On Fri, Jan 9, 2026 at 2:29 AM Crystal Kolipe <[email protected]> wrote: > > Hi, > > On Fri, Jan 09, 2026 at 05:55:10AM +0100, Simon Parrer wrote: > > i hope someone can maybe help me. I get a lot messages from > > PID 37303, this is my xorg server. I use OpenBSD Current MP#192. > > > > Attached you can see my dmgesg and my /var/log/messages > > It's would be better to post the dmesg directly to the list for easier > searching and to keep the information around in the list archives. The second > file is just the exact same error repeated 1542 times, so no need to include > that.
For what it's worth, I think I see the same warning messages since upgrade 7.8 (also on -current). I was hoping to post about this after I tested this AI summarized change, but haven't gotten around to doing so: | The warning message evergreen_packet3_check: radeon 0000:1d:00.0: | vbo resource seems too big for the bo indicates that the kernel | driver for older Radeon GPUs is rejecting a command submission | from userspace due to an invalid vertex buffer object (VBO) size.[2] | This occurs because newer Mesa graphics code is sending a command | submission with a VBO size that exceeds the allocated buffer object | (BO) size, which the kernel driver validates to prevent unauthorized | memory access.[2] Older GPUs lack hardware memory protection, so | the kernel must perform this validation. | | The warning is triggered during the validation of a Packet3 command | submission, specifically when checking the size of a resource that | may be a valid vertex buffer (SQ_TEX_VTX_VALID_BUFFER).[3] | Although the kernel driver limits the size to the buffer's actual | size and continues processing, the repeated warning can flood the | system logs, with reports indicating over 22,000 instances in a | short time.[3] | This excessive logging is considered problematic and potentially | useless to end users, as it does not provide actionable information | and is difficult to diagnose without deep knowledge of the kernel | and Mesa code [3] | | To address this, developers have suggested changing the warning to | warn_once() to prevent repeated messages, as the issue is likely | due to a userspace change in Mesa rather than a hardware or kernel | bug [2] [3] The current ratelimiting mechanism (introduced in | commit 59d76d6bc206) is insufficient to prevent log flooding.[3] | It is recommended to file a bug report with the Mesa project at | https://gitlab.freedesktop.org/mesa/mesa/-/issues/ to address the | root cause in the userspace code. --patrick [2] https://lists.freedesktop.org/archives/amd-gfx/2025-August/129685.html [3] https://lkml.org/lkml/2025/8/29/1457 > > It happens when I do nothing and the system only run. > > Did it start happening after a recent update or has OpenBSD always behaved > like this on your hardware? > > What background processes are running? > > What window manager are you using? > > Is X11 being started by xenodm or startx from the command line? > > A copy of the Xorg.log might also be useful. > > > I can live with the warning but it is a bit annoying when I search > > some in my dmesg. I would be very happy when someone can help me. > > Is graphics rendering actually working correctly or are there visible > glitches? >
