Tuesday, 2023-01-17

openpowerbot[mattermost] <gayatri> Thanks luke. I got it.00:09
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc00:13
octaviusprogrammerjake, if you still have the chroot you've used to compile microwatt.bit for ECP5, can you let me know your yosys, ghdl, ghdl-yosys-plugin commits?01:32
octaviusI tried a yosys one commit behind the one we use, and still have the same issue with01:33
octaviusERROR: cell type '$mem_v2' is unsupported (instantiated as 'soc0.processor_internal.processor.icache_0.itlb_tags')01:33
*** octavius <octavius!~octavius@92.40.169.98.threembb.co.uk> has quit IRC01:38
programmerjakei just used my main dev repo, but i enabled the DOCKER flag when building Microwatt, so i expect it used all the tools in whatever docker image it used, rather than the ones i installed for libre-soc01:54
programmerjake*main dev folders01:54
programmerjakelemme see if i can find which version that docker image was...01:55
programmerjakeactually the flag is USE_DOCKER02:00
programmerjakehere's the images I still have on my computer after removing the obviously unrelated ones:02:01
programmerjakedebian                                                              10                528ac3ebe420   6 weeks ago     114MB02:01
programmerjakehdlc/nextpnr                                                        ecp5              3af0d88e38c4   10 months ago   1.03GB02:01
programmerjakehdlc/ghdl                                                           yosys             b2bb9f1fe687   10 months ago   801MB02:01
programmerjakehdlc/prjtrellis                                                     latest            3c29a6a46c18   10 months ago   924MB02:01
programmerjakedebian                                                              buster            ae8514941ea4   2 years ago     114MB02:01
programmerjakedebian                                                              latest            971452c94376   2 years ago     114MB02:01
programmerjakeheader line:02:02
programmerjakeREPOSITORY                                                          TAG               IMAGE ID       CREATED         SIZE02:02
programmerjakethe git commits I have for the repos outside of docker (they're unlikely to have been used when I built Microwatt):02:09
programmerjakeyosys eb5f9d9de61411d22944c105cf3800f121f9066602:09
programmerjakenextpnr 0a70b9c992c06a7553725b3742052eb95abd5f2002:09
programmerjakeprjtrellis 3ae21cf6a07f3883fafa5bf31e9104dfa6c9a63f02:09
programmerjakeI don't have any ghdl repos cloned02:09
programmerjakeI just used the one from that docker image02:10
programmerjakeyosys version from b2bb9f1fe687:02:16
programmerjakeYosys 0.14+60 (git sha1 08c771078, clang 11.0.1-2 -fPIC -Os)02:16
programmerjakedocker image ^02:16
programmerjakenextpnr-ecp5 version from image 3af0d88e38c4:02:17
programmerjakenextpnr-ecp5 -- Next Generation Place and Route (Version nextpnr-0.2-25-g0a70b9c9)02:17
programmerjakeghdl version from image b2bb9f1fe687:02:19
programmerjakeGHDL 2.0.0-dev (v1.0.0-1014-g480451e0) [Dunoon edition]02:19
lkclprogrammerjake: octavius needs to start from the dev-scripts (and explicitly inform us that that is what he has done)02:29
lkclit is wasting significant time - his, yours, everyone's - not knowing this information and him working from a completely unknown state02:29
lkcl(and him not being able to distinguish that because he doesn't read the error output, or more to the point know what it means)02:30
lkcli'll speak with him about this in the morning (in about 8 hours time) on a conf call02:30
programmerjakek, though knowing commits that I got working would still be useful if the dev scripts are broken somehow...idk if that's the case02:30
lkclplease don't spend time on this - he should only be using the versions installed from the dev-scripts02:30
lkclgood point02:31
lkclalthough they should be precisely-pinned explicit tags02:31
lkcl(except for the nmigen ones, which are deliberately a moving target. we need to sort that, for ls180, at some point, by retrospectively creating specific scripts for reproducible-ls180-FINAL7.)02:32
programmerjakeI'm unable to determine the version of the ghdl yosys plugin by inspecting the docker image, only the .so file is copied to the image when it was made and that file doesn't seem to contain the git commit hash02:37
programmerjakeI'll stop working on that now02:37
programmerjakeif you really need it, a good guess would be taking the commit from master immediately before 2022-03-02T01:57:02.13301508Z when the image was built02:38
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC02:54
*** yambo <yambo!~yambo@069-145-120-113.biz.spectrum.com> has joined #libre-soc03:09
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC03:26
lkclthis is good because you should under no circumstances be advising people to use anything other than the reproducible-build authorised and approved revisions of all and any tools and software03:33
lkclif you do so it is absolutely critical and essential that you inform them at the time that they are on their own, operating with unauthorised software, and that they should expect to be told to stop requesting "support" on any libre-soc forums03:34
lkclit is really, really important that you do not waste your time - or theirs - on anything other than authorised tools03:34
lkclif you find someone using unauthorised tools and it is clear and obvious that they are completely out of their depth your first and ONLY response should be03:35
lkcl"please install the AUTHORISED tools and only get back to this forum once you have done so or you run into difficulties with the devscripts"03:35
lkclif there turn out to be problems with the reproducible-build scripts03:36
lkcland those *problems* are themselves reproducible03:36
lkclTHEN and ONLY then do we step "outside of the box"03:36
lkcloctavius has literally wasted an entire week of his time because he wasn't running the authorised tools03:36
lkclhe's also wasted several hours of your time03:37
lkcli told him a few weeks ago that he is not to attempt to perform upgrades inside the chroots, but only to start again from scratch03:38
lkclbecause the older devscripts created incorrect chroots03:38
lkcl(buster-updates *without* apt-pinning)03:38
lkclthat can only be corrected *if you know what you are doing*03:39
lkcland he doesn't know what he's doing03:39
lkclprogrammerjake, bottom line: you *MUST NOT* advise people to "install this, install that"03:39
lkclthe *only* thing - every time - is for you to ask, "have you run the dev-scripts from scratch" and if the answer is "no", instruct them to do so03:40
lkclthat is the *only* thing you must do03:40
lkclanything else is wasting _your_ time - and theirs03:40
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc04:00
programmerjakek04:04
*** Ritish <Ritish!~Ritish@2401:4900:4aa8:c9ad:197:228a:4b1f:f347> has joined #libre-soc04:21
Ritishahh, Thanks lkcl, on it!04:21
Ritishdidn't notice the message yesterday04:21
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC04:38
*** Ritish75 <Ritish75!~Ritish@2401:4900:4aa8:c9ad:197:228a:4b1f:f347> has joined #libre-soc04:42
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc05:06
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC06:26
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc06:46
*** ghostmansd <ghostmansd!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC07:37
*** Ritish <Ritish!~Ritish@2401:4900:4aa8:c9ad:197:228a:4b1f:f347> has quit IRC07:43
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC07:50
*** octavius <octavius!~octavius@92.40.168.206.threembb.co.uk> has joined #libre-soc08:20
octaviusprogrammerjake, lkcl, I'm sorry for wasting your time.08:26
octaviusInstead of learning how to view the debug logs (or how to direct stdout to file to check the scripts themselves), I persisted  trying to brute-force the flow to work. My stubbornness got the better of me.08:26
octaviusAs for starting with devscripts, that *is* what I'm doing, however as you said Luke, I'm not explaining *exactly* what I'm doing.08:26
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc08:37
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC09:02
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc09:56
*** Ritish <Ritish!~Ritish@2401:4900:4aa8:c9ad:cabe:370f:2b4c:18d4> has joined #libre-soc10:02
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC10:39
*** Ritish <Ritish!~Ritish@2401:4900:4aa8:c9ad:cabe:370f:2b4c:18d4> has quit IRC11:52
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc12:04
*** octavius <octavius!~octavius@92.40.168.206.threembb.co.uk> has quit IRC12:19
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC13:02
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc13:38
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC14:02
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.168.67> has joined #libre-soc14:27
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.168.67> has quit IRC14:55
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.168.67> has joined #libre-soc14:55
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.168.67> has quit IRC15:02
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.55.211> has joined #libre-soc15:02
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@176.59.55.211> has quit IRC15:13
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc15:16
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC15:21
lkclok so if you ran the devscripts and you ended up with nmigen-0.2 from pypi that's a really serious problem with the devscripts that needs fixing.15:38
lkcli'm going to run a new chroot to confirm15:38
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc15:47
* lkcl done chroot, installing deps15:49
lkclah. i wonder if the installation of qemu from buster-backports does anything stupid16:07
lkcl(like, extra stupid that wasn't previously done, some time in the past... 4-5 months)16:07
lkclinstall-hdl-apt-reqs done16:08
* lkcl install-dev-repos next16:08
lkclalthough really should do hdl-tools-yosys first16:08
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC16:09
lkclwtf nmigen-0.1 being installed16:10
lkclpypi is really, *really* getting on my wick16:11
lkclnmigen                 0.1.dev1205+g29dec30 /home/lkcl/src/nmigen16:13
lkclnmigen-boards          0.1.dev217+ga949308  /home/lkcl/src/nmigen-boards16:13
lkclnmigen-soc             0.1.dev53+gfd2aaa3   /home/lkcl/src/nmigen-soc16:13
lkclnmigen-stdio           0.1.dev7+g475cb6f    /home/lkcl/src/nmigen-stdio16:13
lkclhmmm....16:13
lkclok this is ok16:14
lkcl(pip3 list)16:14
lkcllibresoc-ieee754fpu    0.0.1                /home/lkcl/src/ieee754fpu/src16:14
lkcllibresoc-nmutil        0.0.1                /home/lkcl/src/nmutil/src16:14
lkcllibresoc-openpower-isa 0.0.3                /home/lkcl/src/openpower-isa/src16:14
lkclc4m-jtag               0.1.dev152+gf5322d8  /home/lkcl/src/c4m-jtag16:14
lkcloctavius can you confirm that "pip3 list" in the chroot you are using shows those exact versions?16:15
programmerjakelkcl, that's not ok since later dependencies require v0.3...the issue is what we discovered earlier: the scripts delete the v0.2 tag hence the v0.1.0 version16:35
programmerjakefor nmigen specifically ^16:37
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc16:40
*** tplaten <tplaten!~isengaara@d536cd14.access.ecotel.net> has joined #libre-soc16:56
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC17:04
*** ghostmansd <ghostmansd!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc17:08
*** ghostmansd <ghostmansd!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC17:38
*** ghostmansd <ghostmansd!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc17:38
ghostmansdsetvl et al. are so fricking weird wrt non-zero operands17:39
ghostmansdIt feels kinda unnatural and strange to increment variable temporarily17:39
ghostmansdSpent a lot of time trying to understand why sv.setvl behaves differently than I expected (I expected the value to be decremented upon entry to assembly)17:40
ghostmansdHowever, what we do, and what I did, too, for now, is temporarily incrementing the operand... to let it be decremented gracefully17:40
ghostmansdSorry, I meant svstep17:50
ghostmansdBTW our assembly only covers svstep, not any operand which is non-zero.17:51
ghostmansdlkcl, answering `# FIXME(lkcl): should sv.svstep be like svstep?`18:01
ghostmansdLikely it should, otherwise it'll need a different operand class (why?)18:02
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc18:05
lkclprogrammerjake, i removed all dependencies from all setup.py configurations19:16
lkclghostmansd, i know.  however there is only 7 bits - would it be preferable to store the value "128" as 0b0000000? and not be able to store the value "0" at all?19:18
lkcland so on...19:18
ghostmansdYou mean, kinda, overflowing?19:19
ghostmansdWell perhaps this would make more sense...19:19
lkclyosys installed, nmigen, openpower-isa etc. installed19:19
ghostmansdAnyway, I did the same hack, just in a separate operand19:19
lkcloverflow-and-wrap, preventing and prohibiting the value "0" from being even express-able19:19
ghostmansdAnd added a check that this is "svstep", because somehow "setvl" does not follow the same logic19:20
ghostmansdThis is perhaps the strangest part: they both have non-zero operands19:20
ghostmansdBut handle them differently19:20
ghostmansdI think SVxz/SVxy/SVxz are also affected19:20
lkclyes, i went through that thought-experiment 18+ months ago19:22
lkclat the time restricted setvl to only allow values MAXVL=1..12819:22
lkclrather than (now) MAXVL=0..12719:22
lkcland likewise VL=1..12819:23
lkclrather than VL=0..12719:23
lkclbut then it all went to hell in a handbasket when writing the fail-first specification19:23
lkclbecause that *does* allow - require to be allowed - VL=019:24
lkcland i went, "arrrrgh" and reversed the decision, sacrificing the ability to set MAXVL=128 in the process19:24
ghostmansdLet's keep it for a while. Since there is an explicit SVP64Operand subclass which does this hack, it is managable.19:25
lkclSVx, SVy and SVz can all be thought of as "limit of the dimension inclusive in a for-loop"19:25
lkclfor i in [0..3]19:25
ghostmansdSVP64NonZeroOperand19:25
lkclnot "for i in range(4)"19:26
lkclthe assembler mnenomic/decode simply provides a "convenience", but of course requires the rule that the assembler mnemonic must throw a syntax error if the programmer attempts to set SVx=019:27
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC19:37
sadoon[m]Are we meeting tonight?19:53
programmerjakeyes, afaict19:53
programmerjakein 7min19:53
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc20:04
ghostmansdI won't attend today, sorry20:15
programmerjakenp, sleep well20:20
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC20:28
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc20:32
*** lx0 <lx0!~lxo@gateway/tor-sasl/lxo> has quit IRC20:59
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc21:05
*** yambo <yambo!~yambo@069-145-120-113.biz.spectrum.com> has quit IRC21:59
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC22:20
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc22:21
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC22:49
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc23:08
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC23:18
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc23:18
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC23:52

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!