Created attachment 144962 [details]
journal -b -k, the HDMI monitor works after switched to mirror mode

Have you done any changes recently and after that you see this issue? There are no failures in the logs indicating kernel issue.

I didn't change anything, just install the mainline kernel, and toggling the display.
Without drm.debug=0xe, the external monitor always shows blank while in mirror mode.
Is there anything I could try to gather more useful logs? Or any debug patch I can apply to collect meaningful log for you?

(In reply to AceLan Kao from comment #3)
> I didn't change anything, just install the mainline kernel, and toggling the
> display.
> Without drm.debug=0xe, the external monitor always shows blank while in
> mirror mode.
> Is there anything I could try to gather more useful logs? Or any debug patch
> I can apply to collect meaningful log for you?

@Jani, any suggestion here?

Please attach 'xrandr --verbose' output while in mirror mode for starters.

Created attachment 145053 [details]
xrandr --verbose while failed to switch to mirror mode

This is the xrandr --verbose output when fails to switch mirror mode

Created attachment 145055 [details]
xrandr --verbose when switching to mirror mode works

This xrandr --verbose output is the working one that mirror mode works.

So the external display is listed as DP-5 and primary when it works, and DP-2-3 with eDP-1 as primary when it fails.

By toggling the mirror mode, do you mean using a settings dialog, and enabling/disabling mirror mode there?

And, in the same session, you'd sometimes have DP-5 and sometimes DP-2-3?

Can you reproduce this using xrandr on the command line? For example, if you repeat:

$ xrandr --output DP-5 --off --output DP-2-3 --off --output eDP-1 --off; sleep 1; xrandr --auto; xrandr --verbose

would you sometimes get DP-5 in the output, and sometimes DP2-3?

Created attachment 145124 [details]
xrandr --output DP-6 --off --output eDP-1 --off ; sleep 1 ; xrandr --auto; xrandr --verbose

> By toggling the mirror mode, do you mean using a settings dialog, and enabling/disabling mirror mode there?
I use meta key + p to select mirror mode

> And, in the same session, you'd sometimes have DP-5 and sometimes DP-2-3?
No, I have to try it some times to reproduce the issue, so I should have reboot the machines to collect the failed and successful log.

>Can you reproduce this using xrandr on the command line? For example, if you >repeat:

>$ xrandr --output DP-5 --off --output DP-2-3 --off --output eDP-1 --off; sleep >1; xrandr --auto; xrandr --verbose
Sure, here is the failed log, the external monitor turns off after ran the command.

> would you sometimes get DP-5 in the output, and sometimes DP2-3?
I might have switched the cable and reboot the machine to collect good and bad log. But this time I use the same cable and reboot the machine, I got DP-6 and then DP-5

(In reply to AceLan Kao from comment #9)
> >$ xrandr --output DP-5 --off --output DP-2-3 --off --output eDP-1 --off; sleep 1; xrandr --auto; xrandr --verbose
> Sure, here is the failed log, the external monitor turns off after ran the
> command.

Based on that, I think the mirror mode is a red herring here. I think it's just a sporadic failure to enable the external display.

Created attachment 145127 [details]
test script to test&log displays

Here's a small test script to disable and enable displays, log stuff, and ask you whether it succeeded or not.

It uses sudo to log stuff in dmesg using /dev/kmsg. Please check the script yourself before running random stuff from the internets. ;)

It should stop after you've had one success and one fail.

It will write files xrandr.log and dmesg.log into the current directory. Please attach those to the bug after you're done.

Created attachment 145155 [details]
xrandr test xrandr.log

Here is the logs from your script.
It failed 15 times in a raw and success at 16th time.

Created attachment 145156 [details]
xrandr test dmesg.log

(In reply to AceLan Kao from comment #13)
> Created attachment 145156 [details]
> xrandr test dmesg.log

229.236316] modeset 1 start
[  244.097370] modeset 1 result: fails
[  244.187105] modeset 2 start
[  249.417777] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe B FIFO underrun
[  259.287420] modeset 2 result: fails
[  259.374885] modeset 3 start
[  265.603457] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe B FIFO underrun
[  280.548303] modeset 3 result: fails
[  280.635707] modeset 4 start
[  294.783926] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe B FIFO underrun
[  315.220947] modeset 4 result: fails
[  315.308298] modeset 5 start
[  330.976321] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe B FIFO underrun
[  356.236186] modeset 5 result: fails
[  356.323591] modeset 6 start
[  373.220764] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe B FIFO underrun
[  400.658349] modeset 6 result: fails
[  400.745707] modeset 7 start
[  426.821406] modeset 7 result: fails
[  426.910324] modeset 8 start
[  433.805585] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe B FIFO underrun
[  465.152740] modeset 8 result: fails
[  465.241756] modeset 9 start
[  469.096026] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe B FIFO underrun
[  497.074056] modeset 9 result: fails
[  497.161822] modeset 10 start
[  506.738423] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe B FIFO underrun
[  553.592075] modeset 10 result: fails
[  553.679171] modeset 11 start
[  561.929562] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe B FIFO underrun

Setting the priority considering the reproduction rate after adding drm.debug=0xe parameter.

-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/drm/intel/issues/356.