Finish collecting dwarf dependencies

src/filestore.rs

@@ -4,6 +4,7 @@ use std::{
     path::PathBuf, borrow::Cow,
 };
 
+use gimli::{constants, AttributeValue, DW_TAG_compile_unit};
 use serde::{Deserialize, Serialize};
 use sha2::{digest::generic_array::{GenericArray, typenum::U32}, Digest, Sha256};
 use object::{Object, ReadCache, ObjectSection};
@@ -129,14 +130,15 @@ impl FileStore {
         }
 
         let fp = fs::File::open(&filename)?;
-        self.ingest_output(filename, fp)
+        self.ingest_output(filename, fp);
+        Ok(())
     }
 
     pub fn ingest_output(
         &mut self,
         filename: PathBuf,
         mut content: (impl io::Read + io::Seek),
-    ) -> anyhow::Result<()> {
+    ) -> anyhow::Result<usize> {
         let mut h = Sha256::new();
         io::copy(&mut content, &mut h)?;
         let hash = h.finalize().into();
@@ -167,7 +169,7 @@ impl FileStore {
 
         self.filenames.insert(filename, index);
 
-        Ok(())
+        Ok(index)
     }
 
     fn parse_format(&mut self, fp: &mut (impl io::Read + io::Seek)) -> anyhow::Result<FileFormat> {
@@ -189,22 +191,57 @@ impl FileStore {
                 let mut load_section = |id: gimli::SectionId| -> Result<_, _> {
                     load_file_section(id, &elf, endian, &arena_data)
                 };
-                let mut dwarf = gimli::Dwarf::load(&mut load_section).unwrap();
+                let dwarf = gimli::Dwarf::load(&mut load_section).unwrap();
                 let mut units = dwarf.units();
+                let mut inputs = vec![];
                 while let Ok(Some(unit)) = units.next() {
                     let abbrev = dwarf.abbreviations(&unit)?;
                     let mut entries = unit.entries(&abbrev);
-                    while let Ok(Some(entry)) = entries.next_sibling() {
+                    while let Some((_, entry)) = entries.next_dfs()? {
+                        if entry.tag() == DW_TAG_compile_unit {
+                            let mut basename = None;
+                            let mut dirname = None;
+                            if let Some(name) = entry.attr(constants::DW_AT_name)?.map(|a| a.value()) {
+                                if let Ok(name) = dwarf.attr_string(&dwarf.unit(unit)?, name) {
+                                    basename = Some(PathBuf::from(name.to_string()?));
+                                }
+                            }
+                            if let Some(name) = entry.attr(constants::DW_AT_comp_dir)?.map(|a| a.value()) {
+                                if let Ok(name) = dwarf.attr_string(&dwarf.unit(unit)?, name) {
+                                    dirname = Some(PathBuf::from(name.to_string()?));
+                                }
+                            }
+                            if let (Some(dirname), Some(basename)) = (dirname, basename) {
+                                inputs.push(dirname.join(basename));
+                            }
+                        }
                     }
                 }
-                assert!(elf.is_little_endian());
+
+                let references = inputs.into_iter().map(|input| self.ingest_dependency_local(input)).collect::<Result<Vec<_>, _>>()?.into_iter().filter_map(|x| x).collect();
                 FileFormat::ELF {
-                    references: vec![],
+                    references,
                 }
             },
             _ => FileFormat::Other,
         })
     }
+
+    fn ingest_dependency_local(&mut self, filename: PathBuf) -> anyhow::Result<Option<usize>> {
+        // TODO: this needs to try suffixes of the filename against the filepath table to see if it
+        // was moved between compilation and ingestion
+        let metadata = match fs::metadata(&filename) {
+            Ok(m) => m,
+            Err(e) if e.kind() == io::ErrorKind::NotFound => return Ok(None),
+            Err(e) => return Err(e)?,
+        };
+        if !metadata.is_file() {
+            return Ok(None);
+        }
+
+        let fp = fs::File::open(&filename)?;
+        Ok(Some(self.ingest_output(filename, fp)?))
+    }
 }
 
 fn load_file_section<'input, 'arena, Endian: gimli::Endianity, R: object::ReadRef<'input>>(

src/tracer.rs

@@ -585,6 +585,8 @@ impl Tracer {
                                 // So we need to determine whether exec is successful here.
                                 // PTRACE_EVENT_EXEC only happens for successful exec.
                                 p.is_exec_successful = true;
+                                let path = p.pending_syscall_event.iter().find_map(|e| match e { Event::Exec { prog, .. } => Some(prog.clone()), _ => None }).unwrap();
+                                self.report.files.ingest_output_local(path)?;
                                 self.drain_syscall_events(pid.into(), Box::new(|_| {}));
                                 // Don't use seccomp_aware_cont here because that will skip the next syscall exit stop
                                 self.syscall_enter_cont(pid.into())?;