ELF: better handle corrupt files (#2227)

such as when there's a missing symbol table and invalid relocation table.
and then handle when Viv fails to load a workspace.

closes #2226

CHANGELOG.md

@@ -13,6 +13,7 @@
 ### Bug Fixes
 
 - elf: extract import / export symbols from stripped binaries #2096 @ygasparis
+- elf: fix handling of symbols in corrupt ELF files #2226 @williballenthin
 
 ### capa explorer IDA Pro plugin
 

capa/features/extractors/elffile.py

@@ -50,7 +50,12 @@ def extract_file_export_names(elf: ELFFile, **kwargs):
         if not isinstance(segment, DynamicSegment):
             continue
 
-        logger.debug("Dynamic Segment contains %s symbols: ", segment.num_symbols())
+        tab_ptr, tab_offset = segment.get_table_offset("DT_SYMTAB")
+        if tab_ptr is None or tab_offset is None:
+            logger.debug("Dynamic segment doesn't contain DT_SYMTAB")
+            continue
+
+        logger.debug("Dynamic segment contains %s symbols: ", segment.num_symbols())
 
         for symbol in segment.iter_symbols():
             # The following conditions are based on the following article
@@ -76,6 +81,11 @@ def extract_file_import_names(elf: ELFFile, **kwargs):
         if not isinstance(segment, DynamicSegment):
             continue
 
+        tab_ptr, tab_offset = segment.get_table_offset("DT_SYMTAB")
+        if tab_ptr is None or tab_offset is None:
+            logger.debug("Dynamic segment doesn't contain DT_SYMTAB")
+            continue
+
         for _, symbol in enumerate(segment.iter_symbols()):
             # The following conditions are based on the following article
             # http://www.m4b.io/elf/export/binary/analysis/2015/05/25/what-is-an-elf-export.html
@@ -100,7 +110,16 @@ def extract_file_import_names(elf: ELFFile, **kwargs):
         logger.debug("Dynamic Segment contains %s relocation tables:", len(relocation_tables))
 
         for relocation_table in relocation_tables.values():
-            for relocation in relocation_table.iter_relocations():
+            relocations = []
+            for i in range(relocation_table.num_relocations()):
+                try:
+                    relocations.append(relocation_table.get_relocation(i))
+                except TypeError:
+                    # ELF is corrupt and the relocation table is invalid,
+                    # so stop processing it.
+                    break
+
+            for relocation in relocations:
                 # Extract the symbol name from the symbol table using the symbol index in the relocation
                 if relocation["r_info_sym"] not in symbol_names:
                     continue

capa/loader.py

@@ -64,6 +64,10 @@
 BACKEND_FREEZE = "freeze"
 
 
+class CorruptFile(ValueError):
+    pass
+
+
 def is_supported_format(sample: Path) -> bool:
     """
     Return if this is a supported file based on magic header values
@@ -137,21 +141,28 @@ def get_workspace(path: Path, input_format: str, sigpaths: List[Path]):
     import viv_utils.flirt
 
     logger.debug("generating vivisect workspace for: %s", path)
-    if input_format == FORMAT_AUTO:
-        if not is_supported_format(path):
-            raise UnsupportedFormatError()
 
-        # don't analyze, so that we can add our Flirt function analyzer first.
-        vw = viv_utils.getWorkspace(str(path), analyze=False, should_save=False)
-    elif input_format in {FORMAT_PE, FORMAT_ELF}:
-        vw = viv_utils.getWorkspace(str(path), analyze=False, should_save=False)
-    elif input_format == FORMAT_SC32:
-        # these are not analyzed nor saved.
-        vw = viv_utils.getShellcodeWorkspaceFromFile(str(path), arch="i386", analyze=False)
-    elif input_format == FORMAT_SC64:
-        vw = viv_utils.getShellcodeWorkspaceFromFile(str(path), arch="amd64", analyze=False)
-    else:
-        raise ValueError("unexpected format: " + input_format)
+    try:
+        if input_format == FORMAT_AUTO:
+            if not is_supported_format(path):
+                raise UnsupportedFormatError()
+
+            # don't analyze, so that we can add our Flirt function analyzer first.
+            vw = viv_utils.getWorkspace(str(path), analyze=False, should_save=False)
+        elif input_format in {FORMAT_PE, FORMAT_ELF}:
+            vw = viv_utils.getWorkspace(str(path), analyze=False, should_save=False)
+        elif input_format == FORMAT_SC32:
+            # these are not analyzed nor saved.
+            vw = viv_utils.getShellcodeWorkspaceFromFile(str(path), arch="i386", analyze=False)
+        elif input_format == FORMAT_SC64:
+            vw = viv_utils.getShellcodeWorkspaceFromFile(str(path), arch="amd64", analyze=False)
+        else:
+            raise ValueError("unexpected format: " + input_format)
+    except Exception as e:
+        # vivisect raises raw Exception instances, and we don't want
+        # to do a subclass check via isinstance.
+        if type(e) is Exception and "Couldn't convert rva" in e.args[0]:
+            raise CorruptFile(e.args[0]) from e
 
     viv_utils.flirt.register_flirt_signature_analyzers(vw, [str(s) for s in sigpaths])
 

capa/main.py

@@ -753,6 +753,9 @@ def get_extractor_from_cli(args, input_format: str, backend: str) -> FeatureExtr
     except UnsupportedOSError as e:
         log_unsupported_os_error()
         raise ShouldExitError(E_INVALID_FILE_OS) from e
+    except capa.loader.CorruptFile as e:
+        logger.error("Input file '%s' is not a valid file: %s", args.input_file, str(e))
+        raise ShouldExitError(E_CORRUPT_FILE) from e
 
 
 def main(argv: Optional[List[str]] = None):