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, 215 insertions, 0 deletions

diff --git a/fs/nilfs2/ifile.c b/fs/nilfs2/ifile.c
new file mode 100644
index 000000000..c99e1e635
--- /dev/null
+++ b/fs/nilfs2/ifile.c
@@ -0,0 +1,215 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NILFS inode file
+ *
+ * Copyright (C) 2006-2008 Nippon Telegraph and Telephone Corporation.
+ *
+ * Written by Amagai Yoshiji.
+ * Revised by Ryusuke Konishi.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/buffer_head.h>
+#include "nilfs.h"
+#include "mdt.h"
+#include "alloc.h"
+#include "ifile.h"
+#include "cpfile.h"
+
+/**
+ * struct nilfs_ifile_info - on-memory private data of ifile
+ * @mi: on-memory private data of metadata file
+ * @palloc_cache: persistent object allocator cache of ifile
+ */
+struct nilfs_ifile_info {
+	struct nilfs_mdt_info mi;
+	struct nilfs_palloc_cache palloc_cache;
+};
+
+static inline struct nilfs_ifile_info *NILFS_IFILE_I(struct inode *ifile)
+{
+	return (struct nilfs_ifile_info *)NILFS_MDT(ifile);
+}
+
+/**
+ * nilfs_ifile_create_inode - create a new disk inode
+ * @ifile: ifile inode
+ * @out_ino: pointer to a variable to store inode number
+ * @out_bh: buffer_head contains newly allocated disk inode
+ *
+ * nilfs_ifile_create_inode() allocates a new inode in the ifile metadata
+ * file and stores the inode number in the variable pointed to by @out_ino,
+ * as well as storing the ifile's buffer with the disk inode in the location
+ * pointed to by @out_bh.
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOMEM	- Insufficient memory available.
+ * * %-ENOSPC	- No inode left.
+ */
+int nilfs_ifile_create_inode(struct inode *ifile, u64 *out_ino,
+			     struct buffer_head **out_bh)
+{
+	struct nilfs_palloc_req req;
+	int ret;
+
+	req.pr_entry_nr = NILFS_FIRST_INO(ifile->i_sb);
+	req.pr_entry_bh = NULL;
+
+	ret = nilfs_palloc_prepare_alloc_entry(ifile, &req, false);
+	if (!ret) {
+		ret = nilfs_palloc_get_entry_block(ifile, req.pr_entry_nr, 1,
+						   &req.pr_entry_bh);
+		if (ret < 0)
+			nilfs_palloc_abort_alloc_entry(ifile, &req);
+	}
+	if (ret < 0) {
+		brelse(req.pr_entry_bh);
+		return ret;
+	}
+	nilfs_palloc_commit_alloc_entry(ifile, &req);
+	mark_buffer_dirty(req.pr_entry_bh);
+	nilfs_mdt_mark_dirty(ifile);
+	*out_ino = req.pr_entry_nr;
+	*out_bh = req.pr_entry_bh;
+	return 0;
+}
+
+/**
+ * nilfs_ifile_delete_inode - delete a disk inode
+ * @ifile: ifile inode
+ * @ino: inode number
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EIO	- I/O error (including metadata corruption).
+ * * %-ENOENT	- Inode number unallocated.
+ * * %-ENOMEM	- Insufficient memory available.
+ */
+int nilfs_ifile_delete_inode(struct inode *ifile, u64 ino)
+{
+	struct nilfs_palloc_req req = {
+		.pr_entry_nr = ino, .pr_entry_bh = NULL
+	};
+	struct nilfs_inode *raw_inode;
+	size_t offset;
+	int ret;
+
+	ret = nilfs_palloc_prepare_free_entry(ifile, &req);
+	if (!ret) {
+		ret = nilfs_palloc_get_entry_block(ifile, req.pr_entry_nr, 0,
+						   &req.pr_entry_bh);
+		if (ret < 0)
+			nilfs_palloc_abort_free_entry(ifile, &req);
+	}
+	if (ret < 0) {
+		brelse(req.pr_entry_bh);
+		return ret;
+	}
+
+	offset = nilfs_palloc_entry_offset(ifile, req.pr_entry_nr,
+					   req.pr_entry_bh);
+	raw_inode = kmap_local_folio(req.pr_entry_bh->b_folio, offset);
+	raw_inode->i_flags = 0;
+	kunmap_local(raw_inode);
+
+	mark_buffer_dirty(req.pr_entry_bh);
+	brelse(req.pr_entry_bh);
+
+	nilfs_palloc_commit_free_entry(ifile, &req);
+
+	return 0;
+}
+
+int nilfs_ifile_get_inode_block(struct inode *ifile, u64 ino,
+				struct buffer_head **out_bh)
+{
+	struct super_block *sb = ifile->i_sb;
+	int err;
+
+	if (unlikely(!NILFS_VALID_INODE(sb, ino))) {
+		nilfs_error(sb, "bad inode number: %llu", ino);
+		return -EINVAL;
+	}
+
+	err = nilfs_palloc_get_entry_block(ifile, ino, 0, out_bh);
+	if (unlikely(err))
+		nilfs_warn(sb, "error %d reading inode: ino=%llu", err, ino);
+	return err;
+}
+
+/**
+ * nilfs_ifile_count_free_inodes - calculate free inodes count
+ * @ifile: ifile inode
+ * @nmaxinodes: current maximum of available inodes count [out]
+ * @nfreeinodes: free inodes count [out]
+ *
+ * Return: 0 on success, or a negative error code on failure.
+ */
+int nilfs_ifile_count_free_inodes(struct inode *ifile,
+				    u64 *nmaxinodes, u64 *nfreeinodes)
+{
+	u64 nused;
+	int err;
+
+	*nmaxinodes = 0;
+	*nfreeinodes = 0;
+
+	nused = atomic64_read(&NILFS_I(ifile)->i_root->inodes_count);
+	err = nilfs_palloc_count_max_entries(ifile, nused, nmaxinodes);
+	if (likely(!err))
+		*nfreeinodes = *nmaxinodes - nused;
+	return err;
+}
+
+/**
+ * nilfs_ifile_read - read or get ifile inode
+ * @sb: super block instance
+ * @root: root object
+ * @cno: number of checkpoint entry to read
+ * @inode_size: size of an inode
+ *
+ * Return: 0 on success, or one of the following negative error codes on
+ * failure:
+ * * %-EINVAL	- Invalid checkpoint.
+ * * %-ENOMEM	- Insufficient memory available.
+ * * %-EIO	- I/O error (including metadata corruption).
+ */
+int nilfs_ifile_read(struct super_block *sb, struct nilfs_root *root,
+		     __u64 cno, size_t inode_size)
+{
+	struct the_nilfs *nilfs;
+	struct inode *ifile;
+	int err;
+
+	ifile = nilfs_iget_locked(sb, root, NILFS_IFILE_INO);
+	if (unlikely(!ifile))
+		return -ENOMEM;
+	if (!(inode_state_read_once(ifile) & I_NEW))
+		goto out;
+
+	err = nilfs_mdt_init(ifile, NILFS_MDT_GFP,
+			     sizeof(struct nilfs_ifile_info));
+	if (err)
+		goto failed;
+
+	err = nilfs_palloc_init_blockgroup(ifile, inode_size);
+	if (err)
+		goto failed;
+
+	nilfs_palloc_setup_cache(ifile, &NILFS_IFILE_I(ifile)->palloc_cache);
+
+	nilfs = sb->s_fs_info;
+	err = nilfs_cpfile_read_checkpoint(nilfs->ns_cpfile, cno, root, ifile);
+	if (err)
+		goto failed;
+
+	unlock_new_inode(ifile);
+ out:
+	return 0;
+ failed:
+	iget_failed(ifile);
+	return err;
+}