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, 140 insertions, 0 deletions

diff --git a/drivers/parisc/wax.c b/drivers/parisc/wax.c
new file mode 100644
index 000000000..a82a5e88e
--- /dev/null
+++ b/drivers/parisc/wax.c
@@ -0,0 +1,140 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ *	WAX Device Driver
+ *
+ *	(c) Copyright 2000 The Puffin Group Inc.
+ *
+ *	(c) 2000-2023 by Helge Deller <deller@gmx.de>
+ */
+
+#include <linux/errno.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/ioport.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/types.h>
+
+#include <asm/io.h>
+#include <asm/hardware.h>
+
+#include "gsc.h"
+
+#define WAX_GSC_IRQ	7	/* Hardcoded Interrupt for GSC */
+
+static void wax_choose_irq(struct parisc_device *dev, void *ctrl)
+{
+	int irq;
+
+	switch (dev->id.sversion) {
+		case 0x73:	irq =  1; break; /* i8042 General */
+		case 0x8c:	irq =  6; break; /* Serial */
+		case 0x90:	irq = 10; break; /* EISA */
+		default:	return;		 /* Unknown */
+	}
+
+	gsc_asic_assign_irq(ctrl, irq, &dev->irq);
+
+	switch (dev->id.sversion) {
+		case 0x73:	irq =  2; break; /* i8042 High-priority */
+		case 0x90:	irq =  0; break; /* EISA NMI */
+		default:	return;		 /* No secondary IRQ */
+	}
+
+	gsc_asic_assign_irq(ctrl, irq, &dev->aux_irq);
+}
+
+static void __init
+wax_init_irq(struct gsc_asic *wax)
+{
+	unsigned long base = wax->hpa;
+
+	/* Wax-off */
+	gsc_writel(0x00000000, base+OFFSET_IMR);
+
+	/* clear pending interrupts */
+	gsc_readl(base+OFFSET_IRR);
+
+	/* We're not really convinced we want to reset the onboard
+         * devices. Firmware does it for us...
+	 */
+
+	/* Resets */
+//	gsc_writel(0xFFFFFFFF, base+0x1000); /* HIL */
+//	gsc_writel(0xFFFFFFFF, base+0x2000); /* RS232-B on Wax */
+}
+
+static int __init wax_init_chip(struct parisc_device *dev)
+{
+	struct gsc_asic *wax;
+	struct parisc_device *parent;
+	int ret;
+
+	wax = kzalloc_obj(*wax);
+	if (!wax)
+		return -ENOMEM;
+
+	wax->name = "wax";
+	wax->hpa = dev->hpa.start;
+
+	wax->version = 0;   /* gsc_readb(wax->hpa+WAX_VER); */
+	printk(KERN_INFO "%s at 0x%lx found.\n", wax->name, wax->hpa);
+
+	/* Stop wax hissing for a bit */
+	wax_init_irq(wax);
+
+	/* the IRQ wax should use */
+	dev->irq = gsc_claim_irq(&wax->gsc_irq, WAX_GSC_IRQ);
+	if (dev->irq < 0) {
+		printk(KERN_ERR "%s(): cannot get GSC irq\n",
+				__func__);
+		kfree(wax);
+		return -EBUSY;
+	}
+
+	wax->eim = ((u32) wax->gsc_irq.txn_addr) | wax->gsc_irq.txn_data;
+
+	ret = request_irq(wax->gsc_irq.irq, gsc_asic_intr, 0, "wax", wax);
+	if (ret < 0) {
+		kfree(wax);
+		return ret;
+	}
+
+	/* enable IRQ's for devices below WAX */
+	gsc_writel(wax->eim, wax->hpa + OFFSET_IAR);
+
+	/* Done init'ing, register this driver */
+	ret = gsc_common_setup(dev, wax);
+	if (ret) {
+		kfree(wax);
+		return ret;
+	}
+
+	gsc_fixup_irqs(dev, wax, wax_choose_irq);
+	/* On 715-class machines, Wax EISA is a sibling of Wax, not a child. */
+	parent = parisc_parent(dev);
+	if (parent->id.hw_type != HPHW_IOA) {
+		gsc_fixup_irqs(parent, wax, wax_choose_irq);
+	}
+
+	return ret;
+}
+
+static const struct parisc_device_id wax_tbl[] __initconst = {
+	{ HPHW_BA, HVERSION_REV_ANY_ID, HVERSION_ANY_ID, 0x0008e },
+	{ 0, }
+};
+
+MODULE_DEVICE_TABLE(parisc, wax_tbl);
+
+static struct parisc_driver wax_driver __refdata = {
+	.name =		"wax",
+	.id_table =	wax_tbl,
+	.probe =	wax_init_chip,
+};
+
+static int __init wax_init(void)
+{
+	return register_parisc_driver(&wax_driver);
+}
+arch_initcall(wax_init);