Hide Forgot
Created attachment 1506506 [details] Xorg log of failed session Description of problem: The xorg log is filled with dozens of: [ 28.650] (EE) FBDEV(1): FBIOPUTCMAP: Invalid argument Which looks to be similar to another bug traced back to plymouth. In any case Xorg fails to start. 7.5 had no issues. Downgrade to 7.5 packages (under Additional info) restores working condition Version-Release number of selected component (if applicable): xorg-x11-drv-fbdev-0.5.0-1.el7 xorg-x11-drv-nouveau-1.0.15-1.el7 xorg-x11-server-Xorg-1.20.1-5.1.el7 xorg-x11-drv-vesa-2.4.0-1.el7 plymouth-0.8.9-0.31.20140113.el7 How reproducible: 100% on systems with an nVidia video card. 0% on systems with an intel card. Steps to Reproduce: 1.Update to 7.6 on a system with an nVidia card 2.reboot 3.X fails to start Actual results: X fails to load GDM and hangs "waiting for plymouth" systemd service (EE) FBDEV(1): FBIOPUTCMAP: Invalid argument Expected results: System boots to X as before. Additional info: Possibly related https://bugzilla.redhat.com/show_bug.cgi?id=871054 If I run this command: yum downgrade xorg-x11-drv-nouveau xorg-x11-server-Xorg xorg-x11-drv-qxl.x86_64 xorg-x11-drv-vesa xorg-x11-drv-dummy xorg-x11-drv-v4l xorg-x11-drv-ati xorg-x11-drv-vmware xorg-x11-drv-fbdev xorg-x11-drv-intel.x86_64 and reboot X starts back up again as expected. lshw -numeric -C display *-display description: 3D controller product: G98 [Quadro NVS 450] [10DE:6FA] vendor: NVIDIA Corporation [10DE] physical id: 0 bus info: pci@0000:03:00.0 version: a1 width: 64 bits clock: 33MHz capabilities: pm msi pciexpress bus_master cap_list rom configuration: driver=nouveau latency=0 resources: irq:29 memory:f6000000-f6ffffff memory:ec000000-efffffff memory:f4000000-f5ffffff ioport:e000(size=128) memory:f7000000-f701ffff *-display description: VGA compatible controller product: G98 [Quadro NVS 450] [10DE:6FA] vendor: NVIDIA Corporation [10DE] physical id: 0 bus info: pci@0000:04:00.0 version: a1 width: 64 bits clock: 33MHz capabilities: pm msi pciexpress vga_controller bus_master cap_list rom configuration: driver=nouveau latency=0 resources: irq:30 memory:f2000000-f2ffffff memory:e8000000-ebffffff memory:f0000000-f1ffffff ioport:d000(size=128) memory:f3000000-f301ffff
Created attachment 1506507 [details] Xorg log of successful session after yum downgrade
See also https://bugzilla.redhat.com/show_bug.cgi?id=1456010
In general there seems to be a lot of framebuffer bits on the fedora plymouth changelog.
Created attachment 1507316 [details] anaconda log-capture
This bug is not present in Fedora 29.
The bug remains with the elrepo kernels: kernel-ml-4.19.2-1.el7.elrepo.x86_64 kernel-lt-4.4.163-1.el7.elrepo.x86_64
Bug resolved by rebuilding xorg-x11-server-1.20.3-1.fc29 [1] for 7.6. With this in mind it appears to be an Xorg server bug rather than fbdev. [1] https://koji.fedoraproject.org/koji/buildinfo?buildID=1158089 Deps: https://koji.fedoraproject.org/koji/buildinfo?buildID=1140367 https://koji.fedoraproject.org/koji/buildinfo?buildID=1140369
I did a bunch of scratch builds and isolated the issue down to: 0001-linux-Make-platform-device-probe-less-fragile.patch xorg built without that patch works as expected.
Upstream commit https://gitlab.freedesktop.org/xorg/xserver/commit/0816e8fca6194dfb4cc94c3a7fcb2c7f2a921386 Seems to have the same goals. A test build with that patch instead of the RHEL7.6 version works as expected.
Note: this is seen on the Lenovo P50 corporate laptops running RHEL-7.6 CSB release.
I only have access to P50 notebooks (my and in lab) with GM107GLM [Quadro M1000M] [10DE:13B1] graphics. I tried with a fresh installation of RHEL 7.6 (xorg-x11-server-Xorg-1.20.1-3.el7.x86_64) and RHEL 7 CSB (xorg-x11-server-Xorg-1.20.1-6.jx1.el7.x86_64), but I was unable to reproduce the described problem - docked/undocked, with/without external screen - X started every time without problem. So I am unable to confirm whether the fix solved the original issue.