sched_ext: Add scx_layered, a highly configurable multi-layer scheduler

scx_layered allows classifying tasks into multiple layers and applying
different scheduling policies to them. The configuration is specified in
json and composed of two parts - matches and policies.

The matches classify tasks into a layer and different scheduling policies
can be applied to the layer. For example, if a layer is "Confined", its
tasks are confined to the CPUs that are allocated to the layer. The CPU
allocation is dynamically adjusted to target CPU utilization in a specific
range. This can be used to e.g. confine managerial workloads to a small
number of CPUs while leaving the rest for latency sensitive workloads. A
layer can also be configured to preempt other non-preempting layers.

This scheduler is limited in that it assumes homogenous CPUs in a single
node and the load based CPU allocation limit doesn't work well yet. However,
even in the current form, in an experimental deployment over ~1000 machines,
it showed a significant (>5%) bandwidth gain with a large-scale latency
sensitive workload.

As sched_ext scheduler development style and cadence are different from
kernel proper, the plan is to keep simpler example schedulers and one more
generic scheduler (scx_rusty) in tree and publish others in a separate
repository. This scheduler will be a part of that repo once ready.

NOT_FOR_UPSTREAM_INCLUSION

tools/sched_ext/Makefile

@@ -201,7 +201,7 @@ $(c-sched-targets): %: $(BINDIR)/%
 ###################
 # Rust schedulers #
 ###################
-rust-sched-targets := scx_rusty
+rust-sched-targets := scx_rusty scx_layered
 
 # Separate build target that is available for build systems to use to fetch
 # dependencies in a separate step from building. This allows the scheduler

tools/sched_ext/scx_layered/.gitignore

@@ -0,0 +1,3 @@
+src/bpf/.output
+Cargo.lock
+target

tools/sched_ext/scx_layered/Cargo.toml

@@ -0,0 +1,30 @@
+[package]
+name = "scx_layered"
+version = "0.0.1"
+authors = ["Tejun Heo <[email protected]>", "Meta"]
+edition = "2021"
+description = "Userspace scheduling with BPF for Ads"
+license = "GPL-2.0-only"
+
+[dependencies]
+anyhow = "1.0"
+bitvec = "1.0"
+clap = { version = "4.1", features = ["derive", "env", "unicode", "wrap_help"] }
+ctrlc = { version = "3.1", features = ["termination"] }
+fb_procfs = "0.7"
+lazy_static = "1.4"
+libbpf-rs = "0.21"
+libbpf-sys = { version = "1.2.0", features = ["novendor", "static"] }
+libc = "0.2"
+log = "0.4"
+serde = { version = "1.0", features = ["derive"] }
+serde_json = "1.0"
+simplelog = "0.12"
+
+[build-dependencies]
+bindgen = { version = "0.61" }
+libbpf-cargo = "0.21"
+glob = "0.3"
+
+[features]
+enable_backtrace = []

tools/sched_ext/scx_layered/build.rs

@@ -0,0 +1,77 @@
+// Copyright (c) Meta Platforms, Inc. and affiliates.
+
+// This software may be used and distributed according to the terms of the
+// GNU General Public License version 2.
+extern crate bindgen;
+
+use std::env;
+use std::fs::create_dir_all;
+use std::path::Path;
+use std::path::PathBuf;
+
+use glob::glob;
+use libbpf_cargo::SkeletonBuilder;
+
+const HEADER_PATH: &str = "src/bpf/layered.h";
+
+fn bindgen_layered() {
+    // Tell cargo to invalidate the built crate whenever the wrapper changes
+    println!("cargo:rerun-if-changed={}", HEADER_PATH);
+
+    // The bindgen::Builder is the main entry point
+    // to bindgen, and lets you build up options for
+    // the resulting bindings.
+    let bindings = bindgen::Builder::default()
+        // The input header we would like to generate
+        // bindings for.
+        .header(HEADER_PATH)
+        // Tell cargo to invalidate the built crate whenever any of the
+        // included header files changed.
+        .parse_callbacks(Box::new(bindgen::CargoCallbacks))
+        // Finish the builder and generate the bindings.
+        .generate()
+        // Unwrap the Result and panic on failure.
+        .expect("Unable to generate bindings");
+
+    // Write the bindings to the $OUT_DIR/bindings.rs file.
+    let out_path = PathBuf::from(env::var("OUT_DIR").unwrap());
+    bindings
+        .write_to_file(out_path.join("layered_sys.rs"))
+        .expect("Couldn't write bindings!");
+}
+
+fn gen_bpf_sched(name: &str) {
+    let bpf_cflags = env::var("SCX_RUST_BPF_CFLAGS").unwrap();
+    let clang = env::var("SCX_RUST_CLANG").unwrap();
+    eprintln!("{}", clang);
+    let outpath = format!("./src/bpf/.output/{}.skel.rs", name);
+    let skel = Path::new(&outpath);
+    let src = format!("./src/bpf/{}.bpf.c", name);
+    let obj = format!("./src/bpf/.output/{}.bpf.o", name);
+    SkeletonBuilder::new()
+        .source(src.clone())
+	.obj(obj)
+        .clang(clang)
+        .clang_args(bpf_cflags)
+        .build_and_generate(skel)
+        .unwrap();
+
+    // Trigger rebuild if any .[hc] files are changed in the directory.
+    for path in glob("./src/bpf/*.[hc]").unwrap().filter_map(Result::ok) {
+        println!("cargo:rerun-if-changed={}", path.to_str().unwrap());
+    }
+}
+
+fn main() {
+    bindgen_layered();
+    // It's unfortunate we cannot use `OUT_DIR` to store the generated skeleton.
+    // Reasons are because the generated skeleton contains compiler attributes
+    // that cannot be `include!()`ed via macro. And we cannot use the `#[path = "..."]`
+    // trick either because you cannot yet `concat!(env!("OUT_DIR"), "/skel.rs")` inside
+    // the path attribute either (see https://github.com/rust-lang/rust/pull/83366).
+    //
+    // However, there is hope! When the above feature stabilizes we can clean this
+    // all up.
+    create_dir_all("./src/bpf/.output").unwrap();
+    gen_bpf_sched("layered");
+}

tools/sched_ext/scx_layered/rustfmt.toml

@@ -0,0 +1,8 @@
+# Get help on options with `rustfmt --help=config`
+# Please keep these in alphabetical order.
+edition = "2021"
+group_imports = "StdExternalCrate"
+imports_granularity = "Item"
+merge_derives = false
+use_field_init_shorthand = true
+version = "Two"