Hi,
I’ve been enjoying messing around the laptop, especially as a newbie to Linux, but this sleep/suspend/whatever problem is really making me pull my hair out.
Closing the laptop lid (or otherwise sleeping it in any way) and opening it again to wake does not bring the screen back, and whenever it does manage to ““come back””, it appears in a very corrupted state, like what you would see with a failing GPU.
This “corrupted” state has (on graphical mode) a bunch of small white blocks repeating in a pattern, where the cursor is a massive block about ~2 inches square with lots of colored vertical lines squished together. In text mode (tty ctrl+alt+f{1…6}) the entire screen appears like the cursor lines, however it still accepts text input as I was able to blindly log on and run startx to see the graphical-mode corruption mentioned earlier.
(I’ll try to get a picture later)
I noticed in the dmesg log (sudo journalctl -k -b -1 -o short-precise | grep -a2 -b2 amdgpu) that after waking back up (black screen, not corruption) there is a mess of:
2xxxxx:Nov 1 xx:xx:xx.x HOSTNAME kernel: [drm:dc_dumb_srv_wait_idle [amdgpu]] *ERROR* Error waiting for DMUB idle: status=3
2xxxxx:Nov 1 xx:xx:xx.x HOSTNAME kernel: [drm:dc_dumb_srv_cmd_queue [amdgpu]] *ERROR* Error waiting for DMUB idle: status=3
2xxxxx:Nov 1 xx:xx:xx.x HOSTNAME kernel: [drm:dc_dumb_srv_cmd_queue [amdgpu]] *ERROR* Error queuing DMUB command: status=2
(This repeats either every second or every keypress, as it seems through either magic or placebo that spamming keys when waking brings me into the corrupted state more often than just a black screen)
There is also a warning early on about failing to load some firmware:
2xxxxx:Nov 1 xx:xx:xx.x HOSTNAME kernel: amdgpu 0000:c1:00.0: firmware: failed to load amdgpu/gc_11_0_1_mes_2.bin (-2)
I saw a thread somewhere here where they also mentioned this log message and downloaded the file from git.kernel.org (I got linux-firmware-20231030.tar.gz) and moved the file to an unspecified location and they didn’t see any difference; I moved the file to /lib/firmware/amdgpu (according to Firmware - Debian Wiki ) but the message still appears.
I’ve tried every type of solution, like (from memory):
- kernel
nvme.noacpi=1(read later that that might make things worse, but I saw no difference; removed anyways) - kernel
amdgpu.gpu_recovery=1(this makes the corruption effect appear much more often, so I think it’s having some effect in recovering the GPU?) - kernel
amdgpu.sg_display=0(seemingly only affects Fedora uses experiencing a different graphical wake bug) - kernel
amdgpu.ppfeaturemask=0xffffffff(saw this one on the Arch wiki, thought “why the hell not”, this has no effect) - Sideloading the
gc_11_0_1_mes_2.binfile (no effect, not even detected?) - Use “optimal defaults” BIOS settings preset (I had messed with the BIOS settings before even booting into the Debian installer)
- Two other (official Debian) kernels, full list below
- Fedora 39 live (also had black screen on wake)
Using:
- Framework Laptop 13 (batch 2)
- BIOS: 3.03 (update via lvfs)
- CPU: Ryzen 5 7640U
- RAM: 16gb Crucial DDR5-5600
- OS
-
- Main: Debian 12.2
-
-
- Kernels used: 6.1.0-9, 6.1.0-13, 6.4.0-0.deb12.2, 6.5.0-3 (via trixie)
-
-
-
- DE: KDE 5.27.5
-
-
-
- Upgraded from: Debian 12.0.0 DVD (installed via Ventoy) (just what I had at the time)
-
-
- Tested: (live) Fedora 39 beta 1.1
-
-
- DE: KDE
-
Happy to have a framework! 