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/strncpy_from_user.c | 156 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 156 insertions(+) create mode 100644 lib/strncpy_from_user.c (limited to 'lib/strncpy_from_user.c') diff --git a/lib/strncpy_from_user.c b/lib/strncpy_from_user.c new file mode 100644 index 000000000..5bb752ff7 --- /dev/null +++ b/lib/strncpy_from_user.c @@ -0,0 +1,156 @@ +// SPDX-License-Identifier: GPL-2.0 +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#ifdef CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS +#define IS_UNALIGNED(src, dst) 0 +#else +#define IS_UNALIGNED(src, dst) \ + (((long) dst | (long) src) & (sizeof(long) - 1)) +#endif + +/* + * Do a strncpy, return length of string without final '\0'. + * 'count' is the user-supplied count (return 'count' if we + * hit it), 'max' is the address space maximum (and we return + * -EFAULT if we hit it). + */ +static __always_inline long do_strncpy_from_user(char *dst, const char __user *src, + unsigned long count, unsigned long max) +{ + const struct word_at_a_time constants = WORD_AT_A_TIME_CONSTANTS; + unsigned long res = 0; + + if (IS_UNALIGNED(src, dst)) + goto byte_at_a_time; + + while (max >= sizeof(unsigned long)) { + unsigned long c, data, mask; + + /* Fall back to byte-at-a-time if we get a page fault */ + unsafe_get_user(c, (unsigned long __user *)(src+res), byte_at_a_time); + + /* + * Note that we mask out the bytes following the NUL. This is + * important to do because string oblivious code may read past + * the NUL. For those routines, we don't want to give them + * potentially random bytes after the NUL in `src`. + * + * One example of such code is BPF map keys. BPF treats map keys + * as an opaque set of bytes. Without the post-NUL mask, any BPF + * maps keyed by strings returned from strncpy_from_user() may + * have multiple entries for semantically identical strings. + */ + if (has_zero(c, &data, &constants)) { + data = prep_zero_mask(c, data, &constants); + data = create_zero_mask(data); + mask = zero_bytemask(data); + *(unsigned long *)(dst+res) = c & mask; + return res + find_zero(data); + } + + *(unsigned long *)(dst+res) = c; + + res += sizeof(unsigned long); + max -= sizeof(unsigned long); + } + +byte_at_a_time: + while (max) { + char c; + + unsafe_get_user(c,src+res, efault); + dst[res] = c; + if (!c) + return res; + res++; + max--; + } + + /* + * Uhhuh. We hit 'max'. But was that the user-specified maximum + * too? If so, that's ok - we got as much as the user asked for. + */ + if (res >= count) + return res; + + /* + * Nope: we hit the address space limit, and we still had more + * characters the caller would have wanted. That's an EFAULT. + */ +efault: + return -EFAULT; +} + +/** + * strncpy_from_user: - Copy a NUL terminated string from userspace. + * @dst: Destination address, in kernel space. This buffer must be at + * least @count bytes long. + * @src: Source address, in user space. + * @count: Maximum number of bytes to copy, including the trailing NUL. + * + * Copies a NUL-terminated string from userspace to kernel space. + * + * On success, returns the length of the string (not including the trailing + * NUL). + * + * If access to userspace fails, returns -EFAULT (some data may have been + * copied). + * + * If @count is smaller than the length of the string, copies @count bytes + * and returns @count. + */ +long strncpy_from_user(char *dst, const char __user *src, long count) +{ + unsigned long max_addr, src_addr; + + might_fault(); + if (should_fail_usercopy()) + return -EFAULT; + if (unlikely(count <= 0)) + return 0; + + kasan_check_write(dst, count); + check_object_size(dst, count, false); + + if (can_do_masked_user_access()) { + long retval; + + src = masked_user_read_access_begin(src); + retval = do_strncpy_from_user(dst, src, count, count); + user_read_access_end(); + return retval; + } + + max_addr = TASK_SIZE_MAX; + src_addr = (unsigned long)untagged_addr(src); + if (likely(src_addr < max_addr)) { + unsigned long max = max_addr - src_addr; + long retval; + + /* + * Truncate 'max' to the user-specified limit, so that + * we only have one limit we need to check in the loop + */ + if (max > count) + max = count; + + if (user_read_access_begin(src, max)) { + retval = do_strncpy_from_user(dst, src, count, max); + user_read_access_end(); + return retval; + } + } + return -EFAULT; +} +EXPORT_SYMBOL(strncpy_from_user); -- cgit v1.3.1