x86/kprobes: Fix crash when probing CS CALL instructionsgrafted

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) <mhiramat@kernel.org>
Signed-off-by: Jinke Han <jinkehan@didiglobal.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Acked-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000

1 files changed, 351 insertions, 0 deletions

diff --git a/fs/signalfd.c b/fs/signalfd.c
new file mode 100644
index 000000000..dff53745e
--- /dev/null
+++ b/fs/signalfd.c
@@ -0,0 +1,351 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *  fs/signalfd.c
+ *
+ *  Copyright (C) 2003  Linus Torvalds
+ *
+ *  Mon Mar 5, 2007: Davide Libenzi <davidel@xmailserver.org>
+ *      Changed ->read() to return a siginfo strcture instead of signal number.
+ *      Fixed locking in ->poll().
+ *      Added sighand-detach notification.
+ *      Added fd re-use in sys_signalfd() syscall.
+ *      Now using anonymous inode source.
+ *      Thanks to Oleg Nesterov for useful code review and suggestions.
+ *      More comments and suggestions from Arnd Bergmann.
+ *  Sat May 19, 2007: Davi E. M. Arnaut <davi@haxent.com.br>
+ *      Retrieve multiple signals with one read() call
+ *  Sun Jul 15, 2007: Davide Libenzi <davidel@xmailserver.org>
+ *      Attach to the sighand only during read() and poll().
+ */
+
+#include <linux/file.h>
+#include <linux/poll.h>
+#include <linux/init.h>
+#include <linux/fs.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/kernel.h>
+#include <linux/signal.h>
+#include <linux/list.h>
+#include <linux/anon_inodes.h>
+#include <linux/signalfd.h>
+#include <linux/syscalls.h>
+#include <linux/proc_fs.h>
+#include <linux/compat.h>
+
+void signalfd_cleanup(struct sighand_struct *sighand)
+{
+	wake_up_pollfree(&sighand->signalfd_wqh);
+}
+
+struct signalfd_ctx {
+	sigset_t sigmask;
+};
+
+static int signalfd_release(struct inode *inode, struct file *file)
+{
+	kfree(file->private_data);
+	return 0;
+}
+
+static __poll_t signalfd_poll(struct file *file, poll_table *wait)
+{
+	struct signalfd_ctx *ctx = file->private_data;
+	__poll_t events = 0;
+
+	poll_wait(file, &current->sighand->signalfd_wqh, wait);
+
+	spin_lock_irq(&current->sighand->siglock);
+	if (next_signal(&current->pending, &ctx->sigmask) ||
+	    next_signal(&current->signal->shared_pending,
+			&ctx->sigmask))
+		events |= EPOLLIN;
+	spin_unlock_irq(&current->sighand->siglock);
+
+	return events;
+}
+
+/*
+ * Copied from copy_siginfo_to_user() in kernel/signal.c
+ */
+static int signalfd_copyinfo(struct iov_iter *to, kernel_siginfo_t const *kinfo)
+{
+	struct signalfd_siginfo new;
+
+	BUILD_BUG_ON(sizeof(struct signalfd_siginfo) != 128);
+
+	/*
+	 * Unused members should be zero ...
+	 */
+	memset(&new, 0, sizeof(new));
+
+	/*
+	 * If you change siginfo_t structure, please be sure
+	 * this code is fixed accordingly.
+	 */
+	new.ssi_signo = kinfo->si_signo;
+	new.ssi_errno = kinfo->si_errno;
+	new.ssi_code  = kinfo->si_code;
+	switch (siginfo_layout(kinfo->si_signo, kinfo->si_code)) {
+	case SIL_KILL:
+		new.ssi_pid = kinfo->si_pid;
+		new.ssi_uid = kinfo->si_uid;
+		break;
+	case SIL_TIMER:
+		new.ssi_tid = kinfo->si_tid;
+		new.ssi_overrun = kinfo->si_overrun;
+		new.ssi_ptr = (long) kinfo->si_ptr;
+		new.ssi_int = kinfo->si_int;
+		break;
+	case SIL_POLL:
+		new.ssi_band = kinfo->si_band;
+		new.ssi_fd   = kinfo->si_fd;
+		break;
+	case SIL_FAULT_BNDERR:
+	case SIL_FAULT_PKUERR:
+	case SIL_FAULT_PERF_EVENT:
+		/*
+		 * Fall through to the SIL_FAULT case.  SIL_FAULT_BNDERR,
+		 * SIL_FAULT_PKUERR, and SIL_FAULT_PERF_EVENT are only
+		 * generated by faults that deliver them synchronously to
+		 * userspace.  In case someone injects one of these signals
+		 * and signalfd catches it treat it as SIL_FAULT.
+		 */
+	case SIL_FAULT:
+		new.ssi_addr = (long) kinfo->si_addr;
+		break;
+	case SIL_FAULT_TRAPNO:
+		new.ssi_addr = (long) kinfo->si_addr;
+		new.ssi_trapno = kinfo->si_trapno;
+		break;
+	case SIL_FAULT_MCEERR:
+		new.ssi_addr = (long) kinfo->si_addr;
+		new.ssi_addr_lsb = (short) kinfo->si_addr_lsb;
+		break;
+	case SIL_CHLD:
+		new.ssi_pid    = kinfo->si_pid;
+		new.ssi_uid    = kinfo->si_uid;
+		new.ssi_status = kinfo->si_status;
+		new.ssi_utime  = kinfo->si_utime;
+		new.ssi_stime  = kinfo->si_stime;
+		break;
+	case SIL_RT:
+		/*
+		 * This case catches also the signals queued by sigqueue().
+		 */
+		new.ssi_pid = kinfo->si_pid;
+		new.ssi_uid = kinfo->si_uid;
+		new.ssi_ptr = (long) kinfo->si_ptr;
+		new.ssi_int = kinfo->si_int;
+		break;
+	case SIL_SYS:
+		new.ssi_call_addr = (long) kinfo->si_call_addr;
+		new.ssi_syscall   = kinfo->si_syscall;
+		new.ssi_arch      = kinfo->si_arch;
+		break;
+	}
+
+	if (!copy_to_iter_full(&new, sizeof(struct signalfd_siginfo), to))
+		return -EFAULT;
+
+	return sizeof(struct signalfd_siginfo);
+}
+
+static ssize_t signalfd_dequeue(struct signalfd_ctx *ctx, kernel_siginfo_t *info,
+				int nonblock)
+{
+	enum pid_type type;
+	ssize_t ret;
+	DECLARE_WAITQUEUE(wait, current);
+
+	spin_lock_irq(&current->sighand->siglock);
+	ret = dequeue_signal(&ctx->sigmask, info, &type);
+	switch (ret) {
+	case 0:
+		if (!nonblock)
+			break;
+		ret = -EAGAIN;
+		fallthrough;
+	default:
+		spin_unlock_irq(&current->sighand->siglock);
+		return ret;
+	}
+
+	add_wait_queue(&current->sighand->signalfd_wqh, &wait);
+	for (;;) {
+		set_current_state(TASK_INTERRUPTIBLE);
+		ret = dequeue_signal(&ctx->sigmask, info, &type);
+		if (ret != 0)
+			break;
+		if (signal_pending(current)) {
+			ret = -ERESTARTSYS;
+			break;
+		}
+		spin_unlock_irq(&current->sighand->siglock);
+		schedule();
+		spin_lock_irq(&current->sighand->siglock);
+	}
+	spin_unlock_irq(&current->sighand->siglock);
+
+	remove_wait_queue(&current->sighand->signalfd_wqh, &wait);
+	__set_current_state(TASK_RUNNING);
+
+	return ret;
+}
+
+/*
+ * Returns a multiple of the size of a "struct signalfd_siginfo", or a negative
+ * error code. The "count" parameter must be at least the size of a
+ * "struct signalfd_siginfo".
+ */
+static ssize_t signalfd_read_iter(struct kiocb *iocb, struct iov_iter *to)
+{
+	struct file *file = iocb->ki_filp;
+	struct signalfd_ctx *ctx = file->private_data;
+	size_t count = iov_iter_count(to);
+	ssize_t ret, total = 0;
+	kernel_siginfo_t info;
+	bool nonblock;
+
+	count /= sizeof(struct signalfd_siginfo);
+	if (!count)
+		return -EINVAL;
+
+	nonblock = file->f_flags & O_NONBLOCK || iocb->ki_flags & IOCB_NOWAIT;
+	do {
+		ret = signalfd_dequeue(ctx, &info, nonblock);
+		if (unlikely(ret <= 0))
+			break;
+		ret = signalfd_copyinfo(to, &info);
+		if (ret < 0)
+			break;
+		total += ret;
+		nonblock = 1;
+	} while (--count);
+
+	return total ? total: ret;
+}
+
+#ifdef CONFIG_PROC_FS
+static void signalfd_show_fdinfo(struct seq_file *m, struct file *f)
+{
+	struct signalfd_ctx *ctx = f->private_data;
+	sigset_t sigmask;
+
+	sigmask = ctx->sigmask;
+	signotset(&sigmask);
+	render_sigset_t(m, "sigmask:\t", &sigmask);
+}
+#endif
+
+static const struct file_operations signalfd_fops = {
+#ifdef CONFIG_PROC_FS
+	.show_fdinfo	= signalfd_show_fdinfo,
+#endif
+	.release	= signalfd_release,
+	.poll		= signalfd_poll,
+	.read_iter	= signalfd_read_iter,
+	.llseek		= noop_llseek,
+};
+
+static int do_signalfd4(int ufd, sigset_t *mask, int flags)
+{
+	/* Check the SFD_* constants for consistency.  */
+	BUILD_BUG_ON(SFD_CLOEXEC != O_CLOEXEC);
+	BUILD_BUG_ON(SFD_NONBLOCK != O_NONBLOCK);
+
+	if (flags & ~(SFD_CLOEXEC | SFD_NONBLOCK))
+		return -EINVAL;
+
+	sigdelsetmask(mask, sigmask(SIGKILL) | sigmask(SIGSTOP));
+	signotset(mask);
+
+	if (ufd == -1) {
+		int fd;
+		struct signalfd_ctx *ctx __free(kfree) = NULL;
+
+		ctx = kmalloc_obj(*ctx);
+		if (!ctx)
+			return -ENOMEM;
+
+		ctx->sigmask = *mask;
+
+		fd = FD_ADD(flags & O_CLOEXEC,
+			    anon_inode_getfile_fmode(
+				    "[signalfd]", &signalfd_fops, ctx,
+				    O_RDWR | (flags & O_NONBLOCK), FMODE_NOWAIT));
+		if (fd >= 0)
+			retain_and_null_ptr(ctx);
+		return fd;
+	} else {
+		struct signalfd_ctx *ctx;
+
+		CLASS(fd, f)(ufd);
+		if (fd_empty(f))
+			return -EBADF;
+		ctx = fd_file(f)->private_data;
+		if (fd_file(f)->f_op != &signalfd_fops)
+			return -EINVAL;
+		spin_lock_irq(&current->sighand->siglock);
+		ctx->sigmask = *mask;
+		spin_unlock_irq(&current->sighand->siglock);
+
+		wake_up(&current->sighand->signalfd_wqh);
+	}
+
+	return ufd;
+}
+
+SYSCALL_DEFINE4(signalfd4, int, ufd, sigset_t __user *, user_mask,
+		size_t, sizemask, int, flags)
+{
+	sigset_t mask;
+
+	if (sizemask != sizeof(sigset_t))
+		return -EINVAL;
+	if (copy_from_user(&mask, user_mask, sizeof(mask)))
+		return -EFAULT;
+	return do_signalfd4(ufd, &mask, flags);
+}
+
+SYSCALL_DEFINE3(signalfd, int, ufd, sigset_t __user *, user_mask,
+		size_t, sizemask)
+{
+	sigset_t mask;
+
+	if (sizemask != sizeof(sigset_t))
+		return -EINVAL;
+	if (copy_from_user(&mask, user_mask, sizeof(mask)))
+		return -EFAULT;
+	return do_signalfd4(ufd, &mask, 0);
+}
+
+#ifdef CONFIG_COMPAT
+static long do_compat_signalfd4(int ufd,
+			const compat_sigset_t __user *user_mask,
+			compat_size_t sigsetsize, int flags)
+{
+	sigset_t mask;
+
+	if (sigsetsize != sizeof(compat_sigset_t))
+		return -EINVAL;
+	if (get_compat_sigset(&mask, user_mask))
+		return -EFAULT;
+	return do_signalfd4(ufd, &mask, flags);
+}
+
+COMPAT_SYSCALL_DEFINE4(signalfd4, int, ufd,
+		     const compat_sigset_t __user *, user_mask,
+		     compat_size_t, sigsetsize,
+		     int, flags)
+{
+	return do_compat_signalfd4(ufd, user_mask, sigsetsize, flags);
+}
+
+COMPAT_SYSCALL_DEFINE3(signalfd, int, ufd,
+		     const compat_sigset_t __user *, user_mask,
+		     compat_size_t, sigsetsize)
+{
+	return do_compat_signalfd4(ufd, user_mask, sigsetsize, 0);
+}
+#endif