(in EFI mode, although that's maybe not relevant) Using the classical installer, at custom partitioning stage, the /media/win_c and /media/win_d mountpoint are inverted (win_c for D: and win_d for C:) and the guessed letters displayed in partition details are even different from that. Details: * sda1 : "Guessed letter: C". Not mounted, which is ok, but it shouldn't have a guessed letter at all. * sda2 : "Guessed letter: D". Mounted as /boot/EFI, which is ok, but it shouldn't have a guessed letter at all * sda4 : "Guessed letter: E". Not mounted (same comment as sda1) * sda5 : "Guessed letter: F". Mounted as /media/win_d. Ideally should be "Guessed letter C" + mountpoint /media/win_c... This is the partition that contains the Windows 8 system, labelled Boot. Maybe we should use labels for mountpoints when available? Now the second disk: * sdb5: "Guessed letter: C" (it starts from C again, is that a bug?). Mounted as /media/win_c. Expected : "Guessed letter: D", mountpoint /media/win_d. In addition to that, the partitioner also shows the booted USB key as sdc: * sdc1 : "Guessed letter: C" (again). No mountpoint (which is good). Expected: "Guessed letter: E", or no guessed letter at all. Attached are outputs from fdisk and gdisk about my disks. Reproducible: Steps to Reproduce:
Created attachment 6203 [details] Outputs from fdisk and gdisk
You tried an old installer. Please attach your /root/drakx/report.bug.xz for checking. All of these should be fixed as of drakx 16.76
CC: (none) => thierry.vignaud
Hardware: i586 => All
Either the installer currently in boot-nonfree.iso on distrib-coffee is an old one, or it's not fixed. I've downloaded it again and booted it from an usb key. To make things clear: the previous issue, which I had reported on the QA mailing list, is fixed. I mean, only the relevant partitions have mountpoints. The issue that remains is mountpoint naming and displayed guessed drive letters. Not critical, just confusing. Attaching ddebug.log which I copied from /tmp during install.
Created attachment 6214 [details] ddebug.log
(In reply to Samuel VERSCHELDE from comment #3) > Either the installer currently in boot-nonfree.iso on distrib-coffee is an > old one, or it's not fixed. I've downloaded it again and booted it from an > usb key. stage2 is not _in_ boot-nonfree.iso, but stage1 on the iso pulls it in (it is called mdkinst.sqfs and can be found in /distrib/cauldron/<arch>/install/stage2/ > > To make things clear: the previous issue, which I had reported on the QA > mailing list, is fixed. I mean, only the relevant partitions have > mountpoints. The issue that remains is mountpoint naming and displayed > guessed drive letters. Not critical, just confusing. > > Attaching ddebug.log which I copied from /tmp during install. it says * second stage install running (DrakX v16.78) so you did indeed use a newer one
CC: (none) => marja11
Attachment 6214 mime type: text/x-log => text/plain
Comment on attachment 6214 [details] ddebug.log This is useless, you didn't go further so we don't see what the installer would have done...
Attachment 6214 is obsolete: 0 => 1
I don't think you understood by bug report, or I don't understand your comment. My bug report is not about what the installer will do afterwards, but what it does at partitioning stage. I know what it does afterwards. It does what it announces when I select the partitions in the partitioner to see their information: * set /dev/sda2 mountpoint as /boot/EFI (which is OK) * set /dev/sda5 mountpoint as /media/win_d (should be /media/win_c) * set /dev/sdb5 mountpoint as /media/win_c (should be /media/win_d) Doesn't mount other windows partitions anymore, since version 16.78 (which is OK since they are recovery or hidden partitions). And the "Guessed letter" information displayed in each partition's information in the partitionner: * is wrong * uses the letter "C" 3 times * is not coherent with the mountpoint names
I fail to see any "set /dev/" in the report.bug you attached...
Keywords: (none) => NEEDINFO
I first thought you needed the logs for checking the version, since you said I was using an old installer. Now if you need the complete logs until installation, if that helps I'll provide that tonight.
ddebug.log shows you have a /dev/sda6 partition with: /dev/sda6 147666944 198866943 51200000 24.4G Microsoft basic data Shouldn't a drive letter be guessed for this one too and a mount point assigned? As for this bug report, I see two related bugs: 1) "Guessed letter" and the actual mount point are not synced (e.g. "Guessed letter: F" ends up as /media/win_d when it should be /media/win_f. 2) Recovery partitions and stuff like that are properly ignored for the mount point, but they are still taken into account in the "Guessed letter" logic (fixing this should fix (1), but maybe (1) also calls for a check that guessed letter and mount point are the same?)
(In reply to Samuel VERSCHELDE from comment #9) yes I want the complete logs. (In reply to Rémi Verschelde from comment #10) for 2) we might have a missing call to isRecovery() in the partition view
Created attachment 6219 [details] don't guess a drive letter for recovery partition (mga15636) untested
Created attachment 6223 [details] report.bug.xz
(In reply to Rémi Verschelde from comment #10) > ddebug.log shows you have a /dev/sda6 partition with: > /dev/sda6 147666944 198866943 51200000 24.4G Microsoft basic data > > Shouldn't a drive letter be guessed for this one too and a mount point > assigned? sda6 really is an ext4 partition (it's the root partition)
Priority: Normal => LowSeverity: normal => minor
Is this bug report still NEEDINFO because there's a patch to test?
Keywords: (none) => PATCHAssignee: bugsquad => thierry.vignaud
Patch is already applied
Keywords: PATCH => (none)
(In reply to Thierry Vignaud from comment #16) > Patch is already applied You mean attachment 6219 [details] which you described as untested? (I haven't tested myself yet /o\)
Yes: http://gitweb.mageia.org/software/drakx/commit/perl-install/partition_table.pm?id=5efbf00cc1845b67ac7d4a03b9a75ecf18e50004
You can test Mga5 RC or final.
(In reply to Thierry Vignaud from comment #18) > Yes: > http://gitweb.mageia.org/software/drakx/commit/perl-install/partition_table. > pm?id=5efbf00cc1845b67ac7d4a03b9a75ecf18e50004 (In reply to Thierry Vignaud from comment #19) > You can test Mga5 RC or final. No one replied that the fix doesn't work, so assuming the fix works well.
Status: NEW => RESOLVEDResolution: (none) => FIXED