
+ * terminated.
+ *
+ * This routine is a inverse to hfs_mac2triv().
+ * A ':' is replaced by a '/'.
+ */
+void hfs_asc2mac(struct super_block *sb, struct hfs_name *out, const struct qstr *in)
+{
+	struct nls_table *nls_disk = HFS_SB(sb)->nls_disk;
+	struct nls_table *nls_io = HFS_SB(sb)->nls_io;
+	const char *src;
+	char *dst;
+	int srclen, dstlen, size;
+
+	src = in->name;
+	srclen = in->len;
+	dst = out->name;
+	dstlen = HFS_NAMELEN;
+	if (nls_io) {
+		wchar_t ch;
+
+		while (srclen > 0 && dstlen > 0) {
+			size = nls_io->char2uni(src, srclen, &ch);
+			if (size < 0) {
+				ch = '?';
+				size = 1;
+			}
+			src += size;
+			srclen -= size;
+			if (ch == ':')
+				ch = '/';
+			if (nls_disk) {
+				size = nls_disk->uni2char(ch, dst, dstlen);
+				if (size < 0) {
+					if (size == -ENAMETOOLONG)
+						goto out;
+					*dst = '?';
+					size = 1;
+				}
+				dst += size;
+				dstlen -= size;
+			} else {
+				*dst++ = ch > 0xff ? '?' : ch;
+				dstlen--;
+			}
+		}
+	} else {
+		char ch;
+
+		if (dstlen > srclen)
+			dstlen = srclen;
+		while (--dstlen >= 0)
+			*dst++ = (ch = *src++) == ':' ? '/' : ch;
+	}
+out:
+	out->len = dst - (char *)out->name;
+	dstlen = HFS_NAMELEN - out->len;
+	while (--dstlen >= 0)
+		*dst++ = 0;
+}