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, 232 insertions, 0 deletions

diff --git a/drivers/char/adi.c b/drivers/char/adi.c
new file mode 100644
index 000000000..0849d933a
--- /dev/null
+++ b/drivers/char/adi.c
@@ -0,0 +1,232 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Privileged ADI driver for sparc64
+ *
+ * Author: Tom Hromatka <tom.hromatka@oracle.com>
+ */
+#include <linux/kernel.h>
+#include <linux/miscdevice.h>
+#include <linux/module.h>
+#include <linux/proc_fs.h>
+#include <linux/slab.h>
+#include <linux/uaccess.h>
+#include <asm/asi.h>
+
+#define MAX_BUF_SZ	PAGE_SIZE
+
+static int read_mcd_tag(unsigned long addr)
+{
+	long err;
+	int ver;
+
+	__asm__ __volatile__(
+		"1:	ldxa [%[addr]] %[asi], %[ver]\n"
+		"	mov 0, %[err]\n"
+		"2:\n"
+		"	.section .fixup,#alloc,#execinstr\n"
+		"	.align 4\n"
+		"3:	sethi %%hi(2b), %%g1\n"
+		"	jmpl  %%g1 + %%lo(2b), %%g0\n"
+		"	mov %[invalid], %[err]\n"
+		"	.previous\n"
+		"	.section __ex_table, \"a\"\n"
+		"	.align 4\n"
+		"	.word  1b, 3b\n"
+		"	.previous\n"
+		: [ver] "=r" (ver), [err] "=r" (err)
+		: [addr] "r"  (addr), [invalid] "i" (EFAULT),
+		  [asi] "i" (ASI_MCD_REAL)
+		: "memory", "g1"
+		);
+
+	if (err)
+		return -EFAULT;
+	else
+		return ver;
+}
+
+static ssize_t adi_read(struct file *file, char __user *buf,
+			size_t count, loff_t *offp)
+{
+	size_t ver_buf_sz, bytes_read = 0;
+	int ver_buf_idx = 0;
+	loff_t offset;
+	u8 *ver_buf;
+	ssize_t ret;
+
+	ver_buf_sz = min_t(size_t, count, MAX_BUF_SZ);
+	ver_buf = kmalloc(ver_buf_sz, GFP_KERNEL);
+	if (!ver_buf)
+		return -ENOMEM;
+
+	offset = (*offp) * adi_blksize();
+
+	while (bytes_read < count) {
+		ret = read_mcd_tag(offset);
+		if (ret < 0)
+			goto out;
+
+		ver_buf[ver_buf_idx] = (u8)ret;
+		ver_buf_idx++;
+		offset += adi_blksize();
+
+		if (ver_buf_idx >= ver_buf_sz) {
+			if (copy_to_user(buf + bytes_read, ver_buf,
+					 ver_buf_sz)) {
+				ret = -EFAULT;
+				goto out;
+			}
+
+			bytes_read += ver_buf_sz;
+			ver_buf_idx = 0;
+
+			ver_buf_sz = min_t(size_t, count - bytes_read,
+					   MAX_BUF_SZ);
+		}
+	}
+
+	(*offp) += bytes_read;
+	ret = bytes_read;
+out:
+	kfree(ver_buf);
+	return ret;
+}
+
+static int set_mcd_tag(unsigned long addr, u8 ver)
+{
+	long err;
+
+	__asm__ __volatile__(
+		"1:	stxa %[ver], [%[addr]] %[asi]\n"
+		"	mov 0, %[err]\n"
+		"2:\n"
+		"	.section .fixup,#alloc,#execinstr\n"
+		"	.align 4\n"
+		"3:	sethi %%hi(2b), %%g1\n"
+		"	jmpl %%g1 + %%lo(2b), %%g0\n"
+		"	mov %[invalid], %[err]\n"
+		"	.previous\n"
+		"	.section __ex_table, \"a\"\n"
+		"	.align 4\n"
+		"	.word 1b, 3b\n"
+		"	.previous\n"
+		: [err] "=r" (err)
+		: [ver] "r" (ver), [addr] "r"  (addr),
+		  [invalid] "i" (EFAULT), [asi] "i" (ASI_MCD_REAL)
+		: "memory", "g1"
+		);
+
+	if (err)
+		return -EFAULT;
+	else
+		return ver;
+}
+
+static ssize_t adi_write(struct file *file, const char __user *buf,
+			 size_t count, loff_t *offp)
+{
+	size_t ver_buf_sz, bytes_written = 0;
+	loff_t offset;
+	u8 *ver_buf;
+	ssize_t ret;
+	int i;
+
+	if (count == 0)
+		return -EINVAL;
+
+	ver_buf_sz = min_t(size_t, count, MAX_BUF_SZ);
+	ver_buf = kmalloc(ver_buf_sz, GFP_KERNEL);
+	if (!ver_buf)
+		return -ENOMEM;
+
+	offset = (*offp) * adi_blksize();
+
+	do {
+		if (copy_from_user(ver_buf, &buf[bytes_written],
+				   ver_buf_sz)) {
+			ret = -EFAULT;
+			goto out;
+		}
+
+		for (i = 0; i < ver_buf_sz; i++) {
+			ret = set_mcd_tag(offset, ver_buf[i]);
+			if (ret < 0)
+				goto out;
+
+			offset += adi_blksize();
+		}
+
+		bytes_written += ver_buf_sz;
+		ver_buf_sz = min_t(size_t, count - bytes_written, MAX_BUF_SZ);
+	} while (bytes_written < count);
+
+	(*offp) += bytes_written;
+	ret = bytes_written;
+out:
+	__asm__ __volatile__("membar #Sync");
+	kfree(ver_buf);
+	return ret;
+}
+
+static loff_t adi_llseek(struct file *file, loff_t offset, int whence)
+{
+	loff_t ret = -EINVAL;
+
+	switch (whence) {
+	case SEEK_END:
+	case SEEK_DATA:
+	case SEEK_HOLE:
+		/* unsupported */
+		return -EINVAL;
+	case SEEK_CUR:
+		if (offset == 0)
+			return file->f_pos;
+
+		offset += file->f_pos;
+		break;
+	case SEEK_SET:
+		break;
+	}
+
+	if (offset != file->f_pos) {
+		file->f_pos = offset;
+		ret = offset;
+	}
+
+	return ret;
+}
+
+static const struct file_operations adi_fops = {
+	.owner		= THIS_MODULE,
+	.llseek		= adi_llseek,
+	.read		= adi_read,
+	.write		= adi_write,
+	.fop_flags	= FOP_UNSIGNED_OFFSET,
+};
+
+static struct miscdevice adi_miscdev = {
+	.minor = MISC_DYNAMIC_MINOR,
+	.name = KBUILD_MODNAME,
+	.fops = &adi_fops,
+};
+
+static int __init adi_init(void)
+{
+	if (!adi_capable())
+		return -EPERM;
+
+	return misc_register(&adi_miscdev);
+}
+
+static void __exit adi_exit(void)
+{
+	misc_deregister(&adi_miscdev);
+}
+
+module_init(adi_init);
+module_exit(adi_exit);
+
+MODULE_AUTHOR("Tom Hromatka <tom.hromatka@oracle.com>");
+MODULE_DESCRIPTION("Privileged interface to ADI");
+MODULE_VERSION("1.0");
+MODULE_LICENSE("GPL v2");