build reproducibility tracker

[renner0e]

Improvements/workarounds that would make our images more reproducible and avoid unnecessarily downloading things that are already on disk

[renner0e]

here is a diff between the images (initramfs of ghcr.io/renner0e/aurora-composefs:tag) built on July 18th and 19th

sudo podman run -it -v .:/app:Z image:tag lsinitrd --unpack /usr/lib/modules/*/initramfs.img

sudo diff -ru --color latest latest-20260718-dbfe348

diff: latest/etc/systemd/system/initrd.target.wants/dracut-cmdline-ask.service: No such file or directory
diff: latest/etc/systemd/system/initrd.target.wants/memstrack.service: No such file or directory
Binary files latest/lib/modules/7.1.3-201.fc44.x86_64/extra/v4l2loopback/v4l2loopback.ko.xz and latest-20260718-dbfe348/lib/modules/7.1.3-201.fc44.x86_64/extra/v4l2loopback/v4l2loopback.ko.xz differ
Binary files latest/usr/lib/modules/7.1.3-201.fc44.x86_64/extra/v4l2loopback/v4l2loopback.ko.xz and latest-20260718-dbfe348/usr/lib/modules/7.1.3-201.fc44.x86_64/extra/v4l2loopback/v4l2loopback.ko.xz differ
diff: latest/var/lock: No such file or directory

sudo diffoscope latest/lib/modules/7.1.3-201.fc44.x86_64/extra/v4l2loopback/v4l2loopback.ko.xz latest-20260718-dbfe348/lib/modules/7.1.3-201.fc44.x86_64/extra/v4l2loopback/v4l2loopback.ko.xz
--- latest/lib/modules/7.1.3-201.fc44.x86_64/extra/v4l2loopback/v4l2loopback.ko.xz
+++ latest-20260718-dbfe348/lib/modules/7.1.3-201.fc44.x86_64/extra/v4l2loopback/v4l2loopback.ko.xz
├── v4l2loopback.ko
│┄ File has been modified after NT_GNU_BUILD_ID has been applied.
│ ├── readelf --wide --notes {}
│ │ @@ -1,13 +1,13 @@
│ │
│ │  Displaying notes found in: .note.gnu.property
│ │    Owner                Data size   Description
│ │    GNU                  0x00000030  NT_GNU_PROPERTY_TYPE_0        Properties: x86 feature: IBT,x86 feature used: x86, x86 ISA used: x86-64-baseline
│ │
│ │  Displaying notes found in: .note.gnu.build-id
│ │    Owner                Data size   Description
│ │ -  GNU                  0x00000014  NT_GNU_BUILD_ID (unique build ID bitstring)         Build ID: 14e36f4932ee0b0032dbf6dd7309f0f19137a8e9
│ │ +  GNU                  0x00000014  NT_GNU_BUILD_ID (unique build ID bitstring)         Build ID: add4066f0f0846b14ec8fd93342c1411e19ee51c
│ │
│ │  Displaying notes found in: .note.Linux
│ │    Owner                Data size   Description
│ │    Linux                0x00000004  func       description data: 00 00 00 00
│ │    Linux                0x00000016  OPEN       description data: 37 2e 31 2e 33 2d 32 30 31 2e 66 63 34 34 2e 78 38 36 5f 36 34 00
│ ├── strings --all --bytes=8 {}
│ │ @@ -626,19 +626,18 @@
│ │  .rela.gnu.linkonce.this_module
│ │  .note.Linux
│ │  .orc_header
│ │  Universal Blue1
│ │  kernel signing1
│ │  ublue kernel1*0(
│ │  security@universal-blue.org
│ │ -`-F]Bsyw
│ │  ~Module signature appended~
│ │  Universal Blue1
│ │  kernel signing1
│ │  ublue kernel1*0(
│ │  security@universal-blue.org
│ │ -\TLA%{H"
│ │ +H->RY/"#K
│ │  Universal Blue1
│ │  akmods1%0#
│ │  jorge.castro@gmail.com1
│ │  ublue akmods
│ │  ~Module signature appended~

So akmods v4l2loopback has a build id that changes on seemingly every (akmods) image, I'm not sure why that is? Which would explain why some images (with the same akmods images) have the same initramfs. Why do we even still have that?

and there are also broken symlinks which are flagged by diff, I think that is just the way diff works and how lsinitrd extracts the image?