On 18/11/2020 18:25, Davide Marchi wrote: > Hi to All, > finally I've found the right (and tested) procedure in this simple > script, made by sowbug: > https://gist.github.com/sowbug/c7f83140581fbe3e6a9b3ddf24891e77#gistcomment-3531970 The whole problem was that the size of device was wrongly calculated from the debug: > # Device size 16777216, offset 16777216. This is apparently device size covering only LUKS2 header (default LUKS2 takes 16MB), no data at all - then the error is expected. > # All the code from > http://billauer.co.il/blog/2010/10/encrypted-iso-dvd-luks-dm-crypt-fedora-linux/ > > MB_COUNT=100 > VOL_NAME=MyVolName > DIR_TO_COPY=/tmp/mydir > > # Make a 100MB disk image > dd if=/dev/zero of=disk.img bs=1M count=$MB_COUNT All you need to do here is just to allocate requested size for data + 16M for header. (I have no idea why the original post tries to do is such complicated way.) Also you do not need to physically write the file, sparse file creation is much faster here (try sizes in GB), just run (dd can do this using seek=... option too) truncate disk.img -s 100M > > # become root > sudo su > > # write to the image > # ****** NOTE that you'll be prompted three times for a passphrase > losetup /dev/loop1 disk.img && \ This losetup step is not needed, loop is allocated automatically if argument is a file (except for very old kernel where automatic loop detach option is not yet available) just run "cryptsetup luksFormat disk.img" and "cryptsetup open disk.img mybackupdisk" > cryptsetup luksFormat /dev/loop1 && \ > cryptsetup luksOpen /dev/loop1 mybackupdisk && \ > genisoimage -R -J -joliet-long -graft-points -V $VOL_NAME -o > /dev/mapper/mybackupdisk $DIR_TO_COPY > > # close the device > cryptsetup luksClose /dev/mapper/mybackupdisk && \ > losetup -d /dev/loop1 This losetup is automatic on luksClose (so not needed) if you remove explicit loop allocation above. Milan _______________________________________________ dm-crypt mailing list dm-crypt@xxxxxxxx https://www.saout.de/mailman/listinfo/dm-crypt
