cmd-diff: extract files instead of using FUSE

This approach use more disk space but disk access for the diff will
be faster. Files will also survive under after git diff returns in
case manual inspection is desired.

Co-authored-by: Jean-Baptiste Trystram <[email protected]>

Author: dustymabe

src/cmd-diff

@@ -531,72 +531,69 @@ def run_guestfs_mount(image_path, mount_target):
 # the paths to be used for analysis and then clean up once given back
 # control.
 def diff_metal_helper(diff_from, diff_to):
-    metal_from = get_metal_path(diff_from)
-    metal_to = get_metal_path(diff_to)
-
-    mount_dir_from = os.path.join(cache_dir("metal"), diff_from.id)
-    mount_dir_to = os.path.join(cache_dir("metal"), diff_to.id)
-
-    for d in [mount_dir_from, mount_dir_to]:
-        if os.path.exists(d):
-            shutil.rmtree(d)
-        os.makedirs(d)
-
-    # As the libreguest mount call is blocking until unmounted, let's
-    # do that in a separate thread
-    p_from = Process(target=run_guestfs_mount, args=(metal_from, mount_dir_from))
-    p_to = Process(target=run_guestfs_mount, args=(metal_to, mount_dir_to))
-
-    try:
-        p_from.start()
-        p_to.start()
-        # Wait for the FUSE mounts to be ready. We'll check for a known file.
-        for i, d in enumerate([mount_dir_from, mount_dir_to]):
-            p = p_from if i == 0 else p_to
-            timeout = 60  # seconds
-            start_time = time.time()
-            check_file = os.path.join(d, 'ostree')
-            while not os.path.exists(check_file):
-                time.sleep(1)
-                if time.time() - start_time > timeout:
-                    raise Exception(f"Timeout waiting for mount in {d}")
-                if not p.is_alive():
-                    raise Exception(f"A guestfs process for {os.path.basename(d)} died unexpectedly.")
-
-        # Allow the caller to operate on these values
-        yield mount_dir_from, mount_dir_to
-
-    finally:
-        # Unmount the FUSE binds, this will make the guestfs mount calls return
-        runcmd(['fusermount', '-u', mount_dir_from], check=False)
-        runcmd(['fusermount', '-u', mount_dir_to], check=False)
-
-        # Ensure the background processes are terminated
-        def shutdown_process(process):
-            process.join(timeout=5)
-            if process.is_alive():
-                process.terminate()
-                process.join()
-
-        shutdown_process(p_from)
-        shutdown_process(p_to)
+    metal_image_from = get_metal_path(diff_from)
+    metal_image_to = get_metal_path(diff_to)
+
+    diff_dir_from = os.path.join(cache_dir("metal"), diff_from.id)
+    diff_dir_to = os.path.join(cache_dir("metal"), diff_to.id)
+
+    for image_path, diff_dir in [(metal_image_from, diff_dir_from),
+                                 (metal_image_to, diff_dir_to)]:
+        if os.path.exists(diff_dir):
+            # If it exists assume it's cached already and we don't
+            # need to do anything. If it's stale for whatever reason
+            # the user can `cosa diff --gc`.
+            continue
+
+        os.makedirs(diff_dir)
+
+        g = None
+        try:
+            g = guestfs.GuestFS(python_return_dict=True)
+            g.set_backend("direct")
+            g.add_drive_opts(image_path, readonly=1)
+            g.launch()
+
+            # Mount the disks in the guestfs VM
+            root = g.findfs_label("root")
+            g.mount_ro(root, "/")
+            boot = g.findfs_label("boot")
+            g.mount_ro(boot, "/boot")
+            efi = g.findfs_label("EFI-SYSTEM")
+            g.mount_ro(efi, "/boot/efi")
+
+
+            with tempfile.NamedTemporaryFile(suffix=".tar", delete=True) as tmp_tar:
+                g.tar_out("/", tmp_tar.name, xattrs=True, selinux=True, excludes=excludes)
+                # Extract the tarball.
+                runcmd(['tar', '-xf', tmp_tar.name, '-C', diff_dir])
+
+        except Exception as e:
+            print(f"Error in guestfs process for {image_path}: {e}", file=sys.stderr)
+            raise
+        finally:
+            if g:
+                g.close()
+
+    # Allow the caller to operate on these values
+    return diff_dir_from, diff_dir_to
 
 
 def diff_metal(diff_from, diff_to):
-    for mount_dir_from, mount_dir_to in diff_metal_helper(diff_from, diff_to):
-        git_diff(mount_dir_from, mount_dir_to)
+    mount_dir_from, mount_dir_to = diff_metal_helper(diff_from, diff_to)
+    git_diff(mount_dir_from, mount_dir_to)
 
 
 def diff_metal_du(diff_from, diff_to):
-    for mount_dir_from, mount_dir_to in diff_metal_helper(diff_from, diff_to):
-        cmd = ['find', '.', '-type', 'd', '-exec', 'du', '-sh', '{}', ';']
-        diff_cmd_outputs(cmd, mount_dir_from, mount_dir_to, strategy=DiffCmdOutputStrategy.CD)
+    mount_dir_from, mount_dir_to = diff_metal_helper(diff_from, diff_to)
+    cmd = ['find', '.', '-type', 'd', '-exec', 'du', '-sh', '{}', ';']
+    diff_cmd_outputs(cmd, mount_dir_from, mount_dir_to, strategy=DiffCmdOutputStrategy.CD)
 
 
 def diff_metal_ls(diff_from, diff_to):
-    for mount_dir_from, mount_dir_to in diff_metal_helper(diff_from, diff_to):
-        cmd = ['find', '.']
-        diff_cmd_outputs(cmd, mount_dir_from, mount_dir_to, strategy=DiffCmdOutputStrategy.CD)
+    mount_dir_from, mount_dir_to = diff_metal_helper(diff_from, diff_to)
+    cmd = ['find', '.']
+    diff_cmd_outputs(cmd, mount_dir_from, mount_dir_to, strategy=DiffCmdOutputStrategy.CD)
 
 
 def diff_cmd_outputs(cmd, path_from, path_to, strategy: DiffCmdOutputStrategy = DiffCmdOutputStrategy.TEMPLATE):