| bb|hcb | It wouldn't be a problem to get RAM if I knew that I had to... Anyways... | 00:02 |
|---|---|---|
| parabyte | gnarface, well i got all the gpio's specified correctly, i have no activity at all | 00:21 |
| parabyte | gnarface, rather annoying cause i do not believe there is a uart anywhere that i know of without probing | 00:21 |
| gnarface | parabyte: wait, was this one of those pine64 boards? there's definitely uart on them, there's two in fact but one of them crashes it | 00:22 |
| parabyte | gnarface, nah im fooling around with a tegra 2 thingy, "asus tf101" | 00:22 |
| gnarface | ah, right | 00:22 |
| gnarface | dunno much about them | 00:22 |
| rrq | parabyte: there is this "mate-postmarketOS-TF101-v4.0.img" on sourceforge (somewhere) .. unless that's you, it might be of use | 00:25 |
| parabyte | they are all building there targets for old uboot | 00:26 |
| rrq | that one has a dtb for linux 3.? | 00:27 |
| parabyte | nope | 00:27 |
| rrq | hmm at least it has a file named dt.dtb in the first partition :) | 00:28 |
| parabyte | im sure i will figure it out eventually, its a good learning exercise for sure | 00:28 |
| parabyte | i will grow tired of this machine soon | 00:29 |
| parabyte | has a nice ips display in it i may repurposed when grow tired | 00:29 |
| c0rnelius | tried this hub -> https://github.com/clamor-s/u-boot | 00:32 |
| c0rnelius | Thats v2022.01 | 00:33 |
| c0rnelius | found it through this guys fork which is a lil behind - https://github.com/antonialoytorrens/u-boot-asus-tf101 | 00:33 |
| LeePen | Hi. | 12:50 |
| LeePen | I am building the polkit security updates to address CVE-2021-3560. | 12:51 |
| LeePen | In beowulf, the armel build fails. | 12:51 |
| LeePen | I have tracked this down to SIGILL in /usr/bin/highlight. | 12:51 |
| LeePen | http://ix.io/3Nzv | 12:53 |
| LeePen | Is swp not valid? | 12:54 |
| bb|hcb | https://developer.arm.com/documentation/ddi0406/c/Application-Level-Architecture/Instruction-Details/Alphabetical-list-of-instructions/SWP--SWPB?lang=en | 12:59 |
| bb|hcb | This says deprecated/optional | 13:00 |
| LeePen | Hmmm | 13:03 |
| bb|hcb | https://lists.debian.org/debian-arm/2019/02/msg00034.html | 13:04 |
| LeePen | and https://bugs.debian.org/922728 | 13:07 |
| LeePen | I don't really know what to do about that. | 13:10 |
| LeePen | I have skipped the armel build for now. | 13:10 |
| LeePen | Any ideas? | 13:10 |
| bb|hcb | (guessing) Either boost is not built well (and uses swp for spinlocks) or on a real armel the problem is not present | 13:12 |
| LeePen | The buildhost is native arm64 emulating armel (and armhf) | 13:13 |
| LeePen | swp might be emulated in kernel 5.15: https://www.kernel.org/doc/html/v5.15/arm/swp_emulation.html | 13:16 |
| LeePen | But the arm64 builder is beowulf (5.7.2). | 13:16 |
| rrq | https://lwn.net/Articles/614342/ ... suggests there possibly is a sysctl wedging in a software emulation ... (quick reading) | 13:17 |
| LeePen | Yes, I found that too. I think it is more recent kernel. | 13:18 |
| LeePen | I can try to upgrade the builder, but (I think) it is on a boat in Amsterdam if it doesn't reboot! | 13:18 |
| bb|hcb | I only have arm64 installs and that is not easy to test :( | 13:19 |
| bb|hcb | I mean the native armel, adding an armel chroot should be possible | 13:21 |
| LeePen | Actually, I am not sure I can upgrade the kernel. There is nothing in /boot (because it is arm64). | 13:22 |
| LeePen | How do you upgrade the kernel? | 13:22 |
| bb|hcb | What does cat /proc/sys/abi/swp say? | 13:23 |
| LeePen | Not present | 13:23 |
| LeePen | jenkins@devuan-arm64:~$ ls /proc/sys/abi/ | 13:24 |
| LeePen | tagged_addr_disabled | 13:24 |
| bb|hcb | What HW is that? | 13:24 |
| LeePen | Not sure exactly -- the arm64 builder parazyd setup a couple of years ago. | 13:26 |
| LeePen | http://ix.io/3Nzz | 13:26 |
| rrq | so the armel building is not in a qemu VM on that host? | 13:26 |
| LeePen | No, it is 32 bit emulation AFAIU (like i386 can be built on amd64 to my simple mind). | 13:27 |
| rrq | right; I suppose a qemu setup could allow a diffferent kernel.. but that'd be a bit of a walkabout | 13:30 |
| rrq | the (ganeti) armel-builder is in deep coma ... | 13:30 |
| LeePen | And we wenta way from them because they were very slow and also unreliable for low level packages. | 13:31 |
| bb|hcb | That looks like a vm to me... | 13:31 |
| rrq | in that case the bootup kernel and initrd are files on the host; not accessible from within the VM | 13:33 |
| LeePen | OK, so we need parazyd | 13:33 |
| LeePen | Anybody seen him? | 13:34 |
| rrq | a qemu option with an (nested) arm64 VM would probably offer 80% performance .. but that'd still involve all required setup | 13:36 |
| LeePen | OK. Since the rest of the beowulf archs are built, I am going to focus on policykit-1 for ascii first and then come back. | 13:39 |
| LeePen | Thanks all. | 13:39 |
| LeePen | At least the ascii-security armel build is successful! | 14:13 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!