sched_ext: Maintain an online cid mask in the scheduler arenagrafted

Schedulers on the default cid mapping treat [0, nr_online_cids) as the
online set and restart on hotplug. Schedulers that install their own mapping
with scx_bpf_cid_override() have no way to learn which cids are online: the
count no longer identifies members and the CPU-form cpumask is unusable from
cid programs. This is an obvious hole in the cid API.

Add scx_bpf_online_cmask(), a kernel-maintained cmask in the scheduler's
arena, allocated alongside the per-CPU scratch masks and populated after the
cid mapping is finalized and before ops.init(), for child schedulers too.
The pointer stays valid through ops.exit() with no reference to take. It is
the arena offset as a void pointer, the same form struct_ops arena arguments
arrive in. The verifier types the void return as a scalar for the program's
arena cast.

The mask follows the SCX hotplug notifications: seeded from cpu_active_mask
and updated before ops.cid_online/offline() runs, so it lags cpu_online_mask
only inside a hotplug transition. Updates walk the scheduler list under the
lock that also serializes unlinking. Reads are live, not atomic snapshots.
Root initialization excludes hotplug.

v2: Reworded the getter kerneldoc (Andrea Righi).

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>

4 files changed, 420 insertions, 0 deletions

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 <s.hauer@pengutronix.de>
+
+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:
+    $ref: /schemas/types.yaml#/definitions/phandle
+    description:
+      Phandle to the syscon managing the "PMU general register files".
+
+required:
+  - compatible
+  - interrupts
+  - reg
+
+if:
+  properties:
+    compatible:
+      contains:
+        enum:
+          - rockchip,rk3399-dfi
+
+then:
+  required:
+    - clocks
+    - clock-names
+
+additionalProperties: false
+
+examples:
+  - |
+    #include <dt-bindings/interrupt-controller/arm-gic.h>
+    #include <dt-bindings/clock/rk3308-cru.h>
+
+    bus {
+      #address-cells = <2>;
+      #size-cells = <2>;
+
+      dfi: dfi@ff630000 {
+        compatible = "rockchip,rk3399-dfi";
+        reg = <0x00 0xff630000 0x00 0x4000>;
+        interrupts = <GIC_SPI 131 IRQ_TYPE_LEVEL_HIGH 0>;
+        rockchip,pmu = <&pmugrf>;
+        clocks = <&cru PCLK_DDR_MON>;
+        clock-names = "pclk_ddr_mon";
+      };
+    };
diff --git a/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-nocp.yaml b/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-nocp.yaml
new file mode 100644
index 000000000..2bdd05af6
--- /dev/null
+++ b/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-nocp.yaml
@@ -0,0 +1,48 @@
+# SPDX-License-Identifier: GPL-2.0
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/devfreq/event/samsung,exynos-nocp.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Samsung Exynos NoC (Network on Chip) Probe
+
+maintainers:
+  - Chanwoo Choi <cw00.choi@samsung.com>
+  - Krzysztof Kozlowski <krzk@kernel.org>
+
+description: |
+  The Samsung Exynos542x SoC has a NoC (Network on Chip) Probe for NoC bus.
+  NoC provides the primitive values to get the performance data. The packets
+  that the Network on Chip (NoC) probes detects are transported over the
+  network infrastructure to observer units. You can configure probes to capture
+  packets with header or data on the data request response network, or as
+  traffic debug or statistic collectors. Exynos542x bus has multiple NoC probes
+  to provide bandwidth information about behavior of the SoC that you can use
+  while analyzing system performance.
+
+properties:
+  compatible:
+    const: samsung,exynos5420-nocp
+
+  clock-names:
+    items:
+      - const: nocp
+
+  clocks:
+    maxItems: 1
+
+  reg:
+    maxItems: 1
+
+required:
+  - compatible
+  - reg
+
+additionalProperties: false
+
+examples:
+  - |
+    nocp_mem0_0: nocp@10ca1000 {
+        compatible = "samsung,exynos5420-nocp";
+        reg = <0x10ca1000 0x200>;
+    };
diff --git a/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-ppmu.yaml b/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-ppmu.yaml
new file mode 100644
index 000000000..d27dcb2fe
--- /dev/null
+++ b/Documentation/devicetree/bindings/devfreq/event/samsung,exynos-ppmu.yaml
@@ -0,0 +1,169 @@
+# SPDX-License-Identifier: GPL-2.0
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/devfreq/event/samsung,exynos-ppmu.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Samsung Exynos SoC PPMU (Platform Performance Monitoring Unit)
+
+maintainers:
+  - Chanwoo Choi <cw00.choi@samsung.com>
+  - Krzysztof Kozlowski <krzk@kernel.org>
+
+description: |
+  The Samsung Exynos SoC has PPMU (Platform Performance Monitoring Unit) for
+  each IP. PPMU provides the primitive values to get performance data. These
+  PPMU events provide information of the SoC's behaviors so that you may use to
+  analyze system performance, to make behaviors visible and to count usages of
+  each IP (DMC, CPU, RIGHTBUS, LEFTBUS, CAM interface, LCD, G3D, MFC).  The
+  Exynos PPMU driver uses the devfreq-event class to provide event data to
+  various devfreq devices. The devfreq devices would use the event data when
+  determining the current state of each IP.
+
+properties:
+  compatible:
+    enum:
+      - samsung,exynos-ppmu
+      - samsung,exynos-ppmu-v2
+
+  clock-names:
+    items:
+      - const: ppmu
+
+  clocks:
+    maxItems: 1
+
+  reg:
+    maxItems: 1
+
+  events:
+    type: object
+
+    patternProperties:
+      '^ppmu-event[0-9]+(-[a-z0-9]+){,2}$':
+        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 <dt-bindings/clock/exynos3250.h>
+
+    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 <dt-bindings/pmu/exynos_ppmu.h>
+
+    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 <digetx@gmail.com>
+  - Jon Hunter <jonathanh@nvidia.com>
+  - Thierry Reding <thierry.reding@gmail.com>
+
+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 <dt-bindings/memory/tegra30-mc.h>
+
+    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>;
+    };