Bug 30862

Summary: Installer GUI fails to start when run in VirtualBox with EFI boot enabled
Product: Mageia Reporter: Martin Whitaker <mageia>
Component: RPM PackagesAssignee: Kernel and Drivers maintainers <kernel>
Status: RESOLVED FIXED QA Contact:
Severity: major    
Priority: Normal CC: andrewsfarm, fri, marja11, westel, wilcal.int
Version: Cauldron   
Target Milestone: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Source RPM: kernel-5.19.9-1.mga9.src.rpm CVE:
Status comment:
Attachments: Output from installer 'bug' command

Description Martin Whitaker 2022-09-18 21:20:45 CEST
Created attachment 13384 [details]
Output from installer 'bug' command

The X server fails to start, with these errors in /var/log/Xorg.0.log:

[     6.937] (EE) XKB: Could not invoke xkbcomp
[     6.937] (EE) XKB: Couldn't compile keymap
[     6.937] (EE) XKB: Failed to load keymap. Loading default keymap instead.
[     6.938] (EE) XKB: Could not invoke xkbcomp
[     6.938] (EE) XKB: Couldn't compile keymap
[     6.938] XKB: Failed to compile keymap
[     6.938] Keyboard initialization failed. This could be a missing or incorrect setup of xkeyboard-config.
[     6.938] (EE)
Fatal server error:
[     6.938] (EE) Failed to activate virtual core keyboard: 2(EE)

xkbcomp is present and can be run from the debug shell command line.

There is also this error in the system log, which may be the real culprit:

<3>[    6.938686] x86/PAT: Xorg:642 map pfn expected mapping type uncached-minus for [mem 0xe0000000-0xe02fffff], got write-combining
<4>[    6.938741] ------------[ cut here ]------------
<4>[    6.938742] WARNING: CPU: 1 PID: 642 at arch/x86/mm/pat/memtype.c:1107 untrack_pfn+0xb3/0xc0
<4>[    6.938747] Modules linked in: af_packet squashfs loop isofs ...
<4>[    6.938771] CPU: 1 PID: 642 Comm: Xorg Not tainted 5.19.9-desktop-1.mga9 #1
<4>[    6.938773] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
<4>[    6.938774] RIP: 0010:untrack_pfn+0xb3/0xc0
<4>[    6.938776] Code: 65 48 2b 04 ...
<4>[    6.938777] RSP: 0018:ffffb84400617b78 EFLAGS: 00010282
<4>[    6.938779] RAX: 00000000ffffffea RBX: 0000000000000000 RCX: 000000005f8c5067
<4>[    6.938780] RDX: 0000000000000001 RSI: 0000000000000568 RDI: ffffe6dcc17e3168
<4>[    6.938780] RBP: ffff8f3597b267d0 R08: ffffb84400617b78 R09: 00007f2b5f67d000
<4>[    6.938781] R10: ffff8f359fab1690 R11: ffff8f35a3ff91d0 R12: ffffb84400617c58
<4>[    6.938782] R13: 00007f2b5f67d000 R14: ffff8f35803f2268 R15: ffffb84400617be8
<4>[    6.938783] FS:  00007f2b5fa26940(0000) GS:ffff8f359fa80000(0000) knlGS:0000000000000000
<4>[    6.938784] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
<4>[    6.938785] CR2: 00000000019f4028 CR3: 0000000108842000 CR4: 00000000000506e0
<4>[    6.938787] Call Trace:
<4>[    6.938797]  <TASK>
<4>[    6.938799]  unmap_single_vma+0x9f/0x110
<4>[    6.938810]  unmap_vmas+0x92/0x110
<4>[    6.938813]  exit_mmap+0xa3/0x1e0
<4>[    6.938816]  mmput+0x56/0x130
<4>[    6.938819]  dup_mm+0x5c7/0x600
<4>[    6.938822]  copy_process+0xf03/0x1a60
<4>[    6.938826]  kernel_clone+0x9b/0x3e0
<4>[    6.938828]  __do_sys_clone+0x78/0xa0
<4>[    6.938831]  do_syscall_64+0x3a/0x90
<4>[    6.938841]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
<4>[    6.938843] RIP: 0033:0x7f2b601d11c3
<4>[    6.938845] Code: 48 ff ff ff ...
<4>[    6.938846] RSP: 002b:00007ffe7abd3048 EFLAGS: 00000246 ORIG_RAX: 0000000000000038
<4>[    6.938848] RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 00007f2b601d11c3
<4>[    6.938848] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000001200011
<4>[    6.938849] RBP: 0000000000000000 R08: 0000000000000000 R09: 00000000019ad380
<4>[    6.938850] R10: 00007f2b5fa26c10 R11: 0000000000000246 R12: 0000000000000001
<4>[    6.938850] R13: 00007ffe7abd3160 R14: 0000000000638500 R15: 00000000005d8660
<4>[    6.938852]  </TASK>
<4>[    6.938853] ---[ end trace 0000000000000000 ]---
<3>[    6.939521] x86/PAT: Xorg:642 map pfn expected mapping type uncached-minus for [mem 0xe0000000-0xe02fffff], got write-combining

I did not see this bug when I last tested ~6 weeks ago.

There is no error when booting in legacy mode.

Full installer logs attached.
Marja Van Waes 2022-09-19 21:41:14 CEST

CC: (none) => marja11
Assignee: bugsquad => isobuild

Comment 1 Martin Whitaker 2022-09-19 23:15:07 CEST
I can reproduce this bug by rebuilding the Mageia 8 drakx-installer-images using the 5.19 series kernels from backports (either 5.19.3-1.mga8 or 5.19.7-1.mga8). It does not occur when using the 5.18 series kernels (tested with 5.18.7-1.mga8 and 5.18.15-1.mga8). So this looks like a kernel/driver regression.

Component: Installer => RPM Packages
Assignee: isobuild => kernel
Source RPM: (none) => kernel-5.19.9-1.mga9.src.rpm

Comment 2 Morgan Leijström 2022-12-09 14:46:39 CET
Possibly Bug 31241 "no graphics installer in Vbox clients" is a dupe.

I guess it may be long, if ever, to get fixed, so: For Errata.

Keywords: (none) => FOR_ERRATA9
CC: (none) => fri

Comment 3 Lewis Smith 2022-12-13 10:40:13 CET
*** Bug 31241 has been marked as a duplicate of this bug. ***

CC: (none) => wilcal.int

Comment 4 Morgan Leijström 2022-12-13 11:14:34 CET
Discussion thread:
https://ml.mageia.org/l/arc/qa-discuss/2022-12/msg00072.html
Comment 5 Ben McMonagle 2023-02-09 11:01:03 CET
a bit more digging.

if I choose display option *VBoxVGA*  it seems to be more often than not that the graphical stage-2 GUI installer is displayed.

maybe just for my hardware.

after install, switch back to VMSVGA seems to work.

CC: (none) => westel

Comment 6 Thomas Andrews 2023-02-09 20:00:37 CET
Doesn't make any difference on my hardware.

CC: (none) => andrewsfarm

Comment 7 Martin Whitaker 2023-05-08 15:36:49 CEST
This has been fixed in the latest (6.3 series) kernel.

Resolution: (none) => FIXED
Status: NEW => RESOLVED

Comment 8 Morgan Leijström 2023-05-08 16:29:31 CEST
Great then :)

Keywords: FOR_ERRATA9 => (none)