All Projects → HelenOS → Harbours

HelenOS / Harbours

Ported software for HelenOS.

Projects that are alternatives of or similar to Harbours

triton
Triton Operating System
Stars: ✭ 56 (+273.33%)
Mutual labels:  packages, operating-system
Simplefs
基于内存的简易文件系统实现
Stars: ✭ 9 (-40%)
Mutual labels:  operating-system
Nodeos
Lightweight operating system using Node.js as userspace
Stars: ✭ 6,576 (+43740%)
Mutual labels:  operating-system
Tantra
Hobby x86 unix-like kernel. Curiously under development to answer "How to write an OS".
Stars: ✭ 23 (+53.33%)
Mutual labels:  operating-system
Awesome Magento2
Curated list of awesome Magento 2 Extensions, Resources and other Highlights
Stars: ✭ 817 (+5346.67%)
Mutual labels:  packages
Repo
Arch Linux CN Repository
Stars: ✭ 929 (+6093.33%)
Mutual labels:  packages
Darch
A tool for building and booting stateless and immutable images, bare metal.
Stars: ✭ 701 (+4573.33%)
Mutual labels:  operating-system
Helloos
The Operating System is executed on x86 Intel CPU Instruction set. It is written in assembly language only.
Stars: ✭ 12 (-20%)
Mutual labels:  operating-system
Lowlevelprogramming University
How to be low-level programmer
Stars: ✭ 7,224 (+48060%)
Mutual labels:  operating-system
Shinyappdemo
A demo shiny app inside a package
Stars: ✭ 23 (+53.33%)
Mutual labels:  packages
Alchemy
A toy operating system written in Rust.
Stars: ✭ 16 (+6.67%)
Mutual labels:  operating-system
Satis On Heroku
Your private Satis instance on Heroku, just one click away.
Stars: ✭ 5 (-66.67%)
Mutual labels:  packages
Osinysuhomework
🚪Own experiments operating systems🚪
Stars: ✭ 8 (-46.67%)
Mutual labels:  operating-system
Raspberrymatic
🏠 A lightweight, buildroot-based Linux operating system alternative for your CCU3, ELV-Charly or for running your IoT "HomeMatic CCU" as a virtual appliance (using ESXi, Proxmox, VirtualBox, Docker/OCI, Kubernetes/K8s, Home Assistant, etc.) or on your own RaspberryPi, Tinkerboard, etc. SBC devices...
Stars: ✭ 803 (+5253.33%)
Mutual labels:  operating-system
Snacklinux
Linux distribution for lazy people
Stars: ✭ 9 (-40%)
Mutual labels:  operating-system
Cello
Operating System for Enterprise Blockchain
Stars: ✭ 715 (+4666.67%)
Mutual labels:  operating-system
Learning Mit 6.828
something you need in MIT 6.828 Fall 2019
Stars: ✭ 16 (+6.67%)
Mutual labels:  operating-system
Beelzebub
The Lord of Flies - A hobby operating system
Stars: ✭ 24 (+60%)
Mutual labels:  operating-system
Blog os
Writing an OS in Rust
Stars: ✭ 8,120 (+54033.33%)
Mutual labels:  operating-system
Minos
A minimal bare bones operating system based on a monolithic kernel design written purely in Assembly language
Stars: ✭ 12 (-20%)
Mutual labels:  operating-system

HelenOS coastline: build POSIX applications for HelenOS

This repository contains scripts that should simplify porting POSIX applications to run in HelenOS <http://www.helenos.org>_. The motivation for this mini-project is that porting GNU/POSIX applications to HelenOS is mostly about setting correctly the CFLAGS when running ./configure and then copying the generated executables somewhere to uspace/dist. The idea is that this procedure would be recorded in a form of a simple shell script, called harbour (because it is a port). The wrapper script hsct.sh then takes care of downloading the necessary sources and running the user-defined commands from respective HARBOUR file.

The whole idea is highly inspired by makepkg <https://wiki.archlinux.org/index.php/Makepkg>_.

WARNING: some information here is obsoleted. I will try to update it as soon as possible.

Using the coastline

First, clone this repository somewhere on your disk. The examples below would assume ~/helenos/coast.

Next, prepare the directory where the actual building of the POSIX applications would happen. It is recommended to do this outside HelenOS source tree and outside of the coast-line sources. ~/helenos/coast-builds/ia32 is a good choice if you plan to build for ia32 configuration.

Next, you obviously need check-out of HelenOS sources. In the following examples, I would assume ~/helenos/mainline.

To use the coastline, go into the ~/helenos/coast-builds/ia32 and issue the following command::

~/helenos/coast/hsct.sh init ~/helenos/mainline ia32 build

This command would initialize the build directory and forcefully rebuild HelenOS to the defaults of ia32 configuration. Once this command finishes, you can freely play inside ~/helenos/mainline as all the necessary files are already cached in the build directory.

Now you can build some software. Just choose one (for example, msim) and run::

~/helenos/coast/hsct.sh build msim

It may take a while but it shall eventually produce the simulator binary. If you want to copy it to your HelenOS source tree, run::

~/helenos/coast/hsct.sh install msim

If you want to transfer the built files to another machine etc, you may want to run::

~/helenos/coast/hsct.sh archive msim

that produces a TAR.XZ file in archives that you can directly unpack into the uspace/overlay directory.

If you have a multicore machine, you may try setting the variable parallel in hsct.conf inside your build directory to a higher value to allow parallel builds.

Writing your own HARBOUR files

The HARBOUR file is actually a shell script fragment. The coastline script hsct.sh expects to find some variables and functions declared in it. These variables declare URLs of the source tarballs or versions while the functions actually build (or install) the application/library/whatever.

Each HARBOUR is supposed to be in a separate directory. This directory is placed together with the hsct.sh script.

The commands in individual functions are expected to use special variables prepared by the wrapper script that would contain information about selected compiler (i.e. full path to GCC) or flags for the compiler to use. These variables are prefixed with HSCT_ followed by the name commonly used in Makefile\s or in configure scripts (e.g. HSCT_CC contains path to the C compiler).

However, usually it is not possible to write the HARBOUR file directly: for example various arguments to ./configure scripts have to be tried or extra CFLAGS might be necessary.

For testing, you can use the helenos/env.sh script and source it. This script sets all the variables that you can use in the HARBOUR script.

Follows a shortened list of variables available.

  • $HSCT_CC: C compiler to use.
  • $HSCT_CFLAGS: C flags to use.
  • $HSCT_LD: linker to use.
  • $HSCT_LDFLAGS: linker flags
  • $HSCT_LDFLAGS_FOR_CC: linker flags preceded by -Wl,. This is extremely useful when linker is not called explicitly and compiler is used for linking as well. Some of the flags would be recognized during the compilation phase so marking them as linker-specific effectively hides them.
  • $HSCT_GNU_TARGET: Target for which the application is being built. Typically this is the value for the --target option of the configure script.
  • $HSCT_INCLUDE_DIR: Points to directory for header files. This directory is shared by all packages.
  • $HSCT_LIB_DIR: Points to directory for libraries.
  • $HSCT_MY_DIR: Points to installation directory. All files that shall appear in HelenOS must be copied here. The structure of this directory shall mirror the HelenOS one (i.e. the app/ and inc/ directories).

For example, the ./configure script for libgmp <http://gmplib.org/>_ uses the following variables::

run ./configure \
	--disable-shared \
	--host="$HSCT_GNU_TARGET" \
	CC="$HSCT_CC" \
	CFLAGS="$HSCT_CFLAGS $HSCT_LDFLAGS_FOR_CC <more flags>" \
	LD="$HSCT_LD"

Once you know the command sequence that leads to a successful built you should record this sequence into the HARBOUR file. The easiest way is to take an existing one and just change it for the particular application.

The variable shipname declares the package (application or library) name and shall be the same as the directory the HARBOUR is part of.

The variable shipsources contains space separated list of tarballs or other files that needs to be downloaded. Obviously, you can use $shipname inside as shell does the expansion. To simplify updating of the packages, it is a good practice to have variable $shipversion containing the application version and use this variable inside $shipsources. If you need to reference a local file (a patch for example), just write a bare name there. The files are downloaded with wget so make sure the protocol used and the path format is supported by this tool. Git repositories to clone can be specified with the syntax git:localdir:[revision]:url where localdir is the name of the local directory to which the repository is cloned. revision is the revision to check out and url is the URL of the git repository.

The variable shiptugs declares packages this one depends on (the twisted fun is here that tugs are required for the ship to actually leave the harbour). That is a string with space separated list of other ships.

For building is used a build() function. The function is expected to complete the following tasks:

  • unpack the tarballs
  • configure the application or somehow prepare it for building
  • actually build it

If you want to print an informative message to the screen, it is recommended to use msg() function as it would make the message more visible.

To simplify debugging it is recommended to run commands prefixed with function named run. That way the actual command is first printed to the screen and then executed.

Below is an example from libgmp that illustrates a typical build() function::

# Manually extract the files
run tar xjf "${shipname_}-${shipversion}.tar.bz2"

# HelenOS-specific patches are needed
msg "Patching gmp.h..."
patch -p0 <gmp-h.patch

# Run the configure script, notice the extra C flags
cd "${shipname_}-${shipversion}"
run ./configure \
	--disable-shared \
	--host="$HSCT_GNU_TARGET" \
	CC="$HSCT_CC" \
	CFLAGS="$HSCT_CFLAGS $HSCT_LDFLAGS_FOR_CC -D_STDIO_H -DHAVE_STRCHR -Wl,--undefined=longjmp" \
	LD="$HSCT_LD" \
	|| return 1

# The variable $shipfunnels reflects maximum parallelism allowed
# by the HARBOUR and by the current build directory
msg "Building the library..."
run make -j$shipfunnels

# Tests are built and run as one target so this target always fails
# We check that the tests were built by explicitly checking for
# them below.
msg "Building the tests..."
run make check || true
(
	cd tests
	# Check that all tests were built
	find t-bswap t-constants t-count_zeros t-gmpmax t-hightomask \
		t-modlinv t-popc t-parity t-sub
	exit $?
)

After the application is built, it can be either archived or copied to HelenOS source tree. Both these actions requires that the application is packaged first.

The function package() is expected to copy the necessary files outside of the build directory into $HSCT_MY_DIR. If there are some headers or libraries used by other packages, they should be copied into $HSCT_INCLUDE_DIR and $HSCT_LIB_DIR.

Directories $HSCT_INCLUDE_DIR and $HSCT_LIB_DIR behave as standard Unix-like /usr/include and /usr/lib directories, while $HSCT_MY_DIR mirros the HelenOS directory uspace/dist structure. Contents of $HSCT_MY_DIR is copied to uspace/overlay during installation or tarred when archived.

Below is an excerpt from zlib package() function. Notice the usage of the variables and the run() function::

cd "${shipname}-${shipversion}"
run make install DESTDIR=$PWD/PKG

# Copy the headers and static library
run cp PKG/usr/local/include/zlib.h PKG/usr/local/include/zconf.h "$HSCT_INCLUDE_DIR/"
run cp PKG/usr/local/lib/libz.a "$HSCT_LIB_DIR/"

run mkdir -p "$HSCT_MY_DIR/inc/c"
run cp PKG/usr/local/include/zlib.h PKG/usr/local/include/zconf.h "$HSCT_MY_DIR/inc/c"

run mkdir -p "$HSCT_MY_DIR/lib"
run cp PKG/usr/local/lib/libz.a "$HSCT_MY_DIR/lib"
Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].