Live bootstrap to guix#578
Conversation
…compress, and debuginfod client
…mpile via rpath-link and explicit LIBS
feat(steps-guix): add libgcrypt-1.12.1 default build with gcc-detected host and pkg-config path feat(steps-guix): add guile-gcrypt-0.5.0 with dynamic libgcrypt prefix and ld library path
…and end with shared after
… out-of-tree build
…IR for libgcrypt lookup
…IB_SRCDIR and no-git
…refix and checksum
…e-gnutls flow with no-so patch
…fail-safe without panic
…GUILEC load-extension segfault
…ble them in manifest
…time/localtime declarations on linux
|
Currently running a bare metal test with this diff on top of the latest commit: $ git diff
diff --git a/steps-guix/improve/guix-daemon-and-pull.sh b/steps-guix/improve/guix-daemon-and-pull.sh
index c7625474..f61b5d9f 100644
--- a/steps-guix/improve/guix-daemon-and-pull.sh
+++ b/steps-guix/improve/guix-daemon-and-pull.sh
@@ -218,6 +218,9 @@ mkdir -p /proc /sys /dev "${guix_localstate_dir}/daemon-socket" /var/lib/guix /r
mount | grep ' on /proc ' >/dev/null 2>&1 || mount -t proc proc /proc
mount | grep ' on /sys ' >/dev/null 2>&1 || mount -t sysfs sysfs /sys
mount | grep ' on /dev ' >/dev/null 2>&1 || mount -t devtmpfs devtmpfs /dev
+# tmpfs must be unmounted to avoid overfilling memory
+mount | grep ' on /tmp ' >/dev/null 2>&1 && umount /tmp
+test -f /swapfile && swapon /swapfile
if ! mount | grep ' on /dev/pts ' >/dev/null 2>&1; then
mkdir -p /dev/pts
mount -t devpts devpts /dev/pts
diff --git a/steps-guix/jump/linux64.sh b/steps-guix/jump/linux64.sh
index 1bb6ba93..f942c2a1 100644
--- a/steps-guix/jump/linux64.sh
+++ b/steps-guix/jump/linux64.sh
@@ -37,4 +37,8 @@ else
--append="console=ttyS0 root=/dev/sda1 init=/init rw rootwait consoleblank=0"
fi
quiesce_filesystem_for_kexec
-kexec -e
+if [ "${BARE_METAL}" = True ]; then
+ echo b > /proc/sysrq-trigger || true
+else
+ kexec -e
+fiI disabled kexec for bare metal, because I couldn't get the framebuffer to work reliably after kexecing from 32-bit 4.14-openela to 64-bit 6.12-gnu. The other change activates the swapfile, and unmounts /tmp before starting the Guix bootstrap - this was needed to avoid going OOM on my 8GiB RAM bootstrap rig. |
|
I’m also thinking about running a bare-metal setup. Is using Coreboot + SeaBIOS on a modern motherboard a viable option? |
|
If you can run Coreboot on your board, it should be, especially since qemu's default BIOS implementation is SeaBIOS. Note that I had to edit that patch in that previous comment, as I got Bash's conditional syntax wrong (it's if...fi, not if...endif). |
|
Great, I’ll add the patch shortly. |
|
There is currently no reliable method for including Closure in an ISO image and performing a guided installation. |
|
The new patch is malformed: With that fixed, I had a test failure in wmin69qfszbjz0mflj2sya8sm2r5c7bs-glib-2.83.3.drv (GLib 2.83.3), that went away on retry. Maybe we should disable tests for it? |
|
OK, after lots of wrangling... I have an ISO.More to come tomorrow. I've just came home, and am quite tired. And it did took quite some wrangling to get here. |
|
So, finally got around to writing it all up.
And then, some bad news:
|
|
Could you tell in detail why the disk partitioning failed? I'm trying to use a non-default ISO for the setup, specifically, a custom ISO that defines as few components as possible, and I'm adding the |
|
dump.2026-04-28.10.12.04.tar.gz This is the dump I was able to obtain from the failed installation. Looks like maybe the target device isn't being found. With that said, I suspect the failure might be due to some packages being built with chroot off, resulting in contamination from the live-bootstrap host environment. |
|
I tried the Manual partitioning option in Setup, only to get an empty list of potential disks to partition. |
|
what |
|
Looks like I was dumb - the mounted virtual disk had a malformed partition table on it, which confused Setup. A |
Looks like it's looking for the local channel repository inside the ISO. Cloning it should work, though. EDIT: no EDIT 2: it's not actually using substitutes; while it did look for them, what it's actually downloading is all source code, because the bootstrap changes made the upstream substitutes not match. Using the L-B system as a substitute server remains a possibility. |
|
|
Not sure if it was the exact same one, but I did occasionally see a strange failure in the check phase of guix-1.5.0rc1, that wasn't a test failure (which should move on to other tests), but something that instantly interrupted execution of the test suite. It's nondeterministic, and simply retrying helps. |
I suppose part of the concept of reproducible builds is consistency in behavior. I think that if we have time, we should look into the specific cause. As for tests that fail consistently, what I want to do is fix those failing tests. First, I’ll look into why they’re failing, whether it’s because the hashes don’t match or some other reason. |
fixed in 2a58e2c |
| -s i686-linux \ | ||
| - -- guile -c '(exit (string-prefix? "i686" %host-type))' | ||
| + -- guile -c '(exit (string=? "x86_64-pc-linux-muslx32" %host-type))' | ||
| + -- guile -c '(exit (string-prefix? "x86_64" %host-type))' |
There was a problem hiding this comment.
This amounts to effectively disabling the test, as it's meant precisely to check that on a multiarch-capable machine, the correct seed is selected based on the choice of environment (i.e. a 32-bit seed when explicitly building 32-bit on 64-bit hardware).
IMO the right fix is to patch Guile 2.0.9 so it reports the correct "i686-pc-linux-musl" architecture.
3 participants
No description provided.