Description of problem: My server has 2 bttv video capture cards with 8 devices per card. During boot these are detected and initialzed. Using kernel-5.15.74 detection of each chip takes a few seconds, however using kernel-5.15.79 the detection takes well over a minute. So for 16 devices this is around 20mins. I have not actually waited that long as I gave up and killed the power, after about 12mins it had reached device 9 of 16 when the log stopped. The above has been deduced afterwards studying the boot logs. I tested both kernel-server and kernel-desktop flavours of 5.15.79.
Created attachment 13540 [details] Normal boot with 5.15.74
Created attachment 13541 [details] Slow boot with 5.15.79
Thank you for the report, which is clear enough. Assigning to kernel.
Assignee: bugsquad => kernel
Two recommendations unrelated to the actual problem Add lines with SystemMaxUse=200M RuntimeMaxUse=200M to the end of /etc/systemd/journald.conf That will reduce the size of the journal from it's current 10% of the root partition size (2.3GB) to around 200M. On my system that's about 30 days normally. Increase/decrease as needed. Add "audit=0" to the kernel boot parameters. Unless you are using audit the messages clutter the journal.
CC: (none) => davidwhodgins
Thanks Dave - done that, I was a bit concerned about the journal size.
CC: (none) => tmb