On 2/11/2015 4:38 AM, Steve McIntyre wrote: > Hmmm, and just testing with the latest Debian amd64 daily netinst all > works flawlessly here still, using the same empty disk in a KVM setup > as your bug reporter.
Strange.. I'll have to poke around with that. > This suggests the problem is Ubuntu-specific, or there's an issue with > other software. You allude to potential changes in libparted; how do > the versions in Debian and Ubuntu compare here? They are in sync. The change in particular that I think may be related is that in parted3, it now keeps the partition table cached between commands rather than re-reading it every time. This has the affect that when you create a partition and tell parted it will hold an ext2 filesystem, and then print the table, it now reports that it contains an ext2 filesystem since it is no longer re-reading the disk and finding the partition to be empty. Now that I think about it though, this change was in parted proper and so should not affect other libparted clients like parted_server, so that seems to have been a red herring. The thing I saw that struck me as similar to that parted change was that when I put a set +x in the partman-efi script, it appeared to have identified an ext2 filesystem on the disk that partman was formatting with the usual default ext4 root and swap partitions so I kind of assumed it simply had not gotten around to running mke2fs yet but had asked libparted to make an ext2 partition and then partman-efi identified it as a non efi system partition. This script really should be running before the disk has been modified in any way though right? So it should still just see a blank disk or a disk with no partitions and not count it as a non ESP. Hrm... I'll do some more debugging tonight. -- To UNSUBSCRIBE, email to debian-boot-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/54db85a4.9050...@gmail.com
