cli: pakfire-builder: Show more information in build mode

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

libpakfire: Make pakfire_has_flag() public

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

pakfire: Add flag to signal that we are running in build mode

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

cli: pakfire-builder: Remove double error message

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

cli: pakfire-builder: Add other basic commands

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

cli: pakfire-builder: Update help text

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

cli: pakfire-builder: Add repolist command

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

cli: Move pakfire/pakfire-builder into a single file

There is no point to have subdirectory for only one file when everything
else is going into the lib.

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

cli: Move any kind of actions into the library so we can re-use them

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

cli: Include headers of the library in a more explicit way

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

cli: Implement scaffolding for pakfire-builder

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

cli: pakfire: Implement --yes

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

libpakfire: Move loglevel into flags on create for Pakfire

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

libpakfire: Drop old CLI functions

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

cli: Implement interactive selection of solutions

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

parser: Do not compile with -fno-analyzer

The flex/bison-generated code is having too many problems to make this
output any useful.

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

cli: Implement the confirm callback

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

build: Check very early whether we can install the source package

It might still fail, but at least we tried to solve this here before
extracting the build environment and so on.

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

config: Move the strip functions into the string library

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

downloader: Log cURL stuff to the debug log

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

config: Fix read-out-of-bounds when stripping the entire string

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

file: Skip checking files in /usr/lib/valgrind

valgrind is a special case which we cannot harden very well, so we just
skip checking those files.

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

file: Do not fail if we did not find a dynamic section

Some files might not have these. That should not be an error, because
those files will fail their checks later on.

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

file: Silence a message that an interpreter has been found

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

string: Accept NULL as an empty replacement

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

config: Refactor data model and iterating over sections

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

repo: Fail if the baseurl could not be set

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

pakfire_repo_walk: Make sure r is initialized

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

repo: Remove useless logging line

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

downloader: Fix cURL logging

Log lines are not NULL-terminated and so we should not read further than
indicated.

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

logging: Free buffer only if set

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

string: Refactor replace function

The old version was slightly over-engineered and the static analyzer was
not too happy with it. Instead of debugging any problems, I wrote this
version which is a lot easier and should be more bug-free.

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

transaction: Initialize the job queue earlier

It could happen that we try to free the queue when it has not been
initialized.

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

downloader: Correctly initialize cURL

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

transaction: Fix writing the separator line

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

cli: pakfire: Initilize config structs for install/update actions

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

cli: Link against our own libpakfire library

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

cli: pakfire: Use transaction stuff to install packages

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

libpakfire: Fix loading repositories

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

cli: pakfire: Implement enabling/disabling repos on CLI

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

libpakfire: Remove the unused copy_in/copy_out functions

Fixes: #12973
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Bind-mount the test command to the jail

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

libpakfire: Remove pakfire_install/_update/_sync/_erase

These functions used to call a mega function to perform some basic
actions which was getting too complicated to keep alive.

This way, we export the transaction object which breaks down all those
actions into many smaller steps.

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

build: Implement a simple function to install packages in a shell

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

build: Implement installing build dependencies using a transaction

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

pakfire: Re-implement pakfire as a C binary

This patch removes the Python wrapper script which implemented the
"pakfire" command. Therefore we lose the runtime dependency on Pakfire
for basic system administration tasks and can hopefully remove a lot of
Python glue code.

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

libpakfire: Export pakfire_transaction_solve()

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

transaction: Drop DRY_RUN flag

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

transaction: Remove option to perform a dry-run

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

transaction: Automatically solve if not done, yet

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

transactions: Create a pick solution callback

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

transaction: Log any solving problems to the debug logger

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

problems: Refactor fetching problems and solutions

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

transaction: Merge requests into the transaction

It makes a lot more sense to have one large object that wraps the
solving process around libsolv instead of having multiple and passing
things around.

This is the first step to move things over and allow more flexibility
when performing depenedency resolution.

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

request: Obey policy when best flag is set

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

repo: Increase subpriority for commandline repositories

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

transaction: Changing packages is a re-install

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

request: Fix sync

The dist-upgrade job became a noop when the DROP_ORPHANED flag got set,
because it is not a flag but a separate job instead.

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

request: Implement locking the running kernel as a pool job

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

transaction: Silence debugging messaging saying which step to run

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

request: Automatically attempt to update packages on install

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

contrib: Drop priority from development repositories

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

contrib: Make the local repository preferred over stable

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

transactions: Major rewrite of step generation

When we extract from libsolv what steps we need to perform in order to
apply the changes to our system, there was a problem that when updating
packages, the previous version was not properly removed from the
database.

This has now been fixed by adding additional steps which are passive and
will remove the left-over metadata.

This should hopefully make the entire process more robust and easier to
handle in the code as each step has one simple job to do.

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

contrib: Remove old signature configuration

We do not support disabling the signature check any more and we no
longer have any host keys to sign packages.

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

repos: Implement subpriorities to prefer local packages when possible

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

transaction: Don't wipe dump when we change a package

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

transaction: Refactor reading userinstalled packages

This could potentially leak memory and was less robust.

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

dependencies: Don't log parsed dependencies

This is quite verbose...

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

transaction: Refactor code to dump a transaction

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

package: Treat sizes as unsigned long long

It looks like there have been some more casting issues which should
hopefully be fixed now.

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

compress-man-pages: Support already XZ compressed man pages

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

contrib: Disable the bootstrap repository

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

util: Fix casting issue when writing JSON

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

request: Log dependency issues if not in interactive mode

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

build: Build things with a lower nice level

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

jail: Don't abort if /etc/hosts or /etc/resolv.conf cannot be bind-mounted

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

repo: Fix compiler warning due to iterating over the wrong type

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

contrib: Update IPFire 3 repositoriy definitions

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

downloader: Fix JSON reference counting in mirror list parser

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

_pakfire: Correctly handle it no configuration was given

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

pakfire: Fix passing --config option to Pakfire

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

fhs.c: Correct permissions for /var/mail and /tmp

- /var/mail is looking for 0755 but dma is setting 02775
   checking FHS it looks like the settings should be 02775
- /tmp is being checked for 1755 instead of 01755 so the number is
   not being recognised as octal but as decimal
- The /var/mail change confirmed to allow dma to be built in ipfire3.x
   without failing for FHS errors

Tested-by: Adolf Belka <adolf.belka@ipfire.org>
Signed-off-by: Adolf Belka <adolf.belka@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

configure: Bump version to 0.9.29

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

file: Truncate file before rewriting it

It can happen that the file is shorter after we have fixed the
interpreter, which causes that the remaining data is still in the file.

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

_pakfire: util: Fix SEGV when we could not read from file

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

util: Fix potential double-free on error

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

pakfire: Create a ramdisk if no path has been given

This might be more useful when we struggle to cleanup the build
environment due to large IO. The build environment should generally be
small enough to easily fit into memory.

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

util: b64encode: Fix leak of p

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

config: Don't copy section names

This mainly makes the static analyzer happy :)

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

file: Rewrite fixing interpreters

The old version had some weaknesses in finding any arguments that came
after the interpreter which are being fixed here.

This version should also be easier to read.

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

file: Fix check if an interpreter has any additional arguments

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

file: Don't fail on non-implemented CF protection check

See: #13084 - Implement CF protection check
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

util: Fix potential NULL pointer dereference in pattern matching

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

package: Make code clearer where dumping deps

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

*list: Fix potential NULL pointer dereference

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

file: Fix pattern matching for files we don't scan

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

tests: Never compile the test command using the analyzer

We intentionally have some memory leaks and other bad things in here...

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

file: Initialize buffer when dumping a file

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

filter-requires: Remove check if files are on filelist

This does not work in this place because we are streaming the filelist
into the script and the check is being performed in Pakfire.

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