*** ibot is now known as Guest596 | 05:51 | |
Madan_Kartheessa | Which version of Debian to be installed? Debian 10 or Debian 10.10 | 06:05 |
---|---|---|
lkcl | Madan_Kartheessa, by the time security updates are installed it results in debian 10.10 | 07:25 |
lkcl | you arrive at the same point | 07:25 |
Madan_Kartheessa | Thanks Luke. I will install Debian 10.10 | 07:33 |
Las[m] | Which parts of coriolis do you guys use? | 09:25 |
Madan_Kartheessa | I have no idea. | 10:44 |
lkcl | Las[m], alliance-check-toolkit for reference and checking modifications to coriolis2, coriolis2 itself, and Marie-Minerve ran hitas/yagle separately | 13:22 |
lkcl | Las[m], the answer is, strictly speaking, "we use coriolis2", which i appreciate doesn't make much sense :) | 13:23 |
lkcl | or, the answer is, "all of it" :) | 13:23 |
Las[m] | Well then I'll likely have to package all of it | 13:23 |
Las[m] | I am asking because there are 26 components in coriolis | 13:24 |
lkcl | it has several plugins, though, some of which are retired, but... yes | 13:24 |
lkcl | they're typically all used, because they have different purpopses / roles | 13:24 |
lkcl | for example, there's a plugin for creating an IO Ring | 13:25 |
lkcl | but yes, a typical user would need all of it. | 13:26 |
lkcl | alliance is also a dependency (interestingly that's packaged in debian) | 13:28 |
lkcl | and alliance-check-toolkit is useful / used as well | 13:28 |
lkcl | mikolajw: (for when you see the logs) if you can put a bit about yourself on http://libre-soc.org/about_us particularly what you've done (python? c?) etc. and perhaps what you'd _like_ to do, it's easier to find you something | 15:52 |
ghostmansd-pc | lkcl: am I right that mulhw test cases go into src/openpower/test/mul/long_mul_cases.py, not to test_caller.py? | 19:32 |
ghostmansd-pc | Out of curiosity. If so, do we have guidelines on what should go to test_caller.py? Recently I've added a new file test_caller_bcd.py, which tests BCD instructions. Am I right that the current pattern is to decouple the tests outside of test_caller.py? | 19:35 |
ghostmansd-pc | > go into src/openpower/test/mul/long_mul_cases.py | 19:37 |
ghostmansd-pc | It seems this file contains just some meta. What's the actual test case? | 19:37 |
programmerjake | the test cases are here: https://git.libre-soc.org/?p=openpower-isa.git;a=blob;f=src/openpower/test/mul/long_mul_cases.py;h=3247ccc8c8f4afa2fe52a90f35ad57f4703bd23a;hb=HEAD | 19:43 |
ghostmansd-pc | programmerjake: I mean, how to run it? $(python3 src/openpower/test/mul/long_mul_cases.py) does nothing (unsurprisingly, since there's no code aside of classes and other boilerplate). | 19:48 |
programmerjake | you run python3 src/soc/fu/mul/test/test_pipe_caller_long.py | 19:50 |
ghostmansd-pc | programmerjake: could you, please, clarify, where can I find it? $(find openpower-isa -name test_pipe_caller_long.py) yields nothing. | 20:26 |
programmerjake | in soc.git | 20:27 |
programmerjake | the tests were originally all in soc.git | 20:28 |
ghostmansd-pc | ModuleNotFoundError: No module named 'power_instruction_analyzer' | 20:33 |
ghostmansd-pc | IIRC we've discussed some pia-related issues; should I re-install it? | 20:33 |
programmerjake | yes...install using libre-soc-install.sh from git master | 20:35 |
programmerjake | you may need to update your version of Rust, update using `rustup update` | 20:36 |
programmerjake | libre-soc-install.sh is in power-instruction-analyzer's repo; the script in dev-env-setup is out-of-date | 20:44 |
ghostmansd-pc | I'll take a look at it, thank you! | 20:45 |
lkcl | programmerjake, you need to merge that into the script that is already there. | 21:35 |
lkcl | also, the entirety of pia needs to be made optional (and, if not, removed) | 21:35 |
lkcl | and ultimately replaced. | 21:35 |
lkcl | as an experiment to make code critically dependent on an unstable in-development language with no stable targets was a mistake | 21:36 |
lkcl | anything that is running HDL co-simulated against ISACaller *at the moment* is in the soc repo | 21:39 |
lkcl | however if all unit tests are properly abstracted out, it should be possible to (a) run all unit tests against anything and (b) co-simulate them against anything | 21:40 |
lkcl | the list of candidates is: | 21:40 |
lkcl | 1) qemu | 21:40 |
lkcl | 2) ISACaller | 21:40 |
lkcl | 3) power-gem5 | 21:40 |
lkcl | 4) libre-soC HDL | 21:40 |
lkcl | 5) microwatt | 21:41 |
lkcl | 6) native IBM POWER9 | 21:41 |
lkcl | 7) a simulator based on Peter Hsu's cavatools | 21:41 |
lkcl | thus openpower-isa becomes not just a "batch of unit tests" | 21:42 |
lkcl | it becomes a full Conformance Test Suite | 21:42 |
lkcl | suitable for testing *any* Power ISA system | 21:42 |
lkcl | (and at that point, pia is not appropriate to make it a hard critical dependency, and will need to be replaced) | 21:42 |
lkcl | this one https://git.libre-soc.org/?p=dev-env-setup.git;a=blob;f=pia-install;hb=HEAD | 21:45 |
lkcl | programmerjake: this is the script that needs to be updated to a fixed, stable point | 21:46 |
lkcl | i've not heard of libre-soc-install.sh, i have no idea what it is or where it is. | 21:46 |
lkcl | it's not in the dev-env-setup.git repository | 21:46 |
programmerjake | it's in power-instruction-analyzer.git, I did mention it when I closed the last pia bug | 21:59 |
programmerjake | also, Rust is decidedly not an "unstable in-development language with no stable targets" | 22:04 |
lkcl | i was expecting the existing dev-env-setup.git script to be updated. | 22:12 |
lkcl | now there's two scripts, one of which is referred to in all documentation and doesn't work | 22:12 |
lkcl | and another which is not referred to in any of the documentation | 22:13 |
programmerjake | sorry, I didn't think to look in dev-env-setup; I did update the wiki | 22:15 |
programmerjake | https://libre-soc.org/HDL_workflow/#index13h2 | 22:18 |
programmerjake | I had understood the wiki to be our canonical install instructions | 22:19 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!