From a5f7a5bb3b7f28ba7e4fa246775b29a0e5537255 Mon Sep 17 00:00:00 2001 From: Jinke Han Date: Tue, 8 Sep 2026 15:37:42 +0800 Subject: x86/kprobes: Fix crash when probing CS CALL instructions 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) Signed-off-by: Jinke Han Signed-off-by: Ingo Molnar Reviewed-by: Masami Hiramatsu (Google) Acked-by: Yafang Shao Acked-by: Borislav Petkov Cc: Peter Zijlstra Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000 --- fs/ntfs/object_id.c | 158 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 158 insertions(+) create mode 100644 fs/ntfs/object_id.c (limited to 'fs/ntfs/object_id.c') diff --git a/fs/ntfs/object_id.c b/fs/ntfs/object_id.c new file mode 100644 index 000000000..89002c375 --- /dev/null +++ b/fs/ntfs/object_id.c @@ -0,0 +1,158 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * Pocessing of object ids + * + * Part of this file is based on code from the NTFS-3G. + * + * Copyright (c) 2009-2019 Jean-Pierre Andre + * Copyright (c) 2026 LG Electronics Co., Ltd. + */ + +#include "ntfs.h" +#include "index.h" +#include "object_id.h" + +struct object_id_index_key { + union { + u32 alignment; + struct guid guid; + } object_id; +} __packed; + +struct object_id_index_data { + __le64 file_id; + struct guid birth_volume_id; + struct guid birth_object_id; + struct guid domain_id; +} __packed; + +/* Index entry in $Extend/$ObjId */ +struct object_id_index { + struct index_entry_header header; + struct object_id_index_key key; + struct object_id_index_data data; +} __packed; + +__le16 objid_index_name[] = {cpu_to_le16('$'), cpu_to_le16('O'), 0}; + +/* + * open_object_id_index - Open the $Extend/$ObjId file and its index + * @vol: NTFS volume structure + * + * Opens the $ObjId system file and retrieves its index context. + * + * Return: The index context if opened successfully, or NULL if an error + * occurred. + */ +static struct ntfs_index_context *open_object_id_index(struct ntfs_volume *vol) +{ + struct inode *dir_vi, *vi; + struct ntfs_inode *dir_ni; + struct ntfs_index_context *xo = NULL; + struct ntfs_name *name = NULL; + u64 mref; + int uname_len; + __le16 *uname; + + uname_len = ntfs_nlstoucs(vol, "$ObjId", 6, &uname, + NTFS_MAX_NAME_LEN); + if (uname_len < 0) + return NULL; + + /* do not use path_name_to inode - could reopen root */ + dir_vi = ntfs_iget(vol->sb, FILE_Extend); + if (IS_ERR(dir_vi)) { + kmem_cache_free(ntfs_name_cache, uname); + return NULL; + } + dir_ni = NTFS_I(dir_vi); + + mutex_lock_nested(&dir_ni->mrec_lock, NTFS_EXTEND_MUTEX_PARENT); + mref = ntfs_lookup_inode_by_name(dir_ni, uname, uname_len, &name); + mutex_unlock(&dir_ni->mrec_lock); + kfree(name); + kmem_cache_free(ntfs_name_cache, uname); + if (IS_ERR_MREF(mref)) + goto put_dir_vi; + + vi = ntfs_iget(vol->sb, MREF(mref)); + if (IS_ERR(vi)) + goto put_dir_vi; + + xo = ntfs_index_ctx_get(NTFS_I(vi), objid_index_name, 2); + if (!xo) + iput(vi); +put_dir_vi: + iput(dir_vi); + return xo; +} + + +/* + * remove_object_id_index - Remove an object id index entry if attribute present + * @ni: NTFS inode structure containing the attribute + * @xo: Index context for the object id index + * + * Reads the existing object ID attribute and removes it from the index. + * + * Return: 0 on success, or a negative error code on failure. + */ +static int remove_object_id_index(struct ntfs_inode *ni, struct ntfs_index_context *xo) +{ + struct object_id_index_key key = {0}; + s64 size; + + if (ni->data_size == 0) + return -ENODATA; + + /* read the existing object id attribute */ + size = ntfs_inode_attr_pread(VFS_I(ni), 0, sizeof(struct guid), + (char *)&key); + if (size != sizeof(struct guid)) + return -ENODATA; + + if (!ntfs_index_lookup(&key, sizeof(struct object_id_index_key), xo)) + return ntfs_index_rm(xo); + + return 0; +} + +/* + * ntfs_delete_object_id_index - Delete an object_id index entry + * @ni: NTFS inode structure + * + * Opens the object ID index and removes the entry corresponding to the inode. + * + * Return: 0 on success, or a negative error code on failure. + */ +int ntfs_delete_object_id_index(struct ntfs_inode *ni) +{ + struct ntfs_index_context *xo; + struct ntfs_inode *xoni; + struct inode *attr_vi; + int ret = 0; + + attr_vi = ntfs_attr_iget(VFS_I(ni), AT_OBJECT_ID, AT_UNNAMED, 0); + if (IS_ERR(attr_vi)) + return PTR_ERR(attr_vi); + + /* + * read the existing object id and un-index it + */ + xo = open_object_id_index(ni->vol); + if (xo) { + xoni = xo->idx_ni; + mutex_lock_nested(&xoni->mrec_lock, NTFS_EXTEND_MUTEX_PARENT); + ret = remove_object_id_index(NTFS_I(attr_vi), xo); + if (!ret) { + ntfs_index_entry_mark_dirty(xo); + mark_mft_record_dirty(xoni); + } + ntfs_index_ctx_put(xo); + mutex_unlock(&xoni->mrec_lock); + iput(VFS_I(xoni)); + } + + iput(attr_vi); + return ret; +} -- cgit v1.3.1