November 23, 2018

2667 words 13 mins read

gurugio/lowlevelprogramming-university

gurugio/lowlevelprogramming-university

How to be low-level programmer
repo name gurugio/lowlevelprogramming-university
repo link https://github.com/gurugio/lowlevelprogramming-university
homepage
language
size (curr.) 187 kB
stars (curr.) 6081
created 2017-01-01
license GNU General Public License v3.0

NOTICE: Please notice that low-level programming is out of trend and currently there are not many companies hiring low-level developer. It is getting harder for me to find a job. If you does not start professional career yet, I would like to recommend you consider other fields carefully.

Low-Level Programming University

What is it?

I’m inspired by google-interview-university. I’d like to share my experience and show a roadmap to becoming a low-level programmer because I have found that these skills are not as common as they once were. In addition, many students and beginners ask me how they could become low-level programmers and Linux kernel engineers.

This page cannot include every link/book/course. For example, this page introduces Arduino but there is not detailed information about Arduino and embedded systems. You should go further yourself. You have the keyword “Arduino” with which you can start. So your next step is probably googling Arduino, buying a kit, and doing something for yourself, not collecting links or free books. Please remember this page is just a roadmap for beginners.

Low-level programming is a part of computer science. Absolutely it would be much better to get education for computer science first.

What Is the Low-Level?

I classify low-level programming as programming that is very close to the machine, using a lower level programming language like C or assembly. This is in contrast to higher-level programming, typical of user-space applications, using high level languages (e.g. Python, Java).

Yes, systems programming is a very close concept to low-level programming. This page includes the hardware design and firmware development that is not included in systems programming.

Finally, this page includes topics ranging from hardware components to the Linux kernel. That is a huge range of layers. A one page document can never cover the details of all the layers, so the aim of this document is to serve as a starting point for low-level programming.

Theory

There are two background theories to low-level programming:

  • Computer Architecture
  • Operating Systems

I think the best way to learn theory is by taking a course. Reading books is not bad but takes too much time and effort. You can find many good classes on online universities, for instance, Coursera.org and edx.org. Theory is theory. I don’t think you need to get an A+ in the class, just understand the big picture. You’ll get better and better with experience.

Let me introduce several books that I’ve read. They are commonly used as textbooks in universities. If there is no class with these books in your university, it’s worth spending some time reading them.

  • Computer Architecture
    • Computer Architecture, Fifth Edition: A Quantitative Approach
    • Computer Systems: A Programmer’s Perspective
    • Computer Organization and Design, Fourth Edition: The Hardware/Software Interface
  • Operating Systems
    • The Magic Garden Explained: The Internals of UNIX System V Release 4 an Open Systems Design
    • The Design of the UNIX Operating System
    • Operating Systems: Internals and Design Principles by William Stallings
  • Recommended Courses

There is an infinite list of good books. I don’t want to say that you should read many books. Just read one book carefully. Whenever you learn a theory, implement simulation code of it. Implementing one thing is better than knowing one hundred theories.

Languages

Assembly

Choose one between x86 or ARM. No need to know both. It doesn’t matter to know assembly language. The essential thing is understanding the internals of a CPU and computer. So you don’t need to practice the assembly of the latest CPU. Select 8086 or Corex-M.

C language

There is no shortcut. Just read the entire book and solve all the exercises.

If you want to be expert of C programming, visit https://leetcode.com/. Good luck!

Applications

Hardware && Firmware

If you want to be an embedded systems engineer, it would be best to start from a simple hardware kit, rather than starting with the latest ARM chipset.

  • Arduino Start Kit
    • There are many series of Arduinos but “Arduino Start Kit” has the most simple processor(Atmega328P) and guide book
    • Atmega328P has an 8-bit core which is a good place to start digital circuit design and firmware development.
    • You don’t need to know how to draw schematics and layouts and assemble the chips.
    • But you do need to know how to read schematics and understand how the chips are connected.
    • Firmware developers should be able to read the schematics and figure out how to send data to the target device.
    • Follow the guide book!
  • 8086 manual
    • If you’re a beginner to x86 architecture, 8086 is also very good guide for processor architecture and 80x86 assembly
  • 80386 manual

At this point, you should be good to start the latest ARM or x86 processor.

For example, the Raspberry Pi board has a Cortex-A53 Processor that supports a 64-bit instruction set. This allows you to experience a modern processor architecture with rPi. Yes, you can buy it… but… what are you going to do with it? If you have no target project, you would be likely to throw the board into a drawer and forget it like other gadgets you may have bought before.

So, I recommend one project for you.

I’ve made a toy kernel that supports 64-bit long mode, paging and very simple context switching. Making a toy kernel is good way to understand modern computer architecture and hardware control.

In fact, you have already the latest processor and the latest hardware devices. Your laptop! Your desktop! You already have all that you need in order to start! You don’t need to buy anything. The qemu emulator can emulate the latest ARM processors and Intel processors. So everything you need is already on hand. There are many toy kernels and documents you can refer to. Just install qemu emulator and make a tiny kernel that just boots, turns on paging, and prints some messages.

Other toy kernels:

Linux kernel and device driver

You don’t need to make a complete operating system. Join the Linux community and participate in development.

Follow carefully

References

Check when you need something

Other application

Yes, you might not be interested in Linux or firmware. If so, you can find other applications:

  • Windows systems programming & device drivers
  • Security
  • Reverse engineering

I don’t have any knowledge about those applications. Please send me any information for beginners.

Kernels and drivers are not all of low-level programming. One more important application of low-level programming is the software-defined storage or distributed filesystem. Detailed descriptions of them is beyond the scope of this document but there is an excellent course where you can try a simple distributed filesystem.

Future of low-level programming

I do not know the future, but I keep my eye on RUST.

If I could have one week free and alone, I would learn RUST. That is because RUST is the latest language with which I can develop Linux device drivers.

IoT is new trend, so it’s worth to check what OSs are for IoT. ARM, Samsung and some companies has their own realtime OS but sadly many of them are closed source. But Linux Foundation also has a solution: Zephyr

Typical cloud servers have many layers; for instance, host OS, kvm driver, qemu process, guest OS and service application. A container has been developed to provide light virtualization. In the near future, a new concept of OS, a so-called library OS or Unikernel, would replace the typical stack of SW for virtualization.

Big data and cloud computing require bigger and bigger storage. Some disks directly attached to server machines cannot satisfy the required capacity, stability and performance. Therefore there has been research to make huge storage systems with many storage machines connected by a high speed network. It used to be focused on making one huge storage volume. But currently they are providing many volumes dedicated for many virtual machines.

How to start?

I received an email to ask how to start. There are many information about books, courses and projects in this page. It is my mistake to forget to write how to start. Unfortunately there is no King’s Road to King’s Landing. I will just write what I did in order. If you have already done something, please skip it. AGAIN, this is just an example that you could do in order, just in case if you do not know how to start or what to do.

Translation

Please send me the pull request if you’d like to translate this page. I’ll list it here.

Who am I?

I’m inspired by google-interview-university. I’d like to share my experience and show a roadmap to becoming a low-level programmer because I have found that these skills are not as common as they once were. In addition, many students and beginners ask me how they could become low-level programmers and Linux kernel engineers.

FYI, I have over 10 years of experience as a low-level programmer:

  • 80x86 Assembly programming
  • Hardware device with Atmel chip and firmware
  • C language system programming for Unix
  • Device driver in Linux
  • Linux kernel: page allocation
  • Linux kernel: block device driver and md module
programming
comments powered by Disqus
marcobambini/gravity

marcobambini/gravity

November 13, 2018

Gravity Programming Language
CamDavidsonPilon/Probabilistic-Programming-and-Bayesian-Methods-for-Hackers

CamDavidsonPilon/Probabilistic-Programming-and-Bayesian-Methods-for-Hackers

October 29, 2018

aka “Bayesian Methods for Hackers”: An introduction to Bayesian methods + probabilistic programming with a computation/understanding-first, mathematics-second point of view. All in pure Python ;)