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, 418 insertions, 0 deletions

diff --git a/net/ieee802154/trace.h b/net/ieee802154/trace.h
new file mode 100644
index 000000000..591ce0a16
--- /dev/null
+++ b/net/ieee802154/trace.h
@@ -0,0 +1,418 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Based on net/wireless/trace.h */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM cfg802154
+
+#if !defined(__RDEV_CFG802154_OPS_TRACE) || defined(TRACE_HEADER_MULTI_READ)
+#define __RDEV_CFG802154_OPS_TRACE
+
+#include <linux/tracepoint.h>
+
+#include <net/cfg802154.h>
+
+#define MAXNAME		32
+#define WPAN_PHY_ENTRY	__array(char, wpan_phy_name, MAXNAME)
+#define WPAN_PHY_ASSIGN	strscpy(__entry->wpan_phy_name,	 \
+				wpan_phy_name(wpan_phy), \
+				MAXNAME)
+#define WPAN_PHY_PR_FMT	"%s"
+#define WPAN_PHY_PR_ARG	__entry->wpan_phy_name
+
+#define WPAN_DEV_ENTRY	__field(u32, identifier)
+#define WPAN_DEV_ASSIGN	(__entry->identifier) = (!IS_ERR_OR_NULL(wpan_dev) \
+					 ? wpan_dev->identifier : 0)
+#define WPAN_DEV_PR_FMT	"wpan_dev(%u)"
+#define WPAN_DEV_PR_ARG	(__entry->identifier)
+
+#define WPAN_CCA_ENTRY	__field(enum nl802154_cca_modes, cca_mode) \
+			__field(enum nl802154_cca_opts, cca_opt)
+#define WPAN_CCA_ASSIGN \
+	do {					 \
+		(__entry->cca_mode) = cca->mode; \
+		(__entry->cca_opt) = cca->opt;	 \
+	} while (0)
+#define WPAN_CCA_PR_FMT	"cca_mode: %d, cca_opt: %d"
+#define WPAN_CCA_PR_ARG __entry->cca_mode, __entry->cca_opt
+
+#define BOOL_TO_STR(bo) (bo) ? "true" : "false"
+
+/*************************************************************
+ *			rdev->ops traces		     *
+ *************************************************************/
+
+DECLARE_EVENT_CLASS(wpan_phy_only_evt,
+	TP_PROTO(struct wpan_phy *wpan_phy),
+	TP_ARGS(wpan_phy),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+	),
+	TP_printk(WPAN_PHY_PR_FMT, WPAN_PHY_PR_ARG)
+);
+
+DEFINE_EVENT(wpan_phy_only_evt, 802154_rdev_suspend,
+	TP_PROTO(struct wpan_phy *wpan_phy),
+	TP_ARGS(wpan_phy)
+);
+
+DEFINE_EVENT(wpan_phy_only_evt, 802154_rdev_resume,
+	TP_PROTO(struct wpan_phy *wpan_phy),
+	TP_ARGS(wpan_phy)
+);
+
+TRACE_EVENT(802154_rdev_add_virtual_intf,
+	TP_PROTO(struct wpan_phy *wpan_phy, char *name,
+		 enum nl802154_iftype type, __le64 extended_addr),
+	TP_ARGS(wpan_phy, name, type, extended_addr),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		__string(vir_intf_name, name ? name : "<noname>")
+		__field(enum nl802154_iftype, type)
+		__field(__le64, extended_addr)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		__assign_str(vir_intf_name);
+		__entry->type = type;
+		__entry->extended_addr = extended_addr;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", virtual intf name: %s, type: %d, extended addr: 0x%llx",
+		  WPAN_PHY_PR_ARG, __get_str(vir_intf_name), __entry->type,
+		  __le64_to_cpu(__entry->extended_addr))
+);
+
+TRACE_EVENT(802154_rdev_del_virtual_intf,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev),
+	TP_ARGS(wpan_phy, wpan_dev),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT, WPAN_PHY_PR_ARG,
+		  WPAN_DEV_PR_ARG)
+);
+
+TRACE_EVENT(802154_rdev_set_channel,
+	TP_PROTO(struct wpan_phy *wpan_phy, u8 page, u8 channel),
+	TP_ARGS(wpan_phy, page, channel),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		__field(u8, page)
+		__field(u8, channel)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		__entry->page = page;
+		__entry->channel = channel;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", page: %d, channel: %d", WPAN_PHY_PR_ARG,
+		  __entry->page, __entry->channel)
+);
+
+TRACE_EVENT(802154_rdev_set_tx_power,
+	TP_PROTO(struct wpan_phy *wpan_phy, s32 power),
+	TP_ARGS(wpan_phy, power),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		__field(s32, power)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		__entry->power = power;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", mbm: %d", WPAN_PHY_PR_ARG,
+		  __entry->power)
+);
+
+TRACE_EVENT(802154_rdev_set_cca_mode,
+	TP_PROTO(struct wpan_phy *wpan_phy, const struct wpan_phy_cca *cca),
+	TP_ARGS(wpan_phy, cca),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_CCA_ENTRY
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_CCA_ASSIGN;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_CCA_PR_FMT, WPAN_PHY_PR_ARG,
+		  WPAN_CCA_PR_ARG)
+);
+
+TRACE_EVENT(802154_rdev_set_cca_ed_level,
+	TP_PROTO(struct wpan_phy *wpan_phy, s32 ed_level),
+	TP_ARGS(wpan_phy, ed_level),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		__field(s32, ed_level)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		__entry->ed_level = ed_level;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", ed level: %d", WPAN_PHY_PR_ARG,
+		  __entry->ed_level)
+);
+
+DECLARE_EVENT_CLASS(802154_le16_template,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev,
+		 __le16 le16arg),
+	TP_ARGS(wpan_phy, wpan_dev, le16arg),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+		__field(__le16, le16arg)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+		__entry->le16arg = le16arg;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT ", pan id: 0x%04x",
+		  WPAN_PHY_PR_ARG, WPAN_DEV_PR_ARG,
+		  __le16_to_cpu(__entry->le16arg))
+);
+
+DEFINE_EVENT(802154_le16_template, 802154_rdev_set_pan_id,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev,
+		 __le16 le16arg),
+	TP_ARGS(wpan_phy, wpan_dev, le16arg)
+);
+
+DEFINE_EVENT_PRINT(802154_le16_template, 802154_rdev_set_short_addr,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev,
+		 __le16 le16arg),
+	TP_ARGS(wpan_phy, wpan_dev, le16arg),
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT ", short addr: 0x%04x",
+		  WPAN_PHY_PR_ARG, WPAN_DEV_PR_ARG,
+		  __le16_to_cpu(__entry->le16arg))
+);
+
+TRACE_EVENT(802154_rdev_set_backoff_exponent,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev,
+		 u8 min_be, u8 max_be),
+	TP_ARGS(wpan_phy, wpan_dev, min_be, max_be),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+		__field(u8, min_be)
+		__field(u8, max_be)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+		__entry->min_be = min_be;
+		__entry->max_be = max_be;
+	),
+
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT
+		  ", min be: %d, max be: %d", WPAN_PHY_PR_ARG,
+		  WPAN_DEV_PR_ARG, __entry->min_be, __entry->max_be)
+);
+
+TRACE_EVENT(802154_rdev_set_csma_backoffs,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev,
+		 u8 max_csma_backoffs),
+	TP_ARGS(wpan_phy, wpan_dev, max_csma_backoffs),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+		__field(u8, max_csma_backoffs)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+		__entry->max_csma_backoffs = max_csma_backoffs;
+	),
+
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT
+		  ", max csma backoffs: %d", WPAN_PHY_PR_ARG,
+		  WPAN_DEV_PR_ARG, __entry->max_csma_backoffs)
+);
+
+TRACE_EVENT(802154_rdev_set_max_frame_retries,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev,
+		 s8 max_frame_retries),
+	TP_ARGS(wpan_phy, wpan_dev, max_frame_retries),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+		__field(s8, max_frame_retries)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+		__entry->max_frame_retries = max_frame_retries;
+	),
+
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT
+		  ", max frame retries: %d", WPAN_PHY_PR_ARG,
+		  WPAN_DEV_PR_ARG, __entry->max_frame_retries)
+);
+
+TRACE_EVENT(802154_rdev_set_lbt_mode,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev,
+		 bool mode),
+	TP_ARGS(wpan_phy, wpan_dev, mode),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+		__field(bool, mode)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+		__entry->mode = mode;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT
+		", lbt mode: %s", WPAN_PHY_PR_ARG,
+		WPAN_DEV_PR_ARG, BOOL_TO_STR(__entry->mode))
+);
+
+TRACE_EVENT(802154_rdev_set_ackreq_default,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev,
+		 bool ackreq),
+	TP_ARGS(wpan_phy, wpan_dev, ackreq),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+		__field(bool, ackreq)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+		__entry->ackreq = ackreq;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT
+		", ackreq default: %s", WPAN_PHY_PR_ARG,
+		WPAN_DEV_PR_ARG, BOOL_TO_STR(__entry->ackreq))
+);
+
+TRACE_EVENT(802154_rdev_trigger_scan,
+	TP_PROTO(struct wpan_phy *wpan_phy,
+		 struct cfg802154_scan_request *request),
+	TP_ARGS(wpan_phy, request),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		__field(u8, page)
+		__field(u32, channels)
+		__field(u8, duration)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		__entry->page = request->page;
+		__entry->channels = request->channels;
+		__entry->duration = request->duration;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", scan, page: %d, channels: %x, duration %d",
+		  WPAN_PHY_PR_ARG, __entry->page, __entry->channels, __entry->duration)
+);
+
+TRACE_EVENT(802154_rdev_send_beacons,
+	TP_PROTO(struct wpan_phy *wpan_phy,
+		 struct cfg802154_beacon_request *request),
+	TP_ARGS(wpan_phy, request),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		__field(u8, interval)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		__entry->interval = request->interval;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", sending beacons (interval order: %d)",
+		  WPAN_PHY_PR_ARG, __entry->interval)
+);
+
+DECLARE_EVENT_CLASS(802154_wdev_template,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev),
+	TP_ARGS(wpan_phy, wpan_dev),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT,
+		  WPAN_PHY_PR_ARG, WPAN_DEV_PR_ARG)
+);
+
+DEFINE_EVENT(802154_wdev_template, 802154_rdev_abort_scan,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev),
+	TP_ARGS(wpan_phy, wpan_dev)
+);
+
+DEFINE_EVENT(802154_wdev_template, 802154_rdev_stop_beacons,
+	TP_PROTO(struct wpan_phy *wpan_phy, struct wpan_dev *wpan_dev),
+	TP_ARGS(wpan_phy, wpan_dev)
+);
+
+TRACE_EVENT(802154_rdev_associate,
+	TP_PROTO(struct wpan_phy *wpan_phy,
+		 struct wpan_dev *wpan_dev,
+		 struct ieee802154_addr *coord),
+	TP_ARGS(wpan_phy, wpan_dev, coord),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+		__field(__le64, addr)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+		__entry->addr = coord->extended_addr;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT ", associating with: 0x%llx",
+		  WPAN_PHY_PR_ARG, WPAN_DEV_PR_ARG, __entry->addr)
+);
+
+TRACE_EVENT(802154_rdev_disassociate,
+	TP_PROTO(struct wpan_phy *wpan_phy,
+		 struct wpan_dev *wpan_dev,
+		 struct ieee802154_addr *target),
+	TP_ARGS(wpan_phy, wpan_dev, target),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		WPAN_DEV_ENTRY
+		__field(__le64, addr)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		WPAN_DEV_ASSIGN;
+		__entry->addr = target->extended_addr;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", " WPAN_DEV_PR_FMT ", disassociating with: 0x%llx",
+		  WPAN_PHY_PR_ARG, WPAN_DEV_PR_ARG, __entry->addr)
+);
+
+TRACE_EVENT(802154_rdev_return_int,
+	TP_PROTO(struct wpan_phy *wpan_phy, int ret),
+	TP_ARGS(wpan_phy, ret),
+	TP_STRUCT__entry(
+		WPAN_PHY_ENTRY
+		__field(int, ret)
+	),
+	TP_fast_assign(
+		WPAN_PHY_ASSIGN;
+		__entry->ret = ret;
+	),
+	TP_printk(WPAN_PHY_PR_FMT ", returned: %d", WPAN_PHY_PR_ARG,
+		  __entry->ret)
+);
+
+#endif /* !__RDEV_CFG802154_OPS_TRACE || TRACE_HEADER_MULTI_READ */
+
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH .
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE trace
+#include <trace/define_trace.h>