11:30 <farkuhar> heh, the latest tarball for one of my ports was apparently made on a Mac, or some platform that uses extended file attributes. I suppose I could redefine unpack_source in the Pkgfile, adding --no-xattrs to the bsdtar command. But it really would be easier if libarchive had sensible defaults, like not trying to preserve xattrs when unpacking to a filesystem that doesn't support them.

11:47 <farkuhar> we hit a similar problem in python3-wheel (versions before 0.38.2), having to redefine unpack_source to use gnutar rather than bsdtar because of locale issues in the unpacked filenames. Of all the reasons to redefine unpack_source, working around the idiosyncratic behavior of bsdtar is the one I dislike most.

14:18 <jaeger> Understandable

14:31 <farkuhar> Meanwhile I started trying to reproduce blackfrog78's "Very slow setup of ethernet link" with versions 6.4 of Linux and iproute2 (reported on the mailing list). It's been six months since my last kernel update, and my first build of 6.4.1 failed to boot at all! Is anyone else finding kernel 6.4.1 more troublesome than previous stable releases?

14:44 <jaeger> I haven't tried it yet

14:46 <jaeger> What kind of symptoms did you get when it failed to boot?

14:50 <jaeger> I've got a system I need to do hardware upgrades on today, I'll try 6.4 on it as well

15:02 <farkuhar> no visible error messages, just an automatic, silent reboot (as if the kernel couldn't find a rootfs). It would have gone into an endless reboot cycle if my syslinux.cfg did not have a delay allowing me to select a fallback kernel.

15:02 <jaeger> ok

15:18 <jaeger> Wow, this Thermalright Peerless Assassin 120 SE (silly name) cooler has one of the easiest mounting systems I've ever used

15:27 <ukky> Thermalright looks like Noctua CPU cooler, push-pull double fan design

15:28 <jaeger> That's a pretty common design, yeah

15:28 <jaeger> I even put 1 noctua fan on it, heh

15:28 <jaeger> Looks a tad silly with mismatched fans but I only have 1 right now

15:28 <jaeger> Time to see if the CPU swap was successful

15:29 <ukky> My main desktop system has about 15 Noctua fans, in a single system :-)

15:31 <ukky> Make sure your Noctua fat is on proper side, as some models can push air, but some models can only pull air

15:34 <jaeger> I find the NF-F12 fans work well in both, slightly better in push

15:35 <jaeger> And it seems like the CPU swap was good, booted and typing from the system now

15:35 <ukky> Correct, F12 should be pushing.

15:39 <jaeger> At the moment it's pulling, the front side fan sticks up a bit due to RAM clearance. I might move it later if temps are an issue but I do not expect them to be at all

15:40 <jaeger> Especially considering the "case"

15:40 <ukky> There should be no issues, just make sure to apply thermal paste, removing the old one. The "case"?

15:41 <jaeger> Yeah, not my first rodeo by far :) Already done that

15:41 <jaeger> Sec, I'll upload a picture or two

15:43 <jaeger> http://jaeger.morpheus.net/misc/oaf1.jpg http://jaeger.morpheus.net/misc/oaf2.jpg

15:45 <ukky> nice rig. Is this your new SPARC/MIPS motherboard?

15:45 <jaeger> Nah, it's an AM4 platform... ryzen 7 5700X and X570 motherboard

15:46 <jaeger> There are a lot of "open air frame" type things, the only thing that makes this one special, and perhaps only so to me, is that I designed and built it :)

15:47 <ukky> Yeah, I see Radeon graphics, I doubt AMD would have native graphics support for Sparc/Mips architecture

15:47 <jaeger> Probably not, yeah

15:48 <jaeger> So far cooling seems ok, 35C idle and 55-64C load (mostly around 55-56C but spikes)

15:48 <ukky> I know the feeling, as I build my own systems from pieces since circa 2005

15:49 <jaeger> One of these fans has a noticable buzz, though it's not quite loud enough to annoy me yet. Not sure which, will have to test each

15:50 <jaeger> Wouldn't be able to hear it over any music or app with sound

15:52 <ukky> Noctua should not generate any noise, unless running at crazy speed. Noise should be dealt with. Initially you don't notice it, but with time it will annoy you.

15:53 <jaeger> Should not, yeah... but it might be faulty. Not sure, it's been in a box unused for a long time so it might even have been faulty when I received it.

15:53 <jaeger> Definitely one of the fans, the buzz goes away when they slow down without any load

15:55 <ukky> It could be fan on VGA or power supply

16:01 <jaeger> Not in this case, I can get my ear close enough to tell it's one of the CPU fans but not quite distinguish which one without unplugging one of them which I haven't done yet

16:08 <jaeger> Also the GPU fans don't spin at all under only CPU load

16:16 <jaeger> OK, tested each. It's the Thermalright fan buzzing

16:48 <sajcho> farkuhar: linux-6.4.1 without ramdisk boot no problem https://dpaste.com/3JEYH7XV4

17:05 <farkuhar> sajcho: thanks for confirming. I just got a successful boot of 6.4.1 too. But I wasn't able to reproduce blackfrog78's bug on my current machine, which has an Intel ethernet card instead of Realtek.

17:08 <farkuhar> next step: copy the working kernel and modules over to a computer with Realtek NIC, and see whether the ethernet link takes longer than usual to establish.

17:40 <jaeger> building 6.4.1

17:44 <jaeger> 6.4.1 booted for me without any issues and I also do not have any more delay than usual bringing up the network interface. Intel NIC here as well.

18:09 <farkuhar> I tested both old and new kernels on a machine more similar to blackfrog78's hardware, and I saw almost no difference in how long it took to bring up the Realtek NIC. http://sprunge.us/ZlGdim

18:12 <crash_> i'm still on LTS 6.1 and it works perfectly

18:13 <farkuhar> in the interest of science, I might downgrade iproute2 to 6.3 (the version that according to blackfrog78 was faster at establishing a link), and compare the two kernels again.

21:27 <farkuhar> Downgrading iproute2 provided a few more data points, but nothing that clearly replicates blackfrog78's observations. http://sprunge.us/bCu8TB