00:00 <sjg1> Tartarus: Re alex, do you have access to clear caches? Anyway I will take a look tomorrow

00:07 <Tartarus> sjg1: I think you need to (and I just poked some systems to have it too) cron a weekly/monthly docker system prune --all --force --volumes

00:08 <Tartarus> docker doesn't bother cleaning caches automatically and so it just grows over time

09:21 <sjg1> Tartarus: OK I did that on Alex. It only has 1TB which is probably not enough for that many cores

14:11 <Tartarus> How much space did it free? That should be OK for such an otherwise new runner, we shouldn't have that much stale cache around. That said, I did want to see if gitlab is doing something annoying wrt bind mounts and /tmp or otherwise not automatically cleaning up after itself in ways we can do something about

16:01 <Tartarus> ... and rebooting guru, since there doing a "docker system prune --all --force --volumes" managed to leave dangling weird stuff, possibly due to gitlab-runner being in a container and not a stand alone service

16:13 <apalos> you can restart the docker service to 'fix' this

16:13 <apalos> Or at least that's what I did once

16:37 <clamor> Tartarus: yes, edid will work just like before with config disabled. I will provide a detailed answer via email.

16:37 <Tartarus> clamor: Thanks

16:38 <Tartarus> apalos: Good to know esp since it didn't come back up in Linux but I think macOS (it's a mac mini M2 I was hoping would be faster than it is, but still useful enough for a runner)

18:32 <Jones42> Is there anything else I need to do to disable the serial console during boot apart from CONFIG_SILENT_CONSOLE and CONFIG_DISABLE_CONSOLE ? I still get some output on my imx6 serial

19:46 <marex> Jones42: set the iomux and that surely disables it ;)

21:02 <Jones42> marex: hehe, is that how you're supposed to do it? since the mentioned configs don't work, I assume i must have overloooked something...

21:06 <Tartarus> Well, I've fixed a number of small easy Kconfig mistakes / warnings, and found we have one really funny thing that works unintentionally

21:07 <Tartarus> The choice statement in drivers/usb/ulpi/Kconfig is not valid. optional is not a documented keyword, nor one used in the linux kernel

21:07 <marex> Jones42: if you want to be really sure, that's one way

21:08 <Tartarus> But if you have USB_ULPI and USB_ULPI_VIEWPORT or USB_ULPI_VIEWPORT_OMAP in the defconfig, it behaves as expected

21:08 <Tartarus> And if you have neither, it doesn't prompt

21:08 <Tartarus> I need to go and whack that all a bit more to be differently handled, after testing the parts of this series I am pretty sure won't have an unexpected problem

21:10 <Tartarus> How, I don't know. Maybe it's just a prompted question

21:38 <Tartarus> OK, I think not quite the pain I was worried it would be

23:00 <marex> pbrobinson: https://patchwork.ozlabs.org/project/uboot/patch/20250313223947.1711595-1-pbrobinson@gmail.com/ also needs the is_cortex... macro , see end of the file