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, 165 insertions, 0 deletions

diff --git a/fs/jffs2/background.c b/fs/jffs2/background.c
new file mode 100644
index 000000000..bb0ee1a59
--- /dev/null
+++ b/fs/jffs2/background.c
@@ -0,0 +1,165 @@
+/*
+ * JFFS2 -- Journalling Flash File System, Version 2.
+ *
+ * Copyright © 2001-2007 Red Hat, Inc.
+ * Copyright © 2004-2010 David Woodhouse <dwmw2@infradead.org>
+ *
+ * Created by David Woodhouse <dwmw2@infradead.org>
+ *
+ * For licensing information, see the file 'LICENCE' in this directory.
+ *
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/kernel.h>
+#include <linux/jffs2.h>
+#include <linux/mtd/mtd.h>
+#include <linux/completion.h>
+#include <linux/sched/signal.h>
+#include <linux/freezer.h>
+#include <linux/kthread.h>
+#include "nodelist.h"
+
+
+static int jffs2_garbage_collect_thread(void *);
+
+void jffs2_garbage_collect_trigger(struct jffs2_sb_info *c)
+{
+	assert_spin_locked(&c->erase_completion_lock);
+	if (c->gc_task && jffs2_thread_should_wake(c))
+		send_sig(SIGHUP, c->gc_task, 1);
+}
+
+/* This must only ever be called when no GC thread is currently running */
+int jffs2_start_garbage_collect_thread(struct jffs2_sb_info *c)
+{
+	struct task_struct *tsk;
+	int ret = 0;
+
+	BUG_ON(c->gc_task);
+
+	init_completion(&c->gc_thread_start);
+	init_completion(&c->gc_thread_exit);
+
+	tsk = kthread_run(jffs2_garbage_collect_thread, c, "jffs2_gcd_mtd%d", c->mtd->index);
+	if (IS_ERR(tsk)) {
+		pr_warn("fork failed for JFFS2 garbage collect thread: %pe\n",
+			tsk);
+		complete(&c->gc_thread_exit);
+		ret = PTR_ERR(tsk);
+	} else {
+		/* Wait for it... */
+		jffs2_dbg(1, "Garbage collect thread is pid %d\n", tsk->pid);
+		wait_for_completion(&c->gc_thread_start);
+		ret = tsk->pid;
+	}
+
+	return ret;
+}
+
+void jffs2_stop_garbage_collect_thread(struct jffs2_sb_info *c)
+{
+	int wait = 0;
+	spin_lock(&c->erase_completion_lock);
+	if (c->gc_task) {
+		jffs2_dbg(1, "Killing GC task %d\n", c->gc_task->pid);
+		send_sig(SIGKILL, c->gc_task, 1);
+		wait = 1;
+	}
+	spin_unlock(&c->erase_completion_lock);
+	if (wait)
+		wait_for_completion(&c->gc_thread_exit);
+}
+
+static int jffs2_garbage_collect_thread(void *_c)
+{
+	struct jffs2_sb_info *c = _c;
+	sigset_t hupmask;
+
+	siginitset(&hupmask, sigmask(SIGHUP));
+	allow_signal(SIGKILL);
+	allow_signal(SIGSTOP);
+	allow_signal(SIGHUP);
+
+	c->gc_task = current;
+	complete(&c->gc_thread_start);
+
+	set_user_nice(current, 10);
+
+	set_freezable();
+	for (;;) {
+		sigprocmask(SIG_UNBLOCK, &hupmask, NULL);
+	again:
+		spin_lock(&c->erase_completion_lock);
+		if (!jffs2_thread_should_wake(c)) {
+			set_current_state (TASK_INTERRUPTIBLE);
+			spin_unlock(&c->erase_completion_lock);
+			jffs2_dbg(1, "%s(): sleeping...\n", __func__);
+			schedule();
+		} else {
+			spin_unlock(&c->erase_completion_lock);
+		}
+		/* Problem - immediately after bootup, the GCD spends a lot
+		 * of time in places like jffs2_kill_fragtree(); so much so
+		 * that userspace processes (like gdm and X) are starved
+		 * despite plenty of cond_resched()s and renicing.  Yield()
+		 * doesn't help, either (presumably because userspace and GCD
+		 * are generally competing for a higher latency resource -
+		 * disk).
+		 * This forces the GCD to slow the hell down.   Pulling an
+		 * inode in with read_inode() is much preferable to having
+		 * the GC thread get there first. */
+		schedule_timeout_interruptible(msecs_to_jiffies(50));
+
+		if (kthread_should_stop()) {
+			jffs2_dbg(1, "%s(): kthread_stop() called\n", __func__);
+			goto die;
+		}
+
+		/* Put_super will send a SIGKILL and then wait on the sem.
+		 */
+		while (signal_pending(current) || freezing(current)) {
+			unsigned long signr;
+
+			if (try_to_freeze())
+				goto again;
+
+			signr = kernel_dequeue_signal();
+
+			switch(signr) {
+			case SIGSTOP:
+				jffs2_dbg(1, "%s(): SIGSTOP received\n",
+					  __func__);
+				kernel_signal_stop();
+				break;
+
+			case SIGKILL:
+				jffs2_dbg(1, "%s(): SIGKILL received\n",
+					  __func__);
+				goto die;
+
+			case SIGHUP:
+				jffs2_dbg(1, "%s(): SIGHUP received\n",
+					  __func__);
+				break;
+			default:
+				jffs2_dbg(1, "%s(): signal %ld received\n",
+					  __func__, signr);
+			}
+		}
+		/* We don't want SIGHUP to interrupt us. STOP and KILL are OK though. */
+		sigprocmask(SIG_BLOCK, &hupmask, NULL);
+
+		jffs2_dbg(1, "%s(): pass\n", __func__);
+		if (jffs2_garbage_collect_pass(c) == -ENOSPC) {
+			pr_notice("No space for garbage collection. Aborting GC thread\n");
+			goto die;
+		}
+	}
+ die:
+	spin_lock(&c->erase_completion_lock);
+	c->gc_task = NULL;
+	spin_unlock(&c->erase_completion_lock);
+	kthread_complete_and_exit(&c->gc_thread_exit, 0);
+}