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 --- lib/oid_registry.c | 175 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 175 insertions(+) create mode 100644 lib/oid_registry.c (limited to 'lib/oid_registry.c') diff --git a/lib/oid_registry.c b/lib/oid_registry.c new file mode 100644 index 000000000..9b757a117 --- /dev/null +++ b/lib/oid_registry.c @@ -0,0 +1,175 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* ASN.1 Object identifier (OID) registry + * + * Copyright (C) 2012 Red Hat, Inc. All Rights Reserved. + * Written by David Howells (dhowells@redhat.com) + */ + +#include +#include +#include +#include +#include +#include +#include +#include "oid_registry_data.c" + +MODULE_DESCRIPTION("OID Registry"); +MODULE_AUTHOR("Red Hat, Inc."); +MODULE_LICENSE("GPL"); + +/** + * look_up_OID - Find an OID registration for the specified data + * @data: Binary representation of the OID + * @datasize: Size of the binary representation + */ +enum OID look_up_OID(const void *data, size_t datasize) +{ + const unsigned char *octets = data; + enum OID oid; + unsigned char xhash; + unsigned i, j, k, hash; + size_t len; + + /* Hash the OID data */ + hash = datasize - 1; + + for (i = 0; i < datasize; i++) + hash += octets[i] * 33; + hash = (hash >> 24) ^ (hash >> 16) ^ (hash >> 8) ^ hash; + hash &= 0xff; + + /* Binary search the OID registry. OIDs are stored in ascending order + * of hash value then ascending order of size and then in ascending + * order of reverse value. + */ + i = 0; + k = OID__NR; + while (i < k) { + j = (i + k) / 2; + + xhash = oid_search_table[j].hash; + if (xhash > hash) { + k = j; + continue; + } + if (xhash < hash) { + i = j + 1; + continue; + } + + oid = oid_search_table[j].oid; + len = oid_index[oid + 1] - oid_index[oid]; + if (len > datasize) { + k = j; + continue; + } + if (len < datasize) { + i = j + 1; + continue; + } + + /* Variation is most likely to be at the tail end of the + * OID, so do the comparison in reverse. + */ + while (len > 0) { + unsigned char a = oid_data[oid_index[oid] + --len]; + unsigned char b = octets[len]; + if (a > b) { + k = j; + goto next; + } + if (a < b) { + i = j + 1; + goto next; + } + } + return oid; + next: + ; + } + + return OID__NR; +} +EXPORT_SYMBOL_GPL(look_up_OID); + +/** + * parse_OID - Parse an OID from a bytestream + * @data: Binary representation of the header + OID + * @datasize: Size of the binary representation + * @oid: Pointer to oid to return result + * + * Parse an OID from a bytestream that holds the OID in the format + * ASN1_OID | length | oid. The length indicator must equal to datasize - 2. + * -EBADMSG is returned if the bytestream is too short. + */ +int parse_OID(const void *data, size_t datasize, enum OID *oid) +{ + const unsigned char *v = data; + + /* we need 2 bytes of header and at least 1 byte for oid */ + if (datasize < 3 || v[0] != ASN1_OID || v[1] != datasize - 2) + return -EBADMSG; + + *oid = look_up_OID(data + 2, datasize - 2); + return 0; +} +EXPORT_SYMBOL_GPL(parse_OID); + +/* + * sprint_oid - Print an Object Identifier into a buffer + * @data: The encoded OID to print + * @datasize: The size of the encoded OID + * @buffer: The buffer to render into + * @bufsize: The size of the buffer + * + * The OID is rendered into the buffer in "a.b.c.d" format and the number of + * bytes is returned. -EBADMSG is returned if the data could not be interpreted + * and -ENOBUFS if the buffer was too small. + */ +int sprint_oid(const void *data, size_t datasize, char *buffer, size_t bufsize) +{ + const unsigned char *v = data, *end = v + datasize; + unsigned long num; + unsigned char n; + size_t ret; + int count; + + if (v >= end) + goto bad; + + n = *v++; + ret = count = snprintf(buffer, bufsize, "%u.%u", n / 40, n % 40); + if (count >= bufsize) + return -ENOBUFS; + buffer += count; + bufsize -= count; + + while (v < end) { + n = *v++; + if (!(n & 0x80)) { + num = n; + } else { + num = n & 0x7f; + do { + if (v >= end) + goto bad; + n = *v++; + num <<= 7; + num |= n & 0x7f; + } while (n & 0x80); + } + ret += count = snprintf(buffer, bufsize, ".%lu", num); + if (count >= bufsize) + return -ENOBUFS; + buffer += count; + bufsize -= count; + } + + return ret; + +bad: + snprintf(buffer, bufsize, "(bad)"); + return -EBADMSG; +} +EXPORT_SYMBOL_GPL(sprint_oid); -- cgit v1.3.1