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, 405 insertions, 0 deletions

diff --git a/fs/statfs.c b/fs/statfs.c
new file mode 100644
index 000000000..377bcef7a
--- /dev/null
+++ b/fs/statfs.c
@@ -0,0 +1,405 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/syscalls.h>
+#include <linux/export.h>
+#include <linux/fs.h>
+#include <linux/file.h>
+#include <linux/mount.h>
+#include <linux/namei.h>
+#include <linux/statfs.h>
+#include <linux/security.h>
+#include <linux/uaccess.h>
+#include <linux/compat.h>
+#include "internal.h"
+
+static int flags_by_mnt(int mnt_flags)
+{
+	int flags = 0;
+
+	if (mnt_flags & MNT_READONLY)
+		flags |= ST_RDONLY;
+	if (mnt_flags & MNT_NOSUID)
+		flags |= ST_NOSUID;
+	if (mnt_flags & MNT_NODEV)
+		flags |= ST_NODEV;
+	if (mnt_flags & MNT_NOEXEC)
+		flags |= ST_NOEXEC;
+	if (mnt_flags & MNT_NOATIME)
+		flags |= ST_NOATIME;
+	if (mnt_flags & MNT_NODIRATIME)
+		flags |= ST_NODIRATIME;
+	if (mnt_flags & MNT_RELATIME)
+		flags |= ST_RELATIME;
+	if (mnt_flags & MNT_NOSYMFOLLOW)
+		flags |= ST_NOSYMFOLLOW;
+	return flags;
+}
+
+static int flags_by_sb(int s_flags)
+{
+	int flags = 0;
+	if (s_flags & SB_SYNCHRONOUS)
+		flags |= ST_SYNCHRONOUS;
+	if (s_flags & SB_MANDLOCK)
+		flags |= ST_MANDLOCK;
+	if (s_flags & SB_RDONLY)
+		flags |= ST_RDONLY;
+	return flags;
+}
+
+static int calculate_f_flags(struct vfsmount *mnt)
+{
+	return ST_VALID | flags_by_mnt(mnt->mnt_flags) |
+		flags_by_sb(mnt->mnt_sb->s_flags);
+}
+
+static int statfs_by_dentry(struct dentry *dentry, struct kstatfs *buf)
+{
+	int retval;
+
+	if (!dentry->d_sb->s_op->statfs)
+		return -ENOSYS;
+
+	memset(buf, 0, sizeof(*buf));
+	retval = security_sb_statfs(dentry);
+	if (retval)
+		return retval;
+	retval = dentry->d_sb->s_op->statfs(dentry, buf);
+	if (retval == 0 && buf->f_frsize == 0)
+		buf->f_frsize = buf->f_bsize;
+	return retval;
+}
+
+int vfs_get_fsid(struct dentry *dentry, __kernel_fsid_t *fsid)
+{
+	struct kstatfs st;
+	int error;
+
+	error = statfs_by_dentry(dentry, &st);
+	if (error)
+		return error;
+
+	*fsid = st.f_fsid;
+	return 0;
+}
+EXPORT_SYMBOL(vfs_get_fsid);
+
+int vfs_statfs(const struct path *path, struct kstatfs *buf)
+{
+	int error;
+
+	error = statfs_by_dentry(path->dentry, buf);
+	if (!error)
+		buf->f_flags = calculate_f_flags(path->mnt);
+	return error;
+}
+EXPORT_SYMBOL(vfs_statfs);
+
+int user_statfs(const char __user *pathname, struct kstatfs *st)
+{
+	struct path path;
+	int error;
+	unsigned int lookup_flags = LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT;
+	CLASS(filename, name)(pathname);
+retry:
+	error = filename_lookup(AT_FDCWD, name, lookup_flags, &path, NULL);
+	if (!error) {
+		error = vfs_statfs(&path, st);
+		path_put(&path);
+		if (retry_estale(error, lookup_flags)) {
+			lookup_flags |= LOOKUP_REVAL;
+			goto retry;
+		}
+	}
+	return error;
+}
+
+int fd_statfs(int fd, struct kstatfs *st)
+{
+	CLASS(fd_raw, f)(fd);
+
+	if (fd_empty(f))
+		return -EBADF;
+	return vfs_statfs(&fd_file(f)->f_path, st);
+}
+
+static int do_statfs_native(struct kstatfs *st, struct statfs __user *p)
+{
+	struct statfs buf;
+
+	if (sizeof(buf) == sizeof(*st))
+		memcpy(&buf, st, sizeof(*st));
+	else {
+		memset(&buf, 0, sizeof(buf));
+		if (sizeof buf.f_blocks == 4) {
+			if ((st->f_blocks | st->f_bfree | st->f_bavail |
+			     st->f_bsize | st->f_frsize) &
+			    0xffffffff00000000ULL)
+				return -EOVERFLOW;
+			/*
+			 * f_files and f_ffree may be -1; it's okay to stuff
+			 * that into 32 bits
+			 */
+			if (st->f_files != -1 &&
+			    (st->f_files & 0xffffffff00000000ULL))
+				return -EOVERFLOW;
+			if (st->f_ffree != -1 &&
+			    (st->f_ffree & 0xffffffff00000000ULL))
+				return -EOVERFLOW;
+		}
+
+		buf.f_type = st->f_type;
+		buf.f_bsize = st->f_bsize;
+		buf.f_blocks = st->f_blocks;
+		buf.f_bfree = st->f_bfree;
+		buf.f_bavail = st->f_bavail;
+		buf.f_files = st->f_files;
+		buf.f_ffree = st->f_ffree;
+		buf.f_fsid = st->f_fsid;
+		buf.f_namelen = st->f_namelen;
+		buf.f_frsize = st->f_frsize;
+		buf.f_flags = st->f_flags;
+	}
+	if (copy_to_user(p, &buf, sizeof(buf)))
+		return -EFAULT;
+	return 0;
+}
+
+static int do_statfs64(struct kstatfs *st, struct statfs64 __user *p)
+{
+	struct statfs64 buf;
+	if (sizeof(buf) == sizeof(*st))
+		memcpy(&buf, st, sizeof(*st));
+	else {
+		memset(&buf, 0, sizeof(buf));
+		buf.f_type = st->f_type;
+		buf.f_bsize = st->f_bsize;
+		buf.f_blocks = st->f_blocks;
+		buf.f_bfree = st->f_bfree;
+		buf.f_bavail = st->f_bavail;
+		buf.f_files = st->f_files;
+		buf.f_ffree = st->f_ffree;
+		buf.f_fsid = st->f_fsid;
+		buf.f_namelen = st->f_namelen;
+		buf.f_frsize = st->f_frsize;
+		buf.f_flags = st->f_flags;
+	}
+	if (copy_to_user(p, &buf, sizeof(buf)))
+		return -EFAULT;
+	return 0;
+}
+
+SYSCALL_DEFINE2(statfs, const char __user *, pathname, struct statfs __user *, buf)
+{
+	struct kstatfs st;
+	int error = user_statfs(pathname, &st);
+	if (!error)
+		error = do_statfs_native(&st, buf);
+	return error;
+}
+
+SYSCALL_DEFINE3(statfs64, const char __user *, pathname, size_t, sz, struct statfs64 __user *, buf)
+{
+	struct kstatfs st;
+	int error;
+	if (sz != sizeof(*buf))
+		return -EINVAL;
+	error = user_statfs(pathname, &st);
+	if (!error)
+		error = do_statfs64(&st, buf);
+	return error;
+}
+
+SYSCALL_DEFINE2(fstatfs, unsigned int, fd, struct statfs __user *, buf)
+{
+	struct kstatfs st;
+	int error = fd_statfs(fd, &st);
+	if (!error)
+		error = do_statfs_native(&st, buf);
+	return error;
+}
+
+SYSCALL_DEFINE3(fstatfs64, unsigned int, fd, size_t, sz, struct statfs64 __user *, buf)
+{
+	struct kstatfs st;
+	int error;
+
+	if (sz != sizeof(*buf))
+		return -EINVAL;
+
+	error = fd_statfs(fd, &st);
+	if (!error)
+		error = do_statfs64(&st, buf);
+	return error;
+}
+
+static int vfs_ustat(dev_t dev, struct kstatfs *sbuf)
+{
+	struct super_block *s = user_get_super(dev, false);
+	int err;
+	if (!s)
+		return -EINVAL;
+
+	err = statfs_by_dentry(s->s_root, sbuf);
+	drop_super(s);
+	return err;
+}
+
+SYSCALL_DEFINE2(ustat, unsigned, dev, struct ustat __user *, ubuf)
+{
+	struct ustat tmp;
+	struct kstatfs sbuf;
+	int err = vfs_ustat(new_decode_dev(dev), &sbuf);
+	if (err)
+		return err;
+
+	memset(&tmp,0,sizeof(struct ustat));
+	tmp.f_tfree = sbuf.f_bfree;
+	if (IS_ENABLED(CONFIG_ARCH_32BIT_USTAT_F_TINODE))
+		tmp.f_tinode = min_t(u64, sbuf.f_ffree, UINT_MAX);
+	else
+		tmp.f_tinode = sbuf.f_ffree;
+
+	return copy_to_user(ubuf, &tmp, sizeof(struct ustat)) ? -EFAULT : 0;
+}
+
+#ifdef CONFIG_COMPAT
+static int put_compat_statfs(struct compat_statfs __user *ubuf, struct kstatfs *kbuf)
+{
+	struct compat_statfs buf;
+	if (sizeof ubuf->f_blocks == 4) {
+		if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail |
+		     kbuf->f_bsize | kbuf->f_frsize) & 0xffffffff00000000ULL)
+			return -EOVERFLOW;
+		/* f_files and f_ffree may be -1; it's okay
+		 * to stuff that into 32 bits */
+		if (kbuf->f_files != 0xffffffffffffffffULL
+		 && (kbuf->f_files & 0xffffffff00000000ULL))
+			return -EOVERFLOW;
+		if (kbuf->f_ffree != 0xffffffffffffffffULL
+		 && (kbuf->f_ffree & 0xffffffff00000000ULL))
+			return -EOVERFLOW;
+	}
+	memset(&buf, 0, sizeof(struct compat_statfs));
+	buf.f_type = kbuf->f_type;
+	buf.f_bsize = kbuf->f_bsize;
+	buf.f_blocks = kbuf->f_blocks;
+	buf.f_bfree = kbuf->f_bfree;
+	buf.f_bavail = kbuf->f_bavail;
+	buf.f_files = kbuf->f_files;
+	buf.f_ffree = kbuf->f_ffree;
+	buf.f_namelen = kbuf->f_namelen;
+	buf.f_fsid.val[0] = kbuf->f_fsid.val[0];
+	buf.f_fsid.val[1] = kbuf->f_fsid.val[1];
+	buf.f_frsize = kbuf->f_frsize;
+	buf.f_flags = kbuf->f_flags;
+	if (copy_to_user(ubuf, &buf, sizeof(struct compat_statfs)))
+		return -EFAULT;
+	return 0;
+}
+
+/*
+ * The following statfs calls are copies of code from fs/statfs.c and
+ * should be checked against those from time to time
+ */
+COMPAT_SYSCALL_DEFINE2(statfs, const char __user *, pathname, struct compat_statfs __user *, buf)
+{
+	struct kstatfs tmp;
+	int error = user_statfs(pathname, &tmp);
+	if (!error)
+		error = put_compat_statfs(buf, &tmp);
+	return error;
+}
+
+COMPAT_SYSCALL_DEFINE2(fstatfs, unsigned int, fd, struct compat_statfs __user *, buf)
+{
+	struct kstatfs tmp;
+	int error = fd_statfs(fd, &tmp);
+	if (!error)
+		error = put_compat_statfs(buf, &tmp);
+	return error;
+}
+
+static int put_compat_statfs64(struct compat_statfs64 __user *ubuf, struct kstatfs *kbuf)
+{
+	struct compat_statfs64 buf;
+
+	if ((kbuf->f_bsize | kbuf->f_frsize) & 0xffffffff00000000ULL)
+		return -EOVERFLOW;
+
+	memset(&buf, 0, sizeof(struct compat_statfs64));
+	buf.f_type = kbuf->f_type;
+	buf.f_bsize = kbuf->f_bsize;
+	buf.f_blocks = kbuf->f_blocks;
+	buf.f_bfree = kbuf->f_bfree;
+	buf.f_bavail = kbuf->f_bavail;
+	buf.f_files = kbuf->f_files;
+	buf.f_ffree = kbuf->f_ffree;
+	buf.f_namelen = kbuf->f_namelen;
+	buf.f_fsid.val[0] = kbuf->f_fsid.val[0];
+	buf.f_fsid.val[1] = kbuf->f_fsid.val[1];
+	buf.f_frsize = kbuf->f_frsize;
+	buf.f_flags = kbuf->f_flags;
+	if (copy_to_user(ubuf, &buf, sizeof(struct compat_statfs64)))
+		return -EFAULT;
+	return 0;
+}
+
+int kcompat_sys_statfs64(const char __user * pathname, compat_size_t sz, struct compat_statfs64 __user * buf)
+{
+	struct kstatfs tmp;
+	int error;
+
+	if (sz != sizeof(*buf))
+		return -EINVAL;
+
+	error = user_statfs(pathname, &tmp);
+	if (!error)
+		error = put_compat_statfs64(buf, &tmp);
+	return error;
+}
+
+COMPAT_SYSCALL_DEFINE3(statfs64, const char __user *, pathname, compat_size_t, sz, struct compat_statfs64 __user *, buf)
+{
+	return kcompat_sys_statfs64(pathname, sz, buf);
+}
+
+int kcompat_sys_fstatfs64(unsigned int fd, compat_size_t sz, struct compat_statfs64 __user * buf)
+{
+	struct kstatfs tmp;
+	int error;
+
+	if (sz != sizeof(*buf))
+		return -EINVAL;
+
+	error = fd_statfs(fd, &tmp);
+	if (!error)
+		error = put_compat_statfs64(buf, &tmp);
+	return error;
+}
+
+COMPAT_SYSCALL_DEFINE3(fstatfs64, unsigned int, fd, compat_size_t, sz, struct compat_statfs64 __user *, buf)
+{
+	return kcompat_sys_fstatfs64(fd, sz, buf);
+}
+
+/*
+ * This is a copy of sys_ustat, just dealing with a structure layout.
+ * Given how simple this syscall is that apporach is more maintainable
+ * than the various conversion hacks.
+ */
+COMPAT_SYSCALL_DEFINE2(ustat, unsigned, dev, struct compat_ustat __user *, u)
+{
+	struct compat_ustat tmp;
+	struct kstatfs sbuf;
+	int err = vfs_ustat(new_decode_dev(dev), &sbuf);
+	if (err)
+		return err;
+
+	memset(&tmp, 0, sizeof(struct compat_ustat));
+	tmp.f_tfree = sbuf.f_bfree;
+	tmp.f_tinode = sbuf.f_ffree;
+	if (copy_to_user(u, &tmp, sizeof(struct compat_ustat)))
+		return -EFAULT;
+	return 0;
+}
+#endif