From a5f7a5bb3b7f28ba7e4fa246775b29a0e5537255 Mon Sep 17 00:00:00 2001 From: Jinke Han Date: Tue, 8 Sep 2026 15:37:42 +0800 Subject: x86/kprobes: Fix crash when probing CS CALL instructions 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) Signed-off-by: Jinke Han Signed-off-by: Ingo Molnar Reviewed-by: Masami Hiramatsu (Google) Acked-by: Yafang Shao Acked-by: Borislav Petkov Cc: Peter Zijlstra Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000 --- .../bindings/dvfs/performance-domain.yaml | 79 ++++++++++++++++++++++ 1 file changed, 79 insertions(+) create mode 100644 Documentation/devicetree/bindings/dvfs/performance-domain.yaml (limited to 'Documentation/devicetree/bindings/dvfs') diff --git a/Documentation/devicetree/bindings/dvfs/performance-domain.yaml b/Documentation/devicetree/bindings/dvfs/performance-domain.yaml new file mode 100644 index 000000000..09bbd2e33 --- /dev/null +++ b/Documentation/devicetree/bindings/dvfs/performance-domain.yaml @@ -0,0 +1,79 @@ +# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/dvfs/performance-domain.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: Generic performance domains + +maintainers: + - Sudeep Holla + +description: |+ + This binding is intended for performance management of groups of devices or + CPUs that run in the same performance domain. Performance domains must not + be confused with power domains. A performance domain is defined by a set + of devices that always have to run at the same performance level. For a given + performance domain, there is a single point of control that affects all the + devices in the domain, making it impossible to set the performance level of + an individual device in the domain independently from other devices in + that domain. For example, a set of CPUs that share a voltage domain, and + have a common frequency control, is said to be in the same performance + domain. + + This device tree binding can be used to bind performance domain consumer + devices with their performance domains provided by performance domain + providers. A performance domain provider can be represented by any node in + the device tree and can provide one or more performance domains. A consumer + node can refer to the provider by a phandle and a set of phandle arguments + (so called performance domain specifiers) of length specified by the + \#performance-domain-cells property in the performance domain provider node. + +select: true + +properties: + "#performance-domain-cells": + description: + Number of cells in a performance domain specifier. Typically 0 for nodes + representing a single performance domain and 1 for nodes providing + multiple performance domains (e.g. performance controllers), but can be + any value as specified by device tree binding documentation of particular + provider. + enum: [ 0, 1 ] + + performance-domains: + $ref: /schemas/types.yaml#/definitions/phandle-array + description: + A phandle and performance domain specifier as defined by bindings of the + performance controller/provider specified by phandle. + +additionalProperties: true + +examples: + - | + soc { + #address-cells = <2>; + #size-cells = <2>; + + performance: performance-controller@11bc00 { + compatible = "mediatek,cpufreq-hw"; + reg = <0 0x0011bc10 0 0x120>, <0 0x0011bd30 0 0x120>; + + #performance-domain-cells = <1>; + }; + }; + + // The node above defines a performance controller that is a performance + // domain provider and expects one cell as its phandle argument. + + cpus { + #address-cells = <2>; + #size-cells = <0>; + + cpu@0 { + device_type = "cpu"; + compatible = "arm,cortex-a57"; + reg = <0x0 0x0>; + performance-domains = <&performance 1>; + }; + }; -- cgit v1.3.1