ADD EXT2 SUPPORT WOOO

README.md

@@ -1,15 +1,19 @@
 # cursedfs
 
-Make a disk image formatted with both ZFS and FAT, at once.
+Make a disk image formatted with ZFS, ext2, and FAT all at once.
 
 ```bash
 ~/cursedfs% wget 'https://github.com/pcd1193182/cursedfs/releases/download/v1.1/cursed.img'
 ~/cursedfs% sudo mount -o loop -t msdos cursed.img mountpoint/
 ~/cursedfs% ls mountpoint/
 duckroll.jpg
+~/cursedfs% sudo umount mountpoint
+~/cursedfs% sudo mount -o loop -t ext2 cursed.img mountpoint/
+~/cursedfs% ls mountpoint/
+rickroll.jpg
 ~/cursedfs% sudo zpool import cursed -d .
 ~/cursedfs% ls /cursed/
-rickroll.jpg
+rickroll.mp4
 ```
 
 # Why?
@@ -18,13 +22,25 @@ rickroll.jpg
 
 # How?
 
-Bulding on NieDzejkob's great work, since zfs leaves the first 8k of
-the disk empty, you can easily have the two coexist. The FAT
-filesystem lives in zfs's 3.5M of reserved boot space. This can be
-extended with a read-only ext2 filesystem as well; I'll try that
-eventually
+FAT uses the first 512 bytes to store its superblock. ext2 stores its
+metadata starting at the third 512 bytes. ZFS stores its metadata
+starting at the 8k marker. By using a program that creates an ext2
+filesystem that marks many of its blocks unusable, all three of these
+can be made to not overlap their metadata, allowing them to live in
+harmony. This builds on NieDzejkob's great work. The ext2 and FAT
+filesystems live inside ZFS's 3.5MiB boot block space; the ext2
+filesystem lives in the first 512KiB of it, and the FAT filesystem
+lives in the next 3 MiB.
 
 # Can I write to the filesystems?
 
-Yes! Because FAT reserves the sectors used by ZFS's labels and ZFS
-doesn't write to the boot space, both exist in harmony.
+Yes! Because FAT and ext2 reserve the sectors used by ZFS's labels,
+and FAT reserves all space used by ext2, and ZFS doesn't write to the
+boot space, all three exist in harmony.
+
+# Notes
+
+You will need to use the [mkext2
+utility](https://github.com/pcd1193182/mkext2) to build your cursed
+filesystem, since it knows how to reserve blocks when creating an ext2
+filesystem.

mkfs.cursed

@@ -2,10 +2,13 @@
 
 set -euxo pipefail
 FAT_TMP="$(mktemp)"
-trap "rm $FAT_TMP" EXIT
+EXT2_TMP="$(mktemp)"
+trap "rm $FAT_TMP $EXT2_TMP" EXIT
 
 truncate -s 64M cursed.img
 truncate -s 4M $FAT_TMP
-mkfs.fat -R 1024 cursed.img
-dd if=$FAT_TMP of=cursed.img bs=512 skip=1 seek=1 conv=notrunc
+mkext2 $EXT2_TMP 2048 1024
+mkfs.fat -R 2048 $FAT_TMP
+dd if=$EXT2_TMP of=$FAT_TMP bs=512 skip=1 seek=1 conv=notrunc
+dd if=$FAT_TMP of=cursed.img bs=512 conv=notrunc
 sudo zpool create cursed $(pwd)/cursed.img