diff options
Diffstat (limited to 'Fosdem2022')
| -rw-r--r-- | Fosdem2022/TALK.txt | 201 | ||||
| -rw-r--r-- | Fosdem2022/contents.md | 244 | ||||
| -rw-r--r-- | Fosdem2022/img/better_documentation.png | bin | 0 -> 102745 bytes | |||
| -rw-r--r-- | Fosdem2022/img/drake.jpg | bin | 0 -> 93013 bytes | |||
| -rw-r--r-- | Fosdem2022/template.tex | 236 |
5 files changed, 681 insertions, 0 deletions
diff --git a/Fosdem2022/TALK.txt b/Fosdem2022/TALK.txt new file mode 100644 index 0000000..719c001 --- /dev/null +++ b/Fosdem2022/TALK.txt @@ -0,0 +1,201 @@ +Who I am (it's a little bit relevant to understand my previous knowledge) +- telecommunication engineer (EEE equivalent) +- Freelance engineer/programmer at ElenQ.Tech +- Guix user and contributor + +Recently: +- Interested in small computing +- Lisp, specially scheme, specially small scheme implementations + +The early beginnings: +- Started a very small scheme compiler for small machines, and thought + about RISC-V. I decided to target RISC-V assembly (following: + https://compilers.iecc.com/crenshaw/) +- I studied RISC-V via [riscvbook] +- I studied several scheme implementations +- Nothing happened out of this, I gave up when I read about continuations but I + learned some stuff. + +The random events +- One day a friend tells me guix mailing list is looking for help to make a + RISC-V port. I decide to raise my hand. +- Now I'm involved in RISC-V porting effort somehow. +- One day Andy Wingo, the Guile maintainer, mentions it would be interesting + to port Lightening, Guile's JIT code generation library to RISC-V. + +Now I'm working on Lightening: things learned +- Code is data -> I'm a lisp guy, I already knew that! +- Some cool GDB debugging I already forgot (disassemble) +- Now I'm able to hand-assemble instructions too +- Machine code generation is not that complex +- Relocations and large immediates are painful + +Lightening RISC-V is passing all the tests, next step is to give it a serious +try and see if it works! Cross fingers! + +Stage0 +- There was a chance to work on the bootstrap system of Guix via NlNet and the + guys involved in the porting effort helped me prepare a proposal. +- I had to learn about all the steps around so I started with Stage0. + +Stage0's steps: +hex0 -> hex1 -> hex2 -> M0 (macro system) -> M2-Planet [ -> Gnu Mes ] + +Hex0 +- Hex0 are hex files that contain what the ELF executable would, but encoded in + Hexadecimal text. +- I started with hex0, porting it to RISC-V (RV64). Thankfully, I was + developing a RISC-V assembler (I never finished) that helped me generate the + instructions by hand. +- Took the ELF headers from the wikipedia. +- It worked!! + +Effects of this effort +- Other people at Bootstrappable took this work and finished the whole chain + of Stage0! + +A random afternoon: +- I was bored and the Hex0 for RV32 was missing. Looked the headers in the + wikipedia and 20 minutes later it was added to the project. Another starting + point solved! + +Now: +- Took the NlNet grant and now I'm going to contribute to the full source + bootstrap ecosystem. So now this is kind of my job. + +Conclusions: +- This is just random work +- Embrace the randomness of life, stay curious and you may reach interesting + places. + + + +--- + + +My blogposts about this process: +https://ekaitz.elenq.tech/lightening.html +https://ekaitz.elenq.tech/hex0.html +https://ekaitz.elenq.tech/machine-code-generation.html + +RISC-V: +http://riscvbook.com/ + +Bootstrap related: +http://bootstrappable.org +https://github.com/oriansj/stage0 +https://github.com/oriansj/bootstrap-seeds + +GUILE +https://www.gnu.org/software/guile/ +https://gitlab.com/wingo/lightening/-/merge_requests/14/commits + +--- + +We are excited to announce a devroom on Declarative and Minimalistic +Computing at FOSDEM on 5th and 6th of February, 2022, online! + +FOSDEM is one of the most important free software conferences and is +hosted annually at Université libre de Bruxelles in Brussels, +Belgium. Unfortunately again this year FOSDEM will not run a physical +conference but will be online only. Talks will be pre-recorded with +some live content including Q&A sessions and discussion panels. + +We accept talks from languages that attempt to minimize use of hardware +and software while trying to make systems simpler, more robust and more +secure. If you are working on improving today's systems taking +declarative/minimalistic approaches feel free to submit a talk +proposal. Examples include the Scheme/Lisp family of programmings +languages. In past editions, this devroom has received presentations +from a varied number of language communities, including Forth, Guile, +Lua, Nim, Racket, Raku and Tcl as well as several experimental projects +that push minimalism in new directions. + +Minimalism and declarative programming are two important topics for +this devroom. Minimalism matters. Minimalism allows for smaller +systems that take less resources and consume less energy. More +importantly, free and open source minimalism allows for secure systems +that are easy to understand. Declarative programming is a programming +paradigm that expresses the logic of a computation without describing +its control flow. Many languages that apply this style attempt to +minimize or eliminate side effects by describing what the program must +accomplish in terms of the problem domain, rather than describe how to +accomplish it as a sequence of the programming language primitives. + +Finally, in this year's virtual conference we will honor the late Professor +[https://en.wikipedia.org/wiki/John_McCarthy_(computer_scientist) John +McCarthy] as the founder of AI and the inventor of LISP. McCarthy with +his work pioneered artificial intelligence, developed the Lisp +programming language family and kickstarted our modern computing world. + +We want to invite you to submit a talk on declarative and minimalistic +computing that fits that description. We are especially happy to +receive talk submissions from members of groups underrepresented in +free software. + +If you have something you’d like to share with your fellow developers, +please E-mail us! Talks considered for the devroom will have to +be entered in + + - https://penta.fosdem.org/submission/FOSDEM22 + +The deadline for submission is December 20th. If you have a FOSDEM +pentabarf account from a previous year, please use that +account. Otherwise add one on +https://penta.fosdem.org/user/new_account. Reach out to +pjotr.public456@thebird.nl or manolis837@gmail.com if you run into any +trouble. + +When submitting your talk make doubly sure to select "Declarative and +Minimalistic Computing devroom" as track (if you don't we won't find +it), and include the following information: + + * The title and subtitle of your talk + * A short abstract of one paragraph + * A longer description if you wish to do so + * Links to related websites/blogs etc + * Presentations has to be pre-recorded and streamed before the event. + * Start recording early! + +To see what a final talk looks like see + + https://archive.fosdem.org/2021/schedule/event/gnumes/ + +For accepted talks + + * Once your talk was accepted, we will assign you an organizer to +help you to produce the pre-recorded content. + * The organizer will review the content and ensure it has the +required quality. He is also responsible for ensuring the content is +into the system and ready to broadcast. + * During the stream of your talk, you must be available online for +the Q/A session + +Let's make this a fun day! + += Organizers = + +Pjotr Prins, Manolis Ragkousis, Ludovic Courtès, Amirouche Boubekki, +Hisham Muhammad, Jan Nieuwenhuizen, Ricardo Wurmus, Alex +Sassmannshausen, William Byrd, Oliver Propst, Efraim Flashner, Julien +Lepiller + += Code of conduct = + + - https://fosdem.org/2022/practical/conduct/ + += Original proposal = + + - https://libreplanet.org/wiki/FOSDEM2022-devroom-proposal + += Important dates: = + + - Dec 20th 2021: submission deadline for talk proposals + - Dec 31th 2021: announcement of the final schedule + - Jan 14th 2021: submission deadline for recordings + - Feb 5th 2022: FOSDEM! + += Links: = + + - +https://libreplanet.org/wiki/FOSDEM2022-devroom-declarative-and-minimalistic-computing-cfp diff --git a/Fosdem2022/contents.md b/Fosdem2022/contents.md new file mode 100644 index 0000000..e8533b8 --- /dev/null +++ b/Fosdem2022/contents.md @@ -0,0 +1,244 @@ +--- +title: "A year of RISC-V adventures: embracing chaos in your software journey" + +subtitle: How I started from zero and ended up porting a JIT compilation library and assembling files by hand + +license: CC-BY-SA +author: Ekaitz Zárraga +links-as-notes: true +lang: spanish +polyglossia-lang: + name: english +how-to: pandoc -f markdown+smart -t beamer contents.md -o beamer.pdf --pdf-engine=xelatex --template=template.tex +... + +# Who I am +- Telecommunication engineer (EEE equivalent) +- Freelance engineer/programmer at [ElenQ.Tech](https://elenq.tech) +- Guix user and contributor + +## Recently +- Interested in small computing +- Lisp, specially Scheme, more specifically small Scheme implementations. + + + +# The beginnings +- Started a very small scheme compiler for small machines, and thought about + RISC-V. I decided to target RISC-V assembly following ["Let's Build a + Compiler" by Jack Crenshaw](https://compilers.iecc.com/crenshaw/) +- I studied RISC-V with: ["The RISC-V Reader: An Open Architecture + Atlas"](http://riscvbook.com/) +- I studied several scheme implementations and read some papers +- Nothing happened out of this, I gave up when I read about continuations but I + learned some stuff. + + + +# Chaos starts +- One day a friend (the same guy who introduced me to Guix) tells me the Guix + mailing list is looking for help to make a RISC-V port. I decide to raise my + hand. +- Now I'm involved in RISC-V porting effort somehow. +- One day Andy Wingo, the Guile maintainer, mentions it would be interesting to + port Lightening, Guile's JIT code generation library to RISC-V. + + + +# Now I'm working on the Lightening port somehow + +Lightening is a machine code generation library. It's a fork of GNU Lightning, +made for Guile, that aims to be simple. + +## The experience +- My C programming skills are rusty +- 0 documentation: only the one inherited from GNU Lightning +- Dead code +- Andy Wingo helped me take the good direction and made everything easier + +# Lightening: Things learned +- How to assemble instructions by hand * +- Code is data. I'm a lisp guy, I already knew that! * +- Some cool GDB debugging tricks I already forgot +- Machine code generation is not that complex +- Relocations and immediates are painful[^more] + +[^more]: Read more: <https://ekaitz.elenq.tech/machine-code-generation.html> + +# Lightening: Assemble by hand + +``` +addi a0, zero, 56 +``` + +- Opcode `addi`: `0010011` +- Destination register `a0`: `01010` +- `funct3`: `000` +- Source register `zero`: `00000` +- The immediate `56`: `000000111000` + +`000000111000 | 00000 | 000 | 01010 | 0010011` + +All together: `00000011100000000000010100010011` +(In hex: `0x3800513`) + + +# Lightening: Code is data + +``` c +#include<stdint.h> +#include<stdio.h> + +typedef int f0(void); + +int main(int argc, char* argv[]){ + uint32_t instructions[2]; + + instructions[0] = 0x03800513; // addi a0, zero, 56 + instructions[1] = 0x00008067; // jalr zero, ra, 0 + + // Reinterpret the array address as a function + f0 *load_56 = (f0*) instructions; + + int a = load_56(); + printf("%d\n", a); +} +``` + + +# Moving to the bootstrap system + +There was a chance to work on Guix's RISC-V support via NlNet and the +guys involved in the porting effort told me I should send a proposal. + +I had to learn about the full-source bootstrap project for this so I ended up +taking part on it. + +The proposal was rejected, but the learning was already done. *And what now?* + +# Stage0 + +Stage0 is a full source bootstrap system. + +Stage0's steps: + +1. Hex0: Hex encoded raw ELF file with comments +2. Hex1: Hex0 + one character labels and some extras +3. Hex2: Hex1 + proper labels and reasonable basics +4. M0 (macro system): A simple macro system +5. M2-Planet: A C subset that uses M0 as output +6. C compilers (i.e. GNU Mes's `mescc`) + + +# Stage0 + +## The experience + + + +# Hex0 + +Hex0 is a simple assembler written in Hex0. It just converts a Hexadecimal +encoded ELF file to binary. + +It's heavily commented so anyone can decode the instructions and make sure that +it works as expected. It looks like this: + +``` +# :_start ; (0x0600078) + +; Open input file and store FD in s2 +93 08 80 03 # RD_A7 !56 ADDI ; sys_openat +13 05 C0 F9 # RD_A0 !-100 ADDI ; AT_FDCWD +93 05 06 00 # RD_A1 RS1_A2 MV ; input file +13 06 00 00 # RD_A2 MV ; read only +73 00 00 00 # ECALL +13 09 05 00 # RD_S2 RS1_A0 MV ; Save fd + +``` + +# Hex0 + +I wanted to make the first version of the POSIX based Hex0 for RV64. I just +needed to: + +1. Write it in assembly first +2. Assemble everything by hand... + + + + + +# Hex0: Abandoned projects that add entropy: pysc-v + +While I was working on Lightening I started a RISC-V assembler in Python. + +*[I abandoned the project](http://git.elenq.tech/pysc-v/)* + +But the backend was reusable so I could generate all the instructions in the +Hex0 with the format I wanted! + + +# Hex0: disassembly tricks + +I still need to calculate addresses by hand though. + +*But wait!* I learned how to disassemble when I worked on Lightening so I could +obtain each instructions address easily, without all the counting. + +# Hex0 + +## The experience + +- It was rewarding to see that I could write assembly. +- I reused all the GDB tricks from Lightening to check that everything was + working correctly. +- The documentation is weird and there are too many subprojects that make the + project hard to understand for newcomers. It's hard to understand the reasons + behind some decisions and everything looks fragile. +- But `#bootstrappable` at `libera.chat` is a good place to ask questions and + learn from others. + + +# Hex0 extra: embrace boredom + +Sunday morning. I'm bored. + +I reviewed the status of the project and saw RV32 is not ready yet, even if it +is similar to RV64. + +I start a discussion on IRC and in less than half an hour and with the help of +the Wikipedia we replace the ELF headers and make the Hex0 support for RV32. + + +# Status of the projects +- **Lightening**: is passing all the test but needs testing. [The + code][lightening] is available for anyone to read and improve. +- **Stage0**: included [my small RISC-V contribution to Hex0][stage0] and other + contributors expanded my work to every stage. + +[lightening]: https://gitlab.com/wingo/lightening/-/merge_requests/14/commits +[stage0]: https://github.com/oriansj/bootstrap-seeds/pull/2 + + + +# My status + +I tried again with NlNet, with a proposal for some RISC-V porting efforts on +GNU Mes and other bootstrap related projects. + +They look interested on a full-source-bootstrap for RISC-V! + +*I might end up working on a compiler* + + +# Conclusions +- All this is just random work, done without a plan or purpose +- Embrace the chaos of life, stay curious and you may reach interesting places +- If you can, learn from people: it's faster and better +- You don't need to be a genius: most of the things are easy once you know the + context. + +--- + +## Just try, and let it happen diff --git a/Fosdem2022/img/better_documentation.png b/Fosdem2022/img/better_documentation.png Binary files differnew file mode 100644 index 0000000..b8348dc --- /dev/null +++ b/Fosdem2022/img/better_documentation.png diff --git a/Fosdem2022/img/drake.jpg b/Fosdem2022/img/drake.jpg Binary files differnew file mode 100644 index 0000000..10a1e74 --- /dev/null +++ b/Fosdem2022/img/drake.jpg diff --git a/Fosdem2022/template.tex b/Fosdem2022/template.tex new file mode 100644 index 0000000..31b2bf7 --- /dev/null +++ b/Fosdem2022/template.tex @@ -0,0 +1,236 @@ +\documentclass[$if(fontsize)$$fontsize$,$endif$$if(lang)$$babel-lang$,$endif$$if(handout)$handout,$endif$$if(beamer)$ignorenonframetext,$endif$$for(classoption)$$classoption$$sep$,$endfor$]{$documentclass$} +\setbeamertemplate{caption}[numbered] +\setbeamertemplate{caption label separator}{: } +\selectcolormodel{gray} % Color B&W +\beamertemplatenavigationsymbols$if(navigation)$$navigation$$else$empty$endif$ + +\usepackage{amssymb,amsmath} +\usepackage{ifxetex,ifluatex} +\usepackage{fixltx2e} % provides \textsubscript + + +\usefonttheme{professionalfonts} % using non standard fonts for beamer +\usefonttheme{serif} % default family is serif + +% Font +\usepackage{fontspec} +\setmainfont[ + BoldFont = Lato-Bold.ttf, + ItalicFont = Lato-Italic.ttf, + BoldItalicFont = Lato-BoldItalic.ttf +]{Lato-Regular.ttf} +\newcommand{\euro}{€} + +% Images with no captions +\usepackage{caption} +\captionsetup[figure]{labelformat=empty} + +% Bullet style +\useinnertheme{circles} +\newlength{\wideitemsep} +\setlength{\wideitemsep}{\itemsep} +\addtolength{\wideitemsep}{5pt} +\let\olditem\item +\renewcommand{\item}{\setlength{\itemsep}{\wideitemsep}\olditem} + +$if(euro)$ + \newcommand{\euro}{€} +$endif$ + +% Language +\usepackage{polyglossia} +\setmainlanguage[$polyglossia-lang.options$]{$polyglossia-lang.name$} + +\newif\ifbibliography +$if(natbib)$ +\usepackage{natbib} +\bibliographystyle{$if(biblio-style)$$biblio-style$$else$plainnat$endif$} +$endif$ +$if(biblatex)$ +\usepackage[$if(biblio-style)$style=$biblio-style$,$endif$$for(biblatexoptions)$$biblatexoptions$$sep$,$endfor$]{biblatex} +$for(bibliography)$ +\addbibresource{$bibliography$} +$endfor$ +$endif$ + +$if(verbatim-in-note)$ +\usepackage{fancyvrb} +$endif$ + +\hypersetup{ +$if(title-meta)$ + pdftitle={$title-meta$}, +$endif$ +$if(author-meta)$ + pdfauthor={$author-meta$}, +$endif$ +$if(keywords)$ + pdfkeywords={$for(keywords)$$keywords$$sep$, $endfor$}, +$endif$ +$if(colorlinks)$ + colorlinks=true, + linkcolor=$if(linkcolor)$$linkcolor$$else$Maroon$endif$, + citecolor=$if(citecolor)$$citecolor$$else$Blue$endif$, + urlcolor=$if(urlcolor)$$urlcolor$$else$Blue$endif$, +$else$ + pdfborder={0 0 0}, +$endif$ + breaklinks=true} +\urlstyle{same} % don't use monospace font for urls +$if(verbatim-in-note)$ +\VerbatimFootnotes % allows verbatim text in footnotes +$endif$ +$if(listings)$ +\usepackage{listings} +$endif$ +$if(lhs)$ +\lstnewenvironment{code}{\lstset{language=Haskell,basicstyle=\small\ttfamily}}{} +$endif$ +$if(highlighting-macros)$ +$highlighting-macros$ +$endif$ +$if(tables)$ +\usepackage{longtable,booktabs} +\usepackage{caption} +% These lines are needed to make table captions work with longtable: +\makeatletter +\def\fnum@table{\tablename~\thetable} +\makeatother +$endif$ +$if(graphics)$ +\usepackage{graphicx,grffile} +\makeatletter +\def\maxwidth{\ifdim\Gin@nat@width>\linewidth\linewidth\else\Gin@nat@width\fi} +\def\maxheight{\ifdim\Gin@nat@height>\textheight0.8\textheight\else\Gin@nat@height\fi} +\makeatother +% Scale images if necessary, so that they will not overflow the page +% margins by default, and it is still possible to overwrite the defaults +% using explicit options in \includegraphics[width, height, ...]{} +\setkeys{Gin}{width=\maxwidth,height=\maxheight,keepaspectratio} +$endif$ + +% Prevent slide breaks in the middle of a paragraph: +\widowpenalties 1 10000 +\raggedbottom + +$if(section-titles)$ +\AtBeginPart{ + \let\insertpartnumber\relax + \let\partname\relax + \frame{\partpage} +} +\AtBeginSection{ + \ifbibliography + \else + \let\insertsectionnumber\relax + \let\sectionname\relax + \frame{\sectionpage} + \fi +} +\AtBeginSubsection{ + \let\insertsubsectionnumber\relax + \let\subsectionname\relax + \frame{\subsectionpage} +} +$endif$ + +$if(links-as-notes)$ +% Make links footnotes instead of hotlinks: +\renewcommand{\href}[2]{#2\footnote{\url{#1}}} +$endif$ + +$if(strikeout)$ +\usepackage[normalem]{ulem} +% avoid problems with \sout in headers with hyperref: +\pdfstringdefDisableCommands{\renewcommand{\sout}{}} +$endif$ +\setlength{\parindent}{0pt} +\setlength{\parskip}{6pt plus 2pt minus 1pt} +\setlength{\emergencystretch}{3em} % prevent overfull lines +\providecommand{\tightlist}{% + \setlength{\itemsep}{0pt}\setlength{\parskip}{0pt}} +$if(numbersections)$ +\setcounter{secnumdepth}{$if(secnumdepth)$$secnumdepth$$else$5$endif$} +$else$ +\setcounter{secnumdepth}{0} +$endif$ +$if(dir)$ +\ifxetex + % load bidi as late as possible as it modifies e.g. graphicx + $if(latex-dir-rtl)$ + \usepackage[RTLdocument]{bidi} + $else$ + \usepackage{bidi} + $endif$ +\fi +\ifnum 0\ifxetex 1\fi\ifluatex 1\fi=0 % if pdftex + \TeXXeTstate=1 + \newcommand{\RL}[1]{\beginR #1\endR} + \newcommand{\LR}[1]{\beginL #1\endL} + \newenvironment{RTL}{\beginR}{\endR} + \newenvironment{LTR}{\beginL}{\endL} +\fi +$endif$ +$for(header-includes)$ +$header-includes$ +$endfor$ + +$if(title)$ +\title{$title$} +$endif$ +$if(subtitle)$ +\subtitle{$subtitle$} +$endif$ +$if(author)$ +\author{$for(author)$$author$$sep$ \and $endfor$} +$endif$ +$if(institute)$ +\institute{$for(institute)$$institute$$sep$ \and $endfor$} +$endif$ +\date{$date$} + +\begin{document} +$if(title)$ +\frame{\titlepage} +$endif$ + +$for(include-before)$ +$include-before$ + +$endfor$ +$if(toc)$ +\begin{frame} +\tableofcontents[hideallsubsections] +\end{frame} + +$endif$ +$body$ + +$if(natbib)$ +$if(bibliography)$ +$if(biblio-title)$ +$if(book-class)$ +\renewcommand\bibname{$biblio-title$} +$else$ +\renewcommand\refname{$biblio-title$} +$endif$ +$endif$ +\begin{frame}[allowframebreaks]{$biblio-title$} +\bibliographytrue +\bibliography{$for(bibliography)$$bibliography$$sep$,$endfor$} +\end{frame} + +$endif$ +$endif$ +$if(biblatex)$ +\begin{frame}[allowframebreaks]{$biblio-title$} +\bibliographytrue +\printbibliography[heading=none] +\end{frame} + +$endif$ +$for(include-after)$ +$include-after$ + +$endfor$ +\end{document} |
