I installed phosh on my hammeread and after a minute or two from boot the screen freezes. I can login using ssh and looking at dmesg i see is all made of "msm fd900000.mdss: pp done time out, lm=0" messages. I suspect there is something wrong with freedreno
Edited
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information
Child items 0
Show closed items
No child items are currently assigned. Use child items to break down this issue into smaller parts.
Linked items 0
Link issues together to show that they're related.
Learn more.
It is fixed in 5.5 branch https://github.com/masneyb/linux. We need upgrade hammerhead (postmarketos-qcom-msm8974) kernel package. I am unable to do it properly, there are some python errors.
I just installed latest for hammerhead/default and the screen works about as well as it always has for me. My power key might be sticky though. My main problem is getting the screen to turn on reliably with the powerkey. I do see the 'msm fd9000000.mdss: pp done time out, lm=0' messages when I press the powerkey to turn the device back on, but the screen works fine for a while after that. (shelli UI). @MartijnBraam had experienced a screen freeze after 5 minutes or more, workaround for me is to powerkey a few times.
I tested today an installation for hammerhead with phosh (mainline kernel 5.6.0-rc6).
I see the pmOS logo but phosh never shows up, and I get a lot of msm fd900000.mdss: pp done time out, lm=0 errors in dmesg.
I think this issue is not solved yet in the mainline MDP driver.
This issue that apparently affects most MSM8974 phones is related to the async mode of the Qualcomm MDP which is currently enabled in the mainline kernel.
As @dano.kutka pointed some months ago there is a kernel fork from masneyb which temporarily fixes this problem by reverting the commits enabling async mode.
In about 1 of 5 tries, it gets stuck in console during boot, showing drm:adreno_request_fw: *ERROR* failed to load a330_pm4.fw (adreno_gpu.c). The firmware package is installed.
The first two times it booted into Phosh, it crashed shortly afterwards.
Then it did not crash anymore once Phosh was running, even if leaving it running for ten minutes or so and trying to run various applications.
Running applications:
"simple" GTK+3 applications appear to be working fine (the welcome app, the calculator, text editor)
Firefox always crashes directly after opening with an "Alignment trap" (dmesg)
The keys on the virtual keyboard are smaller (#713 (closed))
Other observations (most of it is probably known already, but writing down just in case):
Phosh can't power off the display, so it may look like it crashed when it should get powered off when it is idling
Setting the brightness slider to 0% in the UI makes it go completely dark, which is not useful (you can't see the slider anymore to reverse it). On the PinePhone, at 0% the screen is still somewhat visible.
Battery is always at 0%
CC: @masneyb, maybe there's something useful for you here.
All in all, when Phosh does boot up, I find very impressive how usable it is with a close-to mainline kernel on the Nexus 5. It would be great if the most annoying bugs could be worked out, and then we could add it to the "community" category of our devices.
Not sure if following adds a useful data point, but for what it's worth:
Just tried on Nexus 5 LG Hammerhead device.
I am only interested in ssh access to my device, so after reading the issues people were having with Plasma and Phosh, I instead selected none for UI in pmbootstrap init step.
Lack of UI did not help me ... Like Oliver and others above, I see the message drm:adreno_request_fw: *ERROR* failed to load a330_pm4.fw and then a continuous stream of msm fd900000.mdss: pp done time out, lm=0 on the screen
Do we have any patch for this issues in any postmarketos-linux branch?
I just got modem data connection working on hammerhead, but probably it broken something badly.
Instead of screen freezes also ssh stopped working, so I think it crashes completely :<
So far none, and GPU/display state on msm8974 is getting only worse with each kernel upgrade.
But some work is ongoing in a direction to fix all unfixed issues on msm8974 platform, starting with iommu, gpu, ... maybe we'll get something working better than before eventually.