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, 242 insertions, 0 deletions

diff --git a/drivers/scsi/sim710.c b/drivers/scsi/sim710.c
new file mode 100644
index 000000000..408f37003
--- /dev/null
+++ b/drivers/scsi/sim710.c
@@ -0,0 +1,242 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * sim710.c - Copyright (C) 1999 Richard Hirst <richard@sleepie.demon.co.uk>
+ *
+ *----------------------------------------------------------------------------
+ *----------------------------------------------------------------------------
+ *
+ * MCA card detection code by Trent McNair. (now deleted)
+ * Fixes to not explicitly nul bss data from Xavier Bestel.
+ * Some multiboard fixes from Rolf Eike Beer.
+ * Auto probing of EISA config space from Trevor Hemsley.
+ *
+ * Rewritten to use 53c700.c by James.Bottomley@SteelEye.com
+ */
+
+#include <linux/module.h>
+#include <linux/slab.h>
+
+#include <linux/blkdev.h>
+#include <linux/device.h>
+#include <linux/init.h>
+#include <linux/eisa.h>
+#include <linux/interrupt.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi_transport.h>
+#include <scsi/scsi_transport_spi.h>
+
+#include "53c700.h"
+
+
+/* Must be enough for EISA */
+#define MAX_SLOTS 8
+static __u8 __initdata id_array[MAX_SLOTS] = { [0 ... MAX_SLOTS-1] = 7 };
+
+static char *sim710;		/* command line passed by insmod */
+
+MODULE_AUTHOR("Richard Hirst");
+MODULE_DESCRIPTION("Simple NCR53C710 driver");
+MODULE_LICENSE("GPL");
+
+module_param(sim710, charp, 0);
+
+#ifdef MODULE
+#define ARG_SEP ' '
+#else
+#define ARG_SEP ','
+#endif
+
+static __init int
+param_setup(char *str)
+{
+	char *pos = str, *next;
+	int slot = -1;
+
+	while(pos != NULL && (next = strchr(pos, ':')) != NULL) {
+		int val = (int)simple_strtoul(++next, NULL, 0);
+
+		if(!strncmp(pos, "slot:", 5))
+			slot = val;
+		else if(!strncmp(pos, "id:", 3)) {
+			if(slot == -1) {
+				printk(KERN_WARNING "sim710: Must specify slot for id parameter\n");
+			} else if(slot >= MAX_SLOTS) {
+				printk(KERN_WARNING "sim710: Illegal slot %d for id %d\n", slot, val);
+			} else {
+				id_array[slot] = val;
+			}
+		}
+		if((pos = strchr(pos, ARG_SEP)) != NULL)
+			pos++;
+	}
+	return 1;
+}
+__setup("sim710=", param_setup);
+
+static struct scsi_host_template sim710_driver_template = {
+	.name			= "LSI (Symbios) 710 EISA",
+	.proc_name		= "sim710",
+	.this_id		= 7,
+	.module			= THIS_MODULE,
+};
+
+static int sim710_probe_common(struct device *dev, unsigned long base_addr,
+			       int irq, int clock, int differential,
+			       int scsi_id)
+{
+	struct Scsi_Host * host = NULL;
+	struct NCR_700_Host_Parameters *hostdata =
+		kzalloc_obj(struct NCR_700_Host_Parameters);
+
+	printk(KERN_NOTICE "sim710: %s\n", dev_name(dev));
+	printk(KERN_NOTICE "sim710: irq = %d, clock = %d, base = 0x%lx, scsi_id = %d\n",
+	       irq, clock, base_addr, scsi_id);
+
+	if(hostdata == NULL) {
+		printk(KERN_ERR "sim710: Failed to allocate host data\n");
+		goto out;
+	}
+
+	if(request_region(base_addr, 64, "sim710") == NULL) {
+		printk(KERN_ERR "sim710: Failed to reserve IO region 0x%lx\n",
+		       base_addr);
+		goto out_free;
+	}
+
+	/* Fill in the three required pieces of hostdata */
+	hostdata->base = ioport_map(base_addr, 64);
+	hostdata->differential = differential;
+	hostdata->clock = clock;
+	hostdata->chip710 = 1;
+	hostdata->burst_length = 8;
+
+	/* and register the chip */
+	if((host = NCR_700_detect(&sim710_driver_template, hostdata, dev))
+	   == NULL) {
+		printk(KERN_ERR "sim710: No host detected; card configuration problem?\n");
+		goto out_release;
+	}
+	host->this_id = scsi_id;
+	host->base = base_addr;
+	host->irq = irq;
+	if (request_irq(irq, NCR_700_intr, IRQF_SHARED, "sim710", host)) {
+		printk(KERN_ERR "sim710: request_irq failed\n");
+		goto out_put_host;
+	}
+
+	dev_set_drvdata(dev, host);
+	scsi_scan_host(host);
+
+	return 0;
+
+ out_put_host:
+	scsi_host_put(host);
+ out_release:
+	ioport_unmap(hostdata->base);
+	release_region(base_addr, 64);
+ out_free:
+	kfree(hostdata);
+ out:
+	return -ENODEV;
+}
+
+static int sim710_device_remove(struct device *dev)
+{
+	struct Scsi_Host *host = dev_get_drvdata(dev);
+	struct NCR_700_Host_Parameters *hostdata =
+		(struct NCR_700_Host_Parameters *)host->hostdata[0];
+
+	scsi_remove_host(host);
+	NCR_700_release(host);
+	ioport_unmap(hostdata->base);
+	kfree(hostdata);
+	free_irq(host->irq, host);
+	release_region(host->base, 64);
+	return 0;
+}
+
+#ifdef CONFIG_EISA
+static struct eisa_device_id sim710_eisa_ids[] = {
+	{ "CPQ4410" },
+	{ "CPQ4411" },
+	{ "HWP0C80" },
+	{ "" }
+};
+MODULE_DEVICE_TABLE(eisa, sim710_eisa_ids);
+
+static int sim710_eisa_probe(struct device *dev)
+{
+	struct eisa_device *edev = to_eisa_device(dev);
+	unsigned long io_addr = edev->base_addr;
+	char eisa_cpq_irqs[] = { 11, 14, 15, 10, 9, 0 };
+	char eisa_hwp_irqs[] = { 3, 4, 5, 7, 12, 10, 11, 0};
+	char *eisa_irqs;
+	unsigned char irq_index;
+	unsigned char irq, differential = 0, scsi_id = 7;
+
+	if(strcmp(edev->id.sig, "HWP0C80") == 0) {
+		__u8 val;
+		eisa_irqs =  eisa_hwp_irqs;
+		irq_index = (inb(io_addr + 0xc85) & 0x7) - 1;
+
+		val = inb(io_addr + 0x4);
+		scsi_id = ffs(val) - 1;
+
+		if(scsi_id > 7 || (val & ~(1<<scsi_id)) != 0) {
+			printk(KERN_ERR "sim710.c, EISA card %s has incorrect scsi_id, setting to 7\n", dev_name(dev));
+			scsi_id = 7;
+		}
+	} else {
+		eisa_irqs = eisa_cpq_irqs;
+		irq_index = inb(io_addr + 0xc88) & 0x07;
+	}
+
+	if(irq_index >= strlen(eisa_irqs)) {
+		printk("sim710.c: irq nasty\n");
+		return -ENODEV;
+	}
+
+	irq = eisa_irqs[irq_index];
+		
+	return sim710_probe_common(dev, io_addr, irq, 50,
+				   differential, scsi_id);
+}
+
+static struct eisa_driver sim710_eisa_driver = {
+	.id_table		= sim710_eisa_ids,
+	.driver = {
+		.name		= "sim710",
+		.probe		= sim710_eisa_probe,
+		.remove		= sim710_device_remove,
+	},
+};
+#endif /* CONFIG_EISA */
+
+static int __init sim710_init(void)
+{
+#ifdef MODULE
+	if (sim710)
+		param_setup(sim710);
+#endif
+
+#ifdef CONFIG_EISA
+	/*
+	 * FIXME: We'd really like to return -ENODEV if no devices have actually
+	 * been found.  However eisa_driver_register() only reports problems
+	 * with kobject_register() so simply return success for now.
+	 */
+	eisa_driver_register(&sim710_eisa_driver);
+#endif
+	return 0;
+}
+
+static void __exit sim710_exit(void)
+{
+#ifdef CONFIG_EISA
+	eisa_driver_unregister(&sim710_eisa_driver);
+#endif
+}
+
+module_init(sim710_init);
+module_exit(sim710_exit);