Xvnc, Picom and GPU issue

freezr

freezr

Hi guys, after quarterly updates, Picom and VNC connection are giving me annoying issues, the screen becomes black.

Code: 
# System / kernel
uname -a
freebsd-version -kru
FreeBSD gpc 15.1-RELEASE-p4 FreeBSD 15.1-RELEASE-p4 releng/15.1-n283627-db6066678b33 GENERIC amd64
15.1-RELEASE-p4

# GPU hardware
pciconf -lv | grep -A3 -i vga
vgapci0@pci0:0:2:0:	class=0x030000 rev=0x00 hdr=0x00 vendor=0x8086 device=0x46d1 subvendor=0x0301 subdevice=0x02f3
    vendor     = 'Intel Corporation'
    device     = 'Alder Lake-N [UHD Graphics]'
    class      = display
    subclass   = VGA
none0@pci0:0:10:0:	class=0x118000 rev=0x01 hdr=0x00 vendor=0x8086 device=0x467d subvendor=0x0000 subdevice=0x0000
    vendor     = 'Intel Corporation'
    device     = 'Platform Monitoring Technology'


# Graphics stack packages
pkg info | grep -iE "mesa|picom|drm|libglvnd|xorg-server|xf86-video"
drm-612-kmod-6.12.85.1501000_2 Direct Rendering Manager (DRM) GPU drivers
drm-kmod-20260508              Direct Rendering Manager (DRM) GPU drivers metaport
gpu-firmware-kmod-20260519,1   Firmware modules for the drm-kmod drivers
libdrm-2.4.134,1               Direct Rendering Manager library and headers
libglvnd-1.7.0                 GL Vendor-Neutral Dispatch library
linux-rl9-dri-25.2.7           Mesa libGL runtime libraries (Rocky Linux 9.8)
linux-rl9-libdrm-2.4.128       Rocky Linux Direct Rendering Manager runtime
linux-rl9-libglvnd-1.3.4_1     GL Vendor-Neutral Dispatch library (Rocky Linux 9.6)
mesa-demos-9.0.0               OpenGL demos distributed with Mesa
mesa-dri-26.2.2                OpenGL hardware acceleration drivers for DRI2+
mesa-libs-26.2.2               OpenGL libraries that support GLX and EGL clients
picom-13                       Lightweight X11 compositor
xf86-video-scfb-0.0.7_2        X.Org syscons display driver
xf86-video-vesa-2.6.0          X.Org vesa display driver
xorg-server-21.1.25,1          X.Org X server and related programs

# OpenGL / renderer
glxinfo | grep -E "OpenGL vendor|OpenGL renderer|OpenGL version|direct rendering"
direct rendering: Yes
OpenGL vendor string: Intel
OpenGL renderer string: Mesa Intel(R) Graphics (ADL-N)
OpenGL version string: 4.6 (Compatibility Profile) Mesa 26.2.2

# Xorg
cat /usr/local/etc/X11/xorg.conf.d/20-intel.conf
Section "Device"
	Identifier "Card0"
	Driver     "modesetting"
EndSection


# Loaded GPU kernel modules
kldstat | grep -iE "drm|i915|ttm"
10    1 0xffffffff85200000   24a410 i915kms.ko
11    2 0xffffffff8500f000    9f430 drm.ko
16    1 0xffffffff850c3000     c358 ttm.ko


# GPU hang lines
 dmesg | grep -iE "drm|gpu hang|reset"
[drm] Got Intel graphics stolen memory base 0x7c800000, size 0x3c00000
drmn0: <drmn> on vgapci0
vgapci0: child drmn0 requested pci_enable_io
vgapci0: child drmn0 requested pci_enable_io
drmn0: [drm] Found ALDERLAKE_P/ADL-N (device ID 46d1) display version 13.00 stepping D0
lkpi_iic0: <LinuxKPI I2C> on drmn0
lkpi_iic1: <LinuxKPI I2C> on drmn0
lkpi_iic2: <LinuxKPI I2C> on drmn0
lkpi_iic3: <LinuxKPI I2C> on drmn0
lkpi_iic4: <LinuxKPI I2C> on drmn0
lkpi_iic5: <LinuxKPI I2C> on drmn0
lkpi_iic6: <LinuxKPI I2C> on drmn0
lkpi_iic7: <LinuxKPI I2C> on drmn0
lkpi_iic8: <LinuxKPI I2C> on drmn0
drmn0: successfully loaded firmware image 'i915/adlp_dmc.bin'
drmn0: [drm] Finished loading DMC firmware i915/adlp_dmc.bin (v2.20)
[drm] Initialized i915 1.6.0 for drmn0 on minor 0
VT: Replacing driver "efifb" with new "drmfb".
name=drmn0 id=i915drmfb flags=0x0 stride=7680
drmn0: [drm] GPU HANG: ecode 12:1:859ffffb, in Xvnc [100896]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] Xvnc[100896] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:84df9ffc, in Renderer [101247]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] Renderer[101247] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:859ffffb, in Xvnc [100896]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] Xvnc[100896] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:859ffffb, in Xvnc [100896]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] Xvnc[100896] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dfffff, in picom [100900]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] picom[100900] context reset due to GPU hang
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] Renderer[101247] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffffb, in Renderer [101247]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] Renderer[101247] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffffb, in Renderer [101247]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] MainThread[101148] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffdfb, in MainThread [101148]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] MainThread[101148] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffefb, in MainThread [101148]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] MainThread[101148] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffdfb, in MainThread [101148]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] Renderer[102274] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffffb, in Renderer [102274]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] Renderer[102274] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffffb, in Renderer [102274]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] Renderer[102274] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffffb, in Renderer [102274]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] MainThread[101030] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffdfb, in MainThread [101030]
Fence expiration time out i915-drmn0:MainThread[101030]:8!
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] MainThread[101030] context reset due to GPU hang
drmn0: [drm] GPU HANG: ecode 12:1:85dffefb, in MainThread [101030]
drmn0: [drm] GPU HANG: ecode 12:1:87cabbf2, in MainThread [101030]
drmn0: [drm] Resetting rcs0 for preemption time out
drmn0: [drm] MainThread[101030] context reset due to GPU hang
 
I've seen random crashes with drm-kmod above 61 on Intel HD 530 and UHD 630 with Xfce X11 (I use 515 on 15.1 and 61 16.0), but not sure how ADL works with lower versions.

I don't see mention of GuC firmware and might force loading it in loader.conf. preemption time out might be fixable with increasing a timeout (I'd disable it like Intel suggests but not sure on the udev part for FreeBSD)

Code: 
hw.i915kms.enable_guc="1"
hw.i915kms.request_timeout_ms="200000"
hw.i915kms.enable_hangcheck="0"
Edit: dmesg showed drm warnings about clamping timeouts to 0; apparently setting it to 0 is supported but the warnings still appear (shows the following and repeats 4 times)
Code: 
drmn0: [drm] Warning, clamping heartbeat_interval_ms to 0 to prevent overflow
drmn0: [drm] Warning, clamping preempt_timeout_ms to 0 to prevent overflow
drmn0: [drm] Warning, clamping stop_timeout_ms to 0 to prevent overflow
drmn0: [drm] Warning, clamping timeslice_duration_ms to 0 to prevent overflow
 
I am having the very same issues on my part. Same error messages as well, down to the very same ecodes as yours.
I am also using picom, though I do not know if that would have anything to do with it.
Same message as well when the error occurs: "Resetting rcs0 for preemption time out".

I am on a notebook with a Raptor Lake-P, though (i7 1360p).
Started right after I updated to the newest quarterly.
Nothing I have done managed to fix it thus far (messing with hw.i915kms values or installing upgrades).
Happens on random intervals: sometimes after 5 minutes on Xorg, sometimes after 30 minutes on Xorg.

I can trigger it more reliably when I try opening new tabs on Firefox, though I've managed quite a few times by adding new flashcards to Anki as well.
It completely freezes Xorg and I cannot get back to any interface I had open.

Very first time I have lost work in my whole 4 years of FreeBSD usage.
 
I have the one provided by the meta package, I assumed they were the most updated, I might try, not sure why do we have so many versions.
 
I "solved" my problem by downgrading back to drm-66-kmod, as suggested by bakul.
I still had it as /var/cache/pkg/drm-66-kmod-6.6.25.1501000_11.pkg, so this is what I did as root:

sh: 
pkg delete drm-612-kmod
pkg install /var/cache/pkg/drm-66-kmod-6.6.25.1501000_11.pkg
pkg lock drm-66-kmod

Don't forget to reboot afterwards.

I locked it just to avoid breakages due to updates for the time being.
I am currently under heavy work load and can't stop to debug my OS, so this will have to suffice for now.

My system has been stable for the past 48h, though. It is already way better than the 5-30 minutes complete lockdown cycle I was suffering from.
 
I can confirm the drm-612 driver issue on ThinkPad T480s running 15.1-REL-p5, GPU is 'Kaby Lake-R GT2 [UHD Graphics 620]'. As soon as the driver was updated, the system randomly hangs when loading WebGL-related pages in browsers. Before the update, the system was rock solid. Reverting to drm-66 makes the issues go away.
 
You must log in or register to reply here.
Back
Top