Hi, the new ISO of today looks good so far: https://cdimage.debian.org/cdimage/weekly-live-builds/amd64/iso-hybrid/debian-live-testing-amd64-xfce.iso The file /.disk/mkisofs shows command -compliance no_emul_toc before -map.
File size 2047475712 matches the output of /sbin/isosize. (But it is divisble by 32 * 2048. So it would not have needed any multi-session-emulation padding and thus would not show the problem.) Inspection by xorriso -indev debian-live-testing-amd64-xfce.iso -toc reports that session 1 starts at block 0 TOC layout : Idx , sbsector , Size , Volume Id ISO session : 1 , 0 , 999744s , d-live testing xf amd64 rather than at block 32 in debian-live-9.4.0-amd64-xfce.iso TOC layout : Idx , sbsector , Size , Volume Id ISO session : 1 , 32 , 952816s , d-live 9.4.0 xf amd64 The ultimate practical test happens only when the cylinder size of the ISO is not divisible by 64 KiB. Probability is 50 percent, because cylinder size of the ~2 GiB images will be chosen with sectors-per-head = 32 = 16 KiB. The image gets aligned to an integer number of full cylinders. I.e. Image_size = Cylinder_size * Cylinder_count 25% of the possible Cylinder_sizes are divisible by 64 KiB, 25% need a prime factor of 2 in Cylinder_count (probability 50%), 50% need two prime factors of 2 (probability 25%). So: 25 + 25 * 0.5 + 50 * 0.25 = 50 So a second try: https://cdimage.debian.org/cdimage/weekly-live-builds/amd64/iso-hybrid/debian-live-testing-amd64-lxde.iso has 2091008000 bytes = 31906.25 * 64 KiB. /sbin/isosize yields the same size. Practical test passed. Have a nice day :) Thomas