A Compressor can benefit if the uncompressed size is
known at the start.

Potential benefits are less memory requirements when decoding,
and simpler framing (no "streaming mode").

This changes the pipe to create a new temporary file
containing the main cpio archive.
The final initramfs output file can then be created in
a single write (no appending).

Signed-off-by: Norbert Lange <nolange79@gmail.com>

Signed-off-by: Ben Hutchings <benh@debian.org>

Closes: #970235
Signed-off-by: Ben Hutchings <benh@debian.org>

As these tests require installing an amd64 kernel package, they have
to be skipped on other architectures.  Tthere is currently no way to
explicitly restrict them to specific architectures, so instead add the
skip-not-installable restriction.

Closes: #970184
Signed-off-by: Ben Hutchings <benh@debian.org>

* Change the linux-image-amd64 dependency to -amd64 | -generic:amd64

* Replace the LOCALVERSION variable with a SUPPORTED_FLAVOURS
  variable, and have test-common loop over the flavours

* Change the dpkg-query pipeline that finds the kernel version
  string to avoid assumptions about ordering and versioning of
  dependencies

This should allow the tests to run on Ubuntu or its derivatives
without changes.

Signed-off-by: Ben Hutchings <benh@debian.org>

Signed-off-by: Ben Hutchings <benh@debian.org>

Add firmware for built-in drivers

See merge request kernel-team/initramfs-tools!35

There's an old bug report from 2017 (#857054) complaining that firmware for
built-in drivers is not automatically added to initramfs; and back then,
the kernel did not collect information about such dependencies in a place
where initramfs-tools could access it (#857055).

However, since Linux 5.2 (released in mid-2019), the kernel actually
provides this information via modules.builtin.modinfo; you can see at e.g.
<https://packages.debian.org/bullseye/amd64/linux-image-5.7.0-3-amd64/filelist>
that the buster kernel package ships this file. (There aren't actually any
firmware dependencies listed in there though; I guess the upstream Debian
kernel config just has no built-in code with firmware dependencies?)

Wire it up, so that people with custom kernel configs don't have to waste
time figuring out that their devices don't work properly because the
firmware wasn't available at boot.

Closes: #857054

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

For now, use MODULES=list.

I want to test with MODULES=dep too, but that will require putting a
Debian installation inside the VM so we can run mkinitramfs there.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

This should allow implementing various test cases without too much
repetition.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

Negative timeout values are treated by the kernel as "reboot
immediately" and 0 is treated as "wait forever".  Emulate this
behaviour in the panic() function.

Treat invalid (non-numeric) values the same as 0, which seems to match
what the kernel does.  Previously we would ignore them completely and
open a shell as normal.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

This is needed for testing, and should also be useful for building
more generic initramfs images.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

There are no real sections in this file.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

hppa and m68k have higher soversions for libgcc_s.so, so copy
libgcc_s.so.[1-9] instead of assuming soversion 1.

(I considered copying libgcc_s.so.*, but I'm unsure whether that will
ever match extra development files that we don't want.)

Closes: #959768
Signed-off-by: Ben Hutchings <benh@debian.org>

update-initramfs used to keep track of which initramfs images it
created and "owned", to avoid overwriting the output of another
coinstalled initramfs generator.  As discussed in #765750, this is no
longer a concern and the state checks have been removed.

Now that I'm confident this hasn't caused a regression, also remove
the state updates and the state directory.  After this, there is no
turning back: a downgrade will leave update-initramfs owning nothing.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

The direct invocation of update-initramfs in kernel package maintainer
scripts was removed in:

- Debian linux 2.6.39-1, thus in wheezy
- Ubuntu linux 4.15.0-18.19 (I think), thus in bionic
- kernel-package 12.001, thus in Debian squeeze and Ubuntu lucid

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

un-,mkinitramfs: add support for zstd

Closes: #955469

See merge request kernel-team/initramfs-tools!33

Prevents printf error carry over if wrong console is set

Closes: #960355

See merge request kernel-team/initramfs-tools!30

When overriding configured compressor, warn the user.

See merge request kernel-team/initramfs-tools!31

Add support for zstd compression and decompression.

Always use multithreaded compression, as zstd still
generates identical archives.

The zstd cmdline tool is usually able to decompress
gzip, xz, lzma and lz4 so make sure to try the tool
early.

Signed-off-by: Norbert Lange <nolange79@gmail.com>

Currently the _log_msg() functions is "void" typed - with no return -,
which in terms of shell means it returns whatever its last command
returns. This function is the basic building block for all error/warning
messages in initramfs-tools.

It was noticed [0] that in case of bad console is provided to kernel on
command-line, printf (and apparently all write()-related functions) returns
error, and so this error is carried over in _log_msg(). Happens that
checkfs() function has a loop that runs forever in this scenario (*if* fsck
is not present in initramfs, and obviously if "quiet" is not provided in the
command-line). The situation is easily reproducible and we can find various
reports dating back some years. The reports usually are of the form
"machine can't boot if wrong console is provided" or slightly different
forms of that, almost always relating serial consoles with boot issues.

This patch proposes a pretty simple fix: return zero on _log_msg().
We should definitely not brake the boot due to error log functions;
one could argue we could fix checkfs() and that's true, until eventually
we find another subtle corner case of "misuse" of the _log_msg() return
value (after some debugging), and fix that too, and so on...
W could also argue that printf shouldn't return error in this case,
and although a valid discussion, it's not worth to have users waiting
on a dilemma while boot is quite easy to brake, just by passing a wrong
kernel parameter (or having the underlying serial console device changed
to output to a different port than the previously set on kernel cmdline).

[0] bugs.launchpad.net/cloud-images/+bug/1573095/comments/46
Signed-off-by: Guilherme G. Piccoli <gpiccoli@canonical.com>

The Pinebook LCD disply does not show output until after the initramfs
is loaded, which is too late to enter an encryption passphrase or
debug issues with mounting the rootfs.

Fix this by adding modules to "fb": analogix-anx6345, pwm-sun4i,
sun4i-drm and sun8i-mixer.

I'm now using benh@debian.org in debian/changelog, so make these
consistent.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

documentation.debian-projects.org seems to have gone away many
years ago, but web.archive.org has a copy so link to that.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

This requires a heading row, so add one.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

This is a short document and there is no need to break it up this
much.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

I like to keep up with the official kernel releases. As a result the
firmware they wind up requiring is newer than what is available in
firmware packages.

I placed the firmware from the git repo in /lib/firmware/updates/
which the kernel searches before /lib/firmware.  Kernel can
find them but update-initramfs cannot.

I have attached an initial patch that solves this. I am currently
using the resultng code and it is working fine for me.

The one thing the patch does not do is allow for a custom location
that can be provided to the kernel to search first.

This is because that's a little more complicated and requires a
decision wrt where to specify this location (ie the config file?).

[bwh: Text above is from AP's bug report with minor edits; I added
 the subject line.]
Closes: #956663
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

These all seem to be harmless in practice, as the parameter values
should not contain metacharacters.

In _checkfs_once() *do not* quote $spinner or $force; if these
are empty then we do not want to add arguments for them.  Add a
comment to suppress the warning.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

If we have a generic framebuffer device then there should be no
need to set up any backlight in early boot.

Also, backlight devices are sometimes handled by large drm drivers
like i915 which we don't want to include by default.

Fixes: da851903 ("hook-functions: Include modules for Chromebook ...")
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

It was logging the link target, not the name of the symlink.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

pthread_cancel() in libpthread.so.1 depends on functions in
libgcc_s.so.0, but uses dlopen() to link to it rather than having a
static dependency.  Handle this dynamic dependency in copy_exec so
that it doesn't have to be handled in every affected hook script.

Closes: #950254
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>