7 This section describes the workflow and some best practices for developing
8 the Libre-SOC hardware. We use nmigen, yosys and symbiyosys, and this
9 page is intended not just to help you get set up, it is intended to
10 help advise you of some tricks and practices that will help you become
11 effective team contributors.
13 It is particularly important to bear in mind that we are not just
14 "developing code", here: we are creating a "lasting legacy educational
15 resource" for other people to learn from, and for businesses and students
16 alike to be able to use, learn from and augment for their own purposes.
18 It is also important to appreciate and respect that we are funded under
19 NLNet's Privacy and Enhanced Trust Programme <http://nlnet.nl/PET>. Full
20 transparency, readability, documentation, effective team communication
21 and formal mathematical proofs for all code at all levels is therefore
24 Therefore, we need not only to be "self-sufficient" (absolutely
25 under no circumstances critically reliant on somebody else's servers
26 **or protocols**) we also need to ensure that everything (including
27 all communication such as the mailing list archives) are recorded,
28 replicable, and accessible in perpetuity. Use of slack or a "forum"
29 either actively prevents or makes that much harder.
31 # Collaboration resources
33 The main message here: **use the right tool for the right job**.
35 * mailing list: general communication and discussion.
36 * irc channel #libre-soc: real(ish)-time communication.
37 * bugtracker: task-orientated, goal-orientated *focussed* discussion.
38 * ikiwiki: document store, information store, and (editable) main website
39 * git repositories: code stores (**not binary or auto-generated output store**)
40 * ftp server (<https://ftp.libre-soc.org/>): large (temporary,
41 auto-generated) file store.
43 We will add an IRC channel at some point when there are enough people
44 to warrant having one (and it will be publicly archived)
46 Note also the lack of a "forum" in the above list. this is very
47 deliberate. forums are a serious distraction when it comes to technical
48 heavily goal-orientated development. recent internet users may enjoy
49 looking up the "AOL metoo postings" meme.
51 Note also the complete lack of "social platforms". if we wanted to tell
52 everybody how much better each of us are than anyone else in the team,
53 how many times we made a commit (look at me, look at me, i'm so clever),
54 and how many times we went to the bathroom, we would have installed a
55 social media based project "management" system.
57 ## Main contact method: mailing list
59 To respect the transparency requirements, conversations need to be
60 public and archived (i.e not skype, not telegram, not discord,
61 and anyone seriously suggesting slack will be thrown to the
62 lions). Therefore we have a mailing list. Everything goes through
63 there. <https://lists.libre-soc.org/mailman/listinfo/libre-soc-dev>
64 therefore please do google "mailing list etiquette" and at the very
65 minimum look up and understand the following:
67 * This is a technical mailing list with complex topics. Top posting
68 is completely inappropriate. Don't do it unless you have mitigating
69 circumstances, and even then please apologise and explain ("hello sorry
70 using phone at airport flight soon, v. quick reply: ....")
71 * Always trim context but do not cut excessively to the point where people
72 cannot follow the discussion. Especially do not cut the attribution
73 ("On monday xxx wrote") of something that you are actually replying
75 * Use inline replies i.e. reply at the point in the relevant part of
76 the conversation, as if you were actually having a conversation.
77 * Follow standard IETF reply formatting, using ">" for cascaded
78 indentation of other people's replies. If using gmail, please: SWITCH
79 OFF RICH TEXT EDITING.
80 * Please for god's sake do not use "my replies are in a different
81 colour". Only old and highly regarded people still using AOL are allowed
82 to get away with that (such as Mitch).
83 * Start a new topic with a relevant subject line. If an existing
84 discussion changes direction, change the subject line to reflect the
85 new topic (or start a new conversation entirely, without using the
87 * DMARC is a pain on the neck. Try to avoid GPG signed messages. sigh.
88 * Don't send massive attachments. Put them online (no, not on facebook or
89 google drive or anywhere else that demands privacy violations) and provide
90 the link. Which should not require any kind of login to access. ask the
91 listadmin if you don't have anywhere suitable: FTP access can be arranged.
93 ### Actionable items from mailing list
95 If discussions result in any actionable items, it is important not to
96 lose track of them. Create a bugreport, find the discussion in the
97 archives <https://lists.libre-soc.org/pipermail/libre-soc-dev/>,
98 and put the link actually in the bugtracker as one of the comments.
100 At some point in any discussion, the sudden realisation may dawn on one
101 or more people that this is an "actionable" discussion. at that point
102 it may become better to use <https://bugs.libre-soc.org/>
103 itself to continue the discussion rather than to keep on dropping copies
104 of links into the bugtracker. The bugtracker sends copies of comments
105 *to* the list however this is 'one-way' (note from lkcl: because this
106 involves running an automated perl script from email, on every email,
107 on the server, that is a high security risk, and i'm not doing it. sorry.)
109 ### Mailing list != editable document store
111 Also, please do not use the mailing list as an "information or document
112 store or poor-man's editor". We have the wiki for that. Edit a page and
113 tell people what you did (summarise rather than drop the entire contents
114 at the list) and include the link to the page.
116 Or, if it is more appropriate, commit a document (or source code)
117 into the relevant git repository then look up the link in the gitweb
118 source tree browser and post that (in the bugtracker or mailing list)
119 See <https://git.libre-soc.org/>
121 ### gmail "spam"ifying the list
123 See <https://blog.kittycooper.com/2014/05/keeping-my-mailing-list-emails-out-of-gmails-spam-folder/>
125 Basically it is possible to select any message from the list, create a
126 "filter" (under "More"), and, on the 2nd dialog box, click the "never
127 send this to Spam" option.
131 bugzilla. old and highly effective. sign up in the usual way. any
132 problems, ask on the list.
134 Please do not ask for the project to be transferred to github or other
135 proprietary nonfree service "because it's soooo convenient", as the
136 lions are getting wind and gout from overfeeding on that one.
140 Runs the main libre-soc.org site (including this page). effective,
141 stunningly light on resources, and uses a git repository not a database.
142 That means it can be edited offline.
144 Usual deal: register an account and you can start editing and contributing
147 Hint: to create a new page, find a suitable page that would link to it,
148 first, then put the link in of the page you want to create, as if the
149 page already exists. Save that page, and you will find a question mark
150 next to the new link you created. click that link, and it will fire up a
151 "create new page" editor.
153 Wiki pages are formatted in [[markdown|ikiwiki/markdown]] syntax.
155 Hint again: the wiki is backed by a git repository. Don't go overboard
156 but at the same time do not be afraid that you might "damage" or "lose"
157 pages. Although it would be a minor pain, the pages can always be
158 reverted or edited by the sysadmins to restore things if you get in a tiz.
160 Assistance in creating a much better theme greatly appreciated. e.g.
161 <http://www.math.cmu.edu/~gautam/sj/blog/20140720-ikiwiki-navbar.html>
165 We use git. More on this below. We also use
166 [gitolite3](https://gitolite.com/gitolite/) running on a dedicated server.
167 again, it is extremely effective and low resource utilisation. Reminder:
168 lions are involved if github is mentioned.
170 [gitweb](https://git.wiki.kernel.org/index.php/Gitweb) is provided which
171 does a decent job. <https://git.libre-soc.org/>
173 [Git](https://en.wikipedia.org/wiki/Git) does version control, ie it
174 tracks changes to files so that previous versions can be got back or
177 Checklist page [[HDL_workflow/git_checklist]]
181 <https://ftp.libre-soc.org/> is available for storing large files
182 that do not belong in a git repository, if we have (or ever need)
183 any. Images (etc.) if small and appropriate should go into the
184 wiki, however .tgz archives (etc.) and, at some point, binaries,
185 should be on the ftp server.
187 Ask on the list if you have a file that belongs on the ftp server.
191 As an aside: all this is "old school" and run on a single core 512MB
192 VM with only a 20GB HDD allocation. it costs only 8 GBP per month from
193 mythic-beasts and means that the project is in no way dependent on anyone
194 else - not microsoft, not google, not facebook, not amazon.
196 We tried [gitlab](https://about.gitlab.com/). it didn't go well. please
197 don't ask to replace the above extremely resource-efficient services
202 RAM is the biggest requirement. Minimum 16GB, the more the better (32
203 or 64GB starts to reach "acceptable" levels. Disk space is not hugely
204 critical: 256GB SSD should be more than adequate. Simulations and
205 FPGA compilations however are where raw processing power is a must.
206 High end Graphics Cards are nonessential.
208 What is particularly useful is to have hi-res screens (curved is
209 *strongly* recommended if the LCD is over 24in wide, to avoid eyeballs
210 going "prism" through long term use), and to have several of them: the
211 more the better. Either a DisplayLink UD160A (or more modern variant)
212 or simply using a second machine (lower spec hardware because it will
213 run editors) is really effective.
215 Also it is really recommended to have a UHD monitor (4k - 3840x2160),
216 or at least 2560x1200. If given a choice, 4:3 aspect ratio is better
217 than 16:9 particularly when using several of them. However, caveat
218 (details below): please when editing do not assume that everyone will
219 have access to such high resolution screens.
223 First install and become familiar with
224 [Debian](https://www.debian.org/) ([Ubuntu](https://ubuntu.com/)
226 must) for standardisation cross-team and so that toolchain installation
227 is greatly simplified. yosys in particular warns that trying to use
228 Windows, BSD or MacOS will get you into a world of pain.
230 Only a basic GUI desktop is necessary: fvwm2, xfce4, lxde are perfectly
231 sufficient (alongside wicd-gtk for network management). Other more
232 complex desktops can be used however may consume greater resources.
234 # editors and editing
236 Whilst this is often a personal choice, the fact that many editors are
237 GUI based and run full-screen with the entire right hand side *and* middle
238 *and* the majority of the left side of the hi-res screen entirely unused
239 and bereft of text leaves experienced developers both amused and puzzled.
241 At the point where such full-screen users commit code with line lengths
242 well over 160 characters, that amusement quickly evaporates.
244 Where the problems occur with full-screen editor usage is when a project
245 is split into dozens if not hundreds of small files (as this one is). At
246 that point it becomes pretty much essential to have as many as six to
247 eight files open *and on-screen* at once, without overlaps i.e. not in
248 hidden tabs, next to at least two if not three additional free and clear
249 terminals into which commands are regularly and routinely typed (make,
250 git commit, nosetests3 etc). Illustrated with the following 3840x2160
251 screenshot (click to view full image), where *every one* of those 80x70
252 xterm windows is *relevant to the task at hand*.
254 [[!img 2020-01-24_11-56.png size=640x ]]
256 (hint/tip: fvwm2 set up with "mouse-over to raise focus, rather than
257 additionally requiring a mouse click, can save a huge amount of cumulative
258 development time here, switching between editor terminal(s) and the
261 Once this becomes necessary, it it turn implies that having greater
262 than 80 chars per line - and running editors full-screen - is a severe
263 hinderance to an essential *and highly effective* workflow technique.
265 Additionally, care should be taken to respect that not everyone will have
266 200+ column editor windows and the eyesight of a hawk. They may only have
267 a 1280 x 800 laptop which barely fits two 80x53 xterms side by side.
268 Consequently, having excessively long functions is also a hindrance to
269 others, as such developers with limited screen resources would need to
270 continuously page-up and page-down to read the code even of a single
273 This helps explain in part, below, why compliance with
274 [pep8](https://pep8.org/) is enforced, including its 80 character limit.
275 In short: not everyone has the same "modern" GUI workflow or has access
276 to the same computing resources as you, so please do respect that.
278 More on this concept is
279 [here](https://www.linuxjournal.com/content/line-length-limits).
280 Note *very pointedly* that Linus Torvalds *specifically* states that
281 he does not want Linux kernel development to become the exclusive
282 domain of the "wealthy". That means **no** to assumptions about
283 access to ultra-high resolution screens.
285 # Software prerequisites
287 Whilst many resources online advocate "`sudo`" in front of all root-level
288 commands below, this quickly becomes tiresome. run "`sudo bash`", get a
289 root prompt, and save yourself some typing.
292 * apt-get install vim exuberant-ctags
293 * apt-get install build-essential
294 * apt-get install git python3.7 python3.7-dev python-nosetest3
295 * apt-get install graphviz xdot gtkwave
296 * apt-get install python3-venv
297 * apt-get install python-virtualenv # this is an alternative to python3-venv
298 * apt-get install tcl-dev libreadline-dev bison flex libffi-dev iverilog
299 * return to user prompt (ctrl-d)
301 (The above assumes that you are running Debian.)
303 This will get you python3 and other tools that are
304 needed. [graphviz](https://graphviz.org/) is essential
305 for showing the interconnections between cells, and
306 [gtkwave](http://gtkwave.sourceforge.net/) is essential for debugging.
308 If you would like to save yourself a lot more typing, check out the
309 [dev-env-setup](https://git.libre-soc.org/?p=dev-env-setup.git;a=summary)
310 repository, examine the scripts there and use them to automate much of
313 If you would like just to install only the apt dependencies use
314 [install-hdl-apt-reqs](https://git.libre-soc.org/?p=dev-env-setup.git;a=blob;f=install-hdl-apt-reqs;hb=HEAD) instead.
318 Look up good tutorials on how to use git effectively. There are so many
319 it is hard to recommend one. This is however essential. If you are not
320 comfortable with git, and you let things stay that way, it will seriously
321 impede development progress.
323 If working all day you should expect to be making at least two commits per
324 hour, so should become familiar with it very quickly. If you are *not*
325 doing around 2 commits per hour, something is wrong and you should read
326 the workflow instructions below more carefully, and also ask for advice
329 Worth noting: *this project does not use branches*. All code is committed
330 to master and we *require* that it be either zero-impact additions or that
331 relevant unit tests pass 100%. This ensures that people's work does not
332 get "lost" or isolated and out of touch due to major branch diversion,
333 and that people communicate and coordinate with each other.
337 Follow the source code (git clone) instructions here, do **not** use
338 the "stable" version (do not download the tarball):
339 <http://www.clifford.at/yosys/download.html>
341 Or, alternatively, use the
342 [yosys-et-al](https://git.libre-soc.org/?p=dev-env-setup.git;a=blob;f=yosys-et-al;hb=HEAD)
343 script (which also installs symbiyosys and its dependencies)
345 Do not try to use a fixed revision of yosys (currently 0.9), nmigen is
346 evolving and frequently interacts with yosys.
348 [Yosys](http://www.clifford.at/yosys/) is a framework for Verilog RTL.
349 [Verilog](https://en.wikipedia.org/wiki/Verilog) is a hardware description
351 RTL [Register Transfer
352 Level](https://en.wikipedia.org/wiki/Register-transfer_level)
353 models how data moves between
354 [registers](https://en.wikipedia.org/wiki/Hardware_register).
358 To install follow the [instructions
359 here](https://symbiyosys.readthedocs.io/en/latest/install.html)
360 Once done look at [A simple BMC
361 example](https://symbiyosys.readthedocs.io/en/latest/quickstart.html)
363 You do not have to install all of those (avy, boolector can be left
364 out if desired) however the more that are installed the more effective
365 the formal proof scripts will be (less resource utilisation in certain
368 [SymbiYosys](https://symbiyosys.readthedocs.io/en/latest/) (sby) is a
369 front-end driver program for Yosys-based formal hardware verification
374 [nmigen](https://m-labs.hk/gateware/nmigen/) may be installed as follows:
378 * git clone https://github.com/nmigen/nmigen.git
381 * python3 setup.py develop
384 Testing can then be carried out with "python3 setup.py test"
386 nmigen is a Python toolbox for building complex digital hardware.
388 ## Softfloat and sfpy
390 These are a test suite dependency for the
391 [ieee754fpu](https://www.gaisler.com/index.php/products/ipcores/ieee754fpu)
392 library, and will be changed in the future to use Jacob's
393 [simple-soft-float](https://crates.io/crates/simple-soft-float) library.
394 In the meantime, sfpy can be built as follows:
396 git clone --recursive https://github.com/billzorn/sfpy.git
399 git apply ../softposit_sfpy_build.patch
400 git apply /path/to/ieee754fpu/SoftPosit.patch
401 cd ../berkely-softfloat-3
402 # Note: Do not apply the patch included in sfpy for berkely-softfloat,
403 # it contains the same changes as this one
404 git apply /path/to/ieee754fpu/berkeley-softfloat.patch
407 # prepare a virtual environment for building
410 # or, if you prefer the old way:
411 # virtualenv -p python3 .env
413 # install dependencies
414 source .env/bin/activate
415 pip3 install --upgrade -r requirements.txt
420 make inplace -j$(nproc)
424 deactivate # deactivates venv, optional
425 pip3 install dist/sfpy*.whl
427 You can test your installation by doing the following:
430 >>> from sfpy import *
433 It should print out `Posit8(1.3125)`
435 ## qemu, cross-compilers, gdb
437 As we are doing POWER ISA, POWER ISA compilers, toolchains and
438 emulators are required.
440 Install powerpc64 gcc:
442 apt-get install gcc-9-powerpc64-linux-gnu
446 apt-get install qemu-system-ppc
448 Install gdb from source. Obtain the required tarball matching
449 the version of gcc (9.1) from here <https://ftp.gnu.org/gnu/gdb/>,
452 cd gdb-9.1 (or other location)
455 ../configure --srcdir=.. --host=x86_64-linux --target=powerpc64-linux-gnu
459 [gdb](https://en.wikipedia.org/wiki/GNU_Debugger) lets you debug running
460 programs. [qemu](https://www.qemu.org/) emulates processors, you can
461 run programs under qemu.
463 ## power_instruction_analyzer (pia)
465 We have a custom tool built in rust by programmerjake to help analyze
466 the power instructions execution on *actual* hardware.
468 Note: a very recent version of pip3 is required for this to work.
472 curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
474 Make sure we have the correct and up-to-date rust compiler (rustc):
476 rustup default stable
479 Use rust's package manager *cargo* to install the rust-python build
482 cargo install maturin
484 Install from git source by doing the following:
486 git clone https://salsa.debian.org/Kazan-team/power-instruction-analyzer.git pia
488 maturin build --cargo-extra-args=--features=python-extension
489 python3 -m pip install --user target/wheels/*.whl
491 Note: an ongoing bug in maturin interferes with successful installation.
492 This can be worked around by explicitly installing only the `.whl`
493 files needed rather than installing everything (`\*.whl`).
497 As this is an actual ASIC, we do not rely on an FPGA's JTAG TAP
498 interface, instead require a full complete independent implementation
499 of JTAG. Staf Verhaegen has one, with a full test suite, and it is
500 superb and well-written. The Libre-SOC version includes DMI (Debug
503 git clone https://git.libre-soc.org/git/c4m-jtag.git/
505 python3 setup.py develop
507 Included is an IDCODE tap point, Wishbone Master (for direct memory read
508 and write, fully independent of the core), IOPad redirection and testing,
509 and general purpose shift register capability for any custom use.
511 We added a DMI to JTAG bridge in LibreSOC which is
512 directly connected to the core, to access registers and
513 to be able to start and stop the core and change the PC.
514 In combination with the JTAG Wishbone interface the test
515 [ASIC](https://en.wikipedia.org/wiki/Application-specific_integrated_circuit)
516 can have a bootloader uploaded directly into onboard
517 [SRAM](https://en.wikipedia.org/wiki/Static_random-access_memory) and
520 [Chips4Makers](https://chips4makers.io/) make it possible for makers
521 and hobbyists to make their own open source chips.
523 [JTAG](https://en.wikipedia.org/wiki/JTAG) (Joint Test Action Group) is
524 an industry standard for verifying designs and testing printed circuit
525 boards after manufacture.
528 bus](https://en.wikipedia.org/wiki/Wishbone_%28computer_bus%29) is an open
529 source hardware computer bus intended to let the parts of an integrated
530 circuit communicate with each other.
533 See [[HDL_workflow/coriolis2]] page, for those people doing layout work.
537 A portable FPGA place and route tool.
539 See [[HDL_workflow/nextpnr]] page for installation instructions of nextpnr with ECP5 support for Lattice FPGA ECP5 series.
541 # Registering for git repository access
543 After going through the onboarding process and having agreed to take
544 responsibility for certain tasks, ask on the mailing list for git
545 repository access, sending in a public key (`id_rsa.pub`). If you do
546 not have one then generate it with `ssh-keygen -t rsa`. You will find it
549 NEVER SEND ANYONE THE PRIVATE KEY. By contrast the public key, on
550 account of being public, is perfectly fine to make... err... public.
552 Create a file `~/.ssh/config` with the following lines:
554 Host git.libre-soc.org
557 Wait for the Project Admin to confirm that the ssh key has been added
558 to the required repositories. Once confirmed, you can clone any of the
559 repos at https://git.libre-soc.org/:
561 git clone gitolite3@git.libre-soc.org:REPONAME.git
563 Alternatively, the .ssh/config can be skipped and this used:
565 git clone ssh://gitolite3@git.libre-soc.org:922/REPONAME.git
569 Although there are methods online which describe how (and why) these
570 settings are normally done, honestly it is simpler and easier to open
571 ~/.gitconfig and add them by hand.
573 core.autocrlf is a good idea to ensure that anyone adding DOS-formatted
574 files they don't become a pain. pull.rebase is something that is greatly
575 preferred for this project because it avoids the mess of "multiple
576 extra merge git tree entries", and branch.autosetuprebase=always will,
577 if you want it, always ensure that a new git checkout is set up with rebase.
586 autosetuprebase = always
588 # Checking out the HDL repositories
590 Before running the following, install the
591 dependencies. This is easiest done with this script
592 <https://git.libre-soc.org/?p=dev-env-setup.git;a=blob;f=install-hdl-apt-reqs;hb=HEAD>
596 * git clone gitolite3@git.libre-soc.org:nmigen.git
597 * git clone gitolite3@git.libre-soc.org:c4m-jtag.git
598 * git clone gitolite3@git.libre-soc.org:nmutil.git
599 * git clone gitolite3@git.libre-soc.org:ieee754fpu.git
600 * git clone gitolite3@git.libre-soc.org:nmigen-soc.git
601 * git clone gitolite3@git.libre-soc.org:soc.git
603 In each of these directories, in the order listed, track down the
604 `setup.py` file, then, as root (`sudo bash`), run the following:
606 * python3 setup.py develop
608 The reason for using "develop" mode is that the code may be edited
609 in-place yet still imported "globally". There are variants on this theme
610 for multi-user machine use however it is often just easier to get your
611 own machine these days.
613 The reason for the order is because soc depends on ieee754fpu, and
614 ieee754fpu depends on nmutil
616 If "`python3 setup.py install`" is used it is a pain: edit, then
617 install. edit, then install. It gets extremely tedious, hence why
618 "develop" was created.
620 If you prefer you can use this script instead: of course you checked it
621 in advance and accept full responsibility.
622 <https://git.libre-soc.org/?p=dev-env-setup.git;a=blob;f=hdl-dev-repos;hb=HEAD>
628 * new members, add yourself to the [[about_us]] page and create yourself
629 a home page using someone else's page as a template.
630 * communicate on the mailing list or the bugtracker an intent to take
631 responsibility for a particular task.
632 * assign yourself as the bug's owner
633 * *keep in touch* about what you are doing, and why you are doing it.
634 * edit your home page regularly, particularly to track tasks so that
635 they can be paid by NLNet.
636 * if you cannot do something that you have taken responsibility for,
637 then unless it is a dire personal emergency please say so, on-list. we
638 won't mind. we'll help sort it out.
640 Regarding the above it is important that you read, understand, and agree
641 to the [[charter]] because the charter is about ensuring that we operate
642 as an effective organisation. It's *not* about "setting rules and meting
647 for actual code development
651 * plan in advance to write not just code but a full test suite for
652 that code. **this is not optional**. large python projects that do not
653 have unit tests **FAIL** (see separate section below).
654 * Prioritise writing formal proofs and a single clear unit test that is more
655 like a "worked example".
656 We receive NLNet funds for writing formal proofs, plus they
657 cover corner cases and take far less time to write
659 ### Commit tested or zero-dependent code
661 * only commit code that has been tested (or is presently unused). other
662 people will be depending on you, so do take care not to screw up.
663 not least because, as it says in the [[charter]] it will be your
664 responsibility to fix. that said, do not feel intimidated: ask for help
665 and advice, and you'll get it straight away.
669 * commit often. several times a day, and "git push" it. this is
670 collaboration. if something is left even overnight uncommitted and not
671 pushed so that other people can see it, it is a red flag. if you find
672 yourself thinking "i'll commit it when it's finished" or "i don't want to
673 commit something that people might criticise" *this is not collaboration*,
674 it is making yourself a bottleneck. pair-programming is supposed to help
675 avoid this kind of thing however pair-programming is difficult to organise
676 for remote collaborative libre projects (suggestions welcomed here)
678 ### Enable editor auto-detection of file changes by external programs
680 This is important. "`git pull`" will merge in changes. If you then
681 arbitrarily save a file without re-loading it, you risk destroying
684 ### Absolutely no auto-generated output
686 * **do not commit autogenerated output**. write a shell script and commit
687 that, or add a `Makefile` to run the command that generates the output, but
688 **do not** add the actual output of **any** command to the repository.
689 ever. this is really important. even if it is a human-readable file
690 rather than a binary object file.
691 it is very common to add pdfs (the result of running `latex2pdf`) or
692 configure.in (the result of running `automake`), they are an absolute
693 nuisance and interfere hugely with git diffs, as well as waste hard
694 disk space *and* network bandwidth. don't do it.
696 ### Write commands that do tasks and commit those
698 * if the command needed to create any given autogenerated output is not
699 currently in the list of known project dependencies, first consult on
700 the list if it is okay to make that command become a hard dependency of
701 the project (hint: java, node.js php and .NET commands may cause delays
702 in response time due to other list participants laughing hysterically),
703 and after a decision is made, document the dependency and how its source
704 code is obtained and built (hence why it has to be discussed carefully)
705 * if you find yourself repeating commands regularly, chances are high
706 that someone else will need to run them, too. clearly this includes
707 yourself, therefore, to make everyone's lives easier including your own,
708 put them into a `.sh` shell script (and/or a `Makefile`), commit them to
709 the repository and document them at the very minimum in the README,
710 INSTALL.txt or somewhere in a docs folder as appropriate. if unsure,
711 ask on the mailing list for advice.
713 ### Keep commits single-purpose
715 * edit files making minimal *single purpose* modifications (even if
716 it involves multiple files. Good extreme example: globally changing
717 a function name across an entire codebase is one purpose, one commit,
718 yet hundreds of files. miss out one of those files, requiring multiple
719 commits, and it actually becomes a nuisance).
721 ### Run unit tests prior to commits
723 * prior to committing make sure that relevant unit tests pass, or that
724 the change is a zero-impact addition (no unit tests fail at the minimum)
726 ### Do not break existing code
728 * keep working code working **at all times**. find ways to ensure that
729 this is the case. examples include writing alternative classes that
730 replace existing functionality and adding runtime options to select
731 between old and new code.
733 ### Small commits with relevant commit message
735 * commit no more than around 5 to 10 lines at a time, with a CLEAR message
736 (no "added this" or "changed that").
737 * if as you write you find that the commit message involves a *list* of
738 changes or the word "and", then STOP. do not proceed: it is a "red flag"
739 that the commit has not been properly broken down into separate-purpose
740 commits. ask for advice on-list on how to proceed.
742 ### Exceptions to small commit: atomic single purpose commit
744 * if it is essential to commit large amounts of code, ensure that it
745 is **not** in use **anywhere** by any other code. then make a *small*
746 (single purpose) followup commit which actually puts that code into use.
748 This last rule is kinda flexible, because if you add the code *and* add
749 the unit test *and* added it into the main code *and* ran all relevant
750 unit tests on all cascade-impacted areas by that change, that's perfectly
751 fine too. however if it is the end of a day, and you need to stop and
752 do not have time to run the necessary unit tests, do *not* commit the
753 change which integrates untested code: just commit the new code (only)
754 and follow up the next day *after* running the full relevant unit tests.
756 ### Why such strict rules?
758 The reason for all the above is because python is a dynamically typed
759 language. make one tiny change at the base level of the class hierarchy
760 and the effect may be disastrous.
762 It is therefore worth reiterating: make absolutely certain that you *only*
763 commit working code or zero-impact code.
765 Therefore, if you are absolutely certain that a new addition (new file,
766 new class, new function) is not going to have any side-effects, committing
767 it (a large amount of code) is perfectly fine.
769 As a general rule, however, do not use this an an excuse to write code
770 first then write unit tests as an afterthought. write *less* code *in
771 conjunction* with its (more basic) unit tests, instead. then, folliw up with
772 additions and improvements.
774 The reason for separating out commits to single purpose only becomes
775 obvious (and regretted if not followed) when, months later, a mistake
776 has to be tracked down and reverted. if the commit does not have an
777 easy-to-find message, it cannot even be located, and once found, if the
778 commit confuses several unrelated changes, not only the diff is larger
779 than it should be, the reversion process becomes extremely painful.
783 * all code needs to conform to pep8. use either pep8checker or better
784 run autopep8. however whenever committing whitespace changes, *make a
785 separate commit* with a commit message "whitespace" or "autopep8 cleanup".
786 * pep8 REQUIRES no more than 80 chars per line. this is non-negotiable. if
787 you think you need greater than 80 chars, it *fundamentally* indicates
788 poor code design. split the code down further into smaller classes
791 ### Docstring checker
793 * TBD there is a docstring checker. at the minimum make sure to have
794 an SPD license header, module header docstring, class docstring and
795 function docstrings on at least non-obvious functions.
797 ### Clear code commenting and docstrings
799 * make liberal but not excessive use of comments. describe a group of
800 lines of code, with terse but useful comments describing the purpose,
801 documenting any side-effects, and anything that could trip you or other
802 developers up. unusual coding techniques should *definitely* contain
805 ### Only one class per module (ish)
807 * unless they are very closely related, only have one module (one class)
808 per file. a file only 25 lines long including imports and docstrings
809 is perfectly fine however don't force yourself. again, if unsure,
812 ### File and Directory hierarchy
814 * *keep files short and simple*. see below as to why
815 * create a decent directory hierarchy but do not go mad. ask for advice
820 * please do not use "from module import \*". it is extremely bad practice,
821 causes unnecessary resource utilisation, makes code readability and
822 tracking extremely difficult, and results in unintended side-effects.
824 Example: often you want to find the code from which a class was imported.
825 nirmally you go to the top of the file, check the imports, and you know
826 exactly which file has the class because of the import path. by using
827 wildcards, you have absolutely *no clue* which wildcard imported which
830 Example: sometimes you may accidentally have duplicate code maintained
831 in two or more places. editing one of them you find, puzzlingly, that
832 the code behaves in some files with the old behaviour, but in others it
833 works. after a massive amount of investigation, you find that the working
834 files happen to have a wildcard import of the newer accidental duplicate
835 class **after** the wildcard import of the older class with exactly the
836 same name. if you had used explicit imports, you would have spotted
837 the double import of the class from two separate locations, immediately.
839 Really. don't. use. wildcards.
841 ### Keep file and variables short but clear
843 * try to keep both filenames and variable names short but not ridiculously
844 obtuse. an interesting compromise on imports is "from ridiculousfilename
845 import longsillyname as lsn", and to assign variables as well: "comb =
846 m.d.comb" followed by multiple "comb += nmigen_stmt" lines is a good trick
847 that can reduce code indentation by 6 characters without reducing clarity.
849 Additionally, use comments just above an obtuse variable in order to
850 help explain what it is for. In combination with keeping the the module
851 itself short, other readers will not need to scroll back several pages
852 in order to understand the code.
854 Yes it is tempting to actually use the variables as
855 self-explanatory-comments and generally this can be extremely good
856 practice. the problem comes when the variable is so long that a function
857 with several parameters csn no longer fit on a single line, and takes
858 up five to ten lines rather than one or two. at that point, the length
859 of the code is adversely affected and thus so is readability by forcing
860 readers to scroll through reams of pages.
862 It is a tricky balance: basically use your common sense, or just ask
863 someone else, "can you understand this code?"
865 ### Reasons for code structure
867 Regarding code structure: we decided to go with small modules that are
868 both easy to analyse, as well as fit onto a single page and be readable
869 when displayed as a visual graph on a full UHD monitor. this is done
872 * using the capability of nmigen (TODO crossref to example) output the
873 module to a yosys ilang (.il) file
874 * in a separate terminal window, run yosys
875 * at the yosys prompt type "read_ilang modulename.il"
876 * type "show top" and a graphviz window should appear. note that typing
877 show, then space, then pressing the tab key twice will give a full list
878 of submodules (one of which will be "top")
880 You can now fullsize the graphviz window and scroll around. if it looks
881 reasonably obvious at 100% zoom, i.e the connections can be clearly
882 related in your mind back to the actual code (by matching the graph names
883 against signals and modules in the original nmigen code) and the words are
884 not tiny when zoomed out, and connections are not total incomprehensible
885 spaghetti, then congratulations, you have well-designed code. If not,
886 then this indicates a need to split the code further into submodules
887 and do a bit more work.
889 The reasons for doing a proper modularisation job are several-fold:
891 * firstly, we will not be doing a full automated layout-and-hope
892 using alliance/coriolis2, we will be doing leaf-node thru tree node
893 half-automated half-manual layout, finally getting to the floorplan,
894 then revising and iteratively adjusting.
895 * secondly, examining modules at the gate level (or close to it) is just
896 good practice. poor design creeps in by *not* knowing what the tools
897 are actually doing (word to experienced developers: yes, we know that
898 the yosys graph != final netlist).
899 * thirdly, unit testing, particularly formal proofs, is far easier on
900 small sections of code, and complete in a reasonable time.
902 ## Special warning / alert to vim users!
904 Some time around the beginning of 2019 some bright spark decided that
905 an "auto-recommend-completion-of-stuff" option would be a nice, shiny
906 idea to enable by default from that point onwards.
908 This incredibly annoying "feature" results in tabs (or spaces) being
909 inserted "on your behalf" when you press return on one line, for your
910 "convenience" of not needing to type lots of spaces/tabs just to get
911 to the same indentation level.
913 Of course, this "feature", if you press return on one line in edit
914 mode and then press "escape", leaves a bundle-of-joy extraneous
915 whitespace **exactly** where you don't want it, and didn't ask for it,
916 pooped all over your file.
918 Therefore, *please*: **before** running "git commit", get into the
919 habit of always running "git diff", and at the very minimum
920 speed-skim the entire diff, looking for tell-tale "red squares"
921 (these show up under bash diff colour-syntax-highlighting) that
922 inform you that, without your knowledge or consent, vim has
923 "helpfully" inserted extraneous whitespace.
925 Remove them **before** git committing because they are not part
926 of the actual desired code-modifications, and committing them
927 is a major and constant distraction for reviewers about actual
928 important things like "the code that actually *usefully* was
929 modified for that commit"
931 This has the useful side-effect of ensuring that, right before
932 the commit, you've got the actual diff right in front of you
933 in the xterm window, on which you can base the "commit message".
937 For further reading, see the wikipedia page on
938 [Test-driven Development](https://en.wikipedia.org/wiki/Test-driven_development)
940 This deserves its own special section. It is extremely important to
941 appreciate that without unit tests, python projects are simply unviable.
942 Python itself has over 25,000 individual tests.
944 This can be quite overwhelming to a beginner developer, especially one
945 used to writing scripts of only 100 lines in length.
947 Thanks to Samuel Falvo we learned that writing unit tests as a formal
948 proof is not only shorter, it's also far more readable and also, if
949 written properly, provides 100% coverage of corner-cases that would
950 otherwise be overlooked or require tens to hundreds of thousands of
953 No this is not a joke or even remotely hypothetical, this is an actual
956 The ieee754fpu requires several hundreds of thousands of tests to be
957 run (currently needing several days to run them all), and even then we
958 cannot be absolutely certain that all possible combinations of input have
959 been tested. With 2^128 permutations to try with 2 64 bit FP numbers
960 it is simply impossible to even try.
962 This is where formal proofs come into play.
964 Samuel illustrated to us that "ordinary" unit tests can then be written
965 to *augment* the formal ones, serving the purpose of illustrating how
966 to use the module, more than anything.
968 However it is appreciated that writing formal proofs is a bit of a
969 black art. This is where team collaboration particularly kicks in,
970 so if you need help, ask on the mailing list.
972 ## Don't comment out unit tests: add them first (as failures) and fix code later
974 Unit tests serve an additional critical purpose of keeping track of code
975 that needs to be written. In many cases, you write the unit test *first*,
976 despite knowing full well that the code doesn't even exist or is completely
977 broken. The unit test then serves as a constant and important reminder
978 to actually fix (or write) the code.
980 Therefore, *do not* comment out unit tests just because they "don't work".
981 If you absolutely must stop a unit test from running, **do not delete it**.
982 Simply mark it with an appropriate
983 ["skip" decorator](https://docs.python.org/3/library/unittest.html#skipping-tests-and-expected-failures),
984 preferably with a link to a URL in the [bugtracker](https://bugs.libre-soc.org/)
985 with further details as to why the unit test should not be run.
989 Find appropriate tutorials for nmigen and yosys, as well as symbiyosys.
991 * Robert Baruch's nmigen tutorials look really good:
992 <https://github.com/RobertBaruch/nmigen-tutorial>
993 * Although a verilog example this is very useful to do
994 <https://symbiyosys.readthedocs.io/en/latest/quickstart.html#first-step-a-simple-bmc-example>
995 * This tutorial looks pretty good and will get you started
996 <http://blog.lambdaconcept.com/doku.php?id=nmigen:nmigen_install> and
997 walks not just through simulation, it takes you through using gtkwave
999 * There exist several nmigen examples which are also executable
1000 <https://github.com/m-labs/nmigen/tree/master/examples/> exactly as
1001 described in the above tutorial (python3 filename.py -h)