debvm-create: switch from genext2fs to mkfs.ext4

Why?

genext2fs is slow. It has O(n^2) behaviour, which becomes annoying with
larger filesystems. It also creates an ext2 image and we have to upgrade
it to ext4 in multiple steps. Even then, the image has an inode size of
128, which is incompatible with 2038 and does not allow storing
sub-second precision time stamps.

How?

We use mkfs.ext4 from e2fsprogs (which is required anyway) instead.
Running this from within a --customize-hook does not work, because the
user namespace used for the hook often has no privileges to write to the
output image. Instead, we now ask mmdebstrap to output to a directory
and create the filesystem separately. We construct a special namespace
that has both access to the tree and to the output image and perform the
mkfs.ext4 there. As a consequence, we no longer support any mode but
unshare and since mkfs.ext4 consumes the size parameter, the size syntax
is reduced a bit.

1 files changed, 2 insertions, 3 deletions

diff --git a/README.md b/README.md
index 1ccda36..f757993 100644
--- a/README.md
+++ b/README.md
@@ -32,10 +32,10 @@ A regular user account[^1] suffices. The following Debian packages should be
 installed:
  * `arch-test` (when running foreign images)
  * `e2fsprogs`
- * `genext2fs` (used by `mmdebstrap`)
  * `mmdebstrap`
  * `qemu-system-something`
  * `qemu-user-static` (when creating foreign images)
+ * `uidmap` (for using user namespaces)
 
 What is this image format precisely?
 ====================================
@@ -63,8 +63,7 @@ The other aspect is restricting to Debian-based systems. This allows for a lot
 of simplification of the problem space.
 
 While most similar tools require root privileges at some point, this one works
-with either fakeroot or a subuid allocation for user namespaces, which is often
-available.
+with a subuid allocation for user namespaces, which is often available.
 
 The implementation is so short that it still is feasible to read and understand
 it. Let's see how long that lasts.