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/devfreq/event/rockchip,dfi.yaml | 74 +++++++++ .../devfreq/event/samsung,exynos-nocp.yaml | 48 ++++++ .../devfreq/event/samsung,exynos-ppmu.yaml | 169 +++++++++++++++++++++ .../bindings/devfreq/nvidia,tegra30-actmon.yaml | 129 ++++++++++++++++ 4 files changed, 420 insertions(+) create mode 100644 Documentation/devicetree/bindings/devfreq/event/rockchip,dfi.yaml create mode 100644 Documentation/devicetree/bindings/devfreq/event/samsung,exynos-nocp.yaml create mode 100644 Documentation/devicetree/bindings/devfreq/event/samsung,exynos-ppmu.yaml create mode 100644 Documentation/devicetree/bindings/devfreq/nvidia,tegra30-actmon.yaml (limited to 'Documentation/devicetree/bindings/devfreq') diff --git a/Documentation/devicetree/bindings/devfreq/event/rockchip,dfi.yaml b/Documentation/devicetree/bindings/devfreq/event/rockchip,dfi.yaml new file mode 100644 index 000000000..50d3fabe9 --- /dev/null +++ b/Documentation/devicetree/bindings/devfreq/event/rockchip,dfi.yaml @@ -0,0 +1,74 @@ +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/devfreq/event/rockchip,dfi.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: Rockchip DFI + +maintainers: + - Sascha Hauer + +properties: + compatible: + enum: + - rockchip,rk3399-dfi + - rockchip,rk3568-dfi + - rockchip,rk3588-dfi + + clocks: + maxItems: 1 + + clock-names: + items: + - const: pclk_ddr_mon + + interrupts: + minItems: 1 + maxItems: 4 + + reg: + maxItems: 1 + + rockchip,pmu: + $ref: /schemas/types.yaml#/definitions/phandle + description: + Phandle to the syscon managing the "PMU general register files". + +required: + - compatible + - interrupts + - reg + +if: + properties: + compatible: + contains: + enum: + - rockchip,rk3399-dfi + +then: + required: + - clocks + - clock-names + +additionalProperties: false + +examples: + - | + #include + #include + + bus { + #address-cells = <2>; + #size-cells = <2>; + + dfi: dfi@ff630000 { + compatible = "rockchip,rk3399-dfi"; + reg = <0x00 0xff630000 0x00 0x4000>; + interrupts = ; + rockchip,pmu = <&pmugrf>; + clocks = <&cru PCLK_DDR_MON>; + clock-names = "pclk_ddr_mon"; + }; + }; diff --git a/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-nocp.yaml b/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-nocp.yaml new file mode 100644 index 000000000..2bdd05af6 --- /dev/null +++ b/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-nocp.yaml @@ -0,0 +1,48 @@ +# SPDX-License-Identifier: GPL-2.0 +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/devfreq/event/samsung,exynos-nocp.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: Samsung Exynos NoC (Network on Chip) Probe + +maintainers: + - Chanwoo Choi + - Krzysztof Kozlowski + +description: | + The Samsung Exynos542x SoC has a NoC (Network on Chip) Probe for NoC bus. + NoC provides the primitive values to get the performance data. The packets + that the Network on Chip (NoC) probes detects are transported over the + network infrastructure to observer units. You can configure probes to capture + packets with header or data on the data request response network, or as + traffic debug or statistic collectors. Exynos542x bus has multiple NoC probes + to provide bandwidth information about behavior of the SoC that you can use + while analyzing system performance. + +properties: + compatible: + const: samsung,exynos5420-nocp + + clock-names: + items: + - const: nocp + + clocks: + maxItems: 1 + + reg: + maxItems: 1 + +required: + - compatible + - reg + +additionalProperties: false + +examples: + - | + nocp_mem0_0: nocp@10ca1000 { + compatible = "samsung,exynos5420-nocp"; + reg = <0x10ca1000 0x200>; + }; diff --git a/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-ppmu.yaml b/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-ppmu.yaml new file mode 100644 index 000000000..a3f0cd508 --- /dev/null +++ b/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-ppmu.yaml @@ -0,0 +1,169 @@ +# SPDX-License-Identifier: GPL-2.0 +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/devfreq/event/samsung,exynos-ppmu.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: Samsung Exynos SoC PPMU (Platform Performance Monitoring Unit) + +maintainers: + - Chanwoo Choi + - Krzysztof Kozlowski + +description: | + The Samsung Exynos SoC has PPMU (Platform Performance Monitoring Unit) for + each IP. PPMU provides the primitive values to get performance data. These + PPMU events provide information of the SoC's behaviors so that you may use to + analyze system performance, to make behaviors visible and to count usages of + each IP (DMC, CPU, RIGHTBUS, LEFTBUS, CAM interface, LCD, G3D, MFC). The + Exynos PPMU driver uses the devfreq-event class to provide event data to + various devfreq devices. The devfreq devices would use the event data when + determining the current state of each IP. + +properties: + compatible: + enum: + - samsung,exynos-ppmu + - samsung,exynos-ppmu-v2 + + clock-names: + items: + - const: ppmu + + clocks: + maxItems: 1 + + reg: + maxItems: 1 + + events: + type: object + + patternProperties: + '^ppmu-event[0-9]+(-[a-z0-9]+){0,2}$': + type: object + properties: + event-name: + description: | + The unique event name among PPMU device + $ref: /schemas/types.yaml#/definitions/string + + event-data-type: + description: | + Define the type of data which shell be counted by the counter. + You can check include/dt-bindings/pmu/exynos_ppmu.h for all + possible type, i.e. count read requests, count write data in + bytes, etc. This field is optional and when it is missing, the + driver code will use default data type. + $ref: /schemas/types.yaml#/definitions/uint32 + + required: + - event-name + + additionalProperties: false + + additionalProperties: false + +required: + - compatible + - reg + +additionalProperties: false + +examples: + - | + // PPMUv1 nodes for Exynos3250 (although the board DTS defines events) + #include + + ppmu_dmc0: ppmu@106a0000 { + compatible = "samsung,exynos-ppmu"; + reg = <0x106a0000 0x2000>; + + events { + ppmu_dmc0_3: ppmu-event3-dmc0 { + event-name = "ppmu-event3-dmc0"; + }; + + ppmu_dmc0_2: ppmu-event2-dmc0 { + event-name = "ppmu-event2-dmc0"; + }; + + ppmu_dmc0_1: ppmu-event1-dmc0 { + event-name = "ppmu-event1-dmc0"; + }; + + ppmu_dmc0_0: ppmu-event0-dmc0 { + event-name = "ppmu-event0-dmc0"; + }; + }; + }; + + ppmu_rightbus: ppmu@112a0000 { + compatible = "samsung,exynos-ppmu"; + reg = <0x112a0000 0x2000>; + clocks = <&cmu CLK_PPMURIGHT>; + clock-names = "ppmu"; + + events { + ppmu_rightbus_3: ppmu-event3-rightbus { + event-name = "ppmu-event3-rightbus"; + }; + }; + }; + + - | + // PPMUv2 nodes in Exynos5433 + ppmu_d0_cpu: ppmu@10480000 { + compatible = "samsung,exynos-ppmu-v2"; + reg = <0x10480000 0x2000>; + }; + + ppmu_d0_general: ppmu@10490000 { + compatible = "samsung,exynos-ppmu-v2"; + reg = <0x10490000 0x2000>; + + events { + ppmu_event0_d0_general: ppmu-event0-d0-general { + event-name = "ppmu-event0-d0-general"; + }; + }; + }; + + ppmu_d0_rt: ppmu@104a0000 { + compatible = "samsung,exynos-ppmu-v2"; + reg = <0x104a0000 0x2000>; + }; + + ppmu_d1_cpu: ppmu@104b0000 { + compatible = "samsung,exynos-ppmu-v2"; + reg = <0x104b0000 0x2000>; + }; + + ppmu_d1_general: ppmu@104c0000 { + compatible = "samsung,exynos-ppmu-v2"; + reg = <0x104c0000 0x2000>; + }; + + ppmu_d1_rt: ppmu@104d0000 { + compatible = "samsung,exynos-ppmu-v2"; + reg = <0x104d0000 0x2000>; + }; + + - | + // PPMUv1 nodes with event-data-type for Exynos4412 + #include + + ppmu@106a0000 { + compatible = "samsung,exynos-ppmu"; + reg = <0x106a0000 0x2000>; + clocks = <&clock 400>; + clock-names = "ppmu"; + + events { + ppmu-event3-dmc0 { + event-name = "ppmu-event3-dmc0"; + event-data-type = <(PPMU_RO_DATA_CNT | + PPMU_WO_DATA_CNT)>; + }; + }; + }; diff --git a/Documentation/devicetree/bindings/devfreq/nvidia,tegra30-actmon.yaml b/Documentation/devicetree/bindings/devfreq/nvidia,tegra30-actmon.yaml new file mode 100644 index 000000000..ea1dc86bc --- /dev/null +++ b/Documentation/devicetree/bindings/devfreq/nvidia,tegra30-actmon.yaml @@ -0,0 +1,129 @@ +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/devfreq/nvidia,tegra30-actmon.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: NVIDIA Tegra30 Activity Monitor + +maintainers: + - Dmitry Osipenko + - Jon Hunter + - Thierry Reding + +description: | + The activity monitor block collects statistics about the behaviour of other + components in the system. This information can be used to derive the rate at + which the external memory needs to be clocked in order to serve all requests + from the monitored clients. + +properties: + compatible: + oneOf: + - enum: + - nvidia,tegra30-actmon + - nvidia,tegra114-actmon + - nvidia,tegra124-actmon + - items: + - const: nvidia,tegra210-actmon + - const: nvidia,tegra124-actmon + + reg: + maxItems: 1 + + clocks: + maxItems: 2 + + clock-names: + items: + - const: actmon + - const: emc + + resets: + maxItems: 1 + + reset-names: + items: + - const: actmon + + interrupts: + maxItems: 1 + + interconnects: + minItems: 1 + maxItems: 12 + + interconnect-names: + minItems: 1 + maxItems: 12 + description: + Should include name of the interconnect path for each interconnect + entry. Consult TRM documentation for information about available + memory clients, see MEMORY CONTROLLER and ACTIVITY MONITOR sections. + + operating-points-v2: + description: + Should contain freqs and voltages and opp-supported-hw property, which + is a bitfield indicating SoC speedo ID mask. + + "#cooling-cells": + const: 2 + +required: + - compatible + - reg + - clocks + - clock-names + - resets + - reset-names + - interrupts + - interconnects + - interconnect-names + - operating-points-v2 + - "#cooling-cells" + +additionalProperties: false + +examples: + - | + #include + + mc: memory-controller@7000f000 { + compatible = "nvidia,tegra30-mc"; + reg = <0x7000f000 0x400>; + clocks = <&clk 32>; + clock-names = "mc"; + + interrupts = <0 77 4>; + + #iommu-cells = <1>; + #reset-cells = <1>; + #interconnect-cells = <1>; + }; + + emc: external-memory-controller@7000f400 { + compatible = "nvidia,tegra30-emc"; + reg = <0x7000f400 0x400>; + interrupts = <0 78 4>; + clocks = <&clk 57>; + + nvidia,memory-controller = <&mc>; + operating-points-v2 = <&dvfs_opp_table>; + power-domains = <&domain>; + + #interconnect-cells = <0>; + }; + + actmon@6000c800 { + compatible = "nvidia,tegra30-actmon"; + reg = <0x6000c800 0x400>; + interrupts = <0 45 4>; + clocks = <&clk 119>, <&clk 57>; + clock-names = "actmon", "emc"; + resets = <&rst 119>; + reset-names = "actmon"; + operating-points-v2 = <&dvfs_opp_table>; + interconnects = <&mc TEGRA30_MC_MPCORER &emc>; + interconnect-names = "cpu-read"; + #cooling-cells = <2>; + }; -- cgit v1.3.1