*** tpb has joined #timvideos | 00:00 | |
*** riataman has joined #timvideos | 00:12 | |
*** riataman has quit IRC | 00:18 | |
cr1901_modern | mithro: Made a small bridge.py target (prob not gonna commit). Definitely misalignment in the spi flash | 00:28 |
---|---|---|
cr1901_modern | let's change it and see what happens... | 00:28 |
cr1901_modern | mithro: Okay it's working now up to the stub firmware | 00:49 |
*** tac-tics has quit IRC | 00:50 | |
mithro | cr1901_modern: \o/ | 00:51 |
*** CarlFK has joined #timvideos | 00:57 | |
*** ChanServ sets mode: +v CarlFK | 00:57 | |
cr1901_modern | mithro: ever see the following error? lm32-elf-newlib-gcc: error trying to exec 'cc1': execvp: No such file or directory | 01:03 |
cr1901_modern | I think lm32-elf-newlib-gcc needs to be updated | 01:03 |
* cr1901_modern is running download_env.sh again | 01:05 | |
cr1901_modern | okay download_env.sh didn't fix this, but "conda update gcc-lm32-elf-newlib" is installing a new version | 01:08 |
cr1901_modern | mithro: micropython is working | 01:10 |
cr1901_modern | on ice40hx8k-b-evn | 01:10 |
mithro | \o/ | 01:17 |
cr1901_modern | mithro: I am using the following script as a demo: http://ix.io/1og7 | 01:26 |
cr1901_modern | I assume you have a b-evn board as well> | 01:27 |
cr1901_modern | mithro: When you're ready, I can make PRs for litex (for platform parity w/ litex-buildenv) and litex-buildenv (for adding the base.py target) | 01:32 |
cr1901_modern | mithro: https://github.com/enjoy-digital/litex/pull/112 | 01:37 |
tpb | Title: build/platforms: Add ice40_hx8k_b_evn from Migen. by cr1901 · Pull Request #112 · enjoy-digital/litex · GitHub (at github.com) | 01:37 |
*** aic has joined #timvideos | 01:38 | |
mithro | cr1901_modern: Can you use the "merge_upstream.sh" script in third_party to move things forward? | 01:48 |
cr1901_modern | It didn't do anything when I ran it | 01:49 |
mithro | cr1901_modern: What was the output? | 01:49 |
cr1901_modern | All targets clean, good to update | 01:49 |
mithro | cr1901_modern: Can you past the complete output? | 01:49 |
mithro | cr1901_modern: You need to remove your "Bump litex submodule for platform parity (see #28)." commit first... | 01:50 |
mithro | Have to go now | 01:51 |
mithro | bblr (maybe) | 01:51 |
cr1901_modern | submodules changed but not updated | 01:51 |
cr1901_modern | mithro: What do you want me to do specifically? | 01:51 |
cr1901_modern | ... oh... | 01:52 |
shorne | xobs: Im at downstairs | 01:53 |
*** rohitksingh_work has joined #timvideos | 03:37 | |
*** sb0 has quit IRC | 03:37 | |
*** CarlFK has quit IRC | 04:37 | |
*** fullstopxH has joined #timvideos | 04:57 | |
*** fullstopxH has quit IRC | 05:01 | |
nrossi | mithro: Went to sleep early yesterday so I missed your message. I saw your comments on the doc and have added follow up comments | 05:03 |
mithro | nrossi: will look shortly... | 05:03 |
mithro | nrossi: I'm somewhat confused about the "native" name means in the OE? | 05:05 |
*** rohitksingh has joined #timvideos | 05:06 | |
nrossi | mithro: native means build host, so the host architecture that is running the build | 05:06 |
nrossi | mithro: so when refering to something like python, python-native would be the python that runs on the host with a host specific sysroot and site-packages. Where are python would be refering to the targets python and its associated sysroot | 05:07 |
*** rohitksingh has quit IRC | 05:08 | |
*** thomassgnLp has joined #timvideos | 05:09 | |
mithro | nrossi: For the moment, we could modify make.py to do the generate header step separately | 05:09 |
nrossi | mithro: most of my queries relate more in regards to how you want it to work. Since you can get anything to work without to much hassle | 05:11 |
mithro | nrossi: The make.py / mkimage.py / load.py / flash.py stuff could be moved out of litex-buildenv into a "litex-tools" module and the gateware directory end up as "litex-ugly" module or something? | 05:12 |
nrossi | mithro: I guess the real question is how much you want to move into an OE build and how much you want to keep as litex tooling, because there is a lot of overlap in what each tool (litex vs oe/bitbake) is doing | 05:13 |
mithro | nrossi: There seems to be a bit of overlap between litex tooling verse OE tooling? | 05:13 |
mithro | snap :-P | 05:13 |
nrossi | mithro: For example it makes at lot of sense to have all the software built within OE. But for the gateware parts keeping that within litex makes more sense | 05:14 |
mithro | nrossi: yeah, it might make sense for the BIOS to still be part of litex.... | 05:14 |
nrossi | mithro: maybe, but the issue is then you still have to replicate the software build tooling in both litex and oe | 05:15 |
nrossi | mithro: this is the same mistake Xilinx made with its xsct tools :) | 05:15 |
nrossi | mithro: though with that said i don't see any reason why litex couldn't generate all the code/files/etc of the bios firmware and then bitbake just builds it with the target toolchain | 05:17 |
*** thomassgnLp has quit IRC | 05:17 | |
mithro | nrossi: I think there is some complication by the fact that the toolchain has some interaction with SoC configuration | 05:18 |
nrossi | mithro: yer thats what i was referring to with regards to my "configuration feedback loop" comment | 05:18 |
mithro | btw - can OE provide precompiled / prebuilt packages? | 05:19 |
mithro | rather than an SDK, I mean like so people don't have to compile everything from scratch all the time? | 05:19 |
nrossi | mithro: so you can ship the sstate-cache, which is the build cache. Or you can ship packages (deb/rpm/ipk) of stuff | 05:20 |
nrossi | mithro: I believe the eSDK is maybe the sort of setup you are looking for. Where you ship the build environment with prebuilt content | 05:21 |
mithro | nrossi: for example shipping the gcc toolchain for lm32 is smaller then shipping the gcc source | 05:21 |
nrossi | mithro: sure, but if you go and change any tune configuration (e.g. switch from lite to full soc config) then you would need to rebuilt part of the toolchain (libs/newlib/etc) | 05:22 |
nrossi | mithro: to be honest I have not played a lot with that stuff, but the eSDK is probably what you are looking for, I believe thats exactly how Xilinx ships petalinux (as a form of eSDK) | 05:23 |
mithro | nrossi: well, most of the firmware links against libbase which is built from scratch each time (rather than shipping a C library with the compiler) | 05:23 |
nrossi | mithro: in that case it should be able to get the gcc part of the toolchain for sstate-cache. But if say you linked against libgcc, then you would need to pull the gcc sources to rebuild libgcc | 05:24 |
nrossi | from sstate-cache* | 05:26 |
mithro | nrossi: So, how do we deal with the closed source FPGA toolchains? | 05:26 |
nrossi | mithro: they are a pain from a number of angles, it all really depends on how clean and automated you want it to behave | 05:27 |
nrossi | mithro: but for example, you can't ship those in and eSDK or ship inside sstate since you are restricted from distribution by legal requirements/license agreement | 05:27 |
mithro | yeap | 05:28 |
mithro | they have to be installed manually | 05:28 |
nrossi | installed manually maybe, I haven't used Altera/Intel or Lattices vendor tools but I do remember Xilinx tools could be installed offline via command line | 05:29 |
nrossi | or has that changed | 05:29 |
mithro | nrossi: Hrm, technically they can but you wouldn't really want to be installing it as part of a build.... | 05:30 |
mithro | It's 20+ GB | 05:30 |
nrossi | mithro: i has some notes in the doc about reasons why you might want to do that | 05:30 |
nrossi | mithro: around dependencies and distro comptaiblity being a challenge which OE could wrap | 05:31 |
nrossi | mithro: 20GB is cheap compared to wasting hours farting around with vendor installers :P | 05:31 |
nrossi | mithro: also you could add steps to reduce the size of the install... like removing random peoples home directories hidden in the install ;) | 05:33 |
mithro | nrossi: Well, the less work I can do the better :-P | 05:42 |
nrossi | mithro: I will add the "17GB" download of Vivado to my two pieces of string download queue and have a look at seeing how much can be automated/etc of at least the Xilinx tools | 05:46 |
mithro | https://github.com/SymbiFlow/prjxray-dev-env | 05:47 |
tpb | Title: GitHub - SymbiFlow/prjxray-dev-env: Docker-based standardized development environment for prjxray (at github.com) | 05:47 |
mithro | https://github.com/SymbiFlow/prjxray-dev-env/blob/993a304f9967aae9360e396eb92fefe7a69acb0a/Dockerfile#L36-L53 | 05:48 |
tpb | Title: prjxray-dev-env/Dockerfile at 993a304f9967aae9360e396eb92fefe7a69acb0a · SymbiFlow/prjxray-dev-env · GitHub (at github.com) | 05:48 |
nrossi | Oh i think i have the installer for 2017.2 | 05:50 |
*** MassDebates_ has joined #timvideos | 05:56 | |
*** MassDebates_ has quit IRC | 05:57 | |
mithro | nrossi: bed time for me | 06:05 |
nrossi | mithro: good night | 06:08 |
shorne | mithro: I litex-buildenv to run the simulator, it seems to work too | 08:15 |
shorne | for some reason <enter> is not running so I cant enter commands into the H2U console | 08:27 |
*** futarisIRCcloud has quit IRC | 08:57 | |
*** ryantm has joined #timvideos | 09:00 | |
*** ryantm has quit IRC | 09:04 | |
*** sb0 has joined #timvideos | 09:05 | |
*** puck_ has quit IRC | 09:59 | |
*** puck_ has joined #timvideos | 10:00 | |
*** ErnestG_xc has joined #timvideos | 10:03 | |
*** ErnestG_xc has quit IRC | 10:04 | |
*** refi64Vp has joined #timvideos | 10:14 | |
*** refi64Vp has quit IRC | 10:18 | |
*** Ziemasca has joined #timvideos | 10:41 | |
*** Ziemasca has quit IRC | 10:46 | |
*** samebchasePk has joined #timvideos | 11:07 | |
*** samebchasePk has quit IRC | 11:10 | |
*** bkraptoryo has joined #timvideos | 11:19 | |
*** bkraptoryo has quit IRC | 11:21 | |
*** CarlFK has joined #timvideos | 11:28 | |
*** ChanServ sets mode: +v CarlFK | 11:28 | |
*** kajusgf has joined #timvideos | 11:35 | |
*** metaglogRe has joined #timvideos | 11:42 | |
*** kajusgf has quit IRC | 11:46 | |
*** sb0 has quit IRC | 11:58 | |
*** rohitksingh_work has quit IRC | 12:35 | |
*** rohitksingh has joined #timvideos | 13:25 | |
*** c0rn3j has joined #timvideos | 14:03 | |
*** c0rn3j has quit IRC | 14:10 | |
*** CarlFK has quit IRC | 14:17 | |
*** WizBrightxd has joined #timvideos | 14:26 | |
*** WizBrightxd has quit IRC | 14:26 | |
*** froztbyte has quit IRC | 14:35 | |
*** froztbyte has joined #timvideos | 14:35 | |
*** froztbyte has joined #timvideos | 14:35 | |
*** sb0 has joined #timvideos | 14:37 | |
*** BeerSerclP has joined #timvideos | 14:37 | |
*** BinaryKittenfb has joined #timvideos | 14:59 | |
*** BinaryKittenfb has quit IRC | 15:04 | |
*** samsagaz has quit IRC | 15:06 | |
*** \malex\Hl has joined #timvideos | 16:18 | |
*** \malex\Hl has quit IRC | 16:21 | |
*** rohitksingh has quit IRC | 17:23 | |
cr1901_modern | _florent_: https://github.com/enjoy-digital/litex/pull/112#issuecomment-426901042 Actually there are some subtle issues, but I don't think the icestorm backend is affected | 18:00 |
tpb | Title: build/platforms: Add ice40_hx8k_b_evn from Migen. by cr1901 · Pull Request #112 · enjoy-digital/litex · GitHub (at github.com) | 18:00 |
cr1901_modern | the litex backends don't distinguish between litex.build.generic_platform and migen.build.generic_platform exports, so using a migen target in litex can possibly fail if the backend uses isinstance(Pins), etc | 18:02 |
cr1901_modern | (also, you didn't merge in the platform :P) | 18:02 |
*** rohitksingh has joined #timvideos | 18:04 | |
*** rohitksingh has quit IRC | 18:27 | |
*** rohitksingh has joined #timvideos | 18:38 | |
*** rohitksingh has quit IRC | 19:27 | |
*** rohitksingh has joined #timvideos | 19:28 | |
*** rohitksingh has quit IRC | 19:37 | |
*** namyzarc has joined #timvideos | 19:56 | |
*** CarlFK has joined #timvideos | 20:01 | |
*** ChanServ sets mode: +v CarlFK | 20:01 | |
*** namyzarc has quit IRC | 20:03 | |
cr1901_modern | mithro: Will you be around tonight? | 20:20 |
*** zvedajH has joined #timvideos | 20:25 | |
*** Rashad has joined #timvideos | 20:28 | |
*** zvedajH has quit IRC | 20:30 | |
*** Rashad has quit IRC | 20:32 | |
mithro | cr1901_modern: Maybe? | 20:44 |
cr1901_modern | Maybe we could get b-evn merged tonight? | 20:44 |
mithro | cr1901_modern: Why not right now? | 20:45 |
mithro | cr1901_modern: Looks like it worked on travis... | 20:45 |
cr1901_modern | mithro: Okay, now is fine too | 20:46 |
cr1901_modern | But merge-upstream.sh still did nothing | 20:46 |
mithro | cr1901_modern: Can you paste the complete output? | 20:46 |
mithro | cr1901_modern: Looking at https://github.com/timvideos/litex-buildenv/pull/81 it doesn't seem you need anything updated? | 20:47 |
tpb | Title: Add ICE40HX8K-B-EVN Platform by cr1901 · Pull Request #81 · timvideos/litex-buildenv · GitHub (at github.com) | 20:47 |
cr1901_modern | mithro: Strictly speaking, I don't | 20:48 |
cr1901_modern | but I wanted to add the platform to LiteX for parity w/ build-env | 20:48 |
mithro | cr1901_modern: Great! I'll merge it now then? | 20:48 |
cr1901_modern | If you're okay w/ the lack of parity, then sure | 20:49 |
mithro | cr1901_modern: Merged! | 20:49 |
mithro | cr1901_modern: So, if you do the following "git checkout upstream/master -b litex-roll; git submodule update; cd third_party; ./merge-upstream.sh; git log" | 20:50 |
mithro | cr1901_modern: What happens? | 20:50 |
cr1901_modern | mithro: _Now_ it works and all the submodules are updated | 20:53 |
cr1901_modern | Is there a way to make merge-upstream skip some subdirs (I have "third_party/litesdcard" which of course isn't merged into the tree at all yet) | 20:55 |
*** Kripton has quit IRC | 20:55 | |
cr1901_modern | I get "error: pathspec 'litesdcard' did not match any file(s) known to git." | 20:56 |
cr1901_modern | Well of course :P. I just move it out of the way for now | 20:56 |
*** Kripton has joined #timvideos | 21:02 | |
cr1901_modern | mithro: Thanks for the quick merge. Be back in a few hours. | 21:12 |
mithro | cr1901_modern: you can merge a specific thing by giving it a directory | 21:31 |
mithro | ./merge-upstream.sh litex | 21:31 |
*** puck_ has quit IRC | 21:46 | |
mithro | cr1901_modern: Preference is always to merge ASAP :-) | 21:54 |
mithro | shorne: I've build your or1k toolchain for us to test with here -> https://travis-ci.org/mithro/conda-hdmi2usb-packages/jobs/437354271 | 22:38 |
cr1901_modern | mithro: Ack (still kinda sorta not around) | 22:44 |
*** reaVer has joined #timvideos | 22:56 | |
*** reaVer has quit IRC | 23:00 | |
*** CarlFK has quit IRC | 23:11 |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!