tests: Remove jail tests

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

pakfire: Make the pool status a flag

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

pakfire: Drop the private dir

This is not being used.

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

pakfire: Reorganise initialization

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

tests: Drop the old snapshot tests

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

snapshots: Remove the overly complicated struct

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

pakfire: Make the umount code clearer

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

pakfire: Fix incorrect reference counting

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

snapshot: Ensure we only umount once

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

pakfire: Refactor how we mount the ramdisk, and snapshots

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

libpakfire: Remove non-existant function from headers

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

snapshots: Tell the mount operation where to mount the snapshot

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

pakfire: Tidy up the code to create a new pakfire instance slightly

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

snapshots: Destroy directories after umount

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

snapshots: Remove the old-style snapshots

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

build: Remove old-style snapshots

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

cli: Fix enabling the snapshot

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

pakfire: Experimental implementation to use overlayfs for snapshots

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

tests: Make xfer tests compile and run again

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

repo: Cleanup argument order when downloading packages

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

httpclient: Fix error handling when leaving the main loop early

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

httpclient: Don't run the main loop if there are no transfers

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

httpclient: Don't fall through when trying to reset the timer

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

httpclient: Terminate the event loop if there are no transfers left

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

xfer: Keep a reference to ourselves

This is necessary so that the handle won't be cleaned up when the last
reference to the xfer is gone. It should be good enough to add the xfer
to the HTTP client and then wait for everything to complete.

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

archive: Be less noisy if we cannot open an archive

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

archive: Improve error handling when opening an archive

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

xfer: Only copy the URL once

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

xfer: Fail creation when in offline mode

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

httpclient: Remove references to downloader

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

httpclient: Remove unused cURL import

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

repo: Refactor the xfer creation function

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

xfer: Only have one way to create a new xfer

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

xfer: Decouple more from HTTP client

We used to require a HTTP client to create an xfer but we are actually
not using it any more. That way, we can avoid some more complicated code
paths and we can keep a single request and multiple requests further
apart in the code.

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

ctx: Move the cURL share handle here

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

python: Reformat some code

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

python: Add __version__

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

python: Don't mess with the Gettext configuration

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

python: Drop the native Python code and move the C module

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

pakfire: Simplify storing features

We don't need a struct here because we only store a list of strings.

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

pakfire: Remove the unused locking mechanism

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

build: Tell the dependency callback how many bytes have been consumed

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

packages: Fail when trying to add invalid dependencies

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

pty: Consider ourselves drained if the master fd is closed

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

pty: Initialize/reset file descriptors with EBADF

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

pty: Remove closing an unused file descriptor

This was never used and not initialized, so it closed standard input.

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

log stream: Reset file descriptors to EBADF

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

log stream: Avoid infinite loop

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

cgroup: Initialize file descriptors

It was possible that those closed the standard input.

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

log stream: Remove FILE handle

This is being replaced by our own loop that splits the input by line.

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

jail: Connect parent side of the logger

This is now possible as we have the new event loop.

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

log stream: Correctly check for open file descriptors

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

jail: Wait a little bit later in the child process

That way, we can set for the child process to die before we start
waiting.

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

cgroup: Parse more stats

This is to reduce logging errors when we could not parse a line.

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

parser: Make this less verbose

The parser generates so much logging output that I cannot find anything
in the logs any more.

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

pakfire: Don't fail when looking up empty UID/GIDs

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

pty: Don't try to map empty output buffers

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

jail: If no output handling has been configured dump stuff on the terminal

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

jail: Bring back sending the build output to the logger

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

jail: Setup shell environment in the shell function

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

jail: Always set PS1

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

util: Drop unused function to check if the terminal is a TTY

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

jail: Drop the interactive helper function

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

jail: Make the entire jail private

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

python: Remove Pakfire.execute()

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

jail: Remove an unused buffer struct

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

jail: Replace standard input/output callbacks with the PTY

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

pty: Explicitely request to connect stdin/stdout

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

cli: build: Send all log messages to syslog, too

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

pty: Add callbacks to stream input and output

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

jail: Remove functions that capture the command output

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

configure: Don't check for epoll any more

We don't call this directly any more.

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

pty: De-register events before closing the file descriptors

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

pty: Don't terminate the event loop on error

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

tests: Show return code on failed assertions as well as errno

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

pty: Implement capturing the output

This should be the most efficient version, yet.

We create a memfd buffer which we write to and later copy the output to
the heap. This also reduces the amount of callbacks that we have to pass
around.

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

pty: Refactor forwarding

This should make the process slightly more efficient because we are
checking if there is anything in the buffers before entering the loop.
We also handle any return values from the read()/write() calls better so
that we won't continue to loop too often.

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

arch: Dynamically create an array with the supported arches

Clang did not like the previous solution.

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

tests: Fix string formatting in command.c

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

configure: Remove option to enable the static analyzer

This is not compatible with Clang.

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

configure: Correctly define CURL_HAS_WEBSOCKETS

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

configure: Fix cURL WebSocket check

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

configure: Don't insist on a specific version of JSON-C

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

xfer: Fix check for cURL

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

repo: Fix string formatting

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

jail: Fix formatting strings

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

python: Update function declaration for the log callback

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

python: Update initialization of types

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

python: Correctly initialize and cast keyword arguments

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

cli: Declare command outside the switch statement

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

cli: Fix broken include guard

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

xfer: Compile even if cURL does not have support for WebSockets

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

xfer: Support compiling with older versions of cURL

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

transaction: Don't declare anything after the label

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

libpakfire: Set __attribute__((format(...))) wherever needed

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

build: Replace the bitfield with a state enum

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

file: Don't declare hardlink in the switch statement

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

log stream: Initialize r in case events was zero

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

os: Make more space for the distro information

ArchLinux is using a longer version_id with timestamps which is why we
need some more space here. There is no point in being too tight with
this.

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

configure: Drop dependencies on Python modules

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