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, 239 insertions, 0 deletions

diff --git a/drivers/parisc/lasi.c b/drivers/parisc/lasi.c
new file mode 100644
index 000000000..a5b80cd5c
--- /dev/null
+++ b/drivers/parisc/lasi.c
@@ -0,0 +1,239 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ *	LASI Device Driver
+ *
+ *	(c) Copyright 1999 Red Hat Software
+ *	Portions (c) Copyright 1999 The Puffin Group Inc.
+ *	Portions (c) Copyright 1999 Hewlett-Packard
+ *
+ *	by Alan Cox <alan@redhat.com> and 
+ * 	   Alex deVries <alex@onefishtwo.ca>
+ */
+
+#include <linux/errno.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/pm.h>
+#include <linux/types.h>
+#include <linux/reboot.h>
+
+#include <asm/io.h>
+#include <asm/hardware.h>
+#include <asm/led.h>
+
+#include "gsc.h"
+
+
+#define LASI_VER	0xC008	/* LASI Version */
+
+#define LASI_IO_CONF	0x7FFFE	/* LASI primary configuration register */
+#define LASI_IO_CONF2	0x7FFFF	/* LASI secondary configuration register */
+
+static void lasi_choose_irq(struct parisc_device *dev, void *ctrl)
+{
+	int irq;
+
+	switch (dev->id.sversion) {
+		case 0x74:	irq =  7; break; /* Centronics */
+		case 0x7B:	irq = 13; break; /* Audio */
+		case 0x81:	irq = 14; break; /* Lasi itself */
+		case 0x82:	irq =  9; break; /* SCSI */
+		case 0x83:	irq = 20; break; /* Floppy */
+		case 0x84:	irq = 26; break; /* PS/2 Keyboard */
+		case 0x87:	irq = 18; break; /* ISDN */
+		case 0x8A:	irq =  8; break; /* LAN */
+		case 0x8C:	irq =  5; break; /* RS232 */
+		case 0x8D:	irq = (dev->hw_path == 13) ? 16 : 17; break;
+						 /* Telephone */
+		default: 	return;		 /* unknown */
+	}
+
+	gsc_asic_assign_irq(ctrl, irq, &dev->irq);
+}
+
+static void __init
+lasi_init_irq(struct gsc_asic *this_lasi)
+{
+	unsigned long lasi_base = this_lasi->hpa;
+
+	/* Stop LASI barking for a bit */
+	gsc_writel(0x00000000, lasi_base+OFFSET_IMR);
+
+	/* clear pending interrupts */
+	gsc_readl(lasi_base+OFFSET_IRR);
+
+	/* We're not really convinced we want to reset the onboard
+         * devices. Firmware does it for us...
+	 */
+
+	/* Resets */
+	/* gsc_writel(0xFFFFFFFF, lasi_base+0x2000);*/	/* Parallel */
+	if(pdc_add_valid(lasi_base+0x4004) == PDC_OK)
+		gsc_writel(0xFFFFFFFF, lasi_base+0x4004);	/* Audio */
+	/* gsc_writel(0xFFFFFFFF, lasi_base+0x5000);*/	/* Serial */ 
+	/* gsc_writel(0xFFFFFFFF, lasi_base+0x6000);*/	/* SCSI */
+	gsc_writel(0xFFFFFFFF, lasi_base+0x7000);	/* LAN */
+	gsc_writel(0xFFFFFFFF, lasi_base+0x8000);	/* Keyboard */
+	gsc_writel(0xFFFFFFFF, lasi_base+0xA000);	/* FDC */
+	
+	/* Ok we hit it on the head with a hammer, our Dog is now
+	** comatose and muzzled.  Devices will now unmask LASI
+	** interrupts as they are registered as irq's in the LASI range.
+	*/
+	/* XXX: I thought it was `awks that got `it on the `ead with an
+	 * `ammer.  -- willy
+	 */
+}
+
+
+/*
+   ** lasi_led_init()
+   ** 
+   ** lasi_led_init() initializes the LED controller on the LASI.
+   **
+   ** Since Mirage and Electra machines use a different LED
+   ** address register, we need to check for these machines 
+   ** explicitly.
+ */
+
+#ifndef CONFIG_CHASSIS_LCD_LED
+
+#define lasi_led_init(x)	/* nothing */
+
+#else
+
+static void __init lasi_led_init(unsigned long lasi_hpa)
+{
+	unsigned long datareg;
+
+	switch (CPU_HVERSION) {
+	/* Gecko machines have only one single LED, which can be permanently 
+	   turned on by writing a zero into the power control register. */ 
+	case 0x600:		/* Gecko (712/60) */
+	case 0x601:		/* Gecko (712/80) */
+	case 0x602:		/* Gecko (712/100) */
+	case 0x603:		/* Anole 64 (743/64) */
+	case 0x604:		/* Anole 100 (743/100) */
+	case 0x605:		/* Gecko (712/120) */
+		datareg = lasi_hpa + 0x0000C000;
+		gsc_writeb(0, datareg);
+		return; /* no need to register the LED interrupt-function */  
+
+	/* Mirage and Electra machines need special offsets */
+	case 0x60A:		/* Mirage Jr (715/64) */
+	case 0x60B:		/* Mirage 100 */
+	case 0x60C:		/* Mirage 100+ */
+	case 0x60D:		/* Electra 100 */
+	case 0x60E:		/* Electra 120 */
+		datareg = lasi_hpa - 0x00020000;
+		break;
+
+	default:
+		datareg = lasi_hpa + 0x0000C000;
+		break;
+	}
+
+	register_led_driver(DISPLAY_MODEL_LASI, LED_CMD_REG_NONE, datareg);
+}
+#endif
+
+/*
+ * lasi_power_off
+ *
+ * Function for lasi to turn off the power.  This is accomplished by setting a
+ * 1 to PWR_ON_L in the Power Control Register
+ * 
+ */
+static int lasi_power_off(struct sys_off_data *data)
+{
+	struct gsc_asic *lasi = data->cb_data;
+
+	/* Power down the machine via Power Control Register */
+	gsc_writel(0x02, lasi->hpa + 0x0000C000);
+
+	/* might not be reached: */
+	return NOTIFY_DONE;
+}
+
+static int __init lasi_init_chip(struct parisc_device *dev)
+{
+	struct gsc_asic *lasi;
+	int ret;
+
+	lasi = kzalloc_obj(*lasi);
+	if (!lasi)
+		return -ENOMEM;
+
+	lasi->name = "Lasi";
+	lasi->hpa = dev->hpa.start;
+
+	/* Check the 4-bit (yes, only 4) version register */
+	lasi->version = gsc_readl(lasi->hpa + LASI_VER) & 0xf;
+	printk(KERN_INFO "%s version %d at 0x%lx found.\n",
+		lasi->name, lasi->version, lasi->hpa);
+
+	/* initialize the chassis LEDs really early */ 
+	lasi_led_init(lasi->hpa);
+
+	/* Stop LASI barking for a bit */
+	lasi_init_irq(lasi);
+
+	/* the IRQ lasi should use */
+	dev->irq = gsc_alloc_irq(&lasi->gsc_irq);
+	if (dev->irq < 0) {
+		printk(KERN_ERR "%s(): cannot get GSC irq\n",
+				__func__);
+		kfree(lasi);
+		return -EBUSY;
+	}
+
+	lasi->eim = ((u32) lasi->gsc_irq.txn_addr) | lasi->gsc_irq.txn_data;
+
+	ret = request_irq(lasi->gsc_irq.irq, gsc_asic_intr, 0, "lasi", lasi);
+	if (ret < 0) {
+		goto err_free;
+	}
+
+	/* enable IRQ's for devices below LASI */
+	gsc_writel(lasi->eim, lasi->hpa + OFFSET_IAR);
+
+	/* Done init'ing, register this driver */
+	ret = gsc_common_setup(dev, lasi);
+	if (ret) {
+		goto err_irq;
+	}    
+
+	gsc_fixup_irqs(dev, lasi, lasi_choose_irq);
+
+	/* register the LASI power off function */
+	register_sys_off_handler(SYS_OFF_MODE_POWER_OFF,
+		SYS_OFF_PRIO_DEFAULT, lasi_power_off, lasi);
+
+	return ret;
+
+err_irq:
+	free_irq(lasi->gsc_irq.irq, lasi);
+err_free:
+	kfree(lasi);
+	return ret;
+}
+
+static struct parisc_device_id lasi_tbl[] __initdata = {
+	{ HPHW_BA, HVERSION_REV_ANY_ID, HVERSION_ANY_ID, 0x00081 },
+	{ 0, }
+};
+MODULE_DEVICE_TABLE(parisc, lasi_tbl);
+
+static struct parisc_driver lasi_driver __refdata = {
+	.name =		"lasi",
+	.id_table =	lasi_tbl,
+	.probe =	lasi_init_chip,
+};
+
+static int __init lasi_init(void)
+{
+	return register_parisc_driver(&lasi_driver);
+}
+arch_initcall(lasi_init);