read/elf: include unsized symbols in is_definition() (#866)

It's useful to have these in the symbol map.

Co-authored-by: Peter Collingbourne <pcc@google.com>

crates/examples/testfiles/elf/base-aarch64.objdump

@@ -175,5 +175,7 @@ Symbol map
 0x7f8-0x7fc "__FRAME_END__"
 0x10d80-0x10d88 "__frame_dummy_init_array_entry"
 0x10d88-0x10d90 "__do_global_dtors_aux_fini_array_entry"
+0x11000-0x11008 "__data_start"
 0x11008-0x11010 "__dso_handle"
 0x11010-0x11011 "completed.8500"
+0x11018-0x11018 "__end__"

crates/examples/testfiles/elf/base-relr-i686.objdump

@@ -121,10 +121,13 @@ Symbol map
 0x11d4-0x11e8 "_fini"
 0x2000-0x2004 "_fp_hw"
 0x2004-0x2008 "_IO_stdin_used"
+0x2018-0x204c "__GNU_EH_FRAME_HDR"
 0x2110-0x2114 "__FRAME_END__"
 0x3ec0-0x3ec4 "__frame_dummy_init_array_entry"
 0x3ec4-0x3ec8 "__do_global_dtors_aux_fini_array_entry"
 0x3ec8-0x3fd8 "_DYNAMIC"
 0x3fd8-0x4000 "_GLOBAL_OFFSET_TABLE_"
+0x4000-0x4004 "__data_start"
 0x4004-0x4008 "__dso_handle"
 0x4008-0x4009 "completed.0"
+0x400c-0x400c "_end"

crates/examples/testfiles/elf/base-relr-x86_64.objdump

@@ -112,10 +112,13 @@ Symbol map
 0x1149-0x116c "main"
 0x116c-0x1179 "_fini"
 0x2000-0x2004 "_IO_stdin_used"
+0x2014-0x2048 "__GNU_EH_FRAME_HDR"
 0x20f0-0x20f4 "__FRAME_END__"
 0x3d88-0x3d90 "__frame_dummy_init_array_entry"
 0x3d90-0x3d98 "__do_global_dtors_aux_fini_array_entry"
 0x3d98-0x3fb8 "_DYNAMIC"
 0x3fb8-0x4000 "_GLOBAL_OFFSET_TABLE_"
+0x4000-0x4008 "__data_start"
 0x4008-0x4010 "__dso_handle"
 0x4010-0x4011 "completed.0"
+0x4018-0x4018 "_end"

crates/examples/testfiles/elf/base.objdump

@@ -140,10 +140,13 @@ Symbol map
 0x710-0x712 "__libc_csu_fini"
 0x714-0x71d "_fini"
 0x720-0x724 "_IO_stdin_used"
+0x734-0x770 "__GNU_EH_FRAME_HDR"
 0x874-0x878 "__FRAME_END__"
 0x200da8-0x200db0 "__frame_dummy_init_array_entry"
 0x200db0-0x200db8 "__do_global_dtors_aux_fini_array_entry"
 0x200db8-0x200fb8 "_DYNAMIC"
 0x200fb8-0x201000 "_GLOBAL_OFFSET_TABLE_"
+0x201000-0x201008 "__data_start"
 0x201008-0x201010 "__dso_handle"
 0x201010-0x201011 "completed.7698"
+0x201018-0x201018 "_end"

src/read/elf/file.rs

@@ -432,7 +432,7 @@ where
     fn exports(&self) -> read::Result<Vec<Export<'data>>> {
         let mut exports = Vec::new();
         for symbol in self.dynamic_symbols.iter() {
-            if symbol.is_definition(self.endian) {
+            if symbol.is_definition(self.endian, self.dynamic_symbols.strings()) {
                 let name = symbol.name(self.endian, self.dynamic_symbols.strings())?;
                 let address = symbol.st_value(self.endian).into();
                 exports.push(Export {

src/read/elf/symbol.rs

@@ -206,7 +206,7 @@ impl<'data, Elf: FileHeader, R: ReadRef<'data>> SymbolTable<'data, Elf, R> {
     ) -> SymbolMap<Entry> {
         let mut symbols = Vec::with_capacity(self.symbols.len());
         for symbol in self.symbols {
-            if !symbol.is_definition(endian) {
+            if !symbol.is_definition(endian, self.strings) {
                 continue;
             }
             if let Some(entry) = f(symbol) {
@@ -436,7 +436,8 @@ impl<'data, 'file, Elf: FileHeader, R: ReadRef<'data>> ObjectSymbol<'data>
 
     #[inline]
     fn is_definition(&self) -> bool {
-        self.symbol.is_definition(self.endian)
+        self.symbol
+            .is_definition(self.endian, self.symbols.strings())
     }
 
     #[inline]
@@ -520,14 +521,26 @@ pub trait Sym: Debug + Pod {
     }
 
     /// Return true if the symbol is a definition of a function or data object.
-    fn is_definition(&self, endian: Self::Endian) -> bool {
+    fn is_definition<'data, R: ReadRef<'data>>(
+        &self,
+        endian: Self::Endian,
+        strings: StringTable<'data, R>,
+    ) -> bool {
         let shndx = self.st_shndx(endian);
         if shndx == elf::SHN_UNDEF || (shndx >= elf::SHN_LORESERVE && shndx != elf::SHN_XINDEX) {
             return false;
         }
         match self.st_type() {
-            elf::STT_NOTYPE => self.st_size(endian).into() != 0,
-            elf::STT_FUNC | elf::STT_OBJECT => true,
+            elf::STT_NOTYPE if self.st_bind() == elf::STB_LOCAL => {
+                // Exclude mapping symbols.
+                let Ok(name) = self.name(endian, strings) else {
+                    // Not worth changing API to return an error for this case.
+                    // Caller is likely to call name() themselves.
+                    return true;
+                };
+                !name.starts_with(b"$")
+            }
+            elf::STT_FUNC | elf::STT_OBJECT | elf::STT_NOTYPE => true,
             _ => false,
         }
     }