Description of problem: Home made PC with UEFI. Pre-existing OSes are Windows and Mageia 5, both working fine. Boot on an USB key with Mageia 5 Classical x86_64 dumped with Isodumper, UEFI box not checked. Installation correct until the partitioning step, I chose "Custom disk partitioning" and I can see the ESP is mounted on /mnt/win. I did Alt+CTRL+Home and go on again in the Partitioning step, the ESP is this time correctly mounted on /boot/EFI. I finished the installation but after the first reboot, there is no more Grub and no more Mageia, only Windows starting directly. In the "BIOS set up", I still have two entries in the boot override, Windows that is working (Windows starts) and Mageia that is looping (come back into the BIOS). /report.bug.xz is coming, I need to create a Live USB. Reproducible: Steps to Reproduce:
The USB stick created with Mageia 5 LiveCD-KDE4-en-i586 and Isodumper (UEFI box not checked) isn't seen by the BIOS. I re-created the USB stick with the UEFI box checked. The BIOS can see it, but it isn't bootable, looping like the Mageia entry. So I can't give you the report.
You can start the classical installer without actually performing the install. Just got to tty2, mount your / partition and copy your /root/drakx/report.bug.xz on a USB key
Keywords: (none) => NEEDINFOCC: (none) => thierry.vignaud
Created attachment 6628 [details] report.bug.xz compressed
The same bug is present in the live installer and in diskdrake on a fully updated Cauldron installation. It can be fixed by reverting drakx commit 1e4b61c23a... (display cosmetics when reading back GPT) - in particular the line $_->{pt_type} = 0x0b if $_->{fs_type} eq 'vfat';
CC: (none) => mageia
I see no trace of /dev/sda2 being mounted on /mnt/win. Also we don't use /mnt/win mount point in drakx... (eg your §dev/sda4 is mounted on /media/win_d) I guess you saw /media/win_c instead of /mnt/win?
(In reply to Martin Whitaker from comment #4) > It can be fixed by reverting drakx commit > 1e4b61c23a... (display cosmetics when reading back GPT) - in particular the > line > > $_->{pt_type} = 0x0b if $_->{fs_type} eq 'vfat'; Ah that's interesting. Can you test the attached patch?
Created attachment 6631 [details] set pt_type according to fs_type before flag cases
(In reply to Thierry Vignaud from comment #6) > Ah that's interesting. > Can you test the attached patch? That fixes it in diskdrake and the live installer. It will take me a bit longer to test the classic installer (can't easily replace mdkinst.sqfs now we dd the iso rather than copy it to a FAT partition).
You can put a patch.pl on a USB key & pass "patch" option when booting the CD See https://wiki.mageia.org/en/Drakx-installer_tips_and_tricks#Using_OEM_patch_in_the_tree_.26_drakx-in-chroot eg sg like: use partition::gpt; package partition::gpt; undef *read_one; *read_one = sub { # INSERT HERE THE FULL BODY OF THE PATCHED read_one() }; 1;
commit fd0c659539609be48af980e6201c262bfcd9aa57 Author: Thierry Vignaud <thierry.vignaud@...> Date: Sun May 24 12:11:27 2015 +0200 set pt_type according to fs_type before flag cases (mga#16029) --- Commit Link: http://gitweb.mageia.org/software/drakx/commit/?id=fd0c659539609be48af980e6201c262bfcd9aa57
(In reply to Thierry Vignaud from comment #5) > I see no trace of /dev/sda2 being mounted on /mnt/win. > > Also we don't use /mnt/win mount point in drakx... (eg your §dev/sda4 is > mounted on /media/win_d) > I guess you saw /media/win_c instead of /mnt/win? Yes, you are right it is /media/win_c
For completeness I tested the patch in the classic installer and can confirm it fixes the bug there too.
Keywords: NEEDINFO => PATCHPriority: Normal => release_blockerAssignee: bugsquad => thierry.vignaud
Closed
Status: NEW => RESOLVEDResolution: (none) => FIXED