feat: add unwind data v2 format with base_svma

Author: not-matthias

Cargo.lock

@@ -4,6 +4,8 @@ version = "0.1.0"
 edition = "2024"
 
 [dependencies]
-anyhow = "1.0.100"
-serde = { version = "1.0.225", features = ["derive"] }
-serde_json = "1.0.145"
+anyhow = "1.0"
+serde = { version = "1.0", features = ["derive"] }
+serde_json = "1.0"
+bincode = "1.3"
+log = "0.4"

crates/runner-shared/Cargo.toml

@@ -5,45 +5,153 @@ use core::{
 use serde::{Deserialize, Serialize};
 use std::{hash::DefaultHasher, ops::Range};
 
-/// Unwind data for a single module.
+/// A versioned enum for `UnwindData` to allow for future extensions while maintaining backward compatibility.
 #[derive(Serialize, Deserialize)]
-pub struct UnwindData {
-    pub path: String,
+pub enum UnwindData {
+    V1(UnwindDataV1),
+    V2(UnwindDataV2),
+}
 
-    pub avma_range: Range<u64>,
-    pub base_avma: u64,
+impl UnwindData {
+    pub fn path(&self) -> &str {
+        match self {
+            UnwindData::V1(data) => &data.path,
+            UnwindData::V2(data) => &data.path,
+        }
+    }
 
-    pub eh_frame_hdr: Vec<u8>,
-    pub eh_frame_hdr_svma: Range<u64>,
+    pub fn avma_range(&self) -> &Range<u64> {
+        match self {
+            UnwindData::V1(data) => &data.avma_range,
+            UnwindData::V2(data) => &data.avma_range,
+        }
+    }
 
-    pub eh_frame: Vec<u8>,
-    pub eh_frame_svma: Range<u64>,
+    pub fn base_avma(&self) -> u64 {
+        match self {
+            UnwindData::V1(data) => data.base_avma,
+            UnwindData::V2(data) => data.base_avma,
+        }
+    }
+
+    pub fn base_svma(&self) -> u64 {
+        match self {
+            UnwindData::V1(_) => 0,
+            UnwindData::V2(data) => data.base_svma,
+        }
+    }
+
+    pub fn eh_frame_hdr(&self) -> &Vec<u8> {
+        match self {
+            UnwindData::V1(data) => &data.eh_frame_hdr,
+            UnwindData::V2(data) => &data.eh_frame_hdr,
+        }
+    }
+
+    pub fn eh_frame_hdr_svma(&self) -> &Range<u64> {
+        match self {
+            UnwindData::V1(data) => &data.eh_frame_hdr_svma,
+            UnwindData::V2(data) => &data.eh_frame_hdr_svma,
+        }
+    }
+
+    pub fn eh_frame(&self) -> &Vec<u8> {
+        match self {
+            UnwindData::V1(data) => &data.eh_frame,
+            UnwindData::V2(data) => &data.eh_frame,
+        }
+    }
+
+    pub fn eh_frame_svma(&self) -> &Range<u64> {
+        match self {
+            UnwindData::V1(data) => &data.eh_frame_svma,
+            UnwindData::V2(data) => &data.eh_frame_svma,
+        }
+    }
+
+    pub fn save_to<P: AsRef<std::path::Path>>(&self, folder: P, pid: i32) -> anyhow::Result<()> {
+        let unwind_data_path = folder.as_ref().join(format!(
+            "{}_{:x}_{:x}.unwind2",
+            pid,
+            self.avma_range().start,
+            self.avma_range().end
+        ));
+        self.to_file(unwind_data_path)?;
+
+        Ok(())
+    }
+
+    pub fn to_file<P: AsRef<std::path::Path>>(&self, path: P) -> anyhow::Result<()> {
+        if let Ok(true) = std::fs::exists(path.as_ref()) {
+            log::warn!(
+                "{} already exists, file will be truncated",
+                path.as_ref().display()
+            );
+            log::warn!("{} {:x?}", self.path(), self.avma_range());
+        }
+
+        let mut writer = std::fs::File::create(path.as_ref())?;
+        bincode::serialize_into(&mut writer, self)?;
+        Ok(())
+    }
 }
 
 impl Debug for UnwindData {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         let eh_frame_hdr_hash = {
             let mut hasher = DefaultHasher::new();
-            self.eh_frame_hdr.hash(&mut hasher);
+            self.eh_frame_hdr().hash(&mut hasher);
             hasher.finish()
         };
         let eh_frame_hash = {
             let mut hasher = DefaultHasher::new();
-            self.eh_frame.hash(&mut hasher);
+            self.eh_frame().hash(&mut hasher);
             hasher.finish()
         };
 
         f.debug_struct("UnwindData")
-            .field("path", &self.path)
-            .field("avma_range", &format_args!("{:x?}", self.avma_range))
-            .field("base_avma", &format_args!("{:x}", self.base_avma))
+            .field("path", &self.path())
+            .field("avma_range", &format_args!("{:x?}", self.avma_range()))
+            .field("base_avma", &format_args!("{:x}", self.base_avma()))
+            .field("base_svma", &format_args!("{:x}", self.base_svma()))
             .field(
                 "eh_frame_hdr_svma",
-                &format_args!("{:x?}", self.eh_frame_hdr_svma),
+                &format_args!("{:x?}", self.eh_frame_hdr_svma()),
             )
             .field("eh_frame_hdr_hash", &format_args!("{eh_frame_hdr_hash:x}"))
             .field("eh_frame_hash", &format_args!("{eh_frame_hash:x}"))
-            .field("eh_frame_svma", &format_args!("{:x?}", self.eh_frame_svma))
+            .field(
+                "eh_frame_svma",
+                &format_args!("{:x?}", self.eh_frame_svma()),
+            )
             .finish()
     }
 }
+
+#[derive(Serialize, Deserialize)]
+pub struct UnwindDataV1 {
+    pub path: String,
+
+    pub avma_range: Range<u64>,
+    pub base_avma: u64,
+    pub eh_frame_hdr: Vec<u8>,
+    pub eh_frame_hdr_svma: Range<u64>,
+
+    pub eh_frame: Vec<u8>,
+    pub eh_frame_svma: Range<u64>,
+}
+
+#[derive(Serialize, Deserialize, Clone)]
+pub struct UnwindDataV2 {
+    pub path: String,
+
+    pub avma_range: Range<u64>,
+    pub base_avma: u64,
+    pub base_svma: u64,
+
+    pub eh_frame_hdr: Vec<u8>,
+    pub eh_frame_hdr_svma: Range<u64>,
+
+    pub eh_frame: Vec<u8>,
+    pub eh_frame_svma: Range<u64>,
+}

crates/runner-shared/src/unwind_data.rs

@@ -1,12 +1,9 @@
 use crate::{
     prelude::*,
-    run::runner::wall_time::perf::{
-        perf_map::{ModuleSymbols, Symbol},
-        unwind_data::UnwindDataExt,
-    },
+    run::runner::wall_time::perf::perf_map::{ModuleSymbols, Symbol},
 };
 use linux_perf_data::jitdump::{JitDumpReader, JitDumpRecord};
-use runner_shared::unwind_data::UnwindData;
+use runner_shared::unwind_data::{UnwindData, UnwindDataV2};
 use std::{
     collections::HashSet,
     path::{Path, PathBuf},
@@ -75,15 +72,16 @@ impl JitDump {
                         continue;
                     };
 
-                    jit_unwind_data.push(UnwindData {
+                    jit_unwind_data.push(UnwindData::V2(UnwindDataV2 {
                         path: format!("jit_{name}"),
                         avma_range: avma_start..avma_end,
                         base_avma: 0,
                         eh_frame_hdr,
                         eh_frame_hdr_svma: 0..0,
                         eh_frame,
                         eh_frame_svma: 0..0,
-                    });
+                        base_svma: 0,
+                    }));
                 }
                 JitDumpRecord::CodeUnwindingInfo(record) => {
                     // Store unwind info for the next code loads

src/run/runner/wall_time/perf/jit_dump.rs

@@ -1,12 +1,13 @@
 ---
 source: src/run/runner/wall_time/perf/unwind_data.rs
-expression: "UnwindData::new(MODULE_PATH.as_bytes(), 0x0, start_addr, size, None)"
+expression: "UnwindData::new(MODULE_PATH.as_bytes(), file_offset, start_addr, end_addr,\nNone,)"
 ---
 Ok(
     UnwindData {
         path: "testdata/perf_map/cpp_my_benchmark.bin",
         avma_range: 400000..459000,
         base_avma: 400000,
+        base_svma: 400000,
         eh_frame_hdr_svma: 4577bc..458b30,
         eh_frame_hdr_hash: 4b4eac90f7f5e60d,
         eh_frame_hash: 233bdd4ae9fe4ba4,

src/run/runner/wall_time/perf/snapshots/codspeed__run__runner__wall_time__perf__unwind_data__tests__cpp_unwind_data.snap

@@ -7,6 +7,7 @@ Ok(
         path: "testdata/perf_map/go_fib.bin",
         avma_range: 402000..50f000,
         base_avma: 400000,
+        base_svma: 400000,
         eh_frame_hdr_svma: 6498b0..649b94,
         eh_frame_hdr_hash: f1f69beb959a08d7,
         eh_frame_hash: a8727039dd21b51c,

src/run/runner/wall_time/perf/snapshots/codspeed__run__runner__wall_time__perf__unwind_data__tests__golang_unwind_data.snap

@@ -7,6 +7,7 @@ Ok(
         path: "testdata/perf_map/ty_walltime",
         avma_range: 555555e6d000..555556813000,
         base_avma: 555555554000,
+        base_svma: 0,
         eh_frame_hdr_svma: 7ec298..80f67c,
         eh_frame_hdr_hash: 6d6dd1e2c782318a,
         eh_frame_hash: ee27244db791265a,

src/run/runner/wall_time/perf/snapshots/codspeed__run__runner__wall_time__perf__unwind_data__tests__ruff_unwind_data.snap

@@ -1,12 +1,13 @@
 ---
 source: src/run/runner/wall_time/perf/unwind_data.rs
-expression: "UnwindData::new(MODULE_PATH.as_bytes(), 0x4d000, start_addr, end_addr, None)"
+expression: "UnwindData::new(MODULE_PATH.as_bytes(), file_offset, start_addr, end_addr,\nNone)"
 ---
 Ok(
     UnwindData {
         path: "testdata/perf_map/divan_sleep_benches.bin",
         avma_range: 5555555a2000..555555692000,
         base_avma: 555555554000,
+        base_svma: 0,
         eh_frame_hdr_svma: 2ac74..2ea60,
         eh_frame_hdr_hash: f579da4368e627c1,
         eh_frame_hash: 791501d5a9c438d,

src/run/runner/wall_time/perf/snapshots/codspeed__run__runner__wall_time__perf__unwind_data__tests__rust_divan_unwind_data.snap

@@ -7,6 +7,7 @@ Ok(
         path: "testdata/perf_map/the_algorithms.bin",
         avma_range: 5555555a7000..5555556b0000,
         base_avma: 555555554000,
+        base_svma: 0,
         eh_frame_hdr_svma: 2f0ec..33590,
         eh_frame_hdr_hash: 277fbdb59e6decaa,
         eh_frame_hash: 21d8b0c8da0d1029,

src/run/runner/wall_time/perf/snapshots/codspeed__run__runner__wall_time__perf__unwind_data__tests__the_algorithms_unwind_data.snap

@@ -3,10 +3,10 @@
 use crate::run::runner::wall_time::perf::elf_helper;
 use anyhow::{Context, bail};
 use debugid::CodeId;
-use libc::pid_t;
 use object::Object;
 use object::ObjectSection;
 use runner_shared::unwind_data::UnwindData;
+use runner_shared::unwind_data::UnwindDataV2;
 use std::ops::Range;
 
 pub trait UnwindDataExt {
@@ -19,9 +19,6 @@ pub trait UnwindDataExt {
     ) -> anyhow::Result<Self>
     where
         Self: Sized;
-
-    fn save_to<P: AsRef<std::path::Path>>(&self, folder: P, pid: pid_t) -> anyhow::Result<()>;
-    fn to_file<P: AsRef<std::path::Path>>(&self, path: P) -> anyhow::Result<()>;
 }
 
 impl UnwindDataExt for UnwindData {
@@ -68,6 +65,7 @@ impl UnwindDataExt for UnwindData {
             runtime_file_offset,
             &file,
         )?;
+        let base_svma = elf_helper::relative_address_base(&file);
         let eh_frame = file.section_by_name(".eh_frame");
         let eh_frame_hdr = file.section_by_name(".eh_frame_hdr");
 
@@ -82,10 +80,11 @@ impl UnwindDataExt for UnwindData {
             section.address()..section.address() + section.size()
         }
 
-        Ok(Self {
+        Ok(UnwindData::V2(UnwindDataV2 {
             path,
             avma_range,
             base_avma,
+            base_svma,
             eh_frame_hdr: eh_frame_hdr_data.context("Failed to find eh_frame hdr data")?,
             eh_frame_hdr_svma: eh_frame_hdr
                 .as_ref()
@@ -96,31 +95,7 @@ impl UnwindDataExt for UnwindData {
                 .as_ref()
                 .map(svma_range)
                 .context("Failed to find eh_frame section")?,
-        })
-    }
-
-    fn save_to<P: AsRef<std::path::Path>>(&self, folder: P, pid: pid_t) -> anyhow::Result<()> {
-        let unwind_data_path = folder.as_ref().join(format!(
-            "{}_{:x}_{:x}.unwind",
-            pid, self.avma_range.start, self.avma_range.end
-        ));
-        self.to_file(unwind_data_path)?;
-
-        Ok(())
-    }
-
-    fn to_file<P: AsRef<std::path::Path>>(&self, path: P) -> anyhow::Result<()> {
-        if let Ok(true) = std::fs::exists(path.as_ref()) {
-            log::warn!(
-                "{} already exists, file will be truncated",
-                path.as_ref().display()
-            );
-            log::warn!("{} {:x?}", self.path, self.avma_range);
-        }
-
-        let mut writer = std::fs::File::create(path.as_ref())?;
-        bincode::serialize_into(&mut writer, self)?;
-        Ok(())
+        }))
     }
 }