- Arch-general - lists.archlinux.org

Install guide and UEFI system partition wiki pages suggest different /boot or /efi mount of ESP?
by David C Rankin 15 Sep '26

15 Sep '26

All, I'm trying to work out how to move / from a MBR with spinning disks to NVMe. The bios only boots from NVMe in UEFI mode, but supports booting from both UEFI and MBR in UEFI+Legacy mode. I've formatted the NVMe and set the optimal block size and created the ESP, / and swap partitions. I'm following along Migrate install to new hardware wiki [1] and using the installation guide [2]. The installation guide mounting of the ESP to /boot (/mnt/boot) seems to contradict what is documented in the EFI system partition wiki typical mount points where it describes the disadvantages of mounting ESP to /boot and suggests you mount ESP to /efi instead [3]. The listed disadvantages include: o This prevents setting file-specific permissions and/or extended attributes, o This increases the size requirement for the ESP, o This exposes the kernel and initramfs images to potentially hazardous manipulation from bootable drives or other operating systems when dual-booting, o This makes encrypting /boot impossible, and o This makes root volume snapshots (using Btrfs, Bcachefs, ZFS, LVM) less effective as /boot content would not be included. Currently the system is booting from the MBR hard drives and I'm just setting up the NVMe before booting the Arch install media and then copying / to new NVMe. Going forward, what is the current recommended mount of ESP in this situation, /boot or /efi? [1] Migrate install to new hardware https://wiki.archlinux.org/title/Migrate_installation_to_new_hardware [2] 1.11 Mount the file systems https://wiki.archlinux.org/title/Installation_guide [3] 4.1 Typical mount points https://wiki.archlinux.org/title/EFI_system_partition -- David C. Rankin, J.D.,P.E.

3 8

Mailing list traffic
by Ralf Mardorf 15 Sep '26

15 Sep '26

Hi, the most recent email in the archive is from August 24. Has no one used the mailing list for the past three weeks, or is something wrong? I actually wrote to the mailing list myself on August 29, but it never went through. Regards, Ralf

5 5

Public meetings?
by Vishal Kohli 11 Sep '26

11 Sep '26

Hi I am interested in learning OS development. Are there any public meetings that I can be part of to learn more about Arch Linux / OS development? Thank you

2 1

gitlab Arch - [Code] -> download .zip, .tar, .tar.gz of .tar.bz2 all append .html and download the page instead of repo
by David C Rankin 09 Sep '26

09 Sep '26

Arch gitlab devs, I don't know if this is just the dhcp packge or all, but I just needed to grab a copy of the latest and rather than clone I was just going to download the .tar.bz2. Clicking the [Code] button and selecting the .tar.bz2 format provided the filename with .tar.bz2.html as the extension and the download result was the gitlab page (html) instead to the repository contents. I checked each of the other formats, same result. I ended up just cloning with --depth=1 for the same effect plus the .git config. Thought I would pass this along in case there is a config problem for the download of compressed formats. This was in Firefox 154. Logged in to gitlab or not made no difference (not that it should) -- David C. Rankin, J.D.,P.E.

1 0

IntelliJ IDEA packaging update proposal
by Cynthia Rey 08 Sep '26

08 Sep '26

Hello, I tried to get an account on the GitLab to open an issue on the packaging repo, but since the SSO allowlisting process is quite rocky I'm reaching out directly here. Last year, JetBrains announced [1] a new distribution model for IDEA, ditching the difference between "Community" and "Ultimate" editions in version 2025.3, in favour of a unified build, now known simply as "IntelliJ IDEA". They also formalised a new variant, "IntelliJ IDEA Open Source", clarifying that the officially distributed "Community Edition" was actually not entirely based on the open-source repository, and included proprietary components. Given these changes upstream, my proposal is to rename `intellij-idea-community-edition` to `intellij-idea-oss` moving forward. This would align with the new model, the "OSS" branding being already present both internally and user-facing [2]. Independently, the AUR package distributing `intellij-idea-ultimate-edition` should become simply `intellij-idea`, as this has been suggested on aurweb (the package maintainer is waiting for official repos to make the first move [3]). Cheers ~Cynthia PS: The maintainers in `intellij-idea-community-edition`'s PKGBUILD isn't accurate AFAICT ;) [1] https://blog.jetbrains.com/idea/2025/07/intellij-idea-unified-distribution-… [2] https://github.com/JetBrains/intellij-community/blob/5ada6f537896bdf5d549ac… [3] https://aur.archlinux.org/packages/intellij-idea-ultimate-edition#comment-1…

1 0

Bug Report:[Regression] r8169 RTL8125 speeds tanked after latest update
by Jeronimo Garcia 30 Aug '26

30 Aug '26

Hey, After my last system update on Aug 27 I noticed my download speeds had completely tanked. Took a while to figure out what was going on but managed to narrow it down. My setup: - Realtek RTL8125B 2.5GbE (rev 05) - Driver: r8169 (in-kernel) - Kernel before update: 7.1.3.arch2-1 - Kernel after update: 7.1.9.arch1-2 - linux-firmware-realtek went from 20260622-1 to 20260810-2 on the same day Turns out disabling GRO and TSO with ethtool -K eno1 gro off tso off immediately fixes the speeds, so something with the offload handling is broken. Things I tried to narrow it down: - Booted into linux-lts 6.18.47-1 — same issue, so it's not specific to 7.1.9 - Downgraded linux-firmware-realtek back to 20260622-1 — made no difference - Tried r8125-dkms 9.018.00-1 from AUR — actually made things worse - Disabling GRO/TSO — fixes it every time on both kernels Also had 123,985 dropped RX packets on the interface which I think is related. Since it affects both mainline and LTS I'm guessing the bug has been in r8169 for a while and something just started triggering it consistently. Not sure if it's been reported upstream already but figured it was worth flagging. Cheers

1 0

Where does $HOME/.config/systemd/user/fig-daemon.service come from and why is $HOME/go write only?
by Ralf Mardorf 29 Aug '26

29 Aug '26

Hi, I followed the advice from https://thecybersecguru.com/news/atomic-arch-aur-supply-chain-attack-ebpf-r… [1]. I still have to check a second user account. Right now, I'm trying to figure out where "fig-daemon.service" comes from and why "go" in the home directory has strange permissions. Any ideas? $ /bin/ls /home/rocketmouse/go ls: cannot open directory '/home/rocketmouse/go': Permission denied $ sudo /bin/ls -hdltr /home/rocketmouse/go d-w------- 3 rocketmouse rocketmouse 4.0K May 5 2019 /home/rocketmouse/go $ sudo /bin/ls -hAltr /home/rocketmouse/go/pkg/mod total 16K drwxr-xr-x 4 rocketmouse rocketmouse 4.0K May 5 2019 cache drwxr-xr-x 3 rocketmouse rocketmouse 4.0K Oct 17 2020 golang.org drwxr-xr-x 4 rocketmouse rocketmouse 4.0K Jan 14 2024 gopkg.in drwxr-xr-x 47 rocketmouse rocketmouse 4.0K Feb 4 2026 github.com Regards, Ralf [1] Just so you know: "ls" is an alias for a "ls" substitute, which is why there are those strange characters. $ ls -la /proc/*/exe 2>/dev/null | grep -v "Permission denied" [snip] lrwxrwxrwx rocketmouse rocketmouse 0 B Fri Aug 28 12:10:12 2026 /proc/10524/exe ⇒ /usr/bin/speech-dispatcher [snip] I guess I misremembered that this had something to do with the AUR attacks, but it's actually from the extra repository. $ sudo find /home -path "*/.config/systemd/user/*.service" -exec grep -l "Restart=always" {} \; [sudo] password for rocketmouse: /home/rocketmouse/.config/systemd/user/fig-daemon.service I don't know to what fig-daemon.service belongs to. $ grep -r "Restart=always" /etc/systemd/system/ [empty] $ grep -r "RestartSec=30" /etc/systemd/system/ [empty] $ find /var/lib/ -type f -executable -newer /var/lib/dpkg/info 2>/dev/null [snip] $ sudo ls -la /sys/fs/bpf/ [empty] $ ss -tlnp | grep 127.0.0.1 [nothing suspect] $ sudo netstat -tlnp | grep 127.0.0.1 [nothing suspect] $ sudo find /root/.npm -name "atomic-lockfile" -type d [nothing] $ ls ~/.npm/_cacache/* /home/rocketmouse/.npm/_cacache/content-v2:  sha1  sha512 /home/rocketmouse/.npm/_cacache/index-v5:  00  08  10  18  20  28  30  38  40  48  50  58  60  68  70  78  80  88  90  98  a0  a8  b0  b8  c0  c8  d0  d8  e0  e8  f0  f8  01  09  11  19  21  29  31  39  41  49  51  59  61  69  71  79  81  89  91  99  a1  a9  b1  b9  c1  c9  d1  d9  e1  e9  f1  f9  02  0a  12  1a  22  2a  32  3a  42  4a  52  5a  62  6a  72  7a  82  8a  92  9a  a2  aa  b2  ba  c2  ca  d2  da  e2  ea  f2  fa  03  0b  13  1b  23  2b  33  3b  43  4b  53  5b  63  6b  73  7b  83  8b  93  9b  a3  ab  b3  bb  c3  cb  d3  db  e3  eb  f3  fb  04  0c  14  1c  24  2c  34  3c  44  4c  54  5c  64  6c  74  7c  84  8c  94  9c  a4  ac  b4  bc  c4  cc  d4  dc  e4  ec  f4  fc  05  0d  15  1d  25  2d  35  3d  45  4d  55  5d  65  6d  75  7d  85  8d  95  9d  a5  ad  b5  bd  c5  cd  d5  dd  e5  ed  f5  fd  06  0e  16  1e  26  2e  36  3e  46  4e  56  5e  66  6e  76  7e  86  8e  96  9e  a6  ae  b6  be  c6  ce  d6  de  e6  ee  f6  fe  07  0f  17  1f  27  2f  37  3f  47  4f  57  5f  67  6f  77  7f  87  8f  97  9f  a7  af  b7  bf  c7  cf  d7  df  e7  ef  f7  ff *???* $ sudo find / -type f -size 2961752c -exec sha256sum {} \; 2>/dev/null \ | grep 6144D433F8A0316869877B5F834C801251BBB936E5F1577C5680878C7443C98B [nothing] $ find ~/.npm /tmp /var/tmp -name "deps" -type f 2>/dev/null [nothing] I did not run cat ~/.cache/paru/clone/*, cat ~/.cache/yay/* as I don't use (those) AUR helpers.

1 1

Replace Gzip zcat with zstdcat
by autape 25 Aug '26

25 Aug '26

zcat works only with gzip files. I'm reminded of this on systems with defaults. As far as I know, zstdcat is a drop-in replacement, works with gzip, xz, zstd, and more. In some sense, both `provides` zcat. Pacman packages are zstd, I would assume it is as oblivious on the system as is gzip. GZIP(1): > On some systems, zcat may be installed as gzcat to preserve the original link to compress. This is a petition for a better default: to install Gzip's as gzcat, and symlink zcat to zstdcat. Not exactly sure, where it should go to reach packaging.

3 3

problems 'pacman -Syu'
by Greg Minshall 20 Aug '26

20 Aug '26

hi. trying to update, i kept running into problems, e.g., ---- error: failed retrieving file 'doxygen-1.18.0-1-x86_64.pkg.tar.zst.sig' from losangeles.mirror.pkgbuild.com : Operation too slow. Less than 1 bytes/sec transferred the last 10 seconds ---- but, whichever package in question, it seemed to have downloaded, but maybe not completed whatever last handshake/transaction? ---- doxygen-1.18.0-1-x86_64 6.7 MiB 272 KiB/s 00:25 [##############################################################] 100% ---- then, on the next run it would say that the package was up to date ---- doxygen-1.18.0-1-x86_64 is up to date ---- i commented out `losangeles.mirror.pkgbuild.com` in /etc/pacman.d/mirrorlist, and then i was able to complete pacman -Syu. fwiw. cheers, Greg

4 4

Follow recomendation of ordering for "/usr/lib/modules-load.d/*.conf" files
by EpicTux123 12 Aug '26

12 Aug '26

Hi there, I have noticed that some packages in my system, under "/usr/lib/modules-load.d/", are installed without a numerical prefix. https://man.archlinux.org/man/modules-load.d.5.en says: "It is recommended to prefix all filenames with a two-digit number and a dash to simplify the ordering. It is recommended to use the range 10-40 for configuration files in /usr/ and the range 60-90 for configuration files in /etc/ and /run/, to make sure that local and transient configuration files will always take priority over configuration files shipped by the OS vendor." In my specific system, I have noticed "bluez.conf", "ddcutil.conf", "fwupd-i2c.conf" and "i2c-tools.conf" as files that do not have a prefix in that folder, therefore not following the recommendation. I am not sure of what is the best method to address all of those files through some sort of general communication to all package mantainers, so I am writing to this list, because I think creating an issue on each package would not be a smart idea. My idea here is to get in touch with all relevant Package Mantainers so that they can take a look at their mantained projects, see if they have a configuration file that it is shipped with, and adjust accordingly to add the numerical prefix. Please let me know if there is a better place to do this, if not in this mailing list. Thanks. -EpicTux123

1 0