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, 251 insertions, 0 deletions

diff --git a/drivers/peci/device.c b/drivers/peci/device.c
new file mode 100644
index 000000000..2b9610216
--- /dev/null
+++ b/drivers/peci/device.c
@@ -0,0 +1,251 @@
+// SPDX-License-Identifier: GPL-2.0-only
+// Copyright (c) 2018-2021 Intel Corporation
+
+#include <linux/bitfield.h>
+#include <linux/peci.h>
+#include <linux/peci-cpu.h>
+#include <linux/slab.h>
+
+#include "internal.h"
+
+/*
+ * PECI device can be removed using sysfs, but the removal can also happen as
+ * a result of controller being removed.
+ * Mutex is used to protect PECI device from being double-deleted.
+ */
+static DEFINE_MUTEX(peci_device_del_lock);
+
+#define REVISION_NUM_MASK GENMASK(15, 8)
+static int peci_get_revision(struct peci_device *device, u8 *revision)
+{
+	struct peci_request *req;
+	u64 dib;
+
+	req = peci_xfer_get_dib(device);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
+
+	/*
+	 * PECI device may be in a state where it is unable to return a proper
+	 * DIB, in which case it returns 0 as DIB value.
+	 * Let's treat this as an error to avoid carrying on with the detection
+	 * using invalid revision.
+	 */
+	dib = peci_request_dib_read(req);
+	if (dib == 0) {
+		peci_request_free(req);
+		return -EIO;
+	}
+
+	*revision = FIELD_GET(REVISION_NUM_MASK, dib);
+
+	peci_request_free(req);
+
+	return 0;
+}
+
+static int peci_get_cpu_id(struct peci_device *device, u32 *cpu_id)
+{
+	struct peci_request *req;
+	int ret;
+
+	req = peci_xfer_pkg_cfg_readl(device, PECI_PCS_PKG_ID, PECI_PKG_ID_CPU_ID);
+	if (IS_ERR(req))
+		return PTR_ERR(req);
+
+	ret = peci_request_status(req);
+	if (ret)
+		goto out_req_free;
+
+	*cpu_id = peci_request_data_readl(req);
+out_req_free:
+	peci_request_free(req);
+
+	return ret;
+}
+
+static unsigned int peci_x86_cpu_family(unsigned int sig)
+{
+	unsigned int x86;
+
+	x86 = (sig >> 8) & 0xf;
+
+	if (x86 == 0xf)
+		x86 += (sig >> 20) & 0xff;
+
+	return x86;
+}
+
+static unsigned int peci_x86_cpu_model(unsigned int sig)
+{
+	unsigned int fam, model;
+
+	fam = peci_x86_cpu_family(sig);
+
+	model = (sig >> 4) & 0xf;
+
+	if (fam >= 0x6)
+		model += ((sig >> 16) & 0xf) << 4;
+
+	return model;
+}
+
+static int peci_device_info_init(struct peci_device *device)
+{
+	u8 revision;
+	u32 cpu_id;
+	int ret;
+
+	ret = peci_get_cpu_id(device, &cpu_id);
+	if (ret)
+		return ret;
+
+	device->info.x86_vfm = IFM(peci_x86_cpu_family(cpu_id), peci_x86_cpu_model(cpu_id));
+
+	ret = peci_get_revision(device, &revision);
+	if (ret)
+		return ret;
+	device->info.peci_revision = revision;
+
+	device->info.socket_id = device->addr - PECI_BASE_ADDR;
+
+	return 0;
+}
+
+static int peci_detect(struct peci_controller *controller, u8 addr)
+{
+	/*
+	 * PECI Ping is a command encoded by tx_len = 0, rx_len = 0.
+	 * We expect correct Write FCS if the device at the target address
+	 * is able to respond.
+	 */
+	struct peci_request req = { 0 };
+	int ret;
+
+	mutex_lock(&controller->bus_lock);
+	ret = controller->ops->xfer(controller, addr, &req);
+	mutex_unlock(&controller->bus_lock);
+
+	return ret;
+}
+
+static bool peci_addr_valid(u8 addr)
+{
+	return addr >= PECI_BASE_ADDR && addr < PECI_BASE_ADDR + PECI_DEVICE_NUM_MAX;
+}
+
+static int peci_dev_exists(struct device *dev, void *data)
+{
+	struct peci_device *device = to_peci_device(dev);
+	u8 *addr = data;
+
+	if (device->addr == *addr)
+		return -EBUSY;
+
+	return 0;
+}
+
+int peci_device_create(struct peci_controller *controller, u8 addr)
+{
+	struct peci_device *device;
+	int ret;
+
+	if (!peci_addr_valid(addr))
+		return -EINVAL;
+
+	/* Check if we have already detected this device before. */
+	ret = device_for_each_child(&controller->dev, &addr, peci_dev_exists);
+	if (ret)
+		return 0;
+
+	ret = peci_detect(controller, addr);
+	if (ret) {
+		/*
+		 * Device not present or host state doesn't allow successful
+		 * detection at this time.
+		 */
+		if (ret == -EIO || ret == -ETIMEDOUT)
+			return 0;
+
+		return ret;
+	}
+
+	device = kzalloc_obj(*device);
+	if (!device)
+		return -ENOMEM;
+
+	device_initialize(&device->dev);
+
+	device->addr = addr;
+	device->dev.parent = &controller->dev;
+	device->dev.bus = &peci_bus_type;
+	device->dev.type = &peci_device_type;
+
+	ret = peci_device_info_init(device);
+	if (ret)
+		goto err_put;
+
+	ret = dev_set_name(&device->dev, "%d-%02x", controller->id, device->addr);
+	if (ret)
+		goto err_put;
+
+	ret = device_add(&device->dev);
+	if (ret)
+		goto err_put;
+
+	return 0;
+
+err_put:
+	put_device(&device->dev);
+
+	return ret;
+}
+
+void peci_device_destroy(struct peci_device *device)
+{
+	mutex_lock(&peci_device_del_lock);
+	if (!device->deleted) {
+		device_unregister(&device->dev);
+		device->deleted = true;
+	}
+	mutex_unlock(&peci_device_del_lock);
+}
+
+int __peci_driver_register(struct peci_driver *driver, struct module *owner,
+			   const char *mod_name)
+{
+	driver->driver.bus = &peci_bus_type;
+	driver->driver.owner = owner;
+	driver->driver.mod_name = mod_name;
+
+	if (!driver->probe) {
+		pr_err("peci: trying to register driver without probe callback\n");
+		return -EINVAL;
+	}
+
+	if (!driver->id_table) {
+		pr_err("peci: trying to register driver without device id table\n");
+		return -EINVAL;
+	}
+
+	return driver_register(&driver->driver);
+}
+EXPORT_SYMBOL_NS_GPL(__peci_driver_register, "PECI");
+
+void peci_driver_unregister(struct peci_driver *driver)
+{
+	driver_unregister(&driver->driver);
+}
+EXPORT_SYMBOL_NS_GPL(peci_driver_unregister, "PECI");
+
+static void peci_device_release(struct device *dev)
+{
+	struct peci_device *device = to_peci_device(dev);
+
+	kfree(device);
+}
+
+const struct device_type peci_device_type = {
+	.groups		= peci_device_groups,
+	.release	= peci_device_release,
+};