lb_binary_layered: second take on fixing mtimes

debian/install

@@ -4,3 +4,4 @@ get-ppa-fingerprint usr/share/livecd-rootfs
 minimize-manual usr/share/livecd-rootfs
 checkout-translations-branch usr/share/livecd-rootfs
 update-source-catalog usr/share/livecd-rootfs
+sync-mtime usr/share/livecd-rootfs

live-build/lb_binary_layered

@@ -163,15 +163,24 @@ build_layered_squashfs () {
 			#    (rather than the default which is to skip copies based
 			#    on size + mtime)
 			#  --no-times to not copy mtimes from source to dest (we
-			#    don't care about mtime in the image and want to
+			#    do care about mtime in the image but want to
 			#    deduplicate files that have indentical contents but
-			#    different mtimes)
+			#    different mtimes, and mtime will be fixed below)
 			#  --del because we want to remove files that have been
 			#    deleted in this layer.
 			rsync -aXHAS --checksum --no-times --del chroot/ chroot-2/
 			umount chroot-2
 			rmdir chroot-2
 			overlay_dir="$overlay_dir-2"
+			# We use rsync with --no-times rsync (see above)
+			# for the absolute best size reduction.  But there are
+			# cases where we want mtime preservation to match what
+			# was found in the original archive packages, such as
+			# keeping .py mtime in sync with the matching .pyc.
+			# Operate on the upperdir directly, so that we are only
+			# modifying mtime on files that are actually changed in
+			# this layer. LP: #2107332
+			/usr/share/livecd-rootfs/sync-mtime chroot "$overlay_dir"
 		fi
 
 		create_squashfs "${overlay_dir}" ${squashfs_f}

sync-mtime

@@ -0,0 +1,38 @@
+#!/usr/bin/python3
+
+# usage: sync-mtime src dst
+#
+# synchronize atime/mtime on files between src and dst.
+#
+# src and dst are directories, where dst is expected to contain a subset of the
+# files found in src. for each file present in dst or a subdirectory thereof,
+# if atime/mtime differ between the same file in src and that file in dst,
+# update atime/mtime on the file in dst.
+
+import os
+import sys
+from pathlib import Path
+
+
+def time_eq(a: os.stat_result, b: os.stat_result) -> bool:
+    return (
+        (a.st_mtime_ns == b.st_mtime_ns) and
+        (a.st_atime_ns == b.st_atime_ns)
+    )
+
+
+src, dst = sys.argv[1:]
+for dirpath, dirnames, filenames in Path(dst).walk():
+    for filename in filenames:
+        dst_file = dirpath / filename
+        if not dst_file.is_file():
+            continue
+        src_file = src / dst_file.relative_to(dst)
+
+        src_stat = src_file.stat(follow_symlinks=False)
+        dst_stat = dst_file.stat(follow_symlinks=False)
+        if time_eq(src_stat, dst_stat):
+            continue
+
+        ns = (src_stat.st_atime_ns, src_stat.st_mtime_ns)
+        os.utime(dst_file, ns=ns, follow_symlinks=False)