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, 131 insertions, 0 deletions

diff --git a/drivers/comedi/range.c b/drivers/comedi/range.c
new file mode 100644
index 000000000..5b8f66236
--- /dev/null
+++ b/drivers/comedi/range.c
@@ -0,0 +1,131 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * comedi/range.c
+ * comedi routines for voltage ranges
+ *
+ * COMEDI - Linux Control and Measurement Device Interface
+ * Copyright (C) 1997-8 David A. Schleef <ds@schleef.org>
+ */
+
+#include <linux/uaccess.h>
+#include <linux/comedi/comedidev.h>
+#include "comedi_internal.h"
+
+const struct comedi_lrange range_bipolar10 = { 1, {BIP_RANGE(10)} };
+EXPORT_SYMBOL_GPL(range_bipolar10);
+const struct comedi_lrange range_bipolar5 = { 1, {BIP_RANGE(5)} };
+EXPORT_SYMBOL_GPL(range_bipolar5);
+const struct comedi_lrange range_bipolar2_5 = { 1, {BIP_RANGE(2.5)} };
+EXPORT_SYMBOL_GPL(range_bipolar2_5);
+const struct comedi_lrange range_unipolar10 = { 1, {UNI_RANGE(10)} };
+EXPORT_SYMBOL_GPL(range_unipolar10);
+const struct comedi_lrange range_unipolar5 = { 1, {UNI_RANGE(5)} };
+EXPORT_SYMBOL_GPL(range_unipolar5);
+const struct comedi_lrange range_unipolar2_5 = { 1, {UNI_RANGE(2.5)} };
+EXPORT_SYMBOL_GPL(range_unipolar2_5);
+const struct comedi_lrange range_0_20mA = { 1, {RANGE_mA(0, 20)} };
+EXPORT_SYMBOL_GPL(range_0_20mA);
+const struct comedi_lrange range_4_20mA = { 1, {RANGE_mA(4, 20)} };
+EXPORT_SYMBOL_GPL(range_4_20mA);
+const struct comedi_lrange range_0_32mA = { 1, {RANGE_mA(0, 32)} };
+EXPORT_SYMBOL_GPL(range_0_32mA);
+const struct comedi_lrange range_unknown = { 1, {{0, 1000000, UNIT_none} } };
+EXPORT_SYMBOL_GPL(range_unknown);
+
+/*
+ * COMEDI_RANGEINFO ioctl
+ * range information
+ *
+ * arg:
+ *	pointer to comedi_rangeinfo structure
+ *
+ * reads:
+ *	comedi_rangeinfo structure
+ *
+ * writes:
+ *	array of comedi_krange structures to rangeinfo->range_ptr pointer
+ */
+int do_rangeinfo_ioctl(struct comedi_device *dev,
+		       struct comedi_rangeinfo *it)
+{
+	int subd, chan;
+	const struct comedi_lrange *lr;
+	struct comedi_subdevice *s;
+
+	subd = (it->range_type >> 24) & 0xff;
+	chan = (it->range_type >> 16) & 0xff;
+
+	if (!dev->attached)
+		return -EINVAL;
+	if (subd >= dev->n_subdevices)
+		return -EINVAL;
+	s = &dev->subdevices[subd];
+	if (s->range_table) {
+		lr = s->range_table;
+	} else if (s->range_table_list) {
+		if (chan >= s->n_chan)
+			return -EINVAL;
+		lr = s->range_table_list[chan];
+	} else {
+		return -EINVAL;
+	}
+
+	if (RANGE_LENGTH(it->range_type) != lr->length) {
+		dev_dbg(dev->class_dev,
+			"wrong length %d should be %d (0x%08x)\n",
+			RANGE_LENGTH(it->range_type),
+			lr->length, it->range_type);
+		return -EINVAL;
+	}
+
+	if (copy_to_user(it->range_ptr, lr->range,
+			 sizeof(struct comedi_krange) * lr->length))
+		return -EFAULT;
+
+	return 0;
+}
+
+/**
+ * comedi_check_chanlist() - Validate each element in a chanlist.
+ * @s: comedi_subdevice struct
+ * @n: number of elements in the chanlist
+ * @chanlist: the chanlist to validate
+ *
+ * Each element consists of a channel number, a range index, an analog
+ * reference type and some flags, all packed into an unsigned int.
+ *
+ * This checks that the channel number and range index are supported by
+ * the comedi subdevice.  It does not check whether the analog reference
+ * type and the flags are supported.  Drivers that care should check those
+ * themselves.
+ *
+ * Return: %0 if all @chanlist elements are valid (success),
+ *         %-EINVAL if one or more elements are invalid.
+ */
+int comedi_check_chanlist(struct comedi_subdevice *s, int n,
+			  unsigned int *chanlist)
+{
+	struct comedi_device *dev = s->device;
+	unsigned int chanspec;
+	int chan, range_len, i;
+
+	for (i = 0; i < n; i++) {
+		chanspec = chanlist[i];
+		chan = CR_CHAN(chanspec);
+		if (s->range_table)
+			range_len = s->range_table->length;
+		else if (s->range_table_list && chan < s->n_chan)
+			range_len = s->range_table_list[chan]->length;
+		else
+			range_len = 0;
+		if (chan >= s->n_chan ||
+		    CR_RANGE(chanspec) >= range_len) {
+			dev_warn(dev->class_dev,
+				 "bad chanlist[%d]=0x%08x chan=%d range length=%d\n",
+				 i, chanspec, chan, range_len);
+			return -EINVAL;
+		}
+	}
+	return 0;
+}
+EXPORT_SYMBOL_GPL(comedi_check_chanlist);