Retrieve reply timeout with MST
fm4d opened this issue · 11 comments
EXWM/XELB is crashing with this error https://gist.github.com/fm4d/a6f2c115f1db275e7a82ac2ef40398fb when I am trying to make my 2 daisy-chained external monitors work via following command.
(add-hook 'exwm-randr-screen-change-hook
(lambda ()
(start-process-shell-command
"xrandr" nil "xrandr --output eDP1 --mode 1920x1080 --output DP1-1 --right-of eDP1 --output DP1-2 --right-of DP1-1")))
Xrandr seems to be fine, runniing this command outside of EXWM creates monitor setup as expected.
Could you insert
(message "EXWM:RANDR:OUTPUT: %s" output)
just below Line 72 in exwm-randr.el ((dolist (output outputs)
) and show me the content of the *Messages*
buffer together with the output of
DISPLAY=:0 xrandr --verbose
executed in console?
EXWM:RANDR:OUTPUT: 67
https://gist.github.com/fm4d/e9f193c376800aaed780bca3df084efc
I found out that the bug is not present only with the xrandr setup above but with any setup as long as I have DP1.2 / MST enabled on my monitor and I have these two monitors daisy-chained. For example
xrandr --output eDP1 --mode 1920x1080 --output DP1-1 --right-of eDP1 --auto --output DP1-2 --off
is also timing out with the same error.
EDIT: After some fight I made it work by not calling xrandr in exwm-randr-screen-change-hook but calling exwm-randr--reset manually after setting xrandr via console ( and having workspace-output-plist set accordingly), I will try to look deeper into exwm-randr code later.
I can't find any problem from your last Gist. You first Gist indicates that a GetOutputInfo
request was failed but the inputs for that request were from a preceding GetScreenResources
request which should ensure all inputs are valid. So it's kind of weird.
Besides, which version of Emacs do you use? The RandR code can be problematic on Emacs 24.
25.0.92.1
Or could you insert the following lines at the end of your .emacs:
(add-hook 'exwm-randr-screen-change-hook
(lambda ()
(start-process-shell-command
"randr-test" nil "date >> ~/randr-test.txt; xrandr --verbose >> ~/randr-test.txt")))
and show me the content of ~/randr-test.txt?
Well, there seems to be nothing wrong again. Considered that you can do the refresh manually, what if you set that hook after EXWM has been initialized?
Actually it is totally incosistent, I am only able to make it work manually IF there is the exwm-randr-screen-change-hook that ends with xelb timeout. I actually have a feeling that xrandr could be the source of the problem, linux is just somehow not prepared for DP1.2 because I also experience weird graphical glitches and blackouts etc.
It seems like I fixed it with xf86-video-modesetting instead of xf86-video-intel so feel free to close this as it really seems to be caused by driver.
All right, I'm closing it.