Hello, I’ve started experiencing Xid 62 errors in Ubuntu 22.04 LTS with an RTX 2060. The machine itself is a 2 way supermicro server. The way to reproduce is to induce CPU load on the server. For example, since the server can run 24 threads, start a process with 22 threads leaving 2 threads free …

FYI, I have tried debugging this issue in many ways, all of them unfruitful. I can find no driver that works, and so the box is unusable as is because it’s unstable. In the last crash with the 580 driver, the screen managed to work for a bit while the machine was trying to shutdown cleanly for a r…

You mention seeing, “CTX SWITCH TIMEOUT”, errors with older drivers and Xid 62 is PMU_HALT_ERROR, Internal micro-controller halt (newer drivers). Perhaps you can run the driver process at a higher priority, as it looks like the GPU is unable to be serviced in a timely fashion. This article may of…

If the driver is written in such a way that it has to run at high priority because otherwise the hardware hangs, then it’s the driver’s responsibility to do that. Surely such drivers are not written assuming the system they run on is mostly idle.

With regards to nomodeset, it appears the net effect of adding the nomodeset kernel parameter was to disable the nouveau drivers. If that was not done, then there would be bad side effects as described. It is noted that the nouveau and the nvidia drivers are incompatible with each other, so this m…

FYI since I switched to the nouveau drivers, I’ve noticed mounting circumstantial evidence suggesting the nvidia drivers are implicated (or directly cause) massive kernel memory leaks. The setup is you start Ubuntu 22 LTS, open some tabs in Firefox with e.g. a twitch stream and some oneliner chat, …

Sounds similar indeed. I did not see any other error message before Xid 62. The occurrence is sudden and without any obvious warning. What does slabtop say, do you see an apparent accumulation of kmalloc-rnd objects? In the setup here, only one of the kmalloc-rnd slab allocation categories seeme…

I checked slabinfo, there is no “rnd” anywhere in the file.

Perhaps in Arch Linux with kernel 6.18 slabinfo looks different from Ubuntu 22 LTS with kernel 6.8.0, I do not know. Still I would look at slabinfo and see if any allocation category behaves as if there was an actual memory leak.

Could be because Arch doesn’t enable CONFIG_RANDOM_KMALLOC_CACHES, and probably hasn’t enabled it for some time.

Ubuntu 22, RTX 2060 --- Xid 62 under heavy CPU load

Graphics / Linux Linux

kode54 November 29, 2025, 10:47pm 7

This may be a similar issue, happening with a different system configuration? I have a 6.18 kernel with a patch series that should alleviate CPU usage for the sched-ext schedulers when under total saturation. So maybe this particular hang was GPU related instead of scheduler related:

Topic		Replies	Views
Random Xid 62 error on ML workloads - Titan RTX Linux	0	763	July 8, 2020
Ubuntu 20.04 randomly freezes screen with Xid 62 last message in syslog Linux	2	1406	May 26, 2020
Xid 61 (black screen on startup) Ubuntu 18.04 GTX 1060 mobile Linux	12	3556	August 11, 2020
Hung/frozen machine with X370 board, GTX 1060 card, Ryzen 5 CPU - Xid 32 & 69 - all driver versions Linux	13	1797	December 29, 2018
410.66 crash and system freeze under heavy load (Xid 8, Xid 38) Linux	13	2166	November 15, 2018
Random Xid 61 and Xorg lock-up Linux	406	36421	January 8, 2023
X Server 1.13.1 deadlocks randomly on GeForce GTX680 Linux	6	3214	January 4, 2013
[SOLVED] XID 62: fixeable? Linux	3	4237	November 23, 2017
GSP crashes on 590.48.01 Xid 62 Linux gaming	4	153	April 18, 2026
Dual RTX4000 ADA XID 62 Failure Ubuntu 22.04 Linux cuda , ubuntu	1	178	November 7, 2024

Ubuntu 22, RTX 2060 --- Xid 62 under heavy CPU load

Related topics