feat(runner-shared): add new fields to perf metadata

crates/runner-shared/src/debug_info.rs

@@ -21,6 +21,13 @@ impl std::fmt::Debug for DebugInfo {
     }
 }
 
+/// Per-pid mounting info referencing a deduplicated debug info entry.
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct MappedProcessDebugInfo {
+    pub debug_info_key: String,
+    pub load_bias: u64,
+}
+
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct ModuleDebugInfo {
     /// The path to the object file on disk (e.g. `/usr/lib/libc.so.6`)

crates/runner-shared/src/lib.rs

@@ -2,6 +2,7 @@ pub mod artifacts;
 pub mod debug_info;
 pub mod fifo;
 pub mod metadata;
+pub mod module_symbols;
 pub mod perf_event;
 pub mod unwind_data;
 pub mod walltime_results;

crates/runner-shared/src/metadata.rs

@@ -1,33 +1,74 @@
 use anyhow::Context;
+use libc::pid_t;
 use serde::{Deserialize, Serialize};
+use std::collections::HashMap;
 use std::io::BufWriter;
 use std::path::Path;
+use std::path::PathBuf;
 
-use crate::debug_info::ModuleDebugInfo;
+use crate::debug_info::{MappedProcessDebugInfo, ModuleDebugInfo};
 use crate::fifo::MarkerType;
+use crate::module_symbols::MappedProcessModuleSymbols;
+use crate::unwind_data::MappedProcessUnwindData;
 
-#[derive(Serialize, Deserialize)]
+#[derive(Serialize, Deserialize, Default)]
 pub struct PerfMetadata {
     /// The version of this metadata format.
     pub version: u64,
 
     /// Name and version of the integration
     pub integration: (String, String),
 
+    /// Per-pid modules that should be ignored, with runtime address ranges derived from symbol bounds + load bias
+    #[serde(default, skip_serializing_if = "HashMap::is_empty")]
+    pub ignored_modules_by_pid: HashMap<pid_t, Vec<(String, u64, u64)>>,
+
+    /// Deduplicated debug info entries, keyed by semantic key
+    #[serde(default, skip_serializing_if = "HashMap::is_empty")]
+    pub debug_info: HashMap<String, ModuleDebugInfo>,
+
+    /// Per-pid debug info references, mapping PID to mounted modules' debug info
+    /// Referenced by `path_keys` that point to the deduplicated `debug_info` entries.
+    #[serde(default, skip_serializing_if = "HashMap::is_empty")]
+    pub mapped_process_debug_info_by_pid: HashMap<pid_t, Vec<MappedProcessDebugInfo>>,
+
+    /// Per-pid unwind data references, mapping PID to mounted modules' unwind data
+    /// Referenced by `path_keys` that point to the deduplicated `unwind_data` files on disk.
+    #[serde(default, skip_serializing_if = "HashMap::is_empty")]
+    pub mapped_process_unwind_data_by_pid: HashMap<pid_t, Vec<MappedProcessUnwindData>>,
+
+    /// Per-pid symbol references, mapping PID to its mounted modules' symbols
+    /// Referenced by `path_keys` that point to the deduplicated `symbols.map` files on disk.
+    #[serde(default, skip_serializing_if = "HashMap::is_empty")]
+    pub mapped_process_module_symbols: HashMap<pid_t, Vec<MappedProcessModuleSymbols>>,
+
+    /// Mapping from semantic `path_key` to original binary path on host disk
+    /// Used by `mapped_process_debug_info_by_pid`, `mapped_process_unwind_data_by_pid` and
+    /// `mapped_process_module_symbols` the deduplicated entries
+    ///
+    /// Until now, only kept for traceability, if we ever need to reconstruct the original paths from the keys
+    #[serde(default, skip_serializing_if = "HashMap::is_empty")]
+    pub path_key_to_path: HashMap<String, PathBuf>,
+
+    // Deprecated fields below are kept for backward compatibility, since this struct is used in
+    // the parser and older versions of the runner still generate them
+    //
     /// The URIs of the benchmarks with the timestamps they were executed at.
     #[deprecated(note = "Use ExecutionTimestamps in the 'artifacts' module instead")]
     pub uri_by_ts: Vec<(u64, String)>,
 
     /// Modules that should be ignored and removed from the folded trace and callgraph (e.g. python interpreter)
+    #[deprecated(note = "Use 'ignored_modules_by_pid' instead")]
     pub ignored_modules: Vec<(String, u64, u64)>,
 
     /// Marker for certain regions in the profiling data
     #[deprecated(note = "Use ExecutionTimestamps in the 'artifacts' module instead")]
     pub markers: Vec<MarkerType>,
 
-    /// Debug info for all modules across all processes, mapping PID to module debug info
-    #[serde(default, skip_serializing_if = "std::collections::HashMap::is_empty")]
-    pub debug_info_by_pid: std::collections::HashMap<i32, Vec<ModuleDebugInfo>>,
+    /// Kept for backward compatibility, was used before deduplication of debug info entries.
+    #[serde(default, skip_serializing_if = "HashMap::is_empty")]
+    #[deprecated(note = "Use 'debug_info' + 'mapped_process_debug_info_by_pid' instead")]
+    pub debug_info_by_pid: HashMap<pid_t, Vec<ModuleDebugInfo>>,
 }
 
 impl PerfMetadata {

crates/runner-shared/src/module_symbols.rs

@@ -0,0 +1,11 @@
+use serde::{Deserialize, Serialize};
+
+/// File suffix used when registering module symbols in a PID agnostic way.
+pub const SYMBOLS_MAP_SUFFIX: &str = "symbols.map";
+
+/// Per-pid mounting info referencing a deduplicated perf map entry.
+#[derive(Serialize, Deserialize, Clone, Debug)]
+pub struct MappedProcessModuleSymbols {
+    pub perf_map_key: String,
+    pub load_bias: u64,
+}