feat(profiling): Add Perfetto trace format support

CHANGELOG.md

@@ -4,6 +4,7 @@
 
 **Features**:
 
+- Add Perfetto trace format support for continuous profiling. ([#5659](https://github.com/getsentry/relay/pull/5659))
 - Add metadata support for the `/upload` endpoint. ([#6028](https://github.com/getsentry/relay/pull/6028))
 - Infer user agents and client addresses in the V2 standalone span pipeline. ([#6047](https://github.com/getsentry/relay/pull/6047))
 

Cargo.lock

@@ -4770,6 +4770,7 @@ dependencies = [
  "hashbrown 0.15.5",
  "insta",
  "itertools 0.14.0",
+ "prost 0.14.3",
  "relay-base-schema",
  "relay-dynamic-config",
  "relay-event-schema",

relay-dynamic-config/src/feature.rs

@@ -56,6 +56,14 @@ pub enum Feature {
     /// Serialized as `organizations:continuous-profiling`.
     #[serde(rename = "organizations:continuous-profiling")]
     ContinuousProfiling,
+    /// Enable Perfetto binary trace processing for continuous profiling.
+    ///
+    /// When enabled, compound profile chunk items with `content_type: "perfetto"` are
+    /// expanded from binary Perfetto format into the Sample v2 JSON format.
+    ///
+    /// Serialized as `organizations:continuous-profiling-perfetto`.
+    #[serde(rename = "organizations:continuous-profiling-perfetto")]
+    ContinuousProfilingPerfetto,
     /// Enable log ingestion for our log product (this is not internal logging).
     ///
     /// Serialized as `organizations:ourlogs-ingestion`.
@@ -198,4 +206,18 @@ mod tests {
             r#"["organizations:session-replay"]"#
         );
     }
+
+    #[test]
+    fn test_continuous_profiling_perfetto_serde() {
+        // Verify the serialized name matches what Sentry's backend sends.
+        let serialized = serde_json::to_string(&Feature::ContinuousProfilingPerfetto).unwrap();
+        assert_eq!(
+            serialized,
+            r#""organizations:continuous-profiling-perfetto""#
+        );
+
+        let deserialized: Feature =
+            serde_json::from_str(r#""organizations:continuous-profiling-perfetto""#).unwrap();
+        assert_eq!(deserialized, Feature::ContinuousProfilingPerfetto);
+    }
 }

relay-profiling/Cargo.toml

@@ -19,6 +19,7 @@ chrono = { workspace = true }
 data-encoding = { workspace = true }
 hashbrown = { workspace = true }
 itertools = { workspace = true }
+prost = { workspace = true }
 relay-base-schema = { workspace = true }
 relay-dynamic-config = { workspace = true }
 relay-event-schema = { workspace = true }

relay-profiling/protos/README.md

@@ -0,0 +1,18 @@
+# Perfetto Proto Definitions
+
+`perfetto_trace.proto` contains a minimal subset of the
+[Perfetto trace proto definitions](https://github.com/google/perfetto/tree/master/protos/perfetto/trace)
+needed to decode profiling data. Field numbers match the upstream definitions.
+
+The generated Rust code is checked in at `../src/perfetto/proto.rs`.
+
+## Regenerating
+
+Prerequisites:
+- `protoc`: https://github.com/protocolbuffers/protobuf/releases (or `brew install protobuf`)
+- `protoc-gen-prost`: `cargo install protoc-gen-prost`
+
+Then run:
+```sh
+./relay-profiling/protos/generate.sh
+```

relay-profiling/protos/generate.sh

@@ -0,0 +1,59 @@
+#!/usr/bin/env bash
+#
+# Regenerates the checked-in Rust protobuf bindings for Perfetto trace types
+# using protoc with the protoc-gen-prost plugin.
+#
+# Usage:
+#   ./relay-profiling/protos/generate.sh
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+PROTO_FILE="$SCRIPT_DIR/perfetto_trace.proto"
+OUTPUT_FILE="$SCRIPT_DIR/../src/perfetto/proto.rs"
+
+if ! command -v protoc &>/dev/null; then
+    echo "error: protoc is not installed." >&2
+    echo "       Install it from https://github.com/protocolbuffers/protobuf/releases" >&2
+    echo "       or: brew install protobuf" >&2
+    exit 1
+fi
+echo "Using protoc: $(command -v protoc) ($(protoc --version))"
+
+if ! command -v protoc-gen-prost &>/dev/null; then
+    echo "error: protoc-gen-prost is not installed." >&2
+    echo "       Install it with: cargo install protoc-gen-prost" >&2
+    exit 1
+fi
+echo "Using protoc-gen-prost: $(command -v protoc-gen-prost)"
+
+if [[ ! -f "$PROTO_FILE" ]]; then
+    echo "error: proto file not found at $PROTO_FILE" >&2
+    exit 1
+fi
+
+TMPDIR="$(mktemp -d)"
+trap 'rm -rf "$TMPDIR"' EXIT
+
+echo "Generating Rust bindings..."
+protoc \
+    --prost_out="$TMPDIR" \
+    --proto_path="$SCRIPT_DIR" \
+    "$PROTO_FILE"
+
+# protoc-gen-prost mirrors the proto package path in the output directory.
+GENERATED=$(find "$TMPDIR" -name '*.rs' -type f | head -1)
+
+if [[ -z "$GENERATED" || ! -f "$GENERATED" ]]; then
+    echo "error: no generated .rs file found in $TMPDIR" >&2
+    exit 1
+fi
+
+if [[ ! -s "$GENERATED" ]]; then
+    echo "error: generated file is empty" >&2
+    exit 1
+fi
+
+cp "$GENERATED" "$OUTPUT_FILE"
+echo "Updated $OUTPUT_FILE"
+echo "Done."

relay-profiling/protos/perfetto_trace.proto

@@ -0,0 +1,88 @@
+// Minimal subset of the Perfetto trace proto definitions needed to decode
+// profiling data. Field numbers match the upstream definitions at
+// https://github.com/google/perfetto/tree/master/protos/perfetto/trace
+
+syntax = "proto2";
+package perfetto.protos;
+
+message Trace {
+  repeated TracePacket packet = 1;
+}
+
+message TracePacket {
+  optional uint64 timestamp = 8;
+
+  oneof optional_trusted_packet_sequence_id {
+    uint32 trusted_packet_sequence_id = 10;
+  }
+
+  optional InternedData interned_data = 12;
+  optional uint32 sequence_flags = 13;
+
+  // Only the oneof variants we care about; prost will skip the rest.
+  oneof data {
+    ClockSnapshot clock_snapshot = 6;
+    PerfSample perf_sample = 66;
+  }
+}
+
+// --- clock sync ---------------------------------------------------------------
+
+message ClockSnapshot {
+  message Clock {
+    optional uint32 clock_id = 1;
+    optional uint64 timestamp = 2;
+  }
+  repeated Clock clocks = 1;
+  optional uint32 primary_trace_clock = 2;
+}
+
+// --- interned data -----------------------------------------------------------
+
+message InternedData {
+  repeated InternedString function_names = 5;
+  repeated Frame frames = 6;
+  repeated Callstack callstacks = 7;
+  repeated InternedString build_ids = 16;
+  repeated InternedString mapping_paths = 17;
+  repeated Mapping mappings = 19;
+}
+
+message InternedString {
+  optional uint64 iid = 1;
+  optional bytes str = 2;
+}
+
+// --- profiling common --------------------------------------------------------
+
+message Frame {
+  optional uint64 iid = 1;
+  optional uint64 function_name_id = 2;
+  optional uint64 mapping_id = 3;
+  optional uint64 rel_pc = 4;
+}
+
+message Mapping {
+  optional uint64 iid = 1;
+  optional uint64 build_id = 2;
+  optional uint64 start_offset = 3;
+  optional uint64 start = 4;
+  optional uint64 end = 5;
+  optional uint64 load_bias = 6;
+  repeated uint64 path_string_ids = 7;
+  optional uint64 exact_offset = 8;
+}
+
+message Callstack {
+  optional uint64 iid = 1;
+  repeated uint64 frame_ids = 2;
+}
+
+// --- profiling packets -------------------------------------------------------
+
+message PerfSample {
+  optional uint32 cpu = 1;
+  optional uint32 pid = 2;
+  optional uint32 tid = 3;
+  optional uint64 callstack_iid = 4;
+}

relay-profiling/src/debug_image.rs

@@ -6,40 +6,45 @@ use uuid::{Error as UuidError, Uuid};
 
 use crate::utils;
 
+/// The type of a debug image.
 #[derive(Debug, Serialize, Deserialize, PartialEq, Clone)]
 #[serde(rename_all = "lowercase")]
-enum ImageType {
+pub enum ImageType {
     MachO,
     Symbolic,
     Sourcemap,
     Proguard,
     Jvm,
 }
 
+/// A debug information image referenced by a profile.
 #[derive(Debug, Serialize, Deserialize, PartialEq, Clone)]
 pub struct DebugImage {
+    /// Path or name of the code file (e.g. shared library or executable).
     #[serde(skip_serializing_if = "Option::is_none", alias = "name")]
-    code_file: Option<NativeImagePath>,
+    pub code_file: Option<NativeImagePath>,
+    /// Debug identifier for symbolication.
     #[serde(skip_serializing_if = "Option::is_none", alias = "id")]
-    debug_id: Option<DebugId>,
+    pub debug_id: Option<DebugId>,
+    /// The type of debug image (e.g. `symbolic`, `proguard`).
     #[serde(rename = "type")]
-    image_type: ImageType,
-
+    pub image_type: ImageType,
+    /// Start address of the image in virtual memory.
     #[serde(skip_serializing_if = "Option::is_none")]
-    image_addr: Option<Addr>,
-
+    pub image_addr: Option<Addr>,
+    /// Preferred load address of the image in virtual memory.
     #[serde(skip_serializing_if = "Option::is_none")]
-    image_vmaddr: Option<Addr>,
-
+    pub image_vmaddr: Option<Addr>,
+    /// Size of the image in bytes.
     #[serde(
         default,
         deserialize_with = "utils::deserialize_number_from_string",
         skip_serializing_if = "utils::is_zero"
     )]
-    image_size: u64,
-
+    pub image_size: u64,
+    /// Optional UUID, used as the build ID for proguard images.
     #[serde(skip_serializing_if = "Option::is_none", alias = "build_id")]
-    uuid: Option<Uuid>,
+    pub uuid: Option<Uuid>,
 }
 
 pub fn get_proguard_image(uuid: &str) -> Result<DebugImage, UuidError> {