stripper: Implement copying sources

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

FHS: Permit debug sources

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

filelist: Actually sort when adding files

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

filelist: Drop the sort function again

The filelist was intended to be sorted all the time.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

filelist: Fix return code when we run out of buffer space

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

filelist: Don't overwrite r when using a progress bar

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

stripper: Run the stripping inside the jail

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

stripper: Keep a reference to the jail

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

file: Fix path when opening a file inside the jail

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

ctx: Set a flag when we are running inside the jail

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Change callback signature to include the context instead the inaccessible jail

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Add a function to call a custom callback inside the jail

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

build: Pass the environment to the interactive shell

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Rename pakfire_jail_exec to pakfire_jail_exec_command

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

env: Declare the struct before importing anything

This silences a compiler warning.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Add tests for env

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Merge the basic environment with the custom environment

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

build: Simplify reading the parser environment

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Move the environment into an extra object

This allows us to pass a custom environment to each command that we are
running. This helps us to keep the environment a lot cleaner.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Add struct to pass more stuff to commands

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Create a callback function that is called in the child process

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

Revert "file: Remove ELF detection with libelf"

This reverts commit 65fcf2eb3c4fceb81887bd969e68ceae8613e26e.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

Revert "stripper: Throw this away"

This reverts commit 76774bf62a622da82ca589519331371c71b87bf1.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Add a test to check if the log buffer limit works

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

build: Check for matching files in a loop

When sending the filelist to scripts, we require the event loop to call
us again and again. This seems to cause a deadlock situation and we now
rather search for matching files in a loop until we have found a result
(or not).

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

packager: Sort files before writing the archive

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: log buffer: Test invalid function calls

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

log buffer: Check inputs on dequeue

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

log buffer: Since length is signed, we cannot assign -1

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Add simple tests for the log buffer

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

log buffer: Reset pointers when there is no more data

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: cgroups: Fix return code checks

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Add check for expected return codes

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Cleanup if we cannot create an eventfd

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Report correct error if clone3() failed

We should also clean up.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Recursively delete cgroups

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: string: Check if strings contain whitespace

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Return the correct exit code when a process has dumped

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

build: Ignore if cgroup limits could not be applied

This should ideally not happen in a real environment, but sadly is a
restriction we have in Jenkins when we are running the tests.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

build: Move cgroup accounting here

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Log the status code when a process exits abnormally

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: The static analyzer does not seem to understand how return things

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

path: Make the static analyzer happy again

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

xfer: Don't call fileno() with NULL

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

archive: Don't leak fd if archive could not be opened

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Enable accouting in the parent group

We cannot enable any of this in the child group because then we cannot
fork anything into it any more.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroup: Remove unused function to open root

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

build: Fail properly when we could not open a cgroup

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Implement recursion again

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Check if we received a valid file descriptor for the cgroup

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Don't automatically kill when destroying a cgroup

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Always log the entire path

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Use the correct function to retrieve the path

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Enable coverage tests again

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Log the entire path if we encountered a writing error

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jail: Don't fail if we cannot enable accounting

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Don't try reading memory stats when the memory controller is not enabled

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Fix error codes

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Close the FD after we removed the cgroup

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Refactor the code that creates a cgroup

The previous solution seemed to be very complicated, but this one will
likely have an extra syscall in some cases.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: os: Don't fail if no VERSION is set in /etc/os-release

ArchLinux does not seem to use that field.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Use the child function

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Bring back convenience function to create a child group

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Bring back a function that enables accounting

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Automatically enable necessary controllers

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Remove function to recursively create cgroups

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Don't try to enable already enabled controllers

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Keep bitmap of enabled controllers

This helps us to perform fewer operations when we create a cgroup.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Fix signedness of an integer

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Store the path to the cgroup

This is only used for debugging.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

path: Allow parsing empty paths

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Fix jail tests after changing the cgroups API

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Try to create some recursive dependency chain

This should help us to operate on cgroups since every group now has a
reference to its parent group.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Don't store the path to the root

We should not need this now that we always hold the parent groups.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Keep a reference to the parent group

Since so many operations are recursive, this might be easier to handle.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: cgroup: Update function calls

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

daemon: Create a cgroup for the daemon

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Rename pakfire_cgroup_open to *_create

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Implement recursively enabling controllers

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Remove the overly complicated handling of controllers

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Rely on support for cgroup.kill

This should be available in every kernel that we support now.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

cgroups: Minor code cleanups

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Build Debian packages only when we are in the master branch

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: On bookworm, make sure we have the correct version of cURL

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Don't try to cleanup the workspace

This seems to fail because Jenkins doesn't know on which host to run
this.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Scan for packages in the right place

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Add back --recursive to rsync

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Don't try to --archive because we don't have permissions

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Cleanup the workspace before mastering the repository

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Add principal name with kinit

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Try to cleanup workspace after Debian repo

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Publish files to the master mirror

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Try again to get a Kerberos ticket

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Export the public key

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Use single quotes when handling a credentials variable

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Try signing the Release files

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Skip checking our the source when mastering the repository

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Try importing the GPG key from Jenkins credentials

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Remove the GPG key material as soon as possible

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

jenkins: Try to fetch a fresh Kerberos ticket

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>