Merge pull request #361 from IoT-Inspector/355-fix-cpio-padding

qkaiser · web-flow · commit 736b0b42425b · 2022-05-06T16:52:50.000+02:00
Fix CPIO file padding calculation.
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__input__/sample.nofilepad.cpio-newc b/tests/integration/archive/cpio/cpio_portable_ascii/__input__/sample.nofilepad.cpio-newc
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfaec584f22cd7be300163e7618d371993b7850cd5a29beb0a048d91e3332eb4
+size 860
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__input__/sample.nofilepad.unaligned.cpio-newc b/tests/integration/archive/cpio/cpio_portable_ascii/__input__/sample.nofilepad.unaligned.cpio-newc
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d31b3b73ebe4690094f8838d773739963221d34508cd9243ae1f91125703deeb
+size 861
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.cpio-newc_extract/0-512.cpio_portable_ascii b/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.cpio-newc_extract/0-512.cpio_portable_ascii
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e585bd2adda91197a015975f9bb29e95cfb4dfd4842aca97c4234c2b0565ad8c
+size 512
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.cpio-newc_extract/0-512.cpio_portable_ascii_extract/dev/console b/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.cpio-newc_extract/0-512.cpio_portable_ascii_extract/dev/console
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.cpio-newc_extract/0-512.cpio_portable_ascii_extract/root/.gitkeep b/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.cpio-newc_extract/0-512.cpio_portable_ascii_extract/root/.gitkeep
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.cpio-newc_extract/512-860.unknown b/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.cpio-newc_extract/512-860.unknown
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd13d41bc045c5b2a103d6727c011c1aabb53fd2492539336f49007c29c4571e
+size 348
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/0-1.unknown b/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/0-1.unknown
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:559aead08264d5795d3909718cdd05abd49572e84fe55590eef31a88a08fdffd
+size 1
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/1-513.cpio_portable_ascii b/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/1-513.cpio_portable_ascii
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e585bd2adda91197a015975f9bb29e95cfb4dfd4842aca97c4234c2b0565ad8c
+size 512
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/1-513.cpio_portable_ascii_extract/dev/console b/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/1-513.cpio_portable_ascii_extract/dev/console
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/1-513.cpio_portable_ascii_extract/root/.gitkeep b/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/1-513.cpio_portable_ascii_extract/root/.gitkeep
diff --git a/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/513-861.unknown b/tests/integration/archive/cpio/cpio_portable_ascii/__output__/sample.nofilepad.unaligned.cpio-newc_extract/513-861.unknown
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd13d41bc045c5b2a103d6727c011c1aabb53fd2492539336f49007c29c4571e
+size 348
diff --git a/unblob/handlers/archive/cpio.py b/unblob/handlers/archive/cpio.py
@@ -4,7 +4,7 @@
 from structlog import get_logger
 
 from ...extractors import Command
-from ...file_utils import decode_int, round_up, snull
+from ...file_utils import OffsetFile, decode_int, round_up, snull
 from ...models import File, HexString, StructHandler, ValidChunk
 
 logger = get_logger()
@@ -52,9 +52,11 @@ class _CPIOHandlerBase(StructHandler):
     _FILE_PAD_ALIGN: int = 512
 
     def calculate_chunk(self, file: File, start_offset: int) -> Optional[ValidChunk]:
-        offset = start_offset
+
+        file_with_offset = OffsetFile(file, start_offset)
+        current_offset = start_offset
         while True:
-            file.seek(offset)
+            file.seek(current_offset)
             header = self.parse_header(file)
 
             c_filesize = self._calculate_file_size(header)
@@ -65,17 +67,15 @@ def calculate_chunk(self, file: File, start_offset: int) -> Optional[ValidChunk]
                 return
 
             if c_namesize > 0:
-                file.seek(offset + len(header))
-                tmp_filename = file.read(c_namesize)
+                tmp_filename = file_with_offset.read(c_namesize)
 
                 # heuristics 2: check that filename is null-byte terminated
                 if not tmp_filename.endswith(b"\x00"):
                     return
 
                 filename = snull(tmp_filename)
-
                 if filename == CPIO_TRAILER_NAME:
-                    offset += self._pad_content(header, c_filesize, c_namesize)
+                    current_offset += self._pad_content(header, c_filesize, c_namesize)
                     break
 
             c_mode = self._calculate_mode(header)
@@ -88,27 +88,27 @@ def calculate_chunk(self, file: File, start_offset: int) -> Optional[ValidChunk]
             if not is_valid:
                 return
 
-            file.seek(c_filesize, io.SEEK_CUR)
-            # Rounding up the total of the header size, and the c_filesize, again. Because
-            # some CPIO implementations don't align the first chunk, but do align the 2nd.
-            # In theory, with a "normal" CPIO file, we should just be aligned on the
-            # 4-byte boundary already, but if we are not for some reason, then we just
-            # need to round up again.
-
-            if c_filesize > 0:
-                offset += self._pad_content(header, c_filesize, c_namesize)
-            else:
-                offset = round_up(file.tell(), self._PAD_ALIGN)
-
-        # Add padding that could exists between the cpio trailer and the end-of-file.
-        # cpio aligns the file to 512 bytes
-        offset = round_up(offset, self._FILE_PAD_ALIGN)
+            file_with_offset.seek(c_filesize, io.SEEK_CUR)
+            current_offset += self._pad_content(header, c_filesize, c_namesize)
 
+        end_offset = start_offset + self._pad_file(
+            file_with_offset, current_offset - start_offset
+        )
         return ValidChunk(
             start_offset=start_offset,
-            end_offset=offset,
+            end_offset=end_offset,
         )
 
+    @classmethod
+    def _pad_file(cls, file: OffsetFile, end_offset: int) -> int:
+        """CPIO archives can have a 512 bytes block padding at the end."""
+        file.seek(end_offset, io.SEEK_SET)
+        padded_end_offset = round_up(end_offset, cls._FILE_PAD_ALIGN)
+        padding_size = padded_end_offset - end_offset
+        if file.read(padding_size) == bytes([0]) * padding_size:
+            return padded_end_offset
+        return end_offset
+
     @classmethod
     def _pad_content(cls, header, c_filesize: int, c_namesize: int) -> int:
         """Pad header and content with _PAD_ALIGN bytes."""

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:dfaec584f22cd7be300163e7618d371993b7850cd5a29beb0a048d91e3332eb4`
	`3`	`+size 860`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:d31b3b73ebe4690094f8838d773739963221d34508cd9243ae1f91125703deeb`
	`3`	`+size 861`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:e585bd2adda91197a015975f9bb29e95cfb4dfd4842aca97c4234c2b0565ad8c`
	`3`	`+size 512`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:fd13d41bc045c5b2a103d6727c011c1aabb53fd2492539336f49007c29c4571e`
	`3`	`+size 348`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:559aead08264d5795d3909718cdd05abd49572e84fe55590eef31a88a08fdffd`
	`3`	`+size 1`