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, 168 insertions, 0 deletions

diff --git a/drivers/usb/host/ehci-sysfs.c b/drivers/usb/host/ehci-sysfs.c
new file mode 100644
index 000000000..5e6b545c3
--- /dev/null
+++ b/drivers/usb/host/ehci-sysfs.c
@@ -0,0 +1,168 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2007 by Alan Stern
+ */
+
+/* this file is part of ehci-hcd.c */
+
+
+/* Display the ports dedicated to the companion controller */
+static ssize_t companion_show(struct device *dev,
+			      struct device_attribute *attr,
+			      char *buf)
+{
+	struct ehci_hcd		*ehci;
+	int			nports, index;
+	int			len = 0;
+
+	ehci = hcd_to_ehci(dev_get_drvdata(dev));
+	nports = HCS_N_PORTS(ehci->hcs_params);
+
+	for (index = 0; index < nports; ++index) {
+		if (test_bit(index, &ehci->companion_ports))
+			len += sysfs_emit_at(buf, len, "%d\n", index + 1);
+	}
+	return len;
+}
+
+/*
+ * Dedicate or undedicate a port to the companion controller.
+ * Syntax is "[-]portnum", where a leading '-' sign means
+ * return control of the port to the EHCI controller.
+ */
+static ssize_t companion_store(struct device *dev,
+			       struct device_attribute *attr,
+			       const char *buf, size_t count)
+{
+	struct ehci_hcd		*ehci;
+	int			portnum, new_owner;
+
+	ehci = hcd_to_ehci(dev_get_drvdata(dev));
+	new_owner = PORT_OWNER;		/* Owned by companion */
+	if (sscanf(buf, "%d", &portnum) != 1)
+		return -EINVAL;
+	if (portnum < 0) {
+		portnum = - portnum;
+		new_owner = 0;		/* Owned by EHCI */
+	}
+	if (portnum <= 0 || portnum > HCS_N_PORTS(ehci->hcs_params))
+		return -ENOENT;
+	portnum--;
+	if (new_owner)
+		set_bit(portnum, &ehci->companion_ports);
+	else
+		clear_bit(portnum, &ehci->companion_ports);
+	set_owner(ehci, portnum, new_owner);
+	return count;
+}
+static DEVICE_ATTR_RW(companion);
+
+
+/*
+ * Display / Set uframe_periodic_max
+ */
+static ssize_t uframe_periodic_max_show(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct ehci_hcd		*ehci;
+
+	ehci = hcd_to_ehci(dev_get_drvdata(dev));
+	return sysfs_emit(buf, "%d\n", ehci->uframe_periodic_max);
+}
+
+
+static ssize_t uframe_periodic_max_store(struct device *dev,
+					struct device_attribute *attr,
+					const char *buf, size_t count)
+{
+	struct ehci_hcd		*ehci;
+	unsigned		uframe_periodic_max;
+	unsigned		uframe;
+	unsigned long		flags;
+	ssize_t			ret;
+
+	ehci = hcd_to_ehci(dev_get_drvdata(dev));
+	if (kstrtouint(buf, 0, &uframe_periodic_max) < 0)
+		return -EINVAL;
+
+	if (uframe_periodic_max < 100 || uframe_periodic_max >= 125) {
+		ehci_info(ehci, "rejecting invalid request for "
+				"uframe_periodic_max=%u\n", uframe_periodic_max);
+		return -EINVAL;
+	}
+
+	ret = -EINVAL;
+
+	/*
+	 * lock, so that our checking does not race with possible periodic
+	 * bandwidth allocation through submitting new urbs.
+	 */
+	spin_lock_irqsave (&ehci->lock, flags);
+
+	/*
+	 * for request to decrease max periodic bandwidth, we have to check
+	 * to see whether the decrease is possible.
+	 */
+	if (uframe_periodic_max < ehci->uframe_periodic_max) {
+		u8		allocated_max = 0;
+
+		for (uframe = 0; uframe < EHCI_BANDWIDTH_SIZE; ++uframe)
+			allocated_max = max(allocated_max,
+					ehci->bandwidth[uframe]);
+
+		if (allocated_max > uframe_periodic_max) {
+			ehci_info(ehci,
+				"cannot decrease uframe_periodic_max because "
+				"periodic bandwidth is already allocated "
+				"(%u > %u)\n",
+				allocated_max, uframe_periodic_max);
+			goto out_unlock;
+		}
+	}
+
+	/* increasing is always ok */
+
+	ehci_info(ehci, "setting max periodic bandwidth to %u%% "
+			"(== %u usec/uframe)\n",
+			100*uframe_periodic_max/125, uframe_periodic_max);
+
+	if (uframe_periodic_max != 100)
+		ehci_warn(ehci, "max periodic bandwidth set is non-standard\n");
+
+	ehci->uframe_periodic_max = uframe_periodic_max;
+	ret = count;
+
+out_unlock:
+	spin_unlock_irqrestore (&ehci->lock, flags);
+	return ret;
+}
+static DEVICE_ATTR_RW(uframe_periodic_max);
+
+
+static inline int create_sysfs_files(struct ehci_hcd *ehci)
+{
+	struct device	*controller = ehci_to_hcd(ehci)->self.controller;
+	int	i = 0;
+
+	/* with integrated TT there is no companion! */
+	if (!ehci_is_TDI(ehci))
+		i = device_create_file(controller, &dev_attr_companion);
+	if (i)
+		goto out;
+
+	i = device_create_file(controller, &dev_attr_uframe_periodic_max);
+out:
+	return i;
+}
+
+static inline void remove_sysfs_files(struct ehci_hcd *ehci)
+{
+	struct device	*controller = ehci_to_hcd(ehci)->self.controller;
+
+	/* with integrated TT there is no companion! */
+	if (!ehci_is_TDI(ehci))
+		device_remove_file(controller, &dev_attr_companion);
+
+	device_remove_file(controller, &dev_attr_uframe_periodic_max);
+}