Index: README (this file) | Reference | Short Reference | minimake | makefile | about | functions | functions shortref
CodeQL(master branch):
-
-
TOC: Summary | Security | Thread safety | Branches | About | Dependencies | Usage | Bugs | See also | Notes
A c system library with a focus on size, headeronly, "singlefile", intended for static linking. 187 Bytes for "Hello World", regular elf binary compiled with the standard gcc toolchain.
Currently the latest bundled development version is stored at
https://github.com/michael105/minilib/tree/download
Copyright (c) 2012-2021 Michael (Misc) Myer, misc.myer@zoho.com / www.github.com/michael105 BSD-Style license, with attribution. Static linking is permitted. Commercial use is permitted. Some sourcefiles, not part of milib itself, in the folder contrib do have other licensing terms. "ports" contains other software, not written by me, but ported to compile with minilib. Please look in question into the sources.
Minilib aims to be an incomplete libc replacement,
single header only, as simple and small as possible, KISS,
but supplying ANSI-C89, the most important tools of libc,
and some POSIX/C11 addons (Which don’t add bloat and are optional).
The build can be customized by define switches,
compiled are only the explicitely enabled sources.
The main targets are simplicity and small memory footprint.
Security is in mind as well. Having a small codebase
might be of most importance in this context; the sources can be read through
and checked. (As opposed to, e.g., glibc,
where reading the sources is highly theoretical.)
In order to have all compiled sources combined and dumped to the terminal,
type either SHOWSOURCES=1 make -f Makefile.minilib CONF=yourconfig.conf,
or minimake --config yourconfig.conf --showsources.
Albite the header file is parsed and compiled for every
source file, this takes only fractions of a second
at my development system (amd Turion 2Ghz),
at the same time giving full advantage of the possible
linktime optimizations.
Please be aware that the ANSI and POSIX standards are inherently
insecure. (e.g. scanf, gets, operations on memory in general, ..)
There is the possibility to use gets and other problematic functions with
guarded memory mappings (map_protected).
However, minilib is a system library, so it would imho be neither useful nor
wise trying to fix all 'security problems' by patching problematic functions, andsoon.
You’ve to take care yourself for handling e.g. user input in a secure manner.
For example, using gets for user input, worst case with a variable located
at the stack, is a bad idea. (Buffer overrun, and all sort of stack manipulations are possible).
Using gets for interprocess communication with pipes would be ok,
when the input can be guaranteed to stay within the specifications.
Same with, e.g., configuration files, which obviously shouldn’t
be possible to manipulate.
It cannot be within the targets of a system library trying to prevent developers
from doing anything. This will neccessarily be some sort of patchwork,
but you simply cannot patch all holes. Same time eating up resources,
and limiting the freedom of the developers severely.
(However, I am about to put compiler warnings into minilib for problematic functions
like gets, and a switch to en- and disable the warnings)
And there is the possibilty to put the minilib internal buffer, and the globals,
on (after) the stack with the switch globals_on_stack.
Albite within, e.g., printf, there are tests for buffer overruns,
in some special cases, an overrun could still happen.
To prevent the possiblity of overwriting environmental variables, I’d suggest
reading and copying all environmental variables first,
before processing any user input.
Anyways, it might be the best practice to avoid buffer overruns.
By always checking, for example, user input,
avoiding inherent insecure functions like sprintf, gets, or using these
only in guarded memory areas (map_protected).
Again, when in doubt, showing ALL compiled sources is possible with
minimake --config yourconfig.conf --showsources.
minimake --config yourconfig.conf -S will save the generated assembly into
'sourcefile.s', which still is readable.
Having readable source is IMHO the biggest advantage, especially for security related code.
For the same reason I started to put checksums into the generated configuration files;
to ensure, that some version (which might be already checked, not only by the
security scan of 'codeql', but as well by other hackers) stays the same.
I once tried to debug an internal glibc problem. It’s been a nightmare.
And to be honest, there’s not only the problem, that there could be something hidden
within the library, it is also known that sometimes flaws only show up
with e.g. a combination of gcc and glibc. Worse, there have been flaws
by code, which should ensure security - just to bring in new vulnerabilities
by the added complexity. (e.g. the stack protector/canaries)
Due to its structure, minilib is 'inherent' threadsafe.
(have to document the different possiblities of using threads with minilib)
As long you do not define globals,
(which are only needed for iostream functions, printf, vsprintf, malloc, mallocbrk),
all functions are (obviously) threadsafe.
When using globals, or the global buf, printf, vsprintf, malloc and mallocbrk
are not thread safe, as well as operations on streams. (Might be obvious, as well)
There is the possibility to create a new thread with the flag CLONE_VM,
what will make all functions "threadsafe". Except operations on streams.
Memory is copy on write with CLONE_VM,
there is more documentation at the according linux manpage about clone.
minilib development is separated into several branches.
Within the branch download the assembled ("compiled") 'minilib.h',
a Makefile ('Makefile.minilib') for building with 'minilib.h',
and a script, with 'minilib.h' embedded, 'minimake' are stored.When you’d like to build minilib itself, you’ll
need either the branch master or devel.Within devel-HEAD there are also some ports to minilib, and 'minicore' stored,
these are however work in progress and mainly there for testing minilib.
Not all things stored in devel-HEAD are supposed to build correctly,
or work the way intended; especially the "ports" are work in progress.Linking and compiling with minilib is the same process with all branches.
Minilib evolved out of some assembler experiments,
where I realized how tiny executables can be, if not linked with glibc.
With some more fiddling I got a C "Hello World!" binary with 151 bytes,
compiled and linked with the standard gcc toolchain.
This is more than 3,000 times less than a hello world program
linked statically with glibc,
which shows up with 504kB on my system. (5.4kB linked dynamic)
(On OSX there are some limits,
files with less than 4k aren’t executed,
so that’s hello world’s size there..)
minilib saves a lot of startup overhead,
no linker needed to bind shared libraries,
no lookup tables, less memory fraction;
also (small) binaries might often fit into a single harddisk block;
opposed to a binary and several shared libraries.
This gives an overall more responsive system.
Just for fun, but this minilib also beats klibc, dietlib and musl
in the matter of hello world’s size..
(Linux 32bit:) musl: 7.6k dietlibc: 672 klibc: 367 minilib: 151
(On Linux 64bit I didn’t check the other libraries,
but the bigger registers cost a few bytes:
hello world counts here 185 Bytes.)
Being in syscalls, linker scripts and gcc optimization hell,
I realized, that after I got over the fundamental problems adding
more standard c routines would be cheap.
So I started to port some more little c tools to minilib,
adding more functionality to minilib as well.
Since I hate todays usual overhead in modern libraries,
resulting in the unbelievable size of hundreds of Megabytes
for simple applications as well as minutes for just starting them up,
I also tried to optimize functions like printf where sensible.
The focus is on size, however.
You are able to customize the compilation of the library
and define only the functions you need;
if you are in need of, e.g., a more speed optimized strcmp,
just don’t define the according switch and rule your own.
Albite the header file is parsed and compiled for every
source file, this takes only fractions of a second
at my development system. (amd Turion 2Ghz)
> ~/prog/minilib $ time ./minimake --ldscript onlytext -o hello hello.c ./minimake -o hello hello.c 0.06s user 0.05s system 92% cpu 0.116 total > ~/prog/minilib $ ./tools/shrinkelf hello stripping: hello > ~/prog/minilib $ ls -l hello -rwxr-xr-x 1 micha micha 185 Aug 15 16:16 hello
This whole project exaggerated out of fun and passion.
-
Needs:
gcc / clang
gcc 9.3 works, 10.x seems to sometimes build segfaulting binaries
(I developed yet mainly with gcc 9.3)
clang 11 seems to work. 12 untested. -
Building with Makefile.minilib:
sed, gzip, curl or wget for automatically downloading the gzipped minilib -
When using minimake:
bash, sed, gzip, sort -
Rebuild minilib:
make, perl -
generate documentation:
perl, asciidoc, rst2html5, markdown
Atm, it might be easiest to have a look into the directory examples,
and the Makefile as well as the sources of the different versions of
"hello world" there, as well as "QUICKSTART".
There are several possibilities to use minilib.
Easiest route: use the supplied script minimake, with the argument --autoconf.
minimake --autoconf -o hello hello.c
(or, for Makefiles utilizing the CC environmental setting:))
export CC=minimake --autoconf
This is save only for projects, consisting of single source files.
minimake will create a minilib config within 'minilib.conf', parse this config
and pipe the resulting define flags to gcc.
The source file is parsed and every function used enabled in the minilib config.
However, e.g., conversions used within fprintf aren’t recognized,
and therefore not going to be compiled. (itodec, atodec, for example)
So, in most cases it might be better to create a config file,
edit the config and then use this config file:
Generate a skeleton file, and try to recognize used functions:
minimake --genconf hello.mconf hello.c
Edit the config:
vi hello.mconf (the config file is self documented)
Finally use the config for the compilation:
minimake --config hello.mconf -o hello hello.c
In both cases, it is not neccessarily needed to modify the source files.
"minilib.h", and the config switches are piped to gcc;
the compile and linker flags are passed from minimake.
The standard libc headers are prevented from being included
by defining the according headerguards.
When compiling without --autoconf or --config, you have to
include minilib.h in all sourcefiles, needing definitions from minilib.
You are able to define functions you need yourself, before you include minilib.h.
Example: #define mini_write
(Or) use the config framework, implemented within the script minimake.
The config framework has the big advantage of raising syntax errors for e.g. mistypes,
so instead of having to look for the mistype, it’s shown up even before the compile.
In one of your sourcefiles you also have to define "#define mini_INCLUDESRC",
before the "include "minilib.h" directive.
Either use the supplied Makefile.template by copying it;
or set the environmental variable CC 'export "CC=[path to minilib]/minimake [--config configfile.conf]"'
There is further documentation in the chapter 'configuration' of the reference.
Reference
Currently only Linux 64bit is supported,
and gcc v9.3.
Other versions of gcc result sometimes in segfaulting binaries.
Bugreports and patches, preferred as github issues, are very welcome.
clang 10/11 works mostly, but several combinations of functions and optimization flags
give trouble. Please see the section BUGS.
* 2021/10 Currently trying to get closer to posix/ansic. Did make the mistake of following other implementations, and I didn't think of minilib being able to replace the standard libc. This gets close. The codebase and configuration system gets stable, and I'm trying to keep backwards compatibility. More software is linkable to minilib without changes. Testing is needed. I regard the overall state still as alpha, due to the incompatibilities with posix/ansi. Which are sometimes hard to recognize and track. The "fulldebug" functionality showed up handy. All reports are very welcome. Going to do fuzzy testing and compare different libc implemenatations. (:/ some calculation time would be welcome as well, I'm looking for a sponsor) * 2021/06 About to prepare a release. Adding some security related stuff atm, and I'm trying to prepare a combined package management system. The security scan of CodeQL showed up one (!) flaw, within 30.000 loc's. Which even has been more or less theoretical, and would have most likely resulted in a segfault. I'm trying to get clang compiling minilib. (Also cause of security considerations) * 2021/05 Pushed the devel branch to master. Suffering at the moment some bloating, hello world now comes with 241 Bytes. used to be in the 64bit version around 180 Bytes, 32bit 151 Bytes. I have to look for the problem. Ansi C89 is done to 95%. Posix C 1003.1 around 80%. Several standard library functions for network access added. No stdio buffering yet. Testing, also of master, is needed. * 2021/02 Implemented more of posix. Atm., cleaning up. Guess I need to find a way to clearly separate the minilib standards implementations (posix, ansi) from addons. (eget fgets, ISO c11, and fgetd ( get a digit from stream ) - nonstandard ) Need to rewrite this readme. * 2020/08 A few notes about malloc. The current implementation of malloc aims at being tiny. Very low memory overhead (4Bytes per malloc), no usage of sycalls, no mmap.(!) It uses a preallocated buffer, which has to be defined at compile time. The buffer can be located in either the bss section, so it is allocated by the kernel, along with other globals. Since in most cases there is a page allocated for this section, the space is allocated in each case, and can be used by the "minibuf". The buffer can be located on the stack as well, via the option "globals_on_stack"). Sparse (allocated and free'd) areas are not reused. Instead, when all areas below, say, X, are freed, they are freed in a whole. The intention here is on tiny tools, which only allocate a few variables, where the sizes are more or less known, and below, say, 64kB. It is not the right malloc for other applications. There are two "mallocs" now, which use dynamic allocation: map_protected, which allocates one (or serveral) page(s)(here 4kB per page), and guards the page(s) with a protected page before and after. The intention is having a memory area for user input, where an over- or underflow won't e.g. overwrite return adresses at the stack. And there is "malloc_brk", which allocates space via moving the program's break. malloc_brk, however, doesn't keep track of free'd areas. Here the intention is to 1. Be able to free several allocated areas in one call, by setting the brk back again. And 2., when a malloc'ed area is reallocated, it is possible to simply enlarge the area, without copying. This is used of e.g. in 'scandir', where the contents of a directory are read directly into a mallocated area, and when the allocated buffer shows up to be too small, the allocated area is simply enlarged, until the directory is read in. Since the function sorts the entries (if requested) after the read in, all names have to be kept in memory. There is only one copy of the data done, from kernel to userspace. (When no filter is used). And it is not neccessary to free every direntry, instead, a free of the whole list of dirents can be done with one call. However, again, this obviously will work only, when the whole area can be freed at once. (Subsequent calls to malloc_brk will prevent freeing the areas before). Atm I'm working at a 'usual' malloc. (Pooled, free lists, and so on). When I'm going to finsh this, I cannot say now. I simply do have other obligations as well. However, it is possible to use any malloc implementation, you can find. 1.The only funcion, which uses a malloc within minilib at all, is 'scandir'. (Which uses malloc_brk.) All other functions use the predefined "mini_buf". (e.g. printf, conversions, andsoon) 2. Furthermore the curent malloc implementation of minilib does use the fixed buffer "mini_buf", and no calls to mmap or brk are made at all. Therefore, you can plug in any existing malloc implementation you wish, without having to fear, this will interfere in any way. (If you do not use 'scandir') (E.g. jemalloc, demalloc, and so on. Just do a search at github, there are countless implementations, many very mature). * Seems I finally found the "big bug"; obviously gcc did optimize sometimes(!) parts(!) of system call procedures out of his way. Not anymore. Added an optimization fence. I say, this is a compiler bug. Setting register assignments and assembler statements volatile should execute them, at the very location, no matter of the circumstances. * 2020/07 Ansi C89 is done to about 98%. Not counting floating math and localization functions, which are out of the scope of minilib. The compat headers might be more or less broken, they aren't needed for the usage of minilib and just there for convenience. Possibly I'm going to remove them. Only x64 might compile now. Going to backport as soon more important things have been settled. * 2020/05/23 Rewriting the makefiles. (having both options, either use minimake or minilib.h / makefile.template, makefile.minilib) Pushing forward to ansi-c89 compatibility. linux x64 only, atm. * 2020/01/15 Finally the build system gets more consistent. The whole minilib is now included within the script minimake. So, fetching just this script is enough to build and link with minilib. Documentation might also be better to read, implemented functions are listed in link:mlfunctions-shortref.asc There are still some functions left, but it's close (around 80%) to ansi-C89 conformance. (Without mathematic functions) There are, however, already specialized mathematics libraries out there, so these routines don't have a high priority. To the posix-c specification, there might be around 50% missing. Atm, I'm porting, sometimes writing, basic unix tools to minilib. Many of them depend on posix-c (or have the same functionality), so I'm adding those functions extending ansi-c to minilib as well. Uh. And a happy new year, everyone! ;) * 2019/08/27 Documenting. Preparing a new release, amd64 for now. The copied code from musl for opendir, readdir and family - Obviously bloating. (there's a fixed buffer for the entry names of each dirent, size 2kB. Not tolerable.) So, todo: rewrite. Also todo: write something like config groups. e.g.: fprintf might in most cases also enable atoi and itodec. (%d) Maybe make the config process more verbose. I guess, for now it's better to clean up the master tree and do the release. * 2019/08 restructured the build system. wrote minimake, a wrapper around gcc. Close to be a dropin replacement for glibc, without the need of changing any sources when compiling with minimake and minilib instead of glibc. * 2019/04/02 Overall, minilib gets into a more stable state. The build framework as well as the config framework look good to me. And it's generic enough to don't get into trouble, when there something has to be changed. Ansi C (C89) / is about 75% done. Some functions (fopen/fclose) only trivially tested. Atm, I'm working on getting more tools of the 'suckless linux base' and the 'minutils' compiled with minilib. Besides, I'm fiddling around with some security and performance related fundamental matters. Stack and elf section related; but these questions are optimizations and no fixes. (Or, to be more exact, these questions are about fixes of existing security holes.) * 2019/03/28 Restructuring done, config system written. The documentation about the build process isn't on state. The examples might be self explaining, but I have to rewrite the docs. * 2019/03/19: Ok. about half of ansi c is done. some important functions missing (free, e.g) but close. Restructuring nearly done. Overall - quite close to a first real release. Yet I could also maintain backward compatibility. (Important to me, since I'm using this lib for nearly 7 years myself..) * 01/2019 Got minilib running on OSX (10.11, x64), enabling static linking for small tools. Anyways, being able to link small tools statically on osx is a great thing. This doesn't work with the native libs, and there are reasons, you might want to link statically. Hacking the OS, or executing security critical code, for example. If the kernel has been hacked, well... But statically linking prevents from lib injection. BUT PLEASE BE AWARE, minilib itself is neither checked nor written for security. Although there are some tests, e.g. for buffer overflows, I cannot guarantee any defined functionality. But I worked thoroughly, and I'm using this lib myself for several years.
Minilib aims to be an incomplete libc replacement,
as simple and small as possible, KISS!!!,
but supplying ANSI-C, the most important tools of libc,
some addons (Which don’t add bloat..).
Some syscall wrappers.
One reason to use syscall wrappers is type safety.
(The kernel doesn’t know about type safety,
he simply grabs the arguments he’s fed with.)
Another reason is the problematic of having thecompilers
sometimes parts of the code optimized out,
when working with inline assembly.
Minilib shall be portable.
For now there is Linux x86_64, x86_32,
and OSX Darwin.
(targeted. The ports might not compile now,
I cannot check atm, since my macbook is broken
and other things are important to me as well.)
Android is targeted.
BSD is targeted.
Minilib should stay tiny.
Ideally minilib is going to supply glibc compatible
headers. (sort of)
This library is still under heavy development.
Albite now (2021/06) settling. Ansi C89 is done to 98%
(not counting localization and floating math)
Some of the most important functions of Posix C are implemented.
Testing in different environments is needed.
I developed this library all over all in more than 15 years,
and I’m using it myself in several projects.
However, I cannot give any guarantees for an opensource library.
There is some infomation on the development of minilib itself
within doc/minilib-devel.textile
Adding more syscall wrappers is simple.
Please have a look at include/lseek.h for examples.
As well as include/syscall_stubs.h.
(The file with definitions of simple syscall wrappers)
For bugfixes, please open an issue at github.
(github.com/michael105/minilib)
CLANG:
- strdup doesn’t work with optimize -Os
- fork doesn’t work with -Os,-O2
- signal/abort doesn’t work with -Os,-O2,-O1
- prints (vararg) doesn’t work with -O0 (!??), but O1,O2, Os are ok.
GCC:
versions > 9.3 untested.
→ https://elinux.org/System_Size - Embedded Linux and size
→ http://muppetlabs.com/~breadbox/software/tiny/teensy.html - About creating tiny binaries
→ https://olimex.wordpress.com/2012/04/04/unix-on-pic32-meet-retrobsd-for-duinomite/
- running bsd with 128K RAM (!)
→ https://www.mkompf.com/cplus/cliblist.html - Ansi standard c
→ https://www.etalabs.net/compare_libcs.html - comparison of c/POSIC standard library implementations
tiny c lib implementations
klibc bionic musl uLibc dietlibc minibase embedded artistry libc
c standard libraries
musl uClibc glibc
"It’s sometimes smarter to stay special,
although in this case this means the opposite"
Regarding my (strangely simplicistic) malloc implementation. Which fit's quite well to the minilib in whole. Being smart with utf-anything, don't know which bordercase and so on - is sometimes overall not smart.
So, having a 500kB "hello world" like with glibc, is not soo smart. Especially when it needs to be compiled again at the target, so most bordercases aren't of any interest.
151 Bytes, as accomplished with minilib - There's still some overhead. But it's another dimension.
Since I do need to present myself to non developers as well,
I’d like to say here a few sentences about minilib,
which might be understandable, also if you aren’t engadged in
system development.
Overall, minilib got a quite comprehensive project.
More than 30.000 lines of code,
when counting comments as well,
and only the folders src, include, asm and scripts.
This is quite a lot, pessimistic people do say,
a developer is able to write one line of code per day.
Well, me I’m doing more. What has, to be honest, to do
with the fact, that I didn’t need to exchange and communicate
with other developers. And it is a pessimistic view.
Anyways, if you would like an estimation of the development costs of
something like minilib in money, the number might have 6 0’s, in Dollars.
(Not accounting the fact, more than half of all software projects
do not finish..)
However, to say that again to non opensource developers;
it would not be possible to sell this software for that amount of money.
There are already several other opensource projects with the same target,
and the whole software landscape did change the last decades.
Besides the fact, this project would need to be different, when
having a commercial focus.
So the value on the market might be 0,
noone would pay money for something,
what is neither finished, nor stable,
and available in other (stable and mature) implementations for free
as well.
It’s comparable to my academical "book",
written for the dregree "Magister Artium" in philosophy.
Would be quite expensive to write as an order.
But nearly impossible to sell.
I am however looking for sponsors, or partners,
since it "costs" me more and more to engage myself in the development of minilib.
What on the other hand means some sort of compassion and contemplation to me.
Essentially, the basic libc is half of an operating system.
(Again, not nowadays, all the different hardware has to be accounted for and
needs different drivers, but, as I said,
I’m just trying to make the project understandable)
I have written more than 95% of the sources on my own.
(I know. It is a passion of mine..)
And it is not my first bigger project, this passion accompanies me
for more than 25 years. .. Some people to build mini railways,
me, I’m building software.
A few ideas and solutions are genuin.
E.g., the config system, the "regular expression" engine,
or some parts of howto fit this all together.
But most is just focussed on being of small codesize.
However, in turn, having the possibilities to fine grain the configuration
of the builds, and having executables with less than 200 Bytes
with the standard gcc toolchain didn’t exist yet.
I didn’t do this "just for fun".(Quoting Torvalds)
I did have fun, but it is also a reference, and I did learn some things
about system development and executables in a deep fashion.
(I did some bigger projects before, also commercially,
but this has been more than 15 years ago)
By the way.. My Macbook is broken. :(
If you’d like to sponsor minilib,
a Mackbook (also a used one) would be one of the things
I’m in need of, and help me alot. :)
If you have some other hardware, you’d like to see minilib running at,
just contact me. Most of minilib is portable down to 8 bits.
Something like an arduino, or a banana pi are on my wishlist, too.
I do not know, where this whole project is going to go.
I’m already about to rewrite the basic system tools,
and trying to get a usable "unix" system within 64kB.
(without the kernel)
This is, again, "outstanding". Other projects
with the same focus do have around 1MB.
But again, it is only possible, since I do strip
the functionality down to the real and mostly used basics.
I do have ideas about a linux distribution, or something like that.
But. Who knows. I do have other obligations and ideas as well.
Michael (Misc) Myer, 2012-2021
misc.myer@zoho.com www.github.com/michael105
(my "programming enthusiastic persona"..)
Copyright (c) 2012-2021, Michael (misc) Myer (misc.myer@zoho.com, www.github.com/michael105) Donations welcome. All rights reserved.
I did this minilib out of enthusiasm.
One side, I`d like to have this project being opensource.
On the other hand, I spent really much time with the development.
In an ideal world, someone would sponsor me for this time.
It is, however, not an ideal world we are living in.
Therefore I do have to insist on attributions and what
I`d like to call fair use.
In short, you are allowed to do with this project what you wish,
also commercially, as long you show the copyright notice at a prominent place,
and include the file NOTICE in distributions.
That`s the intention of the license,
which I`d like to name "Fair use with attribution".
Please have a look into the file LICENSE for the exact licensing terms.
Disclaimer:
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL Michael Myer BE LIABLE FOR ANY
DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Please contact me, if you are in need of differing licensing terms.
Some files in the folders headers and contrib do have other licensing terms.
Please look in question into the sources.
"ports" contains other software, not written by me,
but ported to compile with minilib.