Colin Walters [Mon, 21 Aug 2023 21:19:35 +0000 (17:19 -0400)]
composefs: Hard error except on ENOENT even in "optional" case
Since we enabled composefs at build time, the default (non-composefs)
case now always prints
`composefs: Optional support failed: No such file or directory`
But that's normal and expected.
Rework things here so that in the very special case where
we are in "maybe/optional" mode and we get ENOENT, then we
output a much more normal-looking message that doesn't include
the string "failed".
Now on the flip side - if I have explicitly enabled signature
checking, I think we *do* want to make that fatal even if
composefs is in "maybe" mode.
(This part is more debatable; perhaps we should just disallow
the case of "maybe" + signatures at all; but I think this is
an improvement in that direction)
Colin Walters [Mon, 21 Aug 2023 21:12:08 +0000 (17:12 -0400)]
prepare-root: Fold together composefs signature cases
Now that we don't support digest-but-not-signature verification
for composefs, the logic here was unnecessarily complicated.
With a prior prep patch that moved the composefs option
initialization up, we can just have everything related to signature
verification in a single conditonal.
Colin Walters [Mon, 21 Aug 2023 18:05:14 +0000 (14:05 -0400)]
prepare-root: Drop redundant print about signature/digest
We print if we're doing a signature+digest verification; its absence is
sufficient in the other case. The goal here is to avoid polluting
the logs when signatures are not enabled.
Colin Walters [Thu, 24 Aug 2023 23:48:22 +0000 (19:48 -0400)]
build-sys: Really fix composefs check
The ordering of the includes apparently matters...and I didn't
actually check that the previous change enables composefs on c9s.
But I did now. For reals.
While we have the patient open, I switched to `AC_LANG_PROGRAM`
because I originally thought the bug had something to do with that.
As far as I understand, more cleanly separating the includes
from the injected body text is a useful thing in `AC_LANG_PROGRAM`.
Colin Walters [Wed, 23 Aug 2023 20:06:23 +0000 (16:06 -0400)]
Move prepare-root karg helpers into otcore, add unit tests
Add long overdue unit testing coverage for this, which
at least slightly closes out the android boot CI gap.
Actually, this *copies* the karg parsing code into otcore because
it now uses glib, which we're not yet using in the static
prepare-root. It's pretty tempting to drop support for the
static prepare root entirely. But for now we'll live with some
code duplication.
Colin Walters [Thu, 17 Aug 2023 12:26:29 +0000 (08:26 -0400)]
repo: Bump lock timeout to 5 minutes
And update the doc text to talk about having a timeout at all
by default being a mistake.
Timeouts are really best handled at a higher level; if two processes
are contending for the ostree lock and one is actually frozen,
resolving this is something an admin may want to handle and introspect/debug
instead of having the waiter error out.
Most people using ostree are doing it in a way in which they have
higher level timeouts (e.g. on a container pod).
Colin Walters [Wed, 16 Aug 2023 20:16:24 +0000 (16:16 -0400)]
deploy: Add some error prefixing
We saw a bare
`Txn Rebase on <osname> failed: Failed to find kernel in /usr/lib/modules, /usr/lib/ostree-boot or /boot`
which isn't bad, but it'd be better to be a bit more specific.
Colin Walters [Mon, 14 Aug 2023 18:30:42 +0000 (14:30 -0400)]
prepare-root: Check for empty string, not strlen > 0
No point in doing a full strlen, we can just check the first byte.
Also, invert the conditional using `continue` to avoid another
level of indentation.
Colin Walters [Mon, 14 Aug 2023 13:48:49 +0000 (09:48 -0400)]
README.md: Drop dead mailing list, link to GH discussions
While I resisted taking the next step in binding ourselves
more to GH with discussions...it's way, way better than answering
questions out of band in private (also proprietary) chats.
We haven't been successful in using the GNOME discussion forums.
Read composefs configuration from initrd instead of commandline
This drops the `ot-composefs` kernel commandline in favour
of a `[composefs]` section in the `prepare-rootfs.conf` file.
You can set `composefs.enabled` to `signed`, `yes`, `no` or `maybe`,
with `maybe` being the default.
You can also set `composefs.keypath` (or rely on the default
`/etc/ostree/initramfs-root-binding.key`) to point to ed25519 public
keys, one of which which the commit must be signed with, or boot
fails.
The ostree dracut module adds `/etc/ostree/initramfs-root-binding.key`
to the initrd if it exists.
NOTE: This drop the option to define a digest in the commandline.
However, that was currently unused
(i.e. ComposefsConfig.expected_digest was never read).
Additionally it very hard to actually store the composefs digest in
the initrd, as the initrd is typically part of the commit and thus the
composefs. It may be possible to handle this, but lets add it back
when we know exactly how that will work.
Colin Walters [Fri, 4 Aug 2023 02:49:48 +0000 (22:49 -0400)]
deploy: Fix mutex locking for global sync timeout
The locking here was always too long - by holding the mutex
during the `sync()` call, it means `g_cond_wait_until()` can
never wake up (because its API requires the mutex to be locked).
Confusingly though of course we do still print the "timed out"
message, and I think that tricked us when we were doing testing
here.
We only need to lock the mutex when we're manipulating shared
state, which basically boils down to the `gboolean success`.
Colin Walters [Tue, 1 Aug 2023 12:58:27 +0000 (08:58 -0400)]
os-init: Create a mount namespace
Today on anything using readonly sysroot `os-init` fails, because
we don't create a mount namespace if the `UNLOCKED` flag is specified
because we assume it's a readonly operation.
Since technically this is a mutation, let's just lock the sysroot
and use the tested path.
Colin Walters [Sat, 29 Jul 2023 12:47:47 +0000 (08:47 -0400)]
deploy: Support an empty `/etc` and populated `/usr/etc`
In preparation for support for a transient `/etc`:
https://github.com/ostreedev/ostree/issues/2868
particularly in combination with composefs.
Basically it's just much more elegant if we can directly mount
an overlayfs on the *empty* `etc` directory, using `usr/etc` as
the lower.
In the composefs case, we'd have to mount the composefs overlayfs
itself writable (and call `mkdir`) *just* so we can make that
empty `etc` directory which is ugly.
Colin Walters [Fri, 28 Jul 2023 21:54:37 +0000 (17:54 -0400)]
tests/destructive: Turn off global sync()
Let's verify that things work with that off, as they should.
Previously:
https://github.com/ostreedev/ostree/pull/2519/commits/cb731294837736e957ee595ce11ab115277dbb36
"deploy: Add a 5s max timeout on global filesystem sync()"
But we may still have problems even with that, see
https://issues.redhat.com/browse/OCPBUGS-15917
where it might be that even a thread doesn't work because
we're locked in the kernel.
projects / ostree.git / log