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/debug.c | 171 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 171 insertions(+) create mode 100644 fs/ntfs/debug.c (limited to 'fs/ntfs/debug.c') diff --git a/fs/ntfs/debug.c b/fs/ntfs/debug.c new file mode 100644 index 000000000..f241ebc97 --- /dev/null +++ b/fs/ntfs/debug.c @@ -0,0 +1,171 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * NTFS kernel debug support. + * + * Copyright (c) 2001-2004 Anton Altaparmakov + */ +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt +#include "debug.h" + +/* + * __ntfs_warning - output a warning to the syslog + * @function: name of function outputting the warning + * @sb: super block of mounted ntfs filesystem + * @fmt: warning string containing format specifications + * @...: a variable number of arguments specified in @fmt + * + * Outputs a warning to the syslog for the mounted ntfs filesystem described + * by @sb. + * + * @fmt and the corresponding @... is printf style format string containing + * the warning string and the corresponding format arguments, respectively. + * + * @function is the name of the function from which __ntfs_warning is being + * called. + * + * Note, you should be using debug.h::ntfs_warning(@sb, @fmt, @...) instead + * as this provides the @function parameter automatically. + */ +void __ntfs_warning(const char *function, const struct super_block *sb, + const char *fmt, ...) +{ + struct va_format vaf; + va_list args; + int flen = 0; + + if (function) + flen = strlen(function); + va_start(args, fmt); + vaf.fmt = fmt; + vaf.va = &args; +#ifdef DEBUG + if (sb) + pr_warn("(device %s): %s(): %pV\n", + sb->s_id, flen ? function : "", &vaf); + else + pr_warn("%s(): %pV\n", flen ? function : "", &vaf); +#else + if (sb) + pr_warn_ratelimited("(device %s): %s(): %pV\n", + sb->s_id, flen ? function : "", &vaf); + else + pr_warn_ratelimited("%s(): %pV\n", flen ? function : "", &vaf); +#endif + va_end(args); +} + +/* + * __ntfs_error - output an error to the syslog + * @function: name of function outputting the error + * @sb: super block of mounted ntfs filesystem + * @fmt: error string containing format specifications + * @...: a variable number of arguments specified in @fmt + * + * Outputs an error to the syslog for the mounted ntfs filesystem described + * by @sb. + * + * @fmt and the corresponding @... is printf style format string containing + * the error string and the corresponding format arguments, respectively. + * + * @function is the name of the function from which __ntfs_error is being + * called. + * + * Note, you should be using debug.h::ntfs_error(@sb, @fmt, @...) instead + * as this provides the @function parameter automatically. + */ +void __ntfs_error(const char *function, struct super_block *sb, + const char *fmt, ...) +{ + struct va_format vaf; + va_list args; + int flen = 0; + + if (function) + flen = strlen(function); + va_start(args, fmt); + vaf.fmt = fmt; + vaf.va = &args; +#ifdef DEBUG + if (sb) + pr_err("(device %s): %s(): %pV\n", + sb->s_id, flen ? function : "", &vaf); + else + pr_err("%s(): %pV\n", flen ? function : "", &vaf); +#else + if (sb) + pr_err_ratelimited("(device %s): %s(): %pV\n", + sb->s_id, flen ? function : "", &vaf); + else + pr_err_ratelimited("%s(): %pV\n", flen ? function : "", &vaf); +#endif + va_end(args); + + if (sb) + ntfs_handle_error(sb); +} + +#ifdef DEBUG + +/* If 1, output debug messages, and if 0, don't. */ +int debug_msgs; + +void __ntfs_debug(const char *file, int line, const char *function, + const char *fmt, ...) +{ + struct va_format vaf; + va_list args; + int flen = 0; + + if (!debug_msgs) + return; + if (function) + flen = strlen(function); + va_start(args, fmt); + vaf.fmt = fmt; + vaf.va = &args; + pr_debug("(%s, %d): %s(): %pV", file, line, flen ? function : "", &vaf); + va_end(args); +} + +/* Dump a runlist. Caller has to provide synchronisation for @rl. */ +void ntfs_debug_dump_runlist(const struct runlist_element *rl) +{ + int i; + const char *lcn_str[5] = { "LCN_DELALLOC ", "LCN_HOLE ", + "LCN_RL_NOT_MAPPED", "LCN_ENOENT ", + "LCN_unknown " }; + + if (!debug_msgs) + return; + pr_debug("Dumping runlist (values in hex):\n"); + if (!rl) { + pr_debug("Run list not present.\n"); + return; + } + pr_debug("VCN LCN Run length\n"); + for (i = 0; ; i++) { + s64 lcn = (rl + i)->lcn; + + if (lcn < 0) { + int index = -lcn - 1; + + if (index > -LCN_ENOENT - 1) + index = 3; + pr_debug("%-16Lx %s %-16Lx%s\n", + (long long)(rl + i)->vcn, lcn_str[index], + (long long)(rl + i)->length, + (rl + i)->length ? "" : + " (runlist end)"); + } else + pr_debug("%-16Lx %-16Lx %-16Lx%s\n", + (long long)(rl + i)->vcn, + (long long)(rl + i)->lcn, + (long long)(rl + i)->length, + (rl + i)->length ? "" : + " (runlist end)"); + if (!(rl + i)->length) + break; + } +} + +#endif -- cgit v1.3.1