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, 218 insertions, 0 deletions

diff --git a/drivers/mtd/maps/plat-ram.c b/drivers/mtd/maps/plat-ram.c
new file mode 100644
index 000000000..816111031
--- /dev/null
+++ b/drivers/mtd/maps/plat-ram.c
@@ -0,0 +1,218 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* drivers/mtd/maps/plat-ram.c
+ *
+ * (c) 2004-2005 Simtec Electronics
+ *	http://www.simtec.co.uk/products/SWLINUX/
+ *	Ben Dooks <ben@simtec.co.uk>
+ *
+ * Generic platform device based RAM map
+*/
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/ioport.h>
+#include <linux/device.h>
+#include <linux/slab.h>
+#include <linux/platform_device.h>
+
+#include <linux/mtd/mtd.h>
+#include <linux/mtd/map.h>
+#include <linux/mtd/partitions.h>
+#include <linux/mtd/plat-ram.h>
+
+#include <asm/io.h>
+
+/* private structure for each mtd platform ram device created */
+
+struct platram_info {
+	struct device		*dev;
+	struct mtd_info		*mtd;
+	struct map_info		 map;
+	struct platdata_mtd_ram	*pdata;
+};
+
+/* to_platram_info()
+ *
+ * device private data to struct platram_info conversion
+*/
+
+static inline struct platram_info *to_platram_info(struct platform_device *dev)
+{
+	return platform_get_drvdata(dev);
+}
+
+/* platram_setrw
+ *
+ * call the platform device's set rw/ro control
+ *
+ * to = 0 => read-only
+ *    = 1 => read-write
+*/
+
+static inline void platram_setrw(struct platram_info *info, int to)
+{
+	if (info->pdata == NULL)
+		return;
+
+	if (info->pdata->set_rw != NULL)
+		(info->pdata->set_rw)(info->dev, to);
+}
+
+/* platram_remove
+ *
+ * called to remove the device from the driver's control
+*/
+
+static void platram_remove(struct platform_device *pdev)
+{
+	struct platram_info *info = to_platram_info(pdev);
+
+	dev_dbg(&pdev->dev, "removing device\n");
+
+	if (info == NULL)
+		return;
+
+	if (info->mtd) {
+		mtd_device_unregister(info->mtd);
+		map_destroy(info->mtd);
+	}
+
+	/* ensure ram is left read-only */
+
+	platram_setrw(info, PLATRAM_RO);
+
+	kfree(info);
+}
+
+/* platram_probe
+ *
+ * called from device drive system when a device matching our
+ * driver is found.
+*/
+
+static int platram_probe(struct platform_device *pdev)
+{
+	struct platdata_mtd_ram	*pdata;
+	struct platram_info *info;
+	struct resource *res;
+	int err = 0;
+
+	dev_dbg(&pdev->dev, "probe entered\n");
+
+	if (dev_get_platdata(&pdev->dev) == NULL) {
+		dev_err(&pdev->dev, "no platform data supplied\n");
+		err = -ENOENT;
+		goto exit_error;
+	}
+
+	pdata = dev_get_platdata(&pdev->dev);
+
+	info = kzalloc_obj(*info);
+	if (info == NULL) {
+		err = -ENOMEM;
+		goto exit_error;
+	}
+
+	platform_set_drvdata(pdev, info);
+
+	info->dev = &pdev->dev;
+	info->pdata = pdata;
+
+	/* get the resource for the memory mapping */
+	info->map.virt = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
+	if (IS_ERR(info->map.virt)) {
+		err = PTR_ERR(info->map.virt);
+		goto exit_free;
+	}
+
+	dev_dbg(&pdev->dev, "got platform resource %p (0x%llx)\n", res,
+		(unsigned long long)res->start);
+
+	/* setup map parameters */
+
+	info->map.phys = res->start;
+	info->map.size = resource_size(res);
+	info->map.name = pdata->mapname != NULL ?
+			(char *)pdata->mapname : (char *)pdev->name;
+	info->map.bankwidth = pdata->bankwidth;
+
+	dev_dbg(&pdev->dev, "virt %p, %lu bytes\n", info->map.virt, info->map.size);
+
+	simple_map_init(&info->map);
+
+	dev_dbg(&pdev->dev, "initialised map, probing for mtd\n");
+
+	/* probe for the right mtd map driver
+	 * supplied by the platform_data struct */
+
+	if (pdata->map_probes) {
+		const char * const *map_probes = pdata->map_probes;
+
+		for ( ; !info->mtd && *map_probes; map_probes++)
+			info->mtd = do_map_probe(*map_probes , &info->map);
+	}
+	/* fallback to map_ram */
+	else
+		info->mtd = do_map_probe("map_ram", &info->map);
+
+	if (info->mtd == NULL) {
+		dev_err(&pdev->dev, "failed to probe for map_ram\n");
+		err = -ENOMEM;
+		goto exit_free;
+	}
+
+	info->mtd->dev.parent = &pdev->dev;
+
+	platram_setrw(info, PLATRAM_RW);
+
+	/* check to see if there are any available partitions, or whether
+	 * to add this device whole */
+
+	err = mtd_device_parse_register(info->mtd, pdata->probes, NULL,
+					pdata->partitions,
+					pdata->nr_partitions);
+	if (err) {
+		dev_err(&pdev->dev, "failed to register mtd device\n");
+		goto exit_free;
+	}
+
+	dev_info(&pdev->dev, "registered mtd device\n");
+
+	if (pdata->nr_partitions) {
+		/* add the whole device. */
+		err = mtd_device_register(info->mtd, NULL, 0);
+		if (err) {
+			dev_err(&pdev->dev,
+				"failed to register the entire device\n");
+			goto exit_free;
+		}
+	}
+
+	return 0;
+
+ exit_free:
+	platram_remove(pdev);
+ exit_error:
+	return err;
+}
+
+/* device driver info */
+
+/* work with hotplug and coldplug */
+MODULE_ALIAS("platform:mtd-ram");
+
+static struct platform_driver platram_driver = {
+	.probe		= platram_probe,
+	.remove		= platram_remove,
+	.driver		= {
+		.name	= "mtd-ram",
+	},
+};
+
+module_platform_driver(platram_driver);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Ben Dooks <ben@simtec.co.uk>");
+MODULE_DESCRIPTION("MTD platform RAM map driver");