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/dir.c | 180 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 180 insertions(+) create mode 100644 fs/affs/dir.c (limited to 'fs/affs/dir.c') diff --git a/fs/affs/dir.c b/fs/affs/dir.c new file mode 100644 index 000000000..d2b897ebb --- /dev/null +++ b/fs/affs/dir.c @@ -0,0 +1,180 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * linux/fs/affs/dir.c + * + * (c) 1996 Hans-Joachim Widmaier - Rewritten + * + * (C) 1993 Ray Burr - Modified for Amiga FFS filesystem. + * + * (C) 1992 Eric Youngdale Modified for ISO 9660 filesystem. + * + * (C) 1991 Linus Torvalds - minix filesystem + * + * affs directory handling functions + * + */ + +#include +#include +#include "affs.h" + +struct affs_dir_data { + unsigned long ino; + u64 cookie; +}; + +static int affs_readdir(struct file *, struct dir_context *); + +static loff_t affs_dir_llseek(struct file *file, loff_t offset, int whence) +{ + struct affs_dir_data *data = file->private_data; + + return generic_llseek_cookie(file, offset, whence, &data->cookie); +} + +static int affs_dir_open(struct inode *inode, struct file *file) +{ + struct affs_dir_data *data; + + data = kzalloc_obj(struct affs_dir_data); + if (!data) + return -ENOMEM; + file->private_data = data; + return 0; +} + +static int affs_dir_release(struct inode *inode, struct file *file) +{ + kfree(file->private_data); + return 0; +} + +const struct file_operations affs_dir_operations = { + .open = affs_dir_open, + .read = generic_read_dir, + .llseek = affs_dir_llseek, + .iterate_shared = affs_readdir, + .fsync = affs_file_fsync, + .release = affs_dir_release, + .setlease = generic_setlease, +}; + +/* + * directories can handle most operations... + */ +const struct inode_operations affs_dir_inode_operations = { + .create = affs_create, + .lookup = affs_lookup, + .link = affs_link, + .unlink = affs_unlink, + .symlink = affs_symlink, + .mkdir = affs_mkdir, + .rmdir = affs_rmdir, + .rename = affs_rename2, + .setattr = affs_setattr, +}; + +static int +affs_readdir(struct file *file, struct dir_context *ctx) +{ + struct inode *inode = file_inode(file); + struct affs_dir_data *data = file->private_data; + struct super_block *sb = inode->i_sb; + struct buffer_head *dir_bh = NULL; + struct buffer_head *fh_bh = NULL; + unsigned char *name; + int namelen; + u32 i; + int hash_pos; + int chain_pos; + u32 ino; + int error = 0; + + pr_debug("%s(ino=%llu,f_pos=%llx)\n", __func__, inode->i_ino, ctx->pos); + + if (ctx->pos < 2) { + data->ino = 0; + if (!dir_emit_dots(file, ctx)) + return 0; + } + + affs_lock_dir(inode); + chain_pos = (ctx->pos - 2) & 0xffff; + hash_pos = (ctx->pos - 2) >> 16; + if (chain_pos == 0xffff) { + affs_warning(sb, "readdir", "More than 65535 entries in chain"); + chain_pos = 0; + hash_pos++; + ctx->pos = ((hash_pos << 16) | chain_pos) + 2; + } + dir_bh = affs_bread(sb, inode->i_ino); + if (!dir_bh) + goto out_unlock_dir; + + /* If the directory hasn't changed since the last call to readdir(), + * we can jump directly to where we left off. + */ + ino = data->ino; + if (ino && inode_eq_iversion(inode, data->cookie)) { + pr_debug("readdir() left off=%d\n", ino); + goto inside; + } + if (hash_pos >= AFFS_SB(sb)->s_hashsize) + goto done; + + ino = be32_to_cpu(AFFS_HEAD(dir_bh)->table[hash_pos]); + for (i = 0; ino && i < chain_pos; i++) { + fh_bh = affs_bread(sb, ino); + if (!fh_bh) { + affs_error(sb, "readdir","Cannot read block %d", i); + error = -EIO; + goto out_brelse_dir; + } + ino = be32_to_cpu(AFFS_TAIL(sb, fh_bh)->hash_chain); + affs_brelse(fh_bh); + fh_bh = NULL; + } + if (ino) + goto inside; + hash_pos++; + + for (; hash_pos < AFFS_SB(sb)->s_hashsize; hash_pos++) { + ino = be32_to_cpu(AFFS_HEAD(dir_bh)->table[hash_pos]); + if (!ino) + continue; + ctx->pos = (hash_pos << 16) + 2; +inside: + do { + fh_bh = affs_bread(sb, ino); + if (!fh_bh) { + affs_error(sb, "readdir", + "Cannot read block %d", ino); + break; + } + + namelen = min(AFFS_TAIL(sb, fh_bh)->name[0], + (u8)AFFSNAMEMAX); + name = AFFS_TAIL(sb, fh_bh)->name + 1; + pr_debug("readdir(): dir_emit(\"%.*s\", ino=%u), hash=%d, f_pos=%llx\n", + namelen, name, ino, hash_pos, ctx->pos); + + if (!dir_emit(ctx, name, namelen, ino, DT_UNKNOWN)) + goto done; + ctx->pos++; + ino = be32_to_cpu(AFFS_TAIL(sb, fh_bh)->hash_chain); + affs_brelse(fh_bh); + fh_bh = NULL; + } while (ino); + } +done: + data->cookie = inode_query_iversion(inode); + data->ino = ino; + affs_brelse(fh_bh); + +out_brelse_dir: + affs_brelse(dir_bh); + +out_unlock_dir: + affs_unlock_dir(inode); + return error; +} -- cgit v1.3.1