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/affs/inode.c | 417 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 417 insertions(+) create mode 100644 fs/affs/inode.c (limited to 'fs/affs/inode.c') diff --git a/fs/affs/inode.c b/fs/affs/inode.c new file mode 100644 index 000000000..d4a3f381c --- /dev/null +++ b/fs/affs/inode.c @@ -0,0 +1,417 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * linux/fs/affs/inode.c + * + * (c) 1996 Hans-Joachim Widmaier - Rewritten + * + * (C) 1993 Ray Burr - Modified for Amiga FFS filesystem. + * + * (C) 1992 Eric Youngdale Modified for ISO9660 filesystem. + * + * (C) 1991 Linus Torvalds - minix filesystem + */ +#include +#include +#include +#include "affs.h" + +struct inode *affs_iget(struct super_block *sb, unsigned long ino) +{ + struct affs_sb_info *sbi = AFFS_SB(sb); + struct buffer_head *bh; + struct affs_tail *tail; + struct inode *inode; + u32 block; + u32 size; + u32 prot; + u16 id; + + inode = iget_locked(sb, ino); + if (!inode) + return ERR_PTR(-ENOMEM); + if (!(inode_state_read_once(inode) & I_NEW)) + return inode; + + pr_debug("affs_iget(%llu)\n", inode->i_ino); + + block = inode->i_ino; + bh = affs_bread(sb, block); + if (!bh) { + affs_warning(sb, "read_inode", "Cannot read block %d", block); + goto bad_inode; + } + if (affs_checksum_block(sb, bh) || be32_to_cpu(AFFS_HEAD(bh)->ptype) != T_SHORT) { + affs_warning(sb,"read_inode", + "Checksum or type (ptype=%d) error on inode %d", + AFFS_HEAD(bh)->ptype, block); + goto bad_inode; + } + + tail = AFFS_TAIL(sb, bh); + prot = be32_to_cpu(tail->protect); + + inode->i_size = 0; + set_nlink(inode, 1); + inode->i_mode = 0; + AFFS_I(inode)->i_extcnt = 1; + AFFS_I(inode)->i_ext_last = ~1; + AFFS_I(inode)->i_protect = prot; + atomic_set(&AFFS_I(inode)->i_opencnt, 0); + AFFS_I(inode)->i_blkcnt = 0; + AFFS_I(inode)->i_lc = NULL; + AFFS_I(inode)->i_lc_size = 0; + AFFS_I(inode)->i_lc_shift = 0; + AFFS_I(inode)->i_lc_mask = 0; + AFFS_I(inode)->i_ac = NULL; + AFFS_I(inode)->i_ext_bh = NULL; + AFFS_I(inode)->mmu_private = 0; + AFFS_I(inode)->i_lastalloc = 0; + AFFS_I(inode)->i_pa_cnt = 0; + + if (affs_test_opt(sbi->s_flags, SF_SETMODE)) + inode->i_mode = sbi->s_mode; + else + inode->i_mode = affs_prot_to_mode(prot); + + id = be16_to_cpu(tail->uid); + if (id == 0 || affs_test_opt(sbi->s_flags, SF_SETUID)) + inode->i_uid = sbi->s_uid; + else if (id == 0xFFFF && affs_test_opt(sbi->s_flags, SF_MUFS)) + i_uid_write(inode, 0); + else + i_uid_write(inode, id); + + id = be16_to_cpu(tail->gid); + if (id == 0 || affs_test_opt(sbi->s_flags, SF_SETGID)) + inode->i_gid = sbi->s_gid; + else if (id == 0xFFFF && affs_test_opt(sbi->s_flags, SF_MUFS)) + i_gid_write(inode, 0); + else + i_gid_write(inode, id); + + switch (be32_to_cpu(tail->stype)) { + case ST_ROOT: + inode->i_uid = sbi->s_uid; + inode->i_gid = sbi->s_gid; + fallthrough; + case ST_USERDIR: + if (be32_to_cpu(tail->stype) == ST_USERDIR || + affs_test_opt(sbi->s_flags, SF_SETMODE)) { + if (inode->i_mode & S_IRUSR) + inode->i_mode |= S_IXUSR; + if (inode->i_mode & S_IRGRP) + inode->i_mode |= S_IXGRP; + if (inode->i_mode & S_IROTH) + inode->i_mode |= S_IXOTH; + inode->i_mode |= S_IFDIR; + } else + inode->i_mode = S_IRUGO | S_IXUGO | S_IWUSR | S_IFDIR; + /* Maybe it should be controlled by mount parameter? */ + //inode->i_mode |= S_ISVTX; + inode->i_op = &affs_dir_inode_operations; + inode->i_fop = &affs_dir_operations; + break; + case ST_LINKDIR: +#if 0 + affs_warning(sb, "read_inode", "inode is LINKDIR"); + goto bad_inode; +#else + inode->i_mode |= S_IFDIR; + /* ... and leave ->i_op and ->i_fop pointing to empty */ + break; +#endif + case ST_LINKFILE: + affs_warning(sb, "read_inode", "inode is LINKFILE"); + goto bad_inode; + case ST_FILE: + size = be32_to_cpu(tail->size); + inode->i_mode |= S_IFREG; + AFFS_I(inode)->mmu_private = inode->i_size = size; + if (inode->i_size) { + AFFS_I(inode)->i_blkcnt = (size - 1) / + sbi->s_data_blksize + 1; + AFFS_I(inode)->i_extcnt = (AFFS_I(inode)->i_blkcnt - 1) / + sbi->s_hashsize + 1; + } + if (tail->link_chain) + set_nlink(inode, 2); + inode->i_mapping->a_ops = affs_test_opt(sbi->s_flags, SF_OFS) ? + &affs_aops_ofs : &affs_aops; + inode->i_op = &affs_file_inode_operations; + inode->i_fop = &affs_file_operations; + break; + case ST_SOFTLINK: + inode->i_size = strlen((char *)AFFS_HEAD(bh)->table); + inode->i_mode |= S_IFLNK; + inode_nohighmem(inode); + inode->i_op = &affs_symlink_inode_operations; + inode->i_data.a_ops = &affs_symlink_aops; + break; + } + + inode_set_mtime(inode, + inode_set_atime(inode, inode_set_ctime(inode, (be32_to_cpu(tail->change.days) * 86400LL + be32_to_cpu(tail->change.mins) * 60 + be32_to_cpu(tail->change.ticks) / 50 + AFFS_EPOCH_DELTA) + sys_tz.tz_minuteswest * 60, 0).tv_sec, 0).tv_sec, + 0); + affs_brelse(bh); + unlock_new_inode(inode); + return inode; + +bad_inode: + affs_brelse(bh); + iget_failed(inode); + return ERR_PTR(-EIO); +} + +int +affs_write_inode(struct inode *inode, struct writeback_control *wbc) +{ + struct super_block *sb = inode->i_sb; + struct buffer_head *bh; + struct affs_tail *tail; + uid_t uid; + gid_t gid; + + pr_debug("write_inode(%llu)\n", inode->i_ino); + + if (!inode->i_nlink) + // possibly free block + return 0; + bh = affs_bread(sb, inode->i_ino); + if (!bh) { + affs_error(sb, "write_inode", "Cannot read block %llu", inode->i_ino); + return -EIO; + } + tail = AFFS_TAIL(sb, bh); + if (tail->stype == cpu_to_be32(ST_ROOT)) { + affs_secs_to_datestamp(inode_get_mtime_sec(inode), + &AFFS_ROOT_TAIL(sb, bh)->root_change); + } else { + tail->protect = cpu_to_be32(AFFS_I(inode)->i_protect); + tail->size = cpu_to_be32(inode->i_size); + affs_secs_to_datestamp(inode_get_mtime_sec(inode), + &tail->change); + if (!(inode->i_ino == AFFS_SB(sb)->s_root_block)) { + uid = i_uid_read(inode); + gid = i_gid_read(inode); + if (affs_test_opt(AFFS_SB(sb)->s_flags, SF_MUFS)) { + if (uid == 0 || uid == 0xFFFF) + uid = uid ^ ~0; + if (gid == 0 || gid == 0xFFFF) + gid = gid ^ ~0; + } + if (!affs_test_opt(AFFS_SB(sb)->s_flags, SF_SETUID)) + tail->uid = cpu_to_be16(uid); + if (!affs_test_opt(AFFS_SB(sb)->s_flags, SF_SETGID)) + tail->gid = cpu_to_be16(gid); + } + } + affs_fix_checksum(sb, bh); + mark_buffer_dirty(bh); + affs_brelse(bh); + affs_free_prealloc(inode); + return 0; +} + +int +affs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, struct iattr *attr) +{ + struct inode *inode = d_inode(dentry); + int error; + + pr_debug("setattr(%llu,0x%x)\n", inode->i_ino, attr->ia_valid); + + error = setattr_prepare(&nop_mnt_idmap, dentry, attr); + if (error) + goto out; + + if (((attr->ia_valid & ATTR_UID) && + affs_test_opt(AFFS_SB(inode->i_sb)->s_flags, SF_SETUID)) || + ((attr->ia_valid & ATTR_GID) && + affs_test_opt(AFFS_SB(inode->i_sb)->s_flags, SF_SETGID)) || + ((attr->ia_valid & ATTR_MODE) && + (AFFS_SB(inode->i_sb)->s_flags & + (AFFS_MOUNT_SF_SETMODE | AFFS_MOUNT_SF_IMMUTABLE)))) { + if (!affs_test_opt(AFFS_SB(inode->i_sb)->s_flags, SF_QUIET)) + error = -EPERM; + goto out; + } + + if ((attr->ia_valid & ATTR_SIZE) && + attr->ia_size != i_size_read(inode)) { + error = inode_newsize_ok(inode, attr->ia_size); + if (error) + return error; + + truncate_setsize(inode, attr->ia_size); + affs_truncate(inode); + } + + setattr_copy(&nop_mnt_idmap, inode, attr); + mark_inode_dirty(inode); + + if (attr->ia_valid & ATTR_MODE) + affs_mode_to_prot(inode); +out: + return error; +} + +void +affs_evict_inode(struct inode *inode) +{ + unsigned long cache_page; + pr_debug("evict_inode(ino=%llu, nlink=%u)\n", + inode->i_ino, inode->i_nlink); + truncate_inode_pages_final(&inode->i_data); + + if (!inode->i_nlink) { + inode->i_size = 0; + affs_truncate(inode); + } + + clear_inode(inode); + affs_free_prealloc(inode); + cache_page = (unsigned long)AFFS_I(inode)->i_lc; + if (cache_page) { + pr_debug("freeing ext cache\n"); + AFFS_I(inode)->i_lc = NULL; + AFFS_I(inode)->i_ac = NULL; + free_page(cache_page); + } + affs_brelse(AFFS_I(inode)->i_ext_bh); + AFFS_I(inode)->i_ext_last = ~1; + AFFS_I(inode)->i_ext_bh = NULL; + + if (!inode->i_nlink) + affs_free_block(inode->i_sb, inode->i_ino); +} + +struct inode * +affs_new_inode(struct inode *dir) +{ + struct super_block *sb = dir->i_sb; + struct inode *inode; + u32 block; + struct buffer_head *bh; + + if (!(inode = new_inode(sb))) + goto err_inode; + + if (!(block = affs_alloc_block(dir, dir->i_ino))) + goto err_block; + + bh = affs_getzeroblk(sb, block); + if (!bh) + goto err_bh; + mark_buffer_dirty(bh); + affs_brelse(bh); + + inode->i_uid = current_fsuid(); + inode->i_gid = current_fsgid(); + inode->i_ino = block; + set_nlink(inode, 1); + simple_inode_init_ts(inode); + atomic_set(&AFFS_I(inode)->i_opencnt, 0); + AFFS_I(inode)->i_blkcnt = 0; + AFFS_I(inode)->i_lc = NULL; + AFFS_I(inode)->i_lc_size = 0; + AFFS_I(inode)->i_lc_shift = 0; + AFFS_I(inode)->i_lc_mask = 0; + AFFS_I(inode)->i_ac = NULL; + AFFS_I(inode)->i_ext_bh = NULL; + AFFS_I(inode)->mmu_private = 0; + AFFS_I(inode)->i_protect = 0; + AFFS_I(inode)->i_lastalloc = 0; + AFFS_I(inode)->i_pa_cnt = 0; + AFFS_I(inode)->i_extcnt = 1; + AFFS_I(inode)->i_ext_last = ~1; + + insert_inode_hash(inode); + + return inode; + +err_bh: + affs_free_block(sb, block); +err_block: + iput(inode); +err_inode: + return NULL; +} + +/* + * Add an entry to a directory. Create the header block + * and insert it into the hash table. + */ + +int +affs_add_entry(struct inode *dir, struct inode *inode, struct dentry *dentry, s32 type) +{ + struct super_block *sb = dir->i_sb; + struct buffer_head *inode_bh = NULL; + struct buffer_head *bh; + u32 block = 0; + int retval; + + pr_debug("%s(dir=%llu, inode=%llu, \"%pd\", type=%d)\n", __func__, + dir->i_ino, inode->i_ino, dentry, type); + + retval = -EIO; + bh = affs_bread(sb, inode->i_ino); + if (!bh) + goto done; + + affs_lock_link(inode); + switch (type) { + case ST_LINKFILE: + case ST_LINKDIR: + retval = -ENOSPC; + block = affs_alloc_block(dir, dir->i_ino); + if (!block) + goto err; + retval = -EIO; + inode_bh = bh; + bh = affs_getzeroblk(sb, block); + if (!bh) + goto err; + break; + default: + break; + } + + AFFS_HEAD(bh)->ptype = cpu_to_be32(T_SHORT); + AFFS_HEAD(bh)->key = cpu_to_be32(bh->b_blocknr); + affs_copy_name(AFFS_TAIL(sb, bh)->name, dentry); + AFFS_TAIL(sb, bh)->stype = cpu_to_be32(type); + AFFS_TAIL(sb, bh)->parent = cpu_to_be32(dir->i_ino); + + if (inode_bh) { + __be32 chain; + chain = AFFS_TAIL(sb, inode_bh)->link_chain; + AFFS_TAIL(sb, bh)->original = cpu_to_be32(inode->i_ino); + AFFS_TAIL(sb, bh)->link_chain = chain; + AFFS_TAIL(sb, inode_bh)->link_chain = cpu_to_be32(block); + affs_adjust_checksum(inode_bh, block - be32_to_cpu(chain)); + mark_buffer_dirty(inode_bh); + set_nlink(inode, 2); + ihold(inode); + } + affs_fix_checksum(sb, bh); + mark_buffer_dirty(bh); + dentry->d_fsdata = (void *)(long)bh->b_blocknr; + + affs_lock_dir(dir); + retval = affs_insert_hash(dir, bh); + mark_buffer_dirty(bh); + affs_unlock_dir(dir); + affs_unlock_link(inode); + + d_instantiate(dentry, inode); +done: + affs_brelse(inode_bh); + affs_brelse(bh); + return retval; +err: + if (block) + affs_free_block(sb, block); + affs_unlock_link(inode); + goto done; +} -- cgit v1.3.1