all new kernels are unusable with amdgpu - Sapphire Nitro + RX590
coo...@posteo.de
i'am a long time FunToo (GenToo) User.
I bought a Sapphire Nitro + RX590 and thought this will work perfect with Qubes OS.
But there were some hiccups, because the kernel is to old including very old Firmware files.
I could only start installer with ' amdgpu.dpm=0 ' and everything went fine.
After playing around with the System, i decided to reinstall and choose XFS Filesystem (the System is noticeably faster) and just a little SWAP file (1024MB), because my System RAM is 32GB.
After some research i somehow managed the Problem regarding not included ' amdgpu' Driver - just installed it from an old Fedora 25 Repository - and i also downloaded current available Firmware files from Fedora 31 Repository and copied to dom0 ;-).
I was happy to see new Kernels, but none of them are working !
On the Lightdm login screen, there are some graphic glitches, just the backround and after typing my user Password, Lightdm crashs and toss me back to Login menu.
Only working Kernel without glitches are 4.14.74 and 4.14.103 with 'amdgpu.dpm=0 and plymouth.ignore-serial-consoles' grub option.
I also tried to compile a kernel with the config from my Funtoo System, but the behaviour is the same, no login possible or the Desktop just crashs after some time and i'am back to login screen.
I must admit that RPM based System are PITA including SystemD and XFCE is a ugly Desktop Environment - i hope MATE will be available in the Future.
And i also hope you switch to Debian (or maybe to GenToo) for dom0, because it is uncomplicated and APT is very easy to handle.
I'am wondering why all new Kernels don't work on my System - is dom0 just to old?
here's my System:
---
bequiet! Silent Base 800 Black
MSI C236A Workstation - Intel Xeon E3-1245v5 - 32GB RAM Kingston - Sapphire Nitro+ Radeon RX590 8G
Samsung SSD PM961 M.2 128GB - Samsung SSD 840 EVO 120GB - Crucial m4 SSD 128GB
Creative Sound Blaster RX PCIe - Logitech Speaker System Z623 2.1
Logitech G110 Keyboard - Optical Gaming Mouse G400
---
thx
Foppe de Haan
That seems strange. There was an issue with the firmware not supporting the rx 590 properly due to some last-minute changes, but kernels 4.19 and up should support it just fine, so long as you have the newer one.
CooSee
tried yesterday with the current latest kernel, no login possible.
Lightdm kicks me back Login screen.
here's the xsession-errors - with 4.20.3 Kernel :
---
generating cookie with syscall
generating cookie with syscall
generating cookie with syscall
generating cookie with syscall
xfce4-session: No SSH authentication agent found
gpg-agent[5670]: WARNING: "--write-env-file" is an obsolete option - it has no effect
(xfce4-session:5664): xfce4-session-WARNING **: gpg-agent returned no PID in the variables
gpg-agent[5671]: gpg-agent (GnuPG) 2.1.13 started
(xfce4-session:5664): xfce4-session-WARNING **: xfsm_manager_load_session: Something wrong with /home/qubees/.cache/sessions/xfce4-session-dom0:0, Does it exist? Permissions issue?
Screen layout change event received
** (xfdesktop:5681): WARNING **: Thumbnailer failed calling GetFlavors
sys-firewall: Sending monitor layout
asyncio: Task exception was never retrieved
future: <Task finished coro=<GUILauncher.send_monitor_layout() done, defined at /usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py:258> exception=FileNotFoundError(2, 'No such file or directory')>
Traceback (most recent call last):
File "/usr/lib64/python3.5/asyncio/tasks.py", line 240, in _step
result = coro.send(None)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 283, in send_monitor_layout
with open(self.guid_pidfile(vm.xid)) as pidfile:
FileNotFoundError: [Errno 2] No such file or directory: '/var/run/qubes/guid-running.3'
sys-net: Sending monitor layout
asyncio: Task exception was never retrieved
future: <Task finished coro=<GUILauncher.send_monitor_layout() done, defined at /usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py:258> exception=FileNotFoundError(2, 'No such file or directory')>
Traceback (most recent call last):
File "/usr/lib64/python3.5/asyncio/tasks.py", line 240, in _step
result = coro.send(None)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 283, in send_monitor_layout
with open(self.guid_pidfile(vm.xid)) as pidfile:
FileNotFoundError: [Errno 2] No such file or directory: '/var/run/qubes/guid-running.1'
wrapper-1.0: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0.
xfce-polkit: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0.
xfce4-panel: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0.
** (xss-lock:5703): CRITICAL **: X connection lost; exiting.
xfwm4: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0.
Thunar: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0.
xfsettingsd: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0.
XIO: fatal IO error 11 (Resource temporarily unavailable) on X server ":0.0"
after 13 requests (13 known processed) with 0 events remaining.
xfce4-session: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.
xfdesktop: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0.
Gdk-Message: qrexec-policy-agent: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0.
wrapper-1.0: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0.
XIO: fatal IO error 11 (Resource temporarily unavailable) on X server ":0.0"
after 139 requests (139 known processed) with 0 events remaining.
Can't open display :0.0
Invalid MIT-MAGIC-COOKIE-1 keyxprop: unable to open display ':0.0'
asyncio: Task exception was never retrieved
future: <Task finished coro=<GUILauncher.start_gui() done, defined at /usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py:238> exception=CalledProcessError(1, ['xprop', '-root', '-notype', 'KWIN_RUNNING'])>
Traceback (most recent call last):
File "/usr/lib64/python3.5/asyncio/tasks.py", line 240, in _step
result = coro.send(None)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 256, in start_gui
yield from self.start_gui_for_vm(vm, monitor_layout=monitor_layout)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 191, in start_gui_for_vm
guid_cmd = self.common_guid_args(vm)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 173, in common_guid_args
guid_cmd += self.kde_guid_args(vm)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 130, in kde_guid_args
['xprop', '-root', '-notype', 'KWIN_RUNNING']) == \
File "/usr/lib64/python3.5/subprocess.py", line 316, in check_output
**kwargs).stdout
File "/usr/lib64/python3.5/subprocess.py", line 398, in run
output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command '['xprop', '-root', '-notype', 'KWIN_RUNNING']' returned non-zero exit status 1
Invalid MIT-MAGIC-COOKIE-1 keyxprop: unable to open display ':0.0'
asyncio: Task exception was never retrieved
future: <Task finished coro=<GUILauncher.start_gui() done, defined at /usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py:238> exception=CalledProcessError(1, ['xprop', '-root', '-notype', 'KWIN_RUNNING'])>
Traceback (most recent call last):
File "/usr/lib64/python3.5/asyncio/tasks.py", line 240, in _step
result = coro.send(None)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 256, in start_gui
yield from self.start_gui_for_vm(vm, monitor_layout=monitor_layout)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 191, in start_gui_for_vm
guid_cmd = self.common_guid_args(vm)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 173, in common_guid_args
guid_cmd += self.kde_guid_args(vm)
File "/usr/lib/python3.5/site-packages/qubesadmin/tools/qvm_start_gui.py", line 130, in kde_guid_args
['xprop', '-root', '-notype', 'KWIN_RUNNING']) == \
File "/usr/lib64/python3.5/subprocess.py", line 316, in check_output
**kwargs).stdout
File "/usr/lib64/python3.5/subprocess.py", line 398, in run
output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command '['xprop', '-root', '-notype', 'KWIN_RUNNING']' returned non-zero exit status 1
(wrapper-2.0:5761): libnotify-WARNING **: Failed to connect to proxy
Gdk-Message: wrapper-2.0: Fatal IO error 9 (Bad file descriptor) on X server :0.0.
gpg-agent[5671]: SIGTERM received - shutting down ...
gpg-agent[5671]: gpg-agent (GnuPG) 2.1.13 stopped
---
this happens with all new Kernels after 4.14.103 <<< this works, but must use amdgpu.dpm=0 !
booting with >> smt=on reboot=acpi acpi_backlight=vendor
thx
CooSee
---
generating cookie with syscall
generating cookie with syscall
generating cookie with syscall
generating cookie with syscall
xfce4-session: No SSH authentication agent found
gpg-agent[5543]: WARNING: "--write-env-file" is an obsolete option - it has no effect
gpg-agent[5544]: gpg-agent (GnuPG) 2.1.13 started
(xfce4-session:5537): xfce4-session-WARNING **: gpg-agent returned no PID in the variables
(xfce4-session:5537): xfce4-session-WARNING **: xfsm_manager_load_session: Something wrong with /home/qubees/.cache/sessions/xfce4-session-dom0:0, Does it exist? Permissions issue?
** (xfdesktop:5554): WARNING **: Thumbnailer failed calling GetFlavors
Screen layout change event received
sys-firewall: Starting GUI
sys-net: Starting GUI
sys-firewall: Sending monitor layout
sys-net: Sending monitor layout
sys-firewall: Sending monitor layout
sys-net: Sending monitor layout
(wrapper-2.0:5643): Gtk-WARNING **: Negative content width -3 (allocation 1, extents 2x2) while allocating gadget (node button, owner PulseaudioButton)
(xfce4-terminal:5721): Gtk-WARNING **: Allocating size to GtkScrollbar 0x61463788e580 without calling gtk_widget_get_preferred_width/height(). How does the code know the size to allocate?
(xfce4-terminal:5721): Gtk-WARNING **: Allocating size to GtkScrollbar 0x61463788e580 without calling gtk_widget_get_preferred_width/height(). How does the code know the size to allocate?
(xfce4-terminal:5721): Gtk-WARNING **: Allocating size to GtkScrollbar 0x61463788e780 without calling gtk_widget_get_preferred_width/height(). How does the code know the size to allocate?
(Thunar:5552): thunar-WARNING **: Failed to launch the volume manager (Failed to execute child process "thunar-volman" (No such file or directory)), make sure you have the "thunar-volman" package installed.
(Thunar:5552): thunar-WARNING **: Failed to launch the volume manager (Failed to execute child process "thunar-volman" (No such file or directory)), make sure you have the "thunar-volman" package installed.
(Thunar:5552): thunar-WARNING **: Failed to launch the volume manager (Failed to execute child process "thunar-volman" (No such file or directory)), make sure you have the "thunar-volman" package installed.
work-deb: Starting GUI
work-deb: Sending monitor layout
work-deb: Creating appmenus
---
Foppe de Haan
CooSee
> what's ~/.local/share/xorg/Xorg.0.log telling you?
there' no log in this specified directory.
tried right now with kernel-5.0.2-1 - i uploaded some log files - see attachment.
why there's no ' xorg-x11-drv-amdgpu ' package available in Qubes including on install, because without it, Qubes always load the radeon module and Qubes become unusable >> no tear free etc.
all the users with current Hardware can't use Qubes also because of very old Firmware files >> from last year!? - even in 4.1 testing - no offence :-)
how to compile rpms on Qubes - maybe i can use the current available xorg-x11-drv-amdgpu and mybe xorg and mesa too.
after login with new kernels, the desktop is loading just for 1 sec.
there must be an error in the system, because i also tried KDE and Awesome - same behaviour, e.g. Lightdm backround is scrambled with glitches at login screen and it always kicks me back to login screen - using Dual Monitor Setup.
4.14.103 is working fine, but only with ' amdgpu.dpm=0 ' <<< is this bad for the device ?
thx
CooSee
[ 123.712] (EE) glamor0: GL error: GL_OUT_OF_MEMORY in glTexSubImage
[ 123.719] (WW) glamor: Failed to allocate 200x200 FBO due to GL_OUT_OF_MEMORY.
[ 123.719] (WW) glamor: Expect reduced performance.
why 4.14.103-1 Kernel can handle the login and newer Kernels not ?
thx
Foppe de Haan
is there a difference in which driver it loads (lspci -nnk | grep -i vga -A 3)?
Foppe de Haan
> how to compile rpms on Qubes - maybe i can use the current available xorg-x11-drv-amdgpu and mybe xorg and mesa too.
> after login with new kernels, the desktop is loading just for 1 sec.
> there must be an error in the system, because i also tried KDE and Awesome - same behaviour, e.g. Lightdm backround is scrambled with glitches at login screen and it always kicks me back to login screen - using Dual Monitor Setup.
yeah, sounds like there's a driver issue
> 4.14.103 is working fine, but only with ' amdgpu.dpm=0 ' <<< is this bad for the device ?
why did you try that option, specifically? phoronix guide? anyway, as he writes, it's not harmful, it just won't game very well. ;)
CooSee
the amdgpu.dpm=0 option with 4.14 kernel is necessary, because without it the screen goes black and the fans are starting to blow up!
on all kernels, amdgpu is always loaded, no difference.
i updated the pci.ids file for better detection.
changed yum.repos.d and tried, but:
qubes-dom0-update
Using sys-firewall as UpdateVM to download updates for Dom0; this may take some time...
Last metadata expiration check: 0:36:18 ago on Sat Mar 23 12:43:25 2019.
Error:
Problem: The operation would result in removing the following protected packages: dnf, systemd, systemd-udev
(try to add '--skip-broken' to skip uninstallable packages)
with --skip-broken the same answer >> The operation would result in removing the following protected packages: dnf, systemd, systemd-udev
installing xorg-x11-drv-amdgpu from fc29 won't work because it depends on newer version of xorg >> xserver-abi (videodrv-24) and todo so it would brake the system and uninstall dnf!?
this is awful - that's why i never liked RPM based distros - it's PITA handling with packages.
is some one so kind and compile the newest version of xorg-x11-drv-amdgpu for FC25 and make it also officially available for all the users with newer AMD Graphics?
i noticed that the 'amd_iommu_v2' driver is loaded too, but even though, kernel 4.14.103 can handle the desktop but newer kernels not, somehow.
i tried to blacklist to try without - but didn't work at all - it's always loading besides amdgpu module!
dmesg says >> AMD_IOMMUV2 funktionality not available on this system.
thx for your efford :-)
CooSee
the odd thing is with newer kernels that Xorg's starting two times at the same time:
1. Xorg.0.log >> [ 63.475] (II) AMDGPU(0): mem size init: gart size :b7c6e000 vram size: s:1ff2e8000 visible:f2e8000
2. 38.953] (II) AMDGPU(0): mem size init: gart size :b7c6e000 vram size: s:1fe316000 visible:e316000
the second one probing for Monitor goes on and failed with:
[ 62.958] (EE) glamor0: GL error: GL_OUT_OF_MEMORY in glTexSubImage
[ 62.958] (WW) glamor: Failed to allocate 22x22 FBO due to GL_OUT_OF_MEMORY.
[ 62.958] (WW) glamor: Expect reduced performance.
at the end.
please see attachments.
John Mitchell
awokd
>> what's ~/.local/share/xorg/Xorg.0.log telling you?
>
> there' no log in this specified directory.
CooSee
ok, here's the one from working 4.14.103 kernel using a debian-vm -see attachment.
All the other Xorg.logs are from dom0 with newer kernels.
AMD_IOMMUV2 >> is not the problem, because testing a compiled kernel without > same problem > no login possible > crashing back to login and as i said, with newer kernel after 4.14 the login background is scrambled with glitches, but this happens only with newer kernel. With 4.14.103 no problems at all - no glitches and even direct rendering's working fine with the amdgpu :
---
glxinfo | grep rendering
direct rendering: Yes
---
@John Mitchell
thx for the video, but i already managed the Firmware problem in Qubes with using the linux-firmware package from FC31, so it's current!
the error with:
---
[ 62.958] (EE) glamor0: GL error: GL_OUT_OF_MEMORY in glTexSubImage
[ 62.958] (WW) glamor: Failed to allocate 22x22 FBO due to GL_OUT_OF_MEMORY.
[ 62.958] (WW) glamor: Expect reduced performance.
---
was already addressed in newer versions of xorg-server, please see:
https://gitlab.freedesktop.org/xorg/xserver/commits/master?utf8=%E2%9C%93&search=glamor
On my Funtoo System with a manually build kernel, i never got this kind of problems at all - no glitches etc. and everything working fine with RX590.
After installing the RX590 i had to boot with amdgpu.dpm=0 at first, but after recompiling the kernel and also with a better firmware for the Card, i never have to use the boot option anymore.
Now, how to switch or install a newer Version of Xorg-Server and Mesa on Qubes FC25 - the problem is that this packages depends on newer Version of Openssl-1.1 and Glibc-2.26 or higher and Mesa depends on LLVM 7 ?
With FunToo this is so easy :-)
It would be great if there's a solution for this annoying problem, so i can also use newer kernels without booting amdgpu.dpm=0 boot Option.
thx
CooSee
i tried X11 with "Option AccelMethod=none" with newer kernel with no artifacts and login was possible, so the culprit is radeonsi_dri.so from mesa-dri package.
i somehow managed to switch to FC29 and FC29 Updates and installed the newest xorg-x11-drv-amdgpu + xorg-server + mesa-dri-drivers + lightdm, but now lightdm freezes completely.
here are some xorg-log errors:
(WW) AMDGPU(0): gbm_create_device returned NULL, using ShadowFB
[ 1802.400] (EE) AIGLX error: dlopen of /usr/lib64/dri/swrast_dri.so failed (/usr/lib64/dri/swrast_dri.so: undefined symbol: amdgpu_query_sw_info)
[ 1802.400] (EE) AIGLX error: unable to load driver swrast
[ 1802.400] (EE) GLX: could not load software renderer
[ 1802.400] (II) GLX: no usable GL providers found for screen 0
---
lightdm errors:
gbm: failed to open any driver (search paths /usr/lib64/dri)
gbm: Last dlopen error: /usr/lib64/dri/radeonsi_dri.so: undefined symbol: amdgpu_query_sw_info
failed to load driver: radeonsi
gbm: failed to open any driver (search paths /usr/lib64/dri)
gbm: Last dlopen error: /usr/lib64/dri/kms_swrast_dri.so: undefined symbol: amdgpu_query_sw_info
failed to load driver: kms_swrast
gbm: failed to open any driver (search paths /usr/lib64/dri)
gbm: Last dlopen error: /usr/lib64/dri/swrast_dri.so: undefined symbol: amdgpu_query_sw_info
failed to load swrast driver
/bin/X: symbol lookup error: /usr/lib64/xorg/modules/drivers/amdgpu_drv.so: undefined symbol: drmModeListLessees
which packages i have to install to only get X11 working ?
most of packages are not upgraded, e.g python2 and python3!
or can i use some important packages from qubes r4.1 repo ?
this is really bugging me, because the Card is working pefect on FunToo.
i could use onboard Intel-igfx, but then i had to always switch the cables to enjoy perfectly working FunToo with my new RX 590 ;-)
very grateful for any help.
CooSee
with kernel 4.14-103-1 everything's working fine, no artifacts at login, but i can only boot with amdgpu.dpm=0, without the fans going to full speed
and the screen goes black and only reset is possible.
i'll try the 4.14.109 kernel with qubes-linux-kernel builder if there's a possibility to start without amdgpu.dpm=0.
is there a way to use upcoming v4.1 packages?
or should i create all 4.1 packages with qubes-builder and install manually?
maybe it's xfce4 - can i upgrade to fc29 packages without killing the whole system?
sorry for so many questions, i hope someone has a solution.
thx
awokd
if you still get artifacts. You can't really upgrade Fedora in dom0
without a lot of additional work; I'm surprised you managed to get as
far as you did! You lost me a bit on this thread though, why not run the
older kernel for now until it is resolved?
awokd
> however, i assume the combination with very old dom0 + xen + fedora 25 is not really possible to use on newer hardware.
You might be interested in Qubes 4.1 which will run Fedora 29 in dom0.
It's not ready for release yet, but you could keep an eye out or
possibly help develop it.
> Debian has a way better Package Manager and is much smarter then DNF - why not using dom0 with Debian?
See https://github.com/QubesOS/qubes-issues/issues/1919.
> On Qubes the user has no real control of the system, e.g. the Kernel with so much compiled in modules no one needs or will never use.
>
> there should be a possibility to create your own Kernel on Qubes.
Qubes can be built by following
https://www.qubes-os.org/doc/qubes-iso-building/. You can tweak the
kernel to your liking at the appropriate step.
CooSee
thx for reply, even though i deleted this message because i thought it was to offensive :-)
> You might be interested in Qubes 4.1 which will run Fedora 29 in dom0.
> It's not ready for release yet, but you could keep an eye out or
> possibly help develop it.
i already watch the development every day and even try to install/upgrade some fc29-4.1 packages, but i was not able to upgrade to python3-3.7 on dom0 - i don't know much about fedora or yum/dnf - i never used RPM based distros before.
> See https://github.com/QubesOS/qubes-issues/issues/1919.
i hope the day will come - Debian is more user friendly - Debian SLINK was my first Linux on an i486 - IBM / Microchannel-Bus System with 1MB Graphic and 8MB RAM - good old days :-)
> Qubes can be built by following
> https://www.qubes-os.org/doc/qubes-iso-building/. You can tweak the
> kernel to your liking at the appropriate step.
i already build 4.1 with qubes-builder except KDE-Packages, because konqpopupmenutest failed or doesn't go on with second test:
1/3 Test #1: konqmimedatatest ................. Passed 0.12 sec
Start 2: konqpopupmenutest
and i also build a kernel with qubes-builder - without ISDN+WiFi and some other modules which i never need or use, but it's failing with the same behavior, please look at the screen shot - i could login even with artifacts at login screen but i can't use the desktop - the background is full of artifacts.
it's really weird, this artifacts only appear with newer kernels - the only difference is the "gpu_sched" module, which is only loaded with newer kernel.
I never noticed any kind of artifacts on FunToo with my Sapphire RX590.
--
The Installer should give the user a possibility to choose the size of SWAP, because most Systems nowadays has 32GB or even more - for those who also doesn't use Suspend.
Now i have a 10GB Swap partition which will never even touched by the System!
On FunToo i created only 1GB Swap and the System never used it, even if i recompile the whole System with emerge -e @world.
I would love to see a possibility to install or upgrade to the upcoming Qubes 4.1 - building an ISO is not possible with qubes-builder ATM.
I hope the new released Xen-4.12 is better in combination with Qubes.
thx
qubes123
On the artifacts: I have an AMD APU with an older dGPU (HD-7000 series) and since Qubes kernel 4.19.x the radeon (amdgpu) module have some issues with swiotlb buffer overflow (Xen-4.8). Newer kernel versions improved (at least xserver now can start with the APU drivers), but I still see some errors even in the latest 5.x versions. I see no artifacts under xfce, however when I switch to KDE (to achieve GPU offloading), sometimes I have horizontal lines in the background windows, looking really messy....
I plan to do more testing with fc29-dom0 and Xen-4.12 and we'll see if there's already improvement....
CooSee
i assume this has something to do with 'mesa-libwayland-egl' in combination with 'amdgpu' + xen + fc25 and newer kernels.
> https://bugzilla.redhat.com/show_bug.cgi?id=1648475
>
> Olivier Fourdan:
>*But* AFAICS, `glamor_get_pixmap_texture()` is used by Xwayland *only* with EGLStream in `xwl_glamor_eglstream_post_damage()`
>
removing the mesa-libwayland-egl package didn't work - dnf then remove Xorg+Xfce4 etc.
weird package manager - how you can deal with this - dnf is awful just awful, imo!
Qubes R4.1 KDE compilation - i'll try get at least an ISO - is this even possible atm ?
i think one of important packages is not available, therefore no 4.1 ISO for non-developers.
can you maybe provide one?
i'll even install an Alpha Version of 4.1 to get the best of my new RX590 :-)
greetings
qubes123
What I did is to build only the Qubes R4.1 dom0 components (for fc29 - using the ./setup script to configure the builder.conf) in first step. There was an issue with the pungi* packages, but I downloaded and put those to the right folder from the QR4.1 testing repo manually. Then, when the iso was already working, I built additionally the fedora-29 and debian templates for R4.1 (as rpm) as a second step.
Unfortunately I couldn't build the whonix templates, but as a workaround I installed the Qubes 4.0 templates and modified the apt sources (/etc/apt/sources.d/qubes...) pointing to the Qubes R4.1 testing vm-repository (which is available already along with the dom0 tesintg repo).
I might be wrong, but at this stage the templates for QR.4.0 and R4.1 shouldn't be too different (except the qubes update repos...).
Of course you need to configure many things manually (sys-vm-s, AppVMs, dvm-s) and HVM-s are not working (...yet -> no xen-hvm package, this could be and area of future enhancement of R4.1 --> pvhv2?). But you can use PV mode for sys-net for the time being...