Integrate the SVG flame graph generation into OSDK

Makefile

@@ -21,7 +21,7 @@ OSTD_TASK_STACK_SIZE_IN_PAGES ?= 64
 
 # GDB debugging and profiling options.
 GDB_TCP_PORT ?= 1234
-GDB_PROFILE_FORMAT ?= folded
+GDB_PROFILE_FORMAT ?= flame-graph
 GDB_PROFILE_COUNT ?= 200
 GDB_PROFILE_INTERVAL ?= 0.1
 # End of GDB options.

docs/src/osdk/reference/commands/profile.md

@@ -4,9 +4,9 @@
 
 The profile command is used to collect stack traces when running the target
 kernel in QEMU. It attaches to the GDB server initiated with the run subcommand
-and collects the stack trace periodically. The collected data can be
+and collects the stack trace periodically. The collected information can be
-further analyzed using tools like
+used to directly generate a flame graph, or be stored for later analysis using
-[flame graph](https://github.com/brendangregg/FlameGraph).
+[the original flame graph tool](https://github.com/brendangregg/FlameGraph).
 
 ## Options
 
@@ -33,10 +33,11 @@ Parse a collected JSON profile file into other formats.
 Possible values:
     - `json`:   The parsed stack trace log from GDB in JSON.
     - `folded`: The folded stack trace for flame graph.
+    - `flame-graph`: A SVG flame graph.
 
 If the user does not specify the format, it will be inferred from the
 output file extension. If the output file does not have an extension,
-the default format is folded stack traces.
+the default format is flame graph.
 
 `--cpu-mask <CPU_MASK>`:
 

osdk/Cargo.lock

@@ -15,6 +15,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "e89da841a80418a9b391ebaea17f5c112ffaaa96f621d2c285b5174da76b9011"
 dependencies = [
  "cfg-if",
+ "getrandom",
  "once_cell",
  "version_check",
  "zerocopy",
@@ -98,6 +99,12 @@ dependencies = [
  "windows-sys",
 ]
 
+[[package]]
+name = "arrayvec"
+version = "0.7.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7c02d123df017efcdfbd739ef81735b36c5ba83ec3c59c80a9d7ecc718f92e50"
+
 [[package]]
 name = "assert_cmd"
 version = "2.0.14"
@@ -125,6 +132,12 @@ version = "1.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "bef38d45163c2f1dde094a7dfd33ccf595c92905c8f8f4fdc18d06fb1037718a"
 
+[[package]]
+name = "bitflags"
+version = "2.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b048fb63fd8b5923fc5aa7b340d8e156aec7ec02f0c78fa8a6ddc2613f6f71de"
+
 [[package]]
 name = "block-buffer"
 version = "0.10.4"
@@ -181,6 +194,7 @@ dependencies = [
  "env_logger",
  "indexmap",
  "indicatif",
+ "inferno",
  "lazy_static",
  "linux-bzimage-builder",
  "log",
@@ -316,6 +330,21 @@ dependencies = [
  "cfg-if",
 ]
 
+[[package]]
+name = "crossbeam-channel"
+version = "0.5.13"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "33480d6946193aa8033910124896ca395333cae7e2d1113d1fef6c3272217df2"
+dependencies = [
+ "crossbeam-utils",
+]
+
+[[package]]
+name = "crossbeam-utils"
+version = "0.8.20"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "22ec99545bb0ed0ea7bb9b8e1e9122ea386ff8a48c0922e43f36d45ab09e0e80"
+
 [[package]]
 name = "crypto-common"
 version = "0.1.6"
@@ -332,6 +361,20 @@ version = "0.3.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7762d17f1241643615821a8455a0b2c3e803784b058693d990b11f2dce25a0ca"
 
+[[package]]
+name = "dashmap"
+version = "6.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5041cc499144891f3790297212f32a74fb938e5136a14943f338ef9e0ae276cf"
+dependencies = [
+ "cfg-if",
+ "crossbeam-utils",
+ "hashbrown",
+ "lock_api",
+ "once_cell",
+ "parking_lot_core",
+]
+
 [[package]]
 name = "difflib"
 version = "0.4.0"
@@ -399,6 +442,17 @@ dependencies = [
  "version_check",
 ]
 
+[[package]]
+name = "getrandom"
+version = "0.2.15"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c4567c8db10ae91089c99af84c68c38da3ec2f087c3f82960bcdbf3656b6f4d7"
+dependencies = [
+ "cfg-if",
+ "libc",
+ "wasi",
+]
+
 [[package]]
 name = "hashbrown"
 version = "0.14.3"
@@ -415,6 +469,12 @@ version = "0.4.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "95505c38b4572b2d910cecb0281560f54b440a19336cbbcb27bf6ce6adc6f5a8"
 
+[[package]]
+name = "hermit-abi"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fbf6a919d6cf397374f7dfeeea91d974c7c0a7221d0d0f4f20d859d329e53fcc"
+
 [[package]]
 name = "humantime"
 version = "2.1.0"
@@ -467,6 +527,29 @@ dependencies = [
  "unicode-width",
 ]
 
+[[package]]
+name = "inferno"
+version = "0.11.21"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "232929e1d75fe899576a3d5c7416ad0d88dbfbb3c3d6aa00873a7408a50ddb88"
+dependencies = [
+ "ahash",
+ "clap",
+ "crossbeam-channel",
+ "crossbeam-utils",
+ "dashmap",
+ "env_logger",
+ "indexmap",
+ "is-terminal",
+ "itoa",
+ "log",
+ "num-format",
+ "once_cell",
+ "quick-xml",
+ "rgb",
+ "str_stack",
+]
+
 [[package]]
 name = "instant"
 version = "0.1.13"
@@ -476,6 +559,17 @@ dependencies = [
  "cfg-if",
 ]
 
+[[package]]
+name = "is-terminal"
+version = "0.4.13"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "261f68e344040fbd0edea105bef17c66edf46f984ddb1115b775ce31be948f4b"
+dependencies = [
+ "hermit-abi",
+ "libc",
+ "windows-sys",
+]
+
 [[package]]
 name = "itoa"
 version = "1.0.10"
@@ -499,9 +593,9 @@ checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646"
 
 [[package]]
 name = "libc"
-version = "0.2.153"
+version = "0.2.159"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9c198f91728a82281a64e1f4f9eeb25d82cb32a5de251c6bd1b5154d63a8e7bd"
+checksum = "561d97a539a36e26a9a5fad1ea11a3039a67714694aaa379433e580854bc3dc5"
 
 [[package]]
 name = "libflate"
@@ -531,13 +625,23 @@ dependencies = [
 name = "linux-bzimage-builder"
 version = "0.2.0"
 dependencies = [
- "bitflags",
+ "bitflags 1.3.2",
  "bytemuck",
  "libflate",
  "serde",
  "xmas-elf",
 ]
 
+[[package]]
+name = "lock_api"
+version = "0.4.12"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "07af8b9cdd281b7915f413fa73f29ebd5d55d0d3f0155584dade1ff18cea1b17"
+dependencies = [
+ "autocfg",
+ "scopeguard",
+]
+
 [[package]]
 name = "log"
 version = "0.4.20"
@@ -550,6 +654,16 @@ version = "2.7.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "523dc4f511e55ab87b694dc30d0f820d60906ef06413f93d4d7a1385599cc149"
 
+[[package]]
+name = "num-format"
+version = "0.4.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a652d9771a63711fd3c3deb670acfbe5c30a4072e664d7a3bf5a9e1056ac72c3"
+dependencies = [
+ "arrayvec",
+ "itoa",
+]
+
 [[package]]
 name = "num-traits"
 version = "0.2.19"
@@ -571,6 +685,19 @@ version = "1.19.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3fdb12b2476b595f9358c5161aa467c2438859caa136dec86c26fdd2efe17b92"
 
+[[package]]
+name = "parking_lot_core"
+version = "0.9.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1e401f977ab385c9e4e3ab30627d6f26d00e2c73eef317493c4ec6d468726cf8"
+dependencies = [
+ "cfg-if",
+ "libc",
+ "redox_syscall",
+ "smallvec",
+ "windows-targets",
+]
+
 [[package]]
 name = "portable-atomic"
 version = "1.8.0"
@@ -613,6 +740,15 @@ dependencies = [
  "unicode-ident",
 ]
 
+[[package]]
+name = "quick-xml"
+version = "0.26.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7f50b1c63b38611e7d4d7f68b82d3ad0cc71a2ad2e7f61fc10f1328d917c93cd"
+dependencies = [
+ "memchr",
+]
+
 [[package]]
 name = "quote"
 version = "1.0.35"
@@ -622,6 +758,15 @@ dependencies = [
  "proc-macro2",
 ]
 
+[[package]]
+name = "redox_syscall"
+version = "0.5.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "355ae415ccd3a04315d3f8246e86d67689ea74d88d915576e1589a351062a13b"
+dependencies = [
+ "bitflags 2.6.0",
+]
+
 [[package]]
 name = "regex"
 version = "1.10.4"
@@ -660,6 +805,15 @@ dependencies = [
  "memchr",
 ]
 
+[[package]]
+name = "rgb"
+version = "0.8.50"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "57397d16646700483b67d2dd6511d79318f9d057fdbd21a4066aeac8b41d310a"
+dependencies = [
+ "bytemuck",
+]
+
 [[package]]
 name = "rle-decode-fast"
 version = "1.0.3"
@@ -672,6 +826,12 @@ version = "1.0.17"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "e86697c916019a8588c99b5fac3cead74ec0b4b819707a682fd4d23fa0ce1ba1"
 
+[[package]]
+name = "scopeguard"
+version = "1.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "94143f37725109f92c262ed2cf5e59bce7498c01bcc1502d7b9afe439a4e9f49"
+
 [[package]]
 name = "serde"
 version = "1.0.197"
@@ -729,6 +889,18 @@ version = "1.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "0fda2ff0d084019ba4d7c6f371c95d8fd75ce3524c3cb8fb653a3023f6323e64"
 
+[[package]]
+name = "smallvec"
+version = "1.13.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3c5e1a9a646d36c3599cd173a41282daf47c44583ad367b8e6837255952e5c67"
+
+[[package]]
+name = "str_stack"
+version = "0.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9091b6114800a5f2141aee1d1b9d6ca3592ac062dc5decb3764ec5895a47b4eb"
+
 [[package]]
 name = "strsim"
 version = "0.11.0"
@@ -826,6 +998,12 @@ dependencies = [
  "libc",
 ]
 
+[[package]]
+name = "wasi"
+version = "0.11.0+wasi-snapshot-preview1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9c8d87e72b64a3b4db28d11ce29237c246188f4f51057d65a7eab63b7987e423"
+
 [[package]]
 name = "wasm-bindgen"
 version = "0.2.93"

osdk/Cargo.toml

@@ -21,6 +21,7 @@ version = "0.2.0"
 clap = { version = "4.4.17", features = ["cargo", "derive"] }
 chrono = "0.4.38"
 env_logger = "0.11.0"
+inferno = "0.11.21"
 indexmap = "2.2.1"
 indicatif = "0.17.8" # For a commandline progress bar
 lazy_static = "1.4.0"

osdk/src/cli.rs

@@ -254,9 +254,11 @@ pub struct ProfileArgs {
 pub enum ProfileFormat {
     /// The raw stack trace log parsed from GDB in JSON
     Json,
-    /// The folded stack trace for a
+    /// The folded stack trace for generating a flame graph later using
-    /// [flame graph](https://github.com/brendangregg/FlameGraph)
+    /// [the original tool](https://github.com/brendangregg/FlameGraph)
     Folded,
+    /// A SVG flame graph
+    FlameGraph,
 }
 
 impl ProfileFormat {
@@ -264,6 +266,7 @@ impl ProfileFormat {
         match self {
             ProfileFormat::Json => "json",
             ProfileFormat::Folded => "folded",
+            ProfileFormat::FlameGraph => "svg",
         }
     }
 }
@@ -291,17 +294,18 @@ impl DebugProfileOutArgs {
     ///
     /// If the user does not specify the format, it will be inferred from the
     /// output file extension. If the output file does not have an extension,
-    /// the default format is folded stack traces.
+    /// the default format is flame graph.
     pub fn format(&self) -> ProfileFormat {
         self.format.unwrap_or_else(|| {
             if self.output.is_some() {
                 match self.output.as_ref().unwrap().extension() {
                     Some(ext) if ext == "folded" => ProfileFormat::Folded,
                     Some(ext) if ext == "json" => ProfileFormat::Json,
-                    _ => ProfileFormat::Folded,
+                    Some(ext) if ext == "svg" => ProfileFormat::FlameGraph,
+                    _ => ProfileFormat::FlameGraph,
                 }
             } else {
-                ProfileFormat::Folded
+                ProfileFormat::FlameGraph
             }
         })
     }

osdk/src/commands/profile.rs

@@ -8,6 +8,8 @@
 //! further analyzed using tools like
 //! [flame graph](https://github.com/brendangregg/FlameGraph).
 
+use inferno::flamegraph;
+
 use crate::{
     cli::{ProfileArgs, ProfileFormat},
     commands::util::bin_file_name,
@@ -103,51 +105,69 @@ impl Profile {
     fn serialize_to<W: Write>(&self, format: ProfileFormat, cpu_mask: u128, mut target: W) {
         match format {
             ProfileFormat::Folded => {
-                let mut folded = HashMap::new();
+                let folded = self.fold(cpu_mask);
 
-                // Process each stack trace and fold it for flame graph format
+                // Write the folded traces to the target text writer.
-                for capture in &self.stack_traces {
-                    for (cpu_id, stack) in capture {
-                        if *cpu_id >= 128 || cpu_mask & (1u128 << *cpu_id) == 0 {
-                            continue;
-                        }
-
-                        // Fold the stack trace
-                        let folded_key = stack.iter().rev().cloned().collect::<Vec<_>>().join(";");
-                        *folded.entry(folded_key).or_insert(0) += 1;
-                    }
-                }
-
-                // Write the folded traces
                 for (key, count) in folded {
                     writeln!(&mut target, "{} {}", key, count)
                         .expect("Failed to write folded output");
                 }
             }
             ProfileFormat::Json => {
-                // Filter out the stack traces based on the CPU mask
+                let filtered = self.filter_cpu(cpu_mask);
-                let filtered_traces = self
-                    .stack_traces
-                    .iter()
-                    .map(|capture| {
-                        capture
-                            .iter()
-                            .filter(|(cpu_id, _)| {
-                                **cpu_id < 128 && cpu_mask & (1u128 << **cpu_id) != 0
-                            })
-                            .map(|(cpu_id, stack)| (*cpu_id, stack.clone()))
-                            .collect::<HashMap<_, _>>()
-                    })
-                    .collect::<Vec<_>>();
-
-                let filtered = Profile {
-                    stack_traces: filtered_traces,
-                };
 
                 serde_json::to_writer(target, &filtered).expect("Failed to write JSON output");
             }
+            ProfileFormat::FlameGraph => {
+                let folded = self.fold(cpu_mask);
+
+                // Generate the flame graph folded text lines.
+                let lines = folded
+                    .iter()
+                    .map(|(key, count)| format!("{} {}", key, count))
+                    .collect::<Vec<_>>();
+
+                // Generate the flame graph to the target SVG writer.
+                let mut opt = flamegraph::Options::default();
+                flamegraph::from_lines(&mut opt, lines.iter().map(|s| s.as_str()), target).unwrap();
+            }
         }
     }
+
+    fn filter_cpu(&self, cpu_mask: u128) -> Profile {
+        let filtered_traces = self
+            .stack_traces
+            .iter()
+            .map(|capture| {
+                capture
+                    .iter()
+                    .filter(|(cpu_id, _)| **cpu_id < 128 && cpu_mask & (1u128 << **cpu_id) != 0)
+                    .map(|(cpu_id, stack)| (*cpu_id, stack.clone()))
+                    .collect::<HashMap<_, _>>()
+            })
+            .collect::<Vec<_>>();
+
+        Self {
+            stack_traces: filtered_traces,
+        }
+    }
+
+    fn fold(&self, cpu_mask: u128) -> HashMap<String, u32> {
+        let mut folded = HashMap::new();
+
+        for capture in &self.stack_traces {
+            for (cpu_id, stack) in capture {
+                if *cpu_id >= 128 || cpu_mask & (1u128 << *cpu_id) == 0 {
+                    continue;
+                }
+
+                let folded_key = stack.iter().rev().cloned().collect::<Vec<_>>().join(";");
+                *folded.entry(folded_key).or_insert(0) += 1;
+            }
+        }
+
+        folded
+    }
 }
 
 #[derive(Debug)]