[09:15] <alkisg> Hi, on bionic-hwe fully updated, i5-4440 CPU/GPU, and this minimal xorg.conf: Section "Monitor"    Identifier  "HDMI-1"    Option      "Primary" "true" EndSection
[09:15] <alkisg> Xorg ...starts but isn't displayed on screen! vt7 shows a small "bpfilter" message in text mode, and `export DISPLAY=:0; xrandr` works, and lightdm is supposed to be running, but it's not displayed
[09:15] <alkisg> Related Xorg.0.log: https://termbin.com/95di
[09:16] <alkisg> [   982.100] (EE) modeset(0): failed to set mode: Invalid argument
[09:16] <alkisg> I believe this is related to "Primary", as if I remove it, it then works fine
[09:18] <alkisg> As a workaround, I removed "primary" from my xorg.conf (the non-testing one is longer) and everything works fine, and I'm setting primary with xrandr after login
[16:31] <tomreyn> alkisg: if you set "primary" in corg.conf, maybe you need to also have a second  monitor section?
[16:31] <tomreyn> *xorg.conf
[16:32] <tomreyn> https://wiki.archlinux.org/index.php/Multihead#Configuration_using_xorg.conf
[16:48] <alkisg> tomreyn: in my initial test, I had one, and it triggered the error; then I started removing lines until I minimized the problem to the lines I mentioned above
[16:49] <alkisg> Here's my original one: https://termbin.com/qjp6
[16:49] <alkisg> So I played with removing lines, and I saw that it was the "primary" line that breaks xorg
[16:49] <alkisg> This used to work in 16.04; I don't know at which point it broke because for a couple of years I didn't use a xorg.conf
[17:50] <tomreyn> hmm, no idea then. your log shows that those options are parsed, and the parser itself seems to be happy with them.
[17:50] <tomreyn> just the modesetting is not.
[18:02] <alkisg> I do think it's a xorg regression as it was working previously... but I've no idea in which subsystem. I've had intel clients in the past that didn't show a login screen, while VNC was working (!), and updating kernel/xorg fixed them, so I'm guessing it's specific to intel or modesetting...
[18:03] <alkisg> I haven't seen that symptom in non-intel cards (where xorg "works" but isn't shown) 
[18:14] <tjaalton> I can push a new hwe xserver to a ppa tomorrow
[18:15] <tjaalton> mesa is being prepared already
[18:18] <alkisg> tjaalton: ah thank you, I'll stick around, please ping me if it's available for testing
[18:19] <tjaalton> if latest 1.20-branch doesn't fix it, then perhaps needs to be upstreamed
[18:19] <alkisg> Gotcha
[18:20] <tjaalton> and eoan has a git snapshot of it
[18:20] <alkisg> Maybe I could test with the 19.10 live cd too, by putting an init break, and creating a xorg.conf before continuing...
[18:20] <alkisg> I'll do that tomorrow morning
[18:20] <tjaalton> you can restart the session too
[18:20] <tjaalton> from a terminal
[18:20] <alkisg> Ah right
[18:20] <alkisg> Easier