On Mon, Nov 17, 2025 at 03:57:13PM +0100, Walter Alejandro Iglesias wrote: > On Mon, Nov 17, 2025 at 10:45:00AM +0000, Stuart Henderson wrote: > > On 2025/11/17 10:20, Martin Pieuchot wrote: > > > Hello Walter, > > > > > > On 17/11/25(Mon) 07:27, Walter Alejandro Iglesias wrote: > > > > On Tue, Nov 11, 2025 at 12:12:19PM +0100, Walter Alejandro Iglesias > > > > wrote: > > > > > Lately I reported this in ports@: > > > > > > > > > > https://marc.info/?l=openbsd-ports&m=176262090530440&w=2 > > > > > > > > > > I'm moving this here since I don't think it's a problem with blender > > > > > but > > > > > with Xorg and drm (not too long ago Xorg freezed on this machine while > > > > > watching a video with mpv.) > > > > > > > > > > > > > Apparently, on this machine and with OpenBSD, any program that stresses > > > > the CPU eventually causes X11 to become completely unresponsive. We are > > > > talking about a i3 from 2022 with 32GB RAM. Today, running this: > > > > > > > > $ stress -v --cpu 4 --io 4 --vm 2 --vm-bytes 128M > > > > > > Where is this command coming from? > > > > ports/sysutils/stress > > > > it might be educational to turn off some of the process types on > > the 'stress' command line and see if it still occurs with e.g. > > just vm workers, or just io workers, or whether a combination is > > needed. > > I just froze X11 agein with only this: > > $ stress -io 4 > > I tried the other options separately and they didn't even tickle the > system.
I tried the command above one more time but with a time out: $ stress --io 4 --timeout 10s After the ten seconds the system remained frozen, but after several minutes, it miraculously recovered. I was able to switch to tty0 and find the following message: Crash Annotation GraphicsCriticalError: |[0][GFX1]: Error in eglSetDamageRegion: 0x3009 (t=174.577) [GFX]: Error in eglSetDamageRegion: 0x3009 > > > > > > > I almost had to hard switch off the machine. > > > > > > > > I bought this machine recently, so I can't say if these problems were > > > > due to any changes (related to SMP support?) Curiously, on my old > > > > machine, a Core 2 Duo from 2007 with integrated graphics and just 4GB > > > > RAM, the same stress command above affects *much less* and does not > > > > freeze Xorg. > > > > > > What do you see in systat(1) and top(1)? Without more data such bug > > > reports are not helping. > > > > > > Do you see the pdaemon running in "top -S"? Do you see high CPUs %spin? > > > Does your machine use some swap during those tests? > > > > > > Please also always include a dmesg in your report. > > > > > > > > > > -- > Walter -- Walter
