Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby penguinsekai » May 28th, '20, 09:19

I've installed Mageia 7.1 on a PC with an MSI Radeon RX 5700 XT Gaming X graphics card. The Classic Installer worked fine, but selected "Framebuffer" as the video driver, instead of amdgpu (maybe because Mageia 7.1 didn't have the right firmware packages back when the ISO was released?). The installed system now boots up fine, using amdgpu as the kernel driver, and fbdev (that is, software rendering) as the X driver.

When I attempt to use the Mageia Control Center to "Set up the display server", it now correctly detects that I have an "ATI Volcanic Islands or later" card (perhaps because I've now installed updates, and have the updated firmware packages?), and selects amdgpu as the driver. But upon clicking the "Test" button, I get a black screen, with a non-flashing underscore-like symbol in the top-left of the screen. My keyboard seems dead at that point: Ctrl+Alt+F3 etc. doesn't work, and pressing the "Num Lock" or "Caps Lock" keys doesn't change the state of the respective lights. However, if I force the PC to shutdown or reset, it boots back up normally (with fbdev).

The xorg.conf file generated by the Mageia Control Center gets left over in "/etc/X11/", named "xorg.conf.test". I've tried using rescue mode to swap out my current xorg.conf file (with fbdev) for the file generated by the Mageia Control Center (with amdgpu). However, upon rebooting, the keyboard locks up as described above (can't switch to a TTY, can't enable/disable "Num Lock" or "Caps Lock"), and the display freezes. The one thing that Mageia responds to is me single-pressing the power button (it shuts down after exactly 90 seconds, i.e. the default timeout after which systemd kills unresponsive processes during a logout/shutdown). I end up having to use rescue mode (which still works fine) to restore the old fbdev-based xorg.conf that the installer created.

# Troubleshooting steps I've tried

## Installing plasma-workspace-wayland, to bypass X11

If I boot to SDDM successfully (using the fbdev-based xorg.conf) and select "Plasma (Wayland)" as the session, then upon clicking the "Login" button, the display and keyboard lock up as described above.

Note that I had previously used "Plasma (Wayland)" in Mageia 7, on a laptop with hybrid Intel/NVIDIA graphics, controlled by mageia-prime. When I had Mageia 7 in Intel mode, "Plasma (Wayland)" worked fine. Even when Mageia 7 was in NVIDIA mode (the proprietary NVIDIA driver is known to be incompatible with most Wayland compositors), I didn't get the lockups described above - I just got sent back to the login screen.

So this seems to rule out any X-related issues. Wayland shouldn't be looking at xorg.conf, or any other X-related config files.

## Switching to XDM as my display manager

With the fbdev-based xorg.conf file, both SDDM and XDM are able to start and log me in. However, with the Mageia Control Center's amdgpu-based xorg.conf file, XDM gives me almost the same symptoms as with SDDM: the display and keyboard lock up in the same way, but pressing the power button turns off the PC instantly, rather than after 90 seconds (unlike SDDM).

## Pressing "e" to edit the GRUB command line, to remove "splash quiet"

Assuming I'm using the amdgpu-based xorg.conf, Mageia locks up at the same point, with the same symptoms.

## Adding a startup delay to SDDM, as per https://www.reddit.com/r/archlinux/comm ... s_at_boot/

I added a startup delay of 30 seconds, rather than 5 seconds. When removing "splash quiet" from the kernel command line, I could see shorewall messages appearing during the delay period, and the "Num Lock" and "Caps Lock" keys were able to change the status of the lights. However, after the 30 seconds, the display and keyboard locked up, as described above.

## Starting Mageia without an xorg.conf file

Mageia locks up at the same point as with the amdgpu-based xorg.conf file, with the same symptoms.

## Starting the system initially in rescue mode, and then using "systemctl default" to resume normal booting

Assuming I'm using the amdgpu-based xorg.conf, Mageia locks up at the same point, with the same symptoms. I haven't tested what happens without an xorg.conf file.

## Starting Mageia using rEFInd, rather than GRUB

Assuming I'm using the amdgpu-based xorg.conf, Mageia locks up at the same point, with the same symptoms. I haven't tested what happens without an xorg.conf file.

## Using a different monitor

It starts up fine with the fbdev-based xorg.conf file. However, I have the same issues when trying to use the Mageia Control Center to generate an amdgpu-based xorg.conf file. i.e. The same things happen when clicking the "Test" button" as described above, and using rescue mode to swap between "xorg.conf" and xorg.conf.test, leads to the same result.

## Using Linux Mint 19.3

A freshly installed Linux Mint 19.3 system uses fbdev - this is to be expected, as Mint 19.3's installation media ships with a Linux 5.0 kernel (which lacks support for AMD Navi graphics cards). However, after updating the system using the Update Manager (which automatically pulls in a Linux 5.3 kernel), the system now works fine with amdgpu. inxi shows "amdgpu,ati" as the X drivers (rather than fbdev), and several OpenGL games (0 A.D. and MegaGlest) work fine.

# Output of "inxi -Fxxxz" on Mageia 7.1

Code: Select all
System:    Host: ivano-X570-UD-mageia.localdomain Kernel: 5.6.14-desktop-2.mga7 x86_64 bits: 64 compiler: gcc v: 8.4.0
           Desktop: KDE Plasma 5.15.4 tk: Qt 5.12.6 wm: kwin_x11 dm: SDDM Distro: Mageia 7 mga7
Machine:   Type: Desktop System: Gigabyte product: X570 UD v: N/A serial: <filter>
           Mobo: Gigabyte model: X570 UD v: x.x serial: <filter> UEFI: American Megatrends v: F11 date: 12/06/2019
Battery:   Device-1: hidpp_battery_0 model: Logitech Wireless Keyboard PID:0055 serial: N/A charge: 70% (should be ignored)
           rechargeable: yes status: Discharging
           Device-2: hidpp_battery_1 model: Logitech Wireless Mouse PID:0039 serial: N/A charge: 55% (should be ignored)
           rechargeable: yes status: Discharging
CPU:       Topology: 6-Core model: AMD Ryzen 5 3600X bits: 64 type: MT MCP arch: Zen L2 cache: 3072 KiB
           flags: lm nx pae sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3 svm bogomips: 91030
           Speed: 2196 MHz min/max: 2200/3800 MHz boost: enabled Core speeds (MHz): 1: 2195 2: 2195 3: 3948 4: 2055 5: 2192
           6: 2117 7: 4176 8: 2058 9: 2195 10: 2196 11: 2196 12: 2083
Graphics:  Device-1: Advanced Micro Devices [AMD/ATI] Navi 10 [Radeon RX 5600 OEM/5600 XT / 5700/5700 XT]
           vendor: Micro-Star MSI driver: amdgpu v: kernel bus ID: 0b:00.0 chip ID: 1002:731f
           Display: x11 server: Mageia X.org 1.20.8 driver: fbdev,v4l compositor: kwin_x11 resolution: 1920x1080~N/A
           OpenGL: renderer: llvmpipe (LLVM 8.0.0 128 bits) v: 3.3 Mesa 20.0.7 compat-v: 3.1 direct render: Yes
Audio:     Device-1: Advanced Micro Devices [AMD/ATI] Navi 10 HDMI Audio driver: snd_hda_intel v: kernel bus ID: 0b:00.1
           chip ID: 1002:ab38
           Device-2: Advanced Micro Devices [AMD] Starship/Matisse HD Audio vendor: Gigabyte driver: snd_hda_intel v: kernel
           bus ID: 0d:00.4 chip ID: 1022:1487
           Sound Server: ALSA v: k5.6.14-desktop-2.mga7
Network:   Device-1: Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet vendor: Gigabyte driver: r8169 v: kernel
           port: f000 bus ID: 04:00.0 chip ID: 10ec:8168
           IF: enp4s0 state: down mac: <filter>
           Device-2: Intel Wireless 8265 / 8275 driver: iwlwifi v: kernel port: f000 bus ID: 05:00.0 chip ID: 8086:24fd
           IF: wlp5s0 state: up mac: <filter>
Drives:    Local Storage: total: 1.13 TiB used: 58.38 GiB (5.1%)
           ID-1: /dev/nvme0n1 vendor: Corsair model: Force MP510 size: 223.57 GiB speed: 31.6 Gb/s lanes: 4 serial: <filter>
           rev: ECFM22.5 scheme: GPT
           ID-2: /dev/sda vendor: Western Digital model: WD1003FZEX-00K3CA0 size: 931.51 GiB speed: 6.0 Gb/s
           rotation: 7200 rpm serial: <filter> rev: 1A01 scheme: GPT
Partition: ID-1: / size: 58.81 GiB used: 7.53 GiB (12.8%) fs: ext4 dev: /dev/nvme0n1p2
           ID-2: swap-1 size: 8.00 GiB used: 0 KiB (0.0%) fs: swap dev: /dev/sda2
Sensors:   System Temperatures: cpu: 48.9 C mobo: N/A gpu: amdgpu temp: 36 C
           Fan Speeds (RPM): N/A gpu: amdgpu fan: 0
Info:      Processes: 286 Uptime: 2h 52m Memory: 15.65 GiB used: 2.95 GiB (18.8%) Init: systemd v: 241 runlevel: 5
           target: runlevel5.target Compilers: gcc: 8.4.0 alt: 8.4.0 Shell: bash v: 4.4.23 running in: konsole inxi: 3.0.33


# Output of "inxi -Fxxxz" on Linux Mint 19.3 (after installing updates)

Code: Select all
System:    Host: ivano-X570-UD-mint Kernel: 5.3.0-51-generic x86_64 bits: 64 compiler: gcc v: 7.5.0 Desktop: Cinnamon 4.4.8
           wm: muffin 4.4.2 dm: LightDM 1.26.0 Distro: Linux Mint 19.3 Tricia base: Ubuntu 18.04 bionic
Machine:   Type: Desktop System: Gigabyte product: X570 UD v: N/A serial: <filter>
           Mobo: Gigabyte model: X570 UD v: x.x serial: <filter> UEFI: American Megatrends v: F11 date: 12/06/2019
Battery:   Device-1: hidpp_battery_0 model: Logitech Wireless Mouse PID:0039 serial: N/A charge: 55% rechargeable: yes
           status: Discharging
           Device-2: hidpp_battery_1 model: Logitech Wireless Keyboard PID:0055 serial: N/A charge: 55% rechargeable: yes
           status: Discharging
CPU:       Topology: 6-Core model: AMD Ryzen 5 3600X bits: 64 type: MT MCP arch: Zen L2 cache: 3072 KiB
           flags: lm nx pae sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3 svm bogomips: 91030
           Speed: 2196 MHz min/max: 2200/3800 MHz boost: enabled Core speeds (MHz): 1: 2193 2: 2195 3: 2195 4: 2187 5: 2193
           6: 2195 7: 2192 8: 2191 9: 2194 10: 2195 11: 2192 12: 2196
Graphics:  Device-1: Advanced Micro Devices [AMD/ATI] vendor: Micro-Star MSI driver: amdgpu v: kernel bus ID: 0b:00.0
           chip ID: 1002:731f
           Display: x11 server: X.Org 1.20.5 driver: amdgpu,ati unloaded: fbdev,modesetting,radeon,vesa
           resolution: 1920x1080~60Hz
           OpenGL: renderer: AMD NAVI10 (DRM 3.33.0 5.3.0-51-generic LLVM 9.0.0) v: 4.5 Mesa 19.2.8 direct render: Yes
Audio:     Device-1: Advanced Micro Devices [AMD/ATI] driver: snd_hda_intel v: kernel bus ID: 0b:00.1 chip ID: 1002:ab38
           Device-2: Advanced Micro Devices [AMD] vendor: Gigabyte driver: snd_hda_intel v: kernel bus ID: 0d:00.4
           chip ID: 1022:1487
           Sound Server: ALSA v: k5.3.0-51-generic
Network:   Device-1: Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet vendor: Gigabyte driver: r8169 v: kernel
           port: f000 bus ID: 04:00.0 chip ID: 10ec:8168
           IF: enp4s0 state: down mac: <filter>
           Device-2: Intel Wireless 8265 / 8275 driver: iwlwifi v: kernel port: f000 bus ID: 05:00.0 chip ID: 8086:24fd
           IF: wlp5s0 state: up mac: <filter>
Drives:    Local Storage: total: 1.13 TiB used: 28.00 GiB (2.4%)
           ID-1: /dev/nvme0n1 vendor: Corsair model: Force MP510 size: 223.57 GiB speed: 31.6 Gb/s lanes: 4 serial: <filter>
           rev: ECFM22.5 scheme: GPT
           ID-2: /dev/sda vendor: Western Digital model: WD1003FZEX-00K3CA0 size: 931.51 GiB speed: 6.0 Gb/s
           rotation: 7200 rpm serial: <filter> rev: 1A01 scheme: GPT
Partition: ID-1: / size: 58.81 GiB used: 18.44 GiB (31.4%) fs: ext4 dev: /dev/nvme0n1p3
           ID-2: swap-1 size: 8.00 GiB used: 0 KiB (0.0%) fs: swap dev: /dev/sda2
Sensors:   System Temperatures: cpu: 40.9 C mobo: N/A
Use of uninitialized value $val2 in string eq at /usr/bin/inxi line 5759.
Use of uninitialized value $val2 in split at /usr/bin/inxi line 5764.
Use of uninitialized value $val2 in concatenation (.) or string at /usr/bin/inxi line 5766.
Use of uninitialized value $val2 in concatenation (.) or string at /usr/bin/inxi line 5767.
Use of uninitialized value $val2 in concatenation (.) or string at /usr/bin/inxi line 5768.
           Fan Speeds (RPM): N/A gpu: fan: 0
Info:      Processes: 332 Uptime: 1m Memory: 15.65 GiB used: 780.3 MiB (4.9%) Init: systemd v: 237 runlevel: 5 Compilers:
           gcc: 7.5.0 alt: 7 Shell: bash v: 4.4.20 running in: gnome-terminal inxi: 3.0.32


# Output of "lspcidrake -v" (in Mageia 7.1)

Code: Select all
xhci_pci        : Advanced Micro Devices, Inc. [AMD]|Matisse USB 3.0 Host Controller [SERIAL_USB] (vendor:1022 device:149c subv:1022 subd:1486)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse Device 24: Function 5 [BRIDGE_HOST] (vendor:1022 device:1445)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse PCIe Dummy Host Bridge [BRIDGE_HOST] (vendor:1022 device:1482)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse Internal PCIe GPP Bridge 0 to bus[E:B] [BRIDGE_PCI] (vendor:1022 device:1484)
i2c_piix4       : Advanced Micro Devices, Inc. [AMD]|FCH SMBus Controller [SERIAL_SMBUS] (vendor:1022 device:790b subv:1458 subd:5001) (rev: 61)
xhci_pci        : Advanced Micro Devices, Inc. [AMD]|Matisse USB 3.0 Host Controller [SERIAL_USB] (vendor:1022 device:149c subv:1022 subd:148c)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse Device 24: Function 7 [BRIDGE_HOST] (vendor:1022 device:1447)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse PCIe Dummy Host Bridge [BRIDGE_HOST] (vendor:1022 device:1482)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse Reserved SPP (vendor:1022 device:1485)
r8169           : Realtek Semiconductor Co., Ltd.|RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller [NETWORK_ETHERNET] (vendor:10ec device:8168 subv:1458 subd:e000) (rev: 16)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse Internal PCIe GPP Bridge 0 to bus[E:B] [BRIDGE_PCI] (vendor:1022 device:1484)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse PCIe GPP Bridge [BRIDGE_PCI] (vendor:1022 device:57a3)
unknown         : Advanced Micro Devices, Inc. [AMD]|FCH SATA Controller [AHCI mode] [STORAGE_SATA] (vendor:1022 device:7901 subv:1458 subd:b002) (rev: 51)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse Device 24: Function 0 [BRIDGE_HOST] (vendor:1022 device:1440)
iwlwifi         : Intel Corporation|Wireless 8265 / 8275 [NETWORK_OTHER] (vendor:8086 device:24fd subv:8086 subd:1010) (rev: 78)
snd_hda_intel   : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse HD Audio Controller [MULTIMEDIA_AUDIO_DEV] (vendor:1022 device:1487 subv:1458 subd:a184)
snd_hda_intel   : Advanced Micro Devices, Inc. [AMD/ATI]|Navi 10 HDMI Audio [MULTIMEDIA_AUDIO_DEV] (vendor:1002 device:ab38)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse GPP Bridge [BRIDGE_PCI] (vendor:1022 device:1483)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse Device 24: Function 2 [BRIDGE_HOST] (vendor:1022 device:1442)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse PCIe GPP Bridge [BRIDGE_PCI] (vendor:1022 device:57a4)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse PCIe Dummy Host Bridge [BRIDGE_HOST] (vendor:1022 device:1482)
unknown         : Advanced Micro Devices, Inc. [AMD]|FCH SATA Controller [AHCI mode] [STORAGE_SATA] (vendor:1022 device:7901 subv:1458 subd:b002) (rev: 51)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse Reserved SPP (vendor:1022 device:1485)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse Device 24: Function 4 [BRIDGE_HOST] (vendor:1022 device:1444)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse PCIe GPP Bridge [BRIDGE_PCI] (vendor:1022 device:57a4)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse Internal PCIe GPP Bridge 0 to bus[E:B] [BRIDGE_PCI] (vendor:1022 device:1484)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse PCIe GPP Bridge [BRIDGE_PCI] (vendor:1022 device:57a3)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse Root Complex [BRIDGE_HOST] (vendor:1022 device:1480 subv:1022 subd:1450)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse Device 24: Function 6 [BRIDGE_HOST] (vendor:1022 device:1446)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse Internal PCIe GPP Bridge 0 to bus[E:B] [BRIDGE_PCI] (vendor:1022 device:1484)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse IOMMU (vendor:1022 device:1481)
unknown         : Advanced Micro Devices, Inc. [AMD]|FCH SATA Controller [AHCI mode] [STORAGE_SATA] (vendor:1022 device:7901) (rev: 51)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse PCIe Dummy Host Bridge [BRIDGE_HOST] (vendor:1022 device:1482)
unknown         : Advanced Micro Devices, Inc. [AMD/ATI]|Navi 10 XL Downstream Port of PCI Express Switch [BRIDGE_PCI] (vendor:1002 device:1479)
unknown         : Phison Electronics Corporation|E12 NVMe Controller (vendor:1987 device:5012) (rev: 01)
ccp             : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse Cryptographic Coprocessor PSPCPP [CRYPT_OTHER] (vendor:1022 device:1486)
unknown         : Advanced Micro Devices, Inc. [AMD]|FCH LPC Bridge [BRIDGE_ISA] (vendor:1022 device:790e subv:1458 subd:5001) (rev: 51)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse PCIe GPP Bridge [BRIDGE_PCI] (vendor:1022 device:57a4)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse PCIe Dummy Host Bridge [BRIDGE_HOST] (vendor:1022 device:1482)
unknown         : Advanced Micro Devices, Inc. [AMD]|FCH SATA Controller [AHCI mode] [STORAGE_SATA] (vendor:1022 device:7901) (rev: 51)
xhci_pci        : Advanced Micro Devices, Inc. [AMD]|Matisse USB 3.0 Host Controller [SERIAL_USB] (vendor:1022 device:149c subv:1458 subd:5007)
Card:ATI Volcanic Islands and later (amdgpu/fglrx): Advanced Micro Devices, Inc. [AMD/ATI]|Navi 10 [Radeon RX 5600 OEM/5600 XT / 5700/5700 XT] [DISPLAY_VGA] (vendor:1002 device:731f subv:1462 subd:3816) (rev: c1)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse PCIe Dummy Host Bridge [BRIDGE_HOST] (vendor:1022 device:1482)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse Switch Upstream [BRIDGE_PCI] (vendor:1022 device:57ad)
unknown         : Advanced Micro Devices, Inc. [AMD]|Matisse Device 24: Function 1 [BRIDGE_HOST] (vendor:1022 device:1441)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse GPP Bridge [BRIDGE_PCI] (vendor:1022 device:1483)
unknown         : Advanced Micro Devices, Inc. [AMD/ATI]|Navi 10 XL Upstream Port of PCI Express Switch [BRIDGE_PCI] (vendor:1002 device:1478) (rev: c1)
k10temp         : Advanced Micro Devices, Inc. [AMD]|Matisse Device 24: Function 3 [BRIDGE_HOST] (vendor:1022 device:1443)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse PCIe Dummy Host Bridge [BRIDGE_HOST] (vendor:1022 device:1482)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse PCIe Dummy Function (vendor:1022 device:148a)
unknown         : Advanced Micro Devices, Inc. [AMD]|Starship/Matisse GPP Bridge [BRIDGE_PCI] (vendor:1022 device:1483)
hub             : Linux 5.6.14-desktop-2.mga7 xhci-hcd|xHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0002)
usbhid          : Logitech|USB Receiver [Human Interface Device|Boot Interface Subclass|Keyboard] (vendor:046d device:c534)
btusb           : Intel Corp.|Unknown [Wireless|Radio Frequency|Bluetooth] (vendor:8087 device:0a2b)
Mouse:evdev     : Logitech|USB Receiver [Human Interface Device|Boot Interface Subclass|Keyboard] (vendor:046d device:c517)
hub             : Linux 5.6.14-desktop-2.mga7 xhci-hcd|xHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0003)
hub             : Linux 5.6.14-desktop-2.mga7 xhci-hcd|xHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0002)
hub             : Linux 5.6.14-desktop-2.mga7 xhci-hcd|xHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0003)
hub             : Linux 5.6.14-desktop-2.mga7 xhci-hcd|xHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0002)
hub             : Linux 5.6.14-desktop-2.mga7 xhci-hcd|xHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0003)
hid_generic     : Logitech USB Receiver
hid_generic     : Logitech Wireless Mouse
hid_generic     : Logitech USB Receiver
hid_generic     : Logitech USB Receiver
hid_generic     : Logitech Wireless Keyboard PID:0055
hid_generic     : Logitech USB Receiver
hid_generic     : Logitech Wireless Mouse PID:0039


Does anyone have any suggestions?

Also, does anyone else have an AMD Navi graphics card working in Mageia 7 / 7.1?
Attachments
xorg.conf
(fbdev-based) xorg.conf file, written by Mageia 7.1's Classic Installer
(2.41 KiB) Downloaded 136 times
xorg.conf
(amdgpu-based) xorg.conf file, written by the Mageia Control Center
(11.09 KiB) Downloaded 141 times
xorg.conf
(amdgpu-based) xorg.conf file written by the Mageia Control Center, while I was testing the alternative monitor
(3.79 KiB) Downloaded 131 times
penguinsekai
 
Posts: 9
Joined: May 28th, '20, 06:25

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby doktor5000 » May 28th, '20, 17:28

penguinsekai wrote:But upon clicking the "Test" button, I get a black screen, with a non-flashing underscore-like symbol in the top-left of the screen.

Please don't use the test button. Simply change what you want to change, and then reboot.

Also, how do you check what X driver is used ?

Do you have the package radeon-firmware installed, if so what version?
Code: Select all
rpm -qa|grep -iE "kernel|radeon"|sort
Cauldron is not for the faint of heart!
Caution: Hot, bubbling magic inside. May explode or cook your kittens!
----
Disclaimer: Beware of allergic reactions in answer to unconstructive complaint-type posts
User avatar
doktor5000
 
Posts: 17629
Joined: Jun 4th, '11, 10:10
Location: Leipzig, Germany

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby penguinsekai » May 30th, '20, 16:19

penguinsekai wrote:the display and keyboard lock up as described above.

Sorry, I made a confusing statement there.

To clarify: I get the black screen and underscore, only when I click the "Test" button in the Mageia Control Center. In all the other cases, the display freezes on whatever it was displaying before the lockup. So if I start Mageia with the amdgpu-based xorg.conf file and without "splash quiet", the display will be frozen on the startup messages. On the other hand, if I start Mageia with the fbdev-based xorg.conf file and try to start a "Plasma (Wayland)" session, the display will be frozen on the login screen.

doktor5000 wrote:Please don't use the test button. Simply change what you want to change, and then reboot.

Early on, while debugging this issue, I actually did skip the "Test" button once, and just clicked "Ok" instead. It rendered my system unbootable, and I ended up reinstalling Mageia.

When I get the chance, I'll install another copy of Mageia 7.1 in another partition, but without GRUB (to avoid overwriting the "\EFI\mageia\" folder on my EFI System Partition). I'll try clicking "Ok" in the throwaway copy of Mageia, but I expect the same result.

doktor5000 wrote:Also, how do you check what X driver is used ?

I've been running the command "inxi -Fxxxz".

Just now, I generated hardinfo reports in Mageia and Linux Mint, for a second opinion. In Mageia, hardinfo specifies "llvmpipe" as the renderer (as compared to "AMD NAVI10" in Linux Mint), with an OpenGL version of 3.1 (as compared to OpenGL 4.5 in Linux Mint).

Here's the complete output of the "Display" section of the hardinfo reports:

Mageia 7.1

Code: Select all
# Display

## Session

Type   x11

## Wayland

Current Display Name   (Not Available)

## X Server

Current Display Name   :0

Vendor   Mageia

Version   (Unknown)

Release Number   12008000

## Screens

Screen 0   1920x1080 pixels

## Outputs (XRandR)

## OpenGL (GLX)

Vendor   VMware, Inc.

Renderer   llvmpipe (LLVM 8.0.0, 128 bits)

Direct Rendering   Yes

Version (Compatibility)   3.1 Mesa 20.0.7

Shading Language Version (Compatibility)   1.40

Version (Core)   3.3 (Core Profile) Mesa 20.0.7

Shading Language Version (Core)   3.30

Version (ES)   OpenGL ES 3.1 Mesa 20.0.7

Shading Language Version (ES)   OpenGL ES GLSL ES 3.10

GLX Version   1.4


Linux Mint 19.3

Code: Select all
# Display

## Display

Resolution   1920x1080 pixels

Vendor   The X.Org Foundation

Version   1.20.5

Current Display Name   :0

## Monitors

Monitor 0   1920x1080 pixels

## OpenGL

Vendor   X.Org

Renderer   AMD NAVI10 (DRM 3.33.0, 5.3.0-53-generic, LLVM 9.0.0)

Version   4.5 (Compatibility Profile) Mesa 19.2.8

Direct Rendering   Yes

## Extensions

Composite   
DAMAGE   
DOUBLE-BUFFER   
DPMS   
DRI2   
DRI3   
GLX   
Generic Event Extension   
MIT-SCREEN-SAVER   
MIT-SHM   
Present   
RANDR   
RECORD   
RENDER   
SECURITY   
SHAPE   
SYNC   
X-Resource   
XC-MISC   
XFIXES   
XFree86-DGA   
XFree86-VidModeExtension   
XINERAMA   
XInputExtension   
XKEYBOARD   
XTEST   
XVideo   
default screen number: 0   


doktor5000 wrote:Do you have the package radeon-firmware installed, if so what version?

Yes; I've got the package "radeon-firmware-20200416-1.mga7.nonfree".

Here's the complete output of "rpm -qa|grep -iE "kernel|radeon"|sort":

Code: Select all
kernel-desktop-5.6.14-2.mga7-1-1.mga7
kernel-desktop-5.6.6-1.mga7-1-1.mga7
kernel-desktop-5.6.8-1.mga7-1-1.mga7
kernel-desktop-latest-5.6.14-2.mga7
kernel-firmware-20190603-1.mga7
kernel-firmware-nonfree-20200519-1.mga7.nonfree
kernel-userspace-headers-5.6.14-2.mga7
lib64drm_radeon1-2.4.101-2.mga7
radeon-firmware-20200416-1.mga7.nonfree
penguinsekai
 
Posts: 9
Joined: May 28th, '20, 06:25

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby penguinsekai » May 31st, '20, 03:48

doktor5000 wrote:Please don't use the test button. Simply change what you want to change, and then reboot.


penguinsekai wrote:Early on, while debugging this issue, I actually did skip the "Test" button once, and just clicked "Ok" instead. It rendered my system unbootable, and I ended up reinstalling Mageia.

When I get the chance, I'll install another copy of Mageia 7.1 in another partition, but without GRUB (to avoid overwriting the "\EFI\mageia\" folder on my EFI System Partition). I'll try clicking "Ok" in the throwaway copy of Mageia, but I expect the same result.


I've got my throwaway copy of Mageia 7.1 installed now, and I can confirm that configuring the graphics card and clicking "Ok" in the Mageia Control Center, renders my system unbootable after a restart (with the usual display and keyboard lockup).

Conveniently enough, though, despite me ticking the "Do not touch the ESP or MBR" option, Mageia's Classic Installer still did install GRUB after all - just in a different folder ("\EFI\tmp\"), and without registering it in the UEFI's NVRAM. As such, I can use rEFInd to get into the GRUB for the throwaway copy of Mageia, and I can then use that GRUB to boot into recovery mode.

And yes, recovery mode works. If I use it to restore a backup of the old (fbdev-based) xorg.conf file, then I'm able to boot my system again, and repeat the process.

So, the problem (how to get Mageia to utilise my graphics card) isn't solved, but at least I now have a setup where I can rapidly test things and revert, without breaking my main installation of Mageia.

Any ideas on how to proceed from here?
penguinsekai
 
Posts: 9
Joined: May 28th, '20, 06:25

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby penguinsekai » Jun 6th, '20, 05:31

I've tried a number of additional things.

# Using a live USB of Linux Lite 5.0

Linux Lite comes with both inxi and hardinfo on its live USB, and the latest version (Linux Lite 5.0, based on Ubuntu 20.04) has a slightly newer hardware enablement stack compared to Linux Mint 19.3, so I was interested in checking it out.

From the live USB, I ran the command "inxi -Fxxxz", and here is the output:

Code: Select all
System:    Kernel: 5.4.0-33-generic x86_64 bits: 64 compiler: gcc v: 9.3.0 Desktop: Xfce 4.14.2 tk: Gtk 3.24.13
           info: xfce4-panel wm: xfwm4 dm: LightDM 1.30.0 Distro: Linux Lite 5.0 LTS base: Ubuntu 20.04 LTS (Focal Fossa)
Machine:   Type: Desktop System: Gigabyte product: X570 UD v: N/A serial: <filter>
           Mobo: Gigabyte model: X570 UD v: x.x serial: <filter> UEFI: American Megatrends v: F11 date: 12/06/2019
Battery:   Device-1: hidpp_battery_0 model: Logitech Wireless Mouse PID:0039 serial: N/A charge: 55% (should be ignored)
           rechargeable: yes status: Discharging
           Device-2: hidpp_battery_1 model: Logitech Wireless Keyboard PID:0055 serial: N/A charge: 70% (should be ignored)
           rechargeable: yes status: Discharging
CPU:       Topology: 6-Core model: AMD Ryzen 5 3600X bits: 64 type: MT MCP arch: Zen L2 cache: 3072 KiB
           flags: avx avx2 lm nx pae sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3 svm bogomips: 91023
           Speed: 2195 MHz min/max: 2200/3800 MHz boost: enabled Core speeds (MHz): 1: 2196 2: 2194 3: 2190 4: 2192 5: 2195
           6: 2194 7: 2195 8: 2193 9: 2191 10: 2196 11: 2187 12: 2185
Graphics:  Device-1: Advanced Micro Devices [AMD/ATI] Navi 10 [Radeon RX 5600 OEM/5600 XT / 5700/5700 XT]
           vendor: Micro-Star MSI driver: amdgpu v: kernel bus ID: 0b:00.0 chip ID: 1002:731f
           Display: x11 server: X.Org 1.20.8 driver: amdgpu,ati unloaded: fbdev,modesetting,radeon,vesa
           resolution: 1920x1080~60Hz
           OpenGL: renderer: AMD Radeon RX 5700 XT (NAVI10 DRM 3.35.0 5.4.0-33-generic LLVM 9.0.1) v: 4.6 Mesa 20.0.4
           direct render: Yes
Audio:     Device-1: Advanced Micro Devices [AMD/ATI] Navi 10 HDMI Audio driver: snd_hda_intel v: kernel bus ID: 0b:00.1
           chip ID: 1002:ab38
           Device-2: Advanced Micro Devices [AMD] Starship/Matisse HD Audio vendor: Gigabyte driver: snd_hda_intel v: kernel
           bus ID: 0d:00.4 chip ID: 1022:1487
           Sound Server: ALSA v: k5.4.0-33-generic
Network:   Device-1: Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet vendor: Gigabyte driver: r8169 v: kernel
           port: f000 bus ID: 04:00.0 chip ID: 10ec:8168
           IF: enp4s0 state: down mac: <filter>
           Device-2: Intel Wireless 8265 / 8275 driver: iwlwifi v: kernel port: f000 bus ID: 05:00.0 chip ID: 8086:24fd
           IF: wlp5s0 state: down mac: <filter>
Drives:    Local Storage: total: 1.15 TiB used: 7.20 GiB (0.6%)
           ID-1: /dev/nvme0n1 vendor: Corsair model: Force MP510 size: 223.57 GiB speed: 31.6 Gb/s lanes: 4 serial: <filter>
           rev: ECFM22.5 scheme: GPT
           ID-2: /dev/sda vendor: Western Digital model: WD1003FZEX-00K3CA0 size: 931.51 GiB speed: 6.0 Gb/s
           rotation: 7200 rpm serial: <filter> rev: 1A01 temp: 29 C scheme: GPT
           ID-3: /dev/sdb type: USB model: General USB Flash Disk size: 7.47 GiB serial: <filter> rev: 1100 scheme: MBR
           ID-4: /dev/sdc type: USB vendor: SanDisk model: Cruzer Facet size: 14.56 GiB serial: <filter> rev: 1.27 scheme: MBR
Partition: ID-1: / size: 7.83 GiB used: 88.7 MiB (1.1%) fs: overlay source: ERR-102
           ID-2: /var/log size: 6.00 GiB used: 28.3 MiB (0.5%) fs: ext4 dev: /dev/sdb3
           ID-3: swap-1 size: 8.00 GiB used: 0 KiB (0.0%) fs: swap dev: /dev/sda2
Sensors:   System Temperatures: cpu: 45.4 C mobo: 16.8 C gpu: amdgpu temp: 36 C
           Fan Speeds (RPM): N/A gpu: amdgpu fan: 0
Info:      Processes: 311 Uptime: 26m Memory: 15.65 GiB used: 934.1 MiB (5.8%) Init: systemd v: 245 runlevel: 5 Compilers:
           gcc: 9.3.0 alt: 9 Shell: bash v: 5.0.16 running in: xfce4-terminal inxi: 3.0.38


I also generated a hardinfo report. Here's the complete output of the "Display" section:

Code: Select all
# Display

## Display

Resolution   1920x1080 pixels

Vendor   The X.Org Foundation

Version   1.20.8

Current Display Name   :0.0

## Monitors

Monitor 0   1920x1080 pixels

## OpenGL

Vendor   X.Org

Renderer   AMD Radeon RX 5700 XT (NAVI10, DRM 3.35.0, 5.4.0-33-generic, LLVM 9.0.1)

Version   4.6 (Compatibility Profile) Mesa 20.0.4

Direct Rendering   Yes

## Extensions

Composite   
DAMAGE   
DOUBLE-BUFFER   
DPMS   
DRI2   
DRI3   
GLX   
Generic Event Extension   
MIT-SCREEN-SAVER   
MIT-SHM   
Present   
RANDR   
RECORD   
RENDER   
SECURITY   
SHAPE   
SYNC   
X-Resource   
XC-MISC   
XFIXES   
XFree86-DGA   
XFree86-VidModeExtension   
XINERAMA   
XInputExtension   
XKEYBOARD   
XTEST   
XVideo   
default screen number: 0


As you can see, the graphics card appears to be working ("amdgpu,ati" are selected as the X drivers, and OpenGL is operating at version 4.6).

It seems to me that this rules out any issues with Mesa, or any upstream issues with X11, as Mageia 7.1 is using the same version of X11 (1.20.8), and (according to one of my inxi dumps I've got saved in a text file) used to use the same version of Mesa (20.0.4), as Linux Lite 5.0. (Mageia 7.1 has since upgraded to Mesa 20.0.7)

# Testing older kernels, in my throwaway Mageia installation

I noticed that both Linux Mint and Linux Lite have older kernels, compared to Mageia (Mint 19.3 currently has Linux 5.3, Linux Lite 5.0 has Linux 5.4, Mageia 7.1 currently has Linux 5.6).

So, in my throwaway copy of Mageia, I installed a number of older kernels that are still available in the repository:

  • kernel-desktop-5.3.13-2.mga7-1-1.mga7.x86_64
  • kernel-desktop-5.4.17-1.mga7-1-1.mga7.x86_64
  • kernel-desktop-5.5.15-3.mga7-1-1.mga7.x86_64
I then swapped out the fbdev-based xorg.conf file, for the amdgpu-based xorg.conf file, and tried to boot the system using each of those kernels (by selecting the appropriate kernel via GRUB's "Advanced options").

Unfortunately, I just kept having the same issue: the display and keyboard would lock up at a certain stage of the boot process.

This is interesting, as it seems to me that this rules out any upstream kernel regressions.

(Just to be sure that the older kernels still work in Mageia 7.1, I swapped back to the fbdev-based xorg.conf file, and booted into kernel-desktop-5.3.13-2.mga7-1-1.mga7.x86_64. It worked fine.)

# Testing the "linus" (i.e. unpatched) kernels, in my throwaway Mageia installation

In the research I've done, I'm not seeing many current reports online, of AMD Navi graphics support being broken in Linux. On the contrary, I've seen a number of AMD Navi benchmarks carried out on Linux 5.3+ kernels, as well as gamers trying to improve their Linux gaming framerates on AMD Navi.

However, I am yet to see any reports of a Mageia user, getting an AMD Navi graphics card to work. The only report I could find of a Mageia user with an AMD Navi graphics card, is this one, with an AMD Radeon RX 5500 XT: viewtopic.php?f=23&t=13335

korrel wrote:But it crashed my Mageia 7 install. Already tried different video drivers but only the VESA driver works.

That report is also unsolved to this day.

So I started wondering if it could be some Mageia-specific problem, such as an issue with a Mageia-specific kernel patch.

Mageia 7.1 has 3 types of kernels available in its repositories:

  • "desktop" kernels (the default), patched by Mageia, and compiled with optimisations for desktop performance.
  • "linus" kernels, which do not have any patches from Mageia - they are purely based on the upstream source code. Nonetheless, the "linus" kernels are still compiled with the same optimisations for desktop performance as the "desktop" kernels - but I'm pretty sure these optimisations ("1000 Hz, voluntary preempt, CFS cpu scheduler and clq i/o scheduler") are common and well tested across many desktop Linux distributions.
  • "server" kernels, patched by Mageia, and compiled with optimisations for server performance.
To test whether any Mageia-specific patches could be breaking AMD Navi support, I installed the following "linus" kernels:

  • kernel-linus-5.3.11-1.mga7-1-1.mga7.x86_64
  • kernel-linus-5.4.20-1.mga7-1-1.mga7.x86_64
  • kernel-linus-5.5.15-1.mga7-1-1.mga7.x86_64
I tried to install a Linux 5.6 series "linus" kernel as well. However, I gave up, as it was giving me a dependency conflict, claiming that the package "radeon-firmware" would have to be removed.

I swapped the fbdev-based xorg.conf file, for the amdgpu-based xorg.conf file, and attempted to boot into each of the 3 "linus" kernels that I'd installed (by selecting the appropriate kernel via GRUB's "Advanced options"). But I had the same issue as usual, with the display and keyboard locking up at the same point in the boot process.

(Again, just to be sure that the "linus" kernels actually work, I swapped back to the fbdev-based xorg.conf file, and booted into kernel-linus-5.5.15-1.mga7-1-1.mga7.x86_64. It worked fine.)

# Removing "vga=791" from the GRUB kernel command line

By default, at least when installing Mageia using the Classic Installer, Mageia inserts the parameter "vga=791" into GRUB's kernel command line. According to several reports online, this should interfere with Kernel Mode Setting, which is apparently required for the amdgpu driver:

So, I re-tested the following kernels with "vga=791" removed from the command line:

  • kernel-desktop-5.6.14-2.mga7-1-1.mga7.x86_64
  • kernel-linus-5.5.15-1.mga7-1-1.mga7.x86_64
  • kernel-desktop-5.3.13-2.mga7-1-1.mga7.x86_64
But the lockup occurred at the same point as usual.

Does anyone have any further suggestions?

Also, has anyone actually got an AMD Navi graphics card working in Mageia?
penguinsekai
 
Posts: 9
Joined: May 28th, '20, 06:25

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby martinw » Jun 6th, '20, 11:27

If you repeat your test, then reboot in text mode, you can save the system log from the failed boot by running as root:
Code: Select all
journalctl -b -1 > failed-boot.log

and save the X server log files (Xorg.*.log) from /var/log. See if there are any useful error messages in those log files.

One other thing to try is to delete the /etc/X11/xorg.conf file and reboot. Then X will automatically configure itself. That sometimes works better than telling it which driver to use, even if it picks the same driver.

Another option is to edit the xorg.conf file and change the driver to "modesetting" instead of "amdgpu".
martinw
 
Posts: 609
Joined: May 14th, '11, 10:59

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby penguinsekai » Jun 10th, '20, 16:21

martinw wrote:One other thing to try is to delete the /etc/X11/xorg.conf file and reboot.

I've already tried that. It didn't help.

martinw wrote:Another option is to edit the xorg.conf file and change the driver to "modesetting" instead of "amdgpu".

Interesting idea. Unfortunately, when I tested it, the display and keyboard locked up, just the same as usual.

martinw wrote:If you repeat your test, then reboot in text mode, you can save the system log from the failed boot by running as root:
Code: Select all
journalctl -b -1 > failed-boot.log

Thank you for that suggestion. I think I may have found the issue, thanks to that.

Just to ensure that I was testing a clean system (I'd been mucking around with regenerating the initrd before), I reinstalled my throwaway copy of Mageia 7.1. But, for a change, I decided to install GNOME as the desktop environment, rather than KDE - the reason being, that if you pick GNOME as your desktop environment, Mageia will install GNOME Display Manager, rather than Simple Desktop Display Manager. GDM uses Wayland by default, rather than X11 (whereas SDDM still uses X11, even if you're running Plasma Wayland). Hence, with GDM as my display manager, I can ensure that X11 doesn't start up at all, thereby totally ruling out any issues with Mageia's custom X11 configuration.

Sure enough, GDM locked up my screen and keyboard, just like running X11 without the fbdev-based xorg.conf file. Unlike X11, though, GDM was at least nice enough to give up after a while (~45 seconds), leaving me looking at the startup messages on TTY1, and with the ability to switch to a different TTY via the keyboard. So I didn't need to use recovery mode this time.

Anyway, I ran the command "journalctl -b -0 > <FILENAME>", and inspected the contents. I don't want to upload the entire file here (it contains sensitive details, like my MAC addresses), so I've instead piped it through grep, and attached the files below.

I'm not sure what to make of the output of 'grep "amdgpu"', but I found the following line from 'grep "radeon"' far more interesting:

Code: Select all
Jun 10 15:21:10 localhost.localdomain org.gnome.Shell.desktop[2069]: radeonsi: Navi family support requires LLVM 9 or higher

Looking into the LLVM matter, I noticed the following lines in the hardinfo reports from the various distros (and the "inxi -Fxxxz" output is similar):

Linux Mint 19.3 (which works)

Code: Select all
Renderer   AMD NAVI10 (DRM 3.33.0, 5.3.0-53-generic, LLVM 9.0.0)


Linux Lite 5.0 (which works)

Code: Select all
Renderer   AMD Radeon RX 5700 XT (NAVI10, DRM 3.35.0, 5.4.0-33-generic, LLVM 9.0.1)


Mageia 7.1 KDE (with the fbdev-based xorg.conf file)

Code: Select all
Renderer   llvmpipe (LLVM 8.0.0, 128 bits)


So, Mageia is using an outdated version of LLVM (8), whereas Linux Mint and Linux Lite are running a newer version (9).

And a quick DuckDuckGo search yielded the following articles, confirming that LLVM 9 or higher is required, to get amdgpu to work on Navi-series graphics cards:


So, it looks like I'll need to submit a bug report about this.

Thank you martinw, for the suggestion to check the "journalctl" output. And thank you doktor5000, for your earlier efforts to help me. (Is the "Test" button in MCC known to be broken?)

I'll submit a bug report, and once I confirm the graphics card to be working with an LLVM 9 backport, I'll mark this thread as solved.
Attachments
journalctl_GDM_Boot without splash quiet_grep_radeon.log
Output of 'journalctl -b -0', piped into 'grep "radeon"'
(125 Bytes) Downloaded 130 times
journalctl_GDM_Boot without splash quiet_grep_amdgpu.log
Output of 'journalctl -b -0', piped into 'grep "amdgpu"'
(3.47 KiB) Downloaded 123 times
penguinsekai
 
Posts: 9
Joined: May 28th, '20, 06:25

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby doktor5000 » Jun 10th, '20, 23:21

penguinsekai wrote:And thank you doktor5000, for your earlier efforts to help me. (Is the "Test" button in MCC known to be broken?)

You're welcome.
As far as I'm aware, either it does not work because the graphic card is already blocked by the loaded driver.
Or it might leave behind an ~/.Xauthority file which is owned by root and blocks the next X start ( https://bugs.mageia.org/show_bug.cgi?id=20835 )
Although it's been ages since I used it myself, but from most reports I've seen there are still issues with it.
Cauldron is not for the faint of heart!
Caution: Hot, bubbling magic inside. May explode or cook your kittens!
----
Disclaimer: Beware of allergic reactions in answer to unconstructive complaint-type posts
User avatar
doktor5000
 
Posts: 17629
Joined: Jun 4th, '11, 10:10
Location: Leipzig, Germany

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby penguinsekai » Jun 17th, '20, 10:50

I just noticed that I'd forgotten to post a link here to the bug report. You can find the bug report here:

Bug 26765 - amdgpu doesn't work on AMD Navi graphics cards (presumably due to missing LLVM 9)

Again, once I confirm the graphics card to be working with an LLVM 9+ backport, I'll mark this thread as solved.
penguinsekai
 
Posts: 9
Joined: May 28th, '20, 06:25

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby johnpenguin » Jun 20th, '20, 06:33

If I got it right, your installation failed because Mageia 7.1 doesn't come with the latest driver (kernel) necessary to support the card. In that case, using the net-install will pull the latest versions of the selected packages available.
johnpenguin
 
Posts: 37
Joined: Feb 28th, '19, 02:15

Re: Cannot boot Mageia 7.1 with amdgpu (Radeon RX 5700 XT)

Postby penguinsekai » Jun 28th, '20, 23:27

I was bored, so I upgraded my throwaway copy of Mageia 7.1, to Mageia Cauldron. (Note to inexperienced users reading this thread: Cauldron is an unstable development release, that may cause you to lose data. Don't use Cauldron, unless you really know what you're doing.)

Cauldron is currently using LLVM 10.0.0 (Navi-series AMD GPUs apparently require LLVM 9 or newer, whereas Mageia 7.1 currently uses LLVM 8.0.0).

After upgrading to Cauldron, I used the Mageia Control Center's "Set up the display server" module, set my graphics card to "ATI Volcanic Islands or later", and rebooted (just like during my earlier efforts, when I was still running Mageia 7.1).

It works.

KDE's visual effects (transparency et. al.) are working now (they were automatically disabled under fbdev), and working fluidly. inxi and hardinfo both report that I'm using amdgpu as my X11 driver. I can run 0 A.D. (an OpenGL game) at its maximum graphics settings, with a good framerate (55+ during the early game). And the "Plasma (Wayland)" session (which ignores the xorg.conf file) is working now.

Of course, I don't want to use an unstable development release of Mageia as my daily operating system, so I'm still hoping for a fix to be backported to Mageia 7.1.

johnpenguin wrote:If I got it right, your installation failed because Mageia 7.1 doesn't come with the latest driver (kernel) necessary to support the card. In that case, using the net-install will pull the latest versions of the selected packages available.

I don't think my installations of Mageia 7.1 were failing - I was getting the lockups after the installation and updates, when I tried to switch to amdgpu using the Mageia Control Center.

Keep in mind that I've always been using the Classic Installer to install Mageia 7.1. The Classic Installer seems to use the fbdev driver by default, and can apply all updates prior to rebooting into the installed system. So, as far as I can see, using the Classic Installer is no different from using the net installer (provided you enable online updates during the installation process). It mostly just saves bandwidth, when it comes to installing packages that haven't been updated since the distro's release (as they can be pulled from the installation medium).
penguinsekai
 
Posts: 9
Joined: May 28th, '20, 06:25


Return to Video

Who is online

Users browsing this forum: No registered users and 1 guest

cron