swupd getting an oom kill on larger updates

[1490kdrm]

Describe the bug

Hello,

On two recent installs of Clear I've been having the swupd process be killed in the terminal, and majority of the time I'll get a hang if I try to jam the process through again directly after. It'll seemingly download all of the bundles, extract, and then get killed at install, which I find odd. When attempting to install 1-2 bundles at a time it normally is not an issue. Note that this is not an imdepotent issue and really only happens roughly 6-7 times out of ten. I've not observed it happening with specific bundles as it seems random. The system will usually not be running anything outside of the swupd process and desktop (and the litany of accompanying processes within kde and gnome desktop environments) in userspace outside of maybe suricata and NetworkManager.

Having 32GB of RAM I'm unsure why it's wanting kill the process or why it's hanging after I try to run it after it gets killed. Let me know what you'd need from me.

To Reproduce Run swupd bundle-add with < 7-8 bundles

Expected behavior That swupd cleanly executes the bundle-install command by downloading, extracting and installing the said bundles and then reloading the system as normal.

Environment (please complete the following information):

• Clear Linux OS Version: 41950
• Platform: bare-metal - Gigabyte Z790 Aorus Master X v1.0, Intel 13900ks, Nvidia Zotac Gaming 3060ti, Crucial 1TB T700 gen5 m.2 NVMe drive (luks encrypted, Clear OS on disk - installed using the Clear ISO). x3 1TB Samsung 990 m.2 NVMe drives in an mdadm RAID5 array, SK Hynix 2TB Platinum m.2 NVMe drive (with a Windows OS install). G.SKILL Trident Z 32GB 6400 XMP. AIO. Gigabyte 850W Platinum PSU.

[1490kdrm]

swupd_taint.txt

Interestingly enough it's actually showing as a kernel taint for swupd, with a kernel NULL pointer dereference for it and mdadm...

output of dmesg attached.