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/crypto/tests/test-utils.h | 111 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 111 insertions(+) create mode 100644 lib/crypto/tests/test-utils.h (limited to 'lib/crypto/tests/test-utils.h') diff --git a/lib/crypto/tests/test-utils.h b/lib/crypto/tests/test-utils.h new file mode 100644 index 000000000..73ca21f91 --- /dev/null +++ b/lib/crypto/tests/test-utils.h @@ -0,0 +1,111 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * Test utility functions shared by the crypto library tests. + * + * For now this is simply a header that's included into the KUnit test suites + * that need it. If this gets too large it could be made its own translation + * unit and libcrypto_test_utils module, but that seems overkill for now. + */ +#ifndef LIB_CRYPTO_TEST_UTILS_H +#define LIB_CRYPTO_TEST_UTILS_H + +#include +#include +#include +#include +#include + +static u64 random_seed; + +static __maybe_unused void action_free_guarded_buf(void *buf) +{ + vfree(buf); +} + +/* + * Allocate a KUnit-managed buffer that has length @size bytes (> 0) immediately + * followed by an unmapped page, and assert that the allocation succeeds. + */ +static __maybe_unused void *alloc_guarded_buf(struct kunit *test, size_t size) +{ + size_t full_size = round_up(size, PAGE_SIZE); + void *buf = vmalloc(full_size); + + KUNIT_ASSERT_NOT_NULL(test, buf); + KUNIT_ASSERT_EQ(test, 0, + kunit_add_action_or_reset(test, action_free_guarded_buf, + buf)); + return buf + full_size - size; +} + +static __maybe_unused void *alloc_buf(struct kunit *test, size_t size) +{ + void *buf = kunit_kmalloc(test, size, GFP_KERNEL); + + KUNIT_ASSERT_NOT_NULL(test, buf); + return buf; +} + +static __maybe_unused void *memdup_buf(struct kunit *test, const void *src, + size_t size) +{ + void *dst = alloc_buf(test, size); + + return memcpy(dst, src, size); +} + +/* + * This is a simple linear congruential generator. It is used only for testing, + * which does not require cryptographically secure random numbers. A hard-coded + * algorithm is used instead of so that it matches the + * algorithm used by the test vector generation script. This allows the input + * data in random test vectors to be concisely stored as just the seed. + */ +static __maybe_unused u32 rand32(void) +{ + random_seed = (random_seed * 25214903917 + 11) & ((1ULL << 48) - 1); + return random_seed >> 16; +} + +static __maybe_unused void rand_bytes(u8 *out, size_t len) +{ + for (size_t i = 0; i < len; i++) + out[i] = rand32(); +} + +static __maybe_unused void rand_bytes_seeded_from_len(u8 *out, size_t len) +{ + random_seed = len; + rand_bytes(out, len); +} + +static __maybe_unused bool rand_bool(void) +{ + return rand32() % 2; +} + +/* Generate a random length, preferring small lengths. */ +static __maybe_unused size_t rand_length(size_t max_len) +{ + size_t len; + + switch (rand32() % 3) { + case 0: + len = rand32() % 128; + break; + case 1: + len = rand32() % 3072; + break; + default: + len = rand32(); + break; + } + return len % (max_len + 1); +} + +static __maybe_unused size_t rand_offset(size_t max_offset) +{ + return min(rand32() % 128, max_offset); +} + +#endif /* LIB_CRYPTO_TEST_UTILS_H */ -- cgit v1.3.1