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, 594 insertions, 0 deletions

diff --git a/fs/nilfs2/namei.c b/fs/nilfs2/namei.c
new file mode 100644
index 000000000..e037e0c6e
--- /dev/null
+++ b/fs/nilfs2/namei.c
@@ -0,0 +1,594 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS pathname lookup operations.
+ *
+ * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Modified for NILFS by Amagai Yoshiji and Ryusuke Konishi.
+ */
+/*
+ *  linux/fs/ext2/namei.c
+ *
+ * Copyright (C) 1992, 1993, 1994, 1995
+ * Remy Card (card@masi.ibp.fr)
+ * Laboratoire MASI - Institut Blaise Pascal
+ * Universite Pierre et Marie Curie (Paris VI)
+ *
+ *  from
+ *
+ *  linux/fs/minix/namei.c
+ *
+ *  Copyright (C) 1991, 1992  Linus Torvalds
+ *
+ *  Big-endian to little-endian byte-swapping/bitmaps by
+ *        David S. Miller (davem@caip.rutgers.edu), 1995
+ */
+
+#include <linux/pagemap.h>
+#include "nilfs.h"
+#include "export.h"
+
+#define NILFS_FID_SIZE_NON_CONNECTABLE \
+	(offsetof(struct nilfs_fid, parent_gen) / 4)
+#define NILFS_FID_SIZE_CONNECTABLE	(sizeof(struct nilfs_fid) / 4)
+
+static inline int nilfs_add_nondir(struct dentry *dentry, struct inode *inode)
+{
+	int err = nilfs_add_link(dentry, inode);
+
+	if (!err) {
+		d_instantiate_new(dentry, inode);
+		return 0;
+	}
+	inode_dec_link_count(inode);
+	unlock_new_inode(inode);
+	iput(inode);
+	return err;
+}
+
+/*
+ * Methods themselves.
+ */
+
+static struct dentry *
+nilfs_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
+{
+	struct inode *inode;
+	u64 ino;
+	int res;
+
+	if (dentry->d_name.len > NILFS_NAME_LEN)
+		return ERR_PTR(-ENAMETOOLONG);
+
+	res = nilfs_inode_by_name(dir, &dentry->d_name, &ino);
+	if (res) {
+		if (res != -ENOENT)
+			return ERR_PTR(res);
+		inode = NULL;
+	} else {
+		inode = nilfs_iget(dir->i_sb, NILFS_I(dir)->i_root, ino);
+		if (inode == ERR_PTR(-ESTALE)) {
+			nilfs_error(dir->i_sb,
+				"deleted inode referenced: %llu", ino);
+			return ERR_PTR(-EIO);
+		}
+	}
+
+	return d_splice_alias(inode, dentry);
+}
+
+/*
+ * By the time this is called, we already have created
+ * the directory cache entry for the new file, but it
+ * is so far negative - it has no inode.
+ *
+ * If the create succeeds, we fill in the inode information
+ * with d_instantiate().
+ */
+static int nilfs_create(struct mnt_idmap *idmap, struct inode *dir,
+			struct dentry *dentry, umode_t mode)
+{
+	struct inode *inode;
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return err;
+	inode = nilfs_new_inode(dir, mode);
+	err = PTR_ERR(inode);
+	if (!IS_ERR(inode)) {
+		inode->i_op = &nilfs_file_inode_operations;
+		inode->i_fop = &nilfs_file_operations;
+		inode->i_mapping->a_ops = &nilfs_aops;
+		nilfs_mark_inode_dirty(inode);
+		err = nilfs_add_nondir(dentry, inode);
+	}
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+}
+
+static int
+nilfs_mknod(struct mnt_idmap *idmap, struct inode *dir,
+	    struct dentry *dentry, umode_t mode, dev_t rdev)
+{
+	struct inode *inode;
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return err;
+	inode = nilfs_new_inode(dir, mode);
+	err = PTR_ERR(inode);
+	if (!IS_ERR(inode)) {
+		init_special_inode(inode, inode->i_mode, rdev);
+		nilfs_mark_inode_dirty(inode);
+		err = nilfs_add_nondir(dentry, inode);
+	}
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+}
+
+static int nilfs_symlink(struct mnt_idmap *idmap, struct inode *dir,
+			 struct dentry *dentry, const char *symname)
+{
+	struct nilfs_transaction_info ti;
+	struct super_block *sb = dir->i_sb;
+	unsigned int l = strlen(symname) + 1;
+	struct inode *inode;
+	int err;
+
+	if (l > sb->s_blocksize)
+		return -ENAMETOOLONG;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return err;
+
+	inode = nilfs_new_inode(dir, S_IFLNK | 0777);
+	err = PTR_ERR(inode);
+	if (IS_ERR(inode))
+		goto out;
+
+	/* slow symlink */
+	inode->i_op = &nilfs_symlink_inode_operations;
+	inode_nohighmem(inode);
+	mapping_set_gfp_mask(inode->i_mapping,
+			     mapping_gfp_constraint(inode->i_mapping,
+						    ~__GFP_FS));
+	inode->i_mapping->a_ops = &nilfs_aops;
+	err = page_symlink(inode, symname, l);
+	if (err)
+		goto out_fail;
+
+	/* mark_inode_dirty(inode); */
+	/* page_symlink() do this */
+
+	err = nilfs_add_nondir(dentry, inode);
+out:
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+
+out_fail:
+	drop_nlink(inode);
+	nilfs_mark_inode_dirty(inode);
+	unlock_new_inode(inode);
+	iput(inode);
+	goto out;
+}
+
+static int nilfs_link(struct dentry *old_dentry, struct inode *dir,
+		      struct dentry *dentry)
+{
+	struct inode *inode = d_inode(old_dentry);
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return err;
+
+	inode_set_ctime_current(inode);
+	inode_inc_link_count(inode);
+	ihold(inode);
+
+	err = nilfs_add_link(dentry, inode);
+	if (!err) {
+		d_instantiate(dentry, inode);
+		err = nilfs_transaction_commit(dir->i_sb);
+	} else {
+		inode_dec_link_count(inode);
+		iput(inode);
+		nilfs_transaction_abort(dir->i_sb);
+	}
+
+	return err;
+}
+
+static struct dentry *nilfs_mkdir(struct mnt_idmap *idmap, struct inode *dir,
+				  struct dentry *dentry, umode_t mode)
+{
+	struct inode *inode;
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 1);
+	if (err)
+		return ERR_PTR(err);
+
+	inc_nlink(dir);
+
+	inode = nilfs_new_inode(dir, mode);
+	err = PTR_ERR(inode);
+	if (IS_ERR(inode))
+		goto out_dir;
+
+	inode->i_op = &nilfs_dir_inode_operations;
+	inode->i_fop = &nilfs_dir_operations;
+	inode->i_mapping->a_ops = &nilfs_aops;
+
+	inc_nlink(inode);
+
+	err = nilfs_make_empty(inode, dir);
+	if (err)
+		goto out_fail;
+
+	err = nilfs_add_link(dentry, inode);
+	if (err)
+		goto out_fail;
+
+	nilfs_mark_inode_dirty(inode);
+	d_instantiate_new(dentry, inode);
+out:
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err ? ERR_PTR(err) : NULL;
+
+out_fail:
+	drop_nlink(inode);
+	drop_nlink(inode);
+	nilfs_mark_inode_dirty(inode);
+	unlock_new_inode(inode);
+	iput(inode);
+out_dir:
+	drop_nlink(dir);
+	nilfs_mark_inode_dirty(dir);
+	goto out;
+}
+
+static int nilfs_do_unlink(struct inode *dir, struct dentry *dentry)
+{
+	struct inode *inode;
+	struct nilfs_dir_entry *de;
+	struct folio *folio;
+	int err;
+
+	de = nilfs_find_entry(dir, &dentry->d_name, &folio);
+	if (IS_ERR(de)) {
+		err = PTR_ERR(de);
+		goto out;
+	}
+
+	inode = d_inode(dentry);
+	err = -EIO;
+	if (le64_to_cpu(de->inode) != inode->i_ino)
+		goto out;
+
+	if (!inode->i_nlink) {
+		nilfs_warn(inode->i_sb,
+			   "deleting nonexistent file (ino=%llu), %d",
+			   inode->i_ino, inode->i_nlink);
+		set_nlink(inode, 1);
+	}
+	err = nilfs_delete_entry(de, folio);
+	folio_release_kmap(folio, de);
+	if (err)
+		goto out;
+
+	inode_set_ctime_to_ts(inode, inode_get_ctime(dir));
+	drop_nlink(inode);
+	err = 0;
+out:
+	return err;
+}
+
+static int nilfs_unlink(struct inode *dir, struct dentry *dentry)
+{
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 0);
+	if (err)
+		return err;
+
+	err = nilfs_do_unlink(dir, dentry);
+
+	if (!err) {
+		nilfs_mark_inode_dirty(dir);
+		nilfs_mark_inode_dirty(d_inode(dentry));
+		err = nilfs_transaction_commit(dir->i_sb);
+	} else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+}
+
+static int nilfs_rmdir(struct inode *dir, struct dentry *dentry)
+{
+	struct inode *inode = d_inode(dentry);
+	struct nilfs_transaction_info ti;
+	int err;
+
+	err = nilfs_transaction_begin(dir->i_sb, &ti, 0);
+	if (err)
+		return err;
+
+	err = -ENOTEMPTY;
+	if (nilfs_empty_dir(inode)) {
+		err = nilfs_do_unlink(dir, dentry);
+		if (!err) {
+			inode->i_size = 0;
+			drop_nlink(inode);
+			nilfs_mark_inode_dirty(inode);
+			drop_nlink(dir);
+			nilfs_mark_inode_dirty(dir);
+		}
+	}
+	if (!err)
+		err = nilfs_transaction_commit(dir->i_sb);
+	else
+		nilfs_transaction_abort(dir->i_sb);
+
+	return err;
+}
+
+static int nilfs_rename(struct mnt_idmap *idmap,
+			struct inode *old_dir, struct dentry *old_dentry,
+			struct inode *new_dir, struct dentry *new_dentry,
+			unsigned int flags)
+{
+	struct inode *old_inode = d_inode(old_dentry);
+	struct inode *new_inode = d_inode(new_dentry);
+	struct folio *dir_folio = NULL;
+	struct nilfs_dir_entry *dir_de = NULL;
+	struct folio *old_folio;
+	struct nilfs_dir_entry *old_de;
+	struct nilfs_transaction_info ti;
+	bool old_is_dir = S_ISDIR(old_inode->i_mode);
+	int err;
+
+	if (flags & ~RENAME_NOREPLACE)
+		return -EINVAL;
+
+	err = nilfs_transaction_begin(old_dir->i_sb, &ti, 1);
+	if (unlikely(err))
+		return err;
+
+	old_de = nilfs_find_entry(old_dir, &old_dentry->d_name, &old_folio);
+	if (IS_ERR(old_de)) {
+		err = PTR_ERR(old_de);
+		goto out;
+	}
+
+	if (old_is_dir && old_dir != new_dir) {
+		err = -EIO;
+		dir_de = nilfs_dotdot(old_inode, &dir_folio);
+		if (!dir_de)
+			goto out_old;
+	}
+
+	if (new_inode) {
+		struct folio *new_folio;
+		struct nilfs_dir_entry *new_de;
+
+		err = -ENOTEMPTY;
+		if (old_is_dir && !nilfs_empty_dir(new_inode))
+			goto out_dir;
+
+		new_de = nilfs_find_entry(new_dir, &new_dentry->d_name,
+					  &new_folio);
+		if (IS_ERR(new_de)) {
+			err = PTR_ERR(new_de);
+			goto out_dir;
+		}
+		err = nilfs_set_link(new_dir, new_de, new_folio, old_inode);
+		folio_release_kmap(new_folio, new_de);
+		if (unlikely(err))
+			goto out_dir;
+		nilfs_mark_inode_dirty(new_dir);
+		inode_set_ctime_current(new_inode);
+		if (old_is_dir)
+			drop_nlink(new_inode);
+		drop_nlink(new_inode);
+		nilfs_mark_inode_dirty(new_inode);
+	} else {
+		err = nilfs_add_link(new_dentry, old_inode);
+		if (err)
+			goto out_dir;
+		if (old_is_dir) {
+			inc_nlink(new_dir);
+			nilfs_mark_inode_dirty(new_dir);
+		}
+	}
+
+	/*
+	 * Like most other Unix systems, set the ctime for inodes on a
+	 * rename.
+	 */
+	inode_set_ctime_current(old_inode);
+
+	err = nilfs_delete_entry(old_de, old_folio);
+	if (likely(!err)) {
+		if (old_is_dir) {
+			if (old_dir != new_dir)
+				err = nilfs_set_link(old_inode, dir_de,
+						     dir_folio, new_dir);
+			drop_nlink(old_dir);
+		}
+		nilfs_mark_inode_dirty(old_dir);
+	}
+	nilfs_mark_inode_dirty(old_inode);
+
+out_dir:
+	if (dir_de)
+		folio_release_kmap(dir_folio, dir_de);
+out_old:
+	folio_release_kmap(old_folio, old_de);
+out:
+	if (likely(!err))
+		err = nilfs_transaction_commit(old_dir->i_sb);
+	else
+		nilfs_transaction_abort(old_dir->i_sb);
+	return err;
+}
+
+/*
+ * Export operations
+ */
+static struct dentry *nilfs_get_parent(struct dentry *child)
+{
+	u64 ino;
+	int res;
+	struct nilfs_root *root;
+
+	res = nilfs_inode_by_name(d_inode(child), &dotdot_name, &ino);
+	if (res)
+		return ERR_PTR(res);
+
+	root = NILFS_I(d_inode(child))->i_root;
+
+	return d_obtain_alias(nilfs_iget(child->d_sb, root, ino));
+}
+
+static struct dentry *nilfs_get_dentry(struct super_block *sb, u64 cno,
+				       u64 ino, u32 gen)
+{
+	struct nilfs_root *root;
+	struct inode *inode;
+
+	if (ino < NILFS_FIRST_INO(sb) && ino != NILFS_ROOT_INO)
+		return ERR_PTR(-ESTALE);
+
+	root = nilfs_lookup_root(sb->s_fs_info, cno);
+	if (!root)
+		return ERR_PTR(-ESTALE);
+
+	inode = nilfs_iget(sb, root, ino);
+	nilfs_put_root(root);
+
+	if (IS_ERR(inode))
+		return ERR_CAST(inode);
+	if (gen && inode->i_generation != gen) {
+		iput(inode);
+		return ERR_PTR(-ESTALE);
+	}
+	return d_obtain_alias(inode);
+}
+
+static struct dentry *nilfs_fh_to_dentry(struct super_block *sb, struct fid *fh,
+					 int fh_len, int fh_type)
+{
+	struct nilfs_fid *fid = (struct nilfs_fid *)fh;
+
+	if (fh_len < NILFS_FID_SIZE_NON_CONNECTABLE ||
+	    (fh_type != FILEID_NILFS_WITH_PARENT &&
+	     fh_type != FILEID_NILFS_WITHOUT_PARENT))
+		return NULL;
+
+	return nilfs_get_dentry(sb, fid->cno, fid->ino, fid->gen);
+}
+
+static struct dentry *nilfs_fh_to_parent(struct super_block *sb, struct fid *fh,
+					 int fh_len, int fh_type)
+{
+	struct nilfs_fid *fid = (struct nilfs_fid *)fh;
+
+	if (fh_len < NILFS_FID_SIZE_CONNECTABLE ||
+	    fh_type != FILEID_NILFS_WITH_PARENT)
+		return NULL;
+
+	return nilfs_get_dentry(sb, fid->cno, fid->parent_ino, fid->parent_gen);
+}
+
+static int nilfs_encode_fh(struct inode *inode, __u32 *fh, int *lenp,
+			   struct inode *parent)
+{
+	struct nilfs_fid *fid = (struct nilfs_fid *)fh;
+	struct nilfs_root *root = NILFS_I(inode)->i_root;
+	int type;
+
+	if (parent && *lenp < NILFS_FID_SIZE_CONNECTABLE) {
+		*lenp = NILFS_FID_SIZE_CONNECTABLE;
+		return FILEID_INVALID;
+	}
+	if (*lenp < NILFS_FID_SIZE_NON_CONNECTABLE) {
+		*lenp = NILFS_FID_SIZE_NON_CONNECTABLE;
+		return FILEID_INVALID;
+	}
+
+	fid->cno = root->cno;
+	fid->ino = inode->i_ino;
+	fid->gen = inode->i_generation;
+
+	if (parent) {
+		fid->parent_ino = parent->i_ino;
+		fid->parent_gen = parent->i_generation;
+		type = FILEID_NILFS_WITH_PARENT;
+		*lenp = NILFS_FID_SIZE_CONNECTABLE;
+	} else {
+		type = FILEID_NILFS_WITHOUT_PARENT;
+		*lenp = NILFS_FID_SIZE_NON_CONNECTABLE;
+	}
+
+	return type;
+}
+
+const struct inode_operations nilfs_dir_inode_operations = {
+	.create		= nilfs_create,
+	.lookup		= nilfs_lookup,
+	.link		= nilfs_link,
+	.unlink		= nilfs_unlink,
+	.symlink	= nilfs_symlink,
+	.mkdir		= nilfs_mkdir,
+	.rmdir		= nilfs_rmdir,
+	.mknod		= nilfs_mknod,
+	.rename		= nilfs_rename,
+	.setattr	= nilfs_setattr,
+	.permission	= nilfs_permission,
+	.fiemap		= nilfs_fiemap,
+	.fileattr_get	= nilfs_fileattr_get,
+	.fileattr_set	= nilfs_fileattr_set,
+};
+
+const struct inode_operations nilfs_special_inode_operations = {
+	.setattr	= nilfs_setattr,
+	.permission	= nilfs_permission,
+};
+
+const struct inode_operations nilfs_symlink_inode_operations = {
+	.get_link	= page_get_link,
+	.permission     = nilfs_permission,
+};
+
+const struct export_operations nilfs_export_ops = {
+	.encode_fh = nilfs_encode_fh,
+	.fh_to_dentry = nilfs_fh_to_dentry,
+	.fh_to_parent = nilfs_fh_to_parent,
+	.get_parent = nilfs_get_parent,
+};