ISOs generated by Cubic are not bootable
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Cubic |
Fix Released
|
Critical
|
Cubic PPA | ||
Classic-development |
Fix Released
|
Critical
|
Cubic PPA | ||
Release-2018-07-34 |
Fix Released
|
Critical
|
Cubic PPA |
Bug Description
From comments posted by Jim Narey (jimnarey) in Bug #1778578
Interestingly, this setup creates an ISO which Etcher on the Mac says does not have a partition table and is not bootable. It boots in a VM but not when flashed to a USB. If I flash it using Rufus in Windows, which copies the ISO file-by-file rather than bit-by-bit and then has its own process for making the USB bootable, it boots fine.
I'm using rev 28 in part because it lets you choose the kernel but also, with the latest version (and the dev version you pointed me to) the ISOs would not boot whether I tried to change the kernel or not. They would get to the GRUB menu (in EFI mode) or the GUI with Try Xubuntu/Install Xubuntu under BIOS mode but then go to a black screen and hang, whatever was chosen.
I haven't had much luck with changing the capser/ kernel in any version of Cubic. I've had mixed success with manually changing it and using mkisofs to build the ISO. Whether or not ISOs boot successfully like this seems to depend on a combination of the 'host' major version, the chroot major version, kernel versions, whether an EFI host is used and possibly more.
Changed in cubic: | |
assignee: | nobody → Cubic PPA (cubic-wizard) |
importance: | Undecided → High |
status: | New → Confirmed |
Hi - thank you for looking into this.
I've worked out that the issue in the first para isn't a bug. I had a look in utilities.py for the mkisofs commands and looked for the files the various formulations refer to and couldn't see either /usr/lib/ ISOLINUX/ isohdpfx. bin or /usr/lib/ syslinux/ isohdpfx. bin. I installed isolinux with apt, ran the command from utilities.py in the terminal and I have an image that Etcher sees as having a partition table.
I'll try to move over to the latest version on a clean VM soon and start looking at the second issue, which is the reason I'm currently using rev 28.
Thanks.