Thursday, 2022-06-09

programmerjakementioning here because the email is being slow: https://lists.libre-soc.org/pipermail/libre-soc-bugs/2022-June/006579.html06:41
programmerjakeThe totals were goofed today, I double checked the history and fixed it, assuming the amounts assigned to the child tasks are correct.06:42
programmerjakelkcl: can you check the calculations and if i forgot stuff that the fpga-fund paid for? #84810:02
programmerjakethx10:05
lkcljust round it up, to take into consideration your time. and also keep from having dumb amounts in the database10:06
programmerjakeyeah, i already did that, i'd like a double check on the rest of it though10:23
Veera[m]lkcl: will binutils-gdb-install be proper name? Or binutils-gdb-powerpc64le-install or binutils-gdb-ppc64le-install12:28
Veera[m]lkcl: For Bug: 84712:34
ghostmansd|2OK it seems I've finally fixed mp3 assembly13:47
*** ghostmansd <ghostmansd!~ghostmans@broadband-188-32-220-29.ip.moscow.rt.ru> has left #libre-soc13:47
*** ghostmansd|2 is now known as ghostmansd13:47
ghostmansdOnce Veera[m] makes this toolchain the primary one and I add support for macros, I think we can drop this intermediate .sv file13:50
lkclbinutils-gdb-install is fine, it will only ever be ppc64le14:18
ghostmansdlkcl, raised https://bugs.libre-soc.org/show_bug.cgi?id=84914:38
ghostmansdcould you please adopt it so that it considers your parts?14:38
ghostmansdI haven't touched 550 yet, no need to edit it twice :-)14:39
ghostmansdI checked binutils code today, it'll be quite a big task, if we really intend to re-use binutils machinery; but profitable, for sure14:39
ghostmansdyes, one of the culprits is that binutils don't have a standalone pass on macro substitution, this happens inside md_assemble, on-the-fly, so to speak14:40
ghostmansdanother one is that binutils operands know only insn value14:41
ghostmansdalso some changes to lexer, perhaps another operand flag...14:41
ghostmansdall in all, this will be a big task, but at least we won't have to invent everything all over again14:41
ghostmansdactually I kinda like that we found it now; the parsing I did was a really quick and dirty, almost the translation of pysvp64asm, so it might have missed more from binutils point of view14:42
ghostmansdI guess once 849 is done we can begin migrating all stuff to binutils14:43
ghostmansd(if there're no errors at all in translation, and this is yet to be checked)14:43
lkclghostmansd[m], added to 849.16:18
lkclbtw please do put links to commits in https://bugs.libre-soc.org/show_bug.cgi?id=84416:18
lkclfound one16:19
lkclhttps://git.libre-soc.org/?p=binutils-gdb.git;a=commitdiff;h=eea73a560f1d34ba8644f7f127c8d75c7ec6276116:19
lkclif there's any other do add it16:19
Veera[m]<ghostmansd> "Once Veera makes this toolchain..." <- Hi, what ya prefer host: x86_64 or ppc64le(talos)?16:24
lkclVeera[m], both.16:26
ghostmansd[m]Both yeah. Via config param I suppose.16:27
ghostmansd[m]Or perhaps via environment variable.16:28
Veera[m]I will make script for both. Immediately what you need?16:28
Veera[m]I thinking of first for talos (ppc64le) and then x86_64!16:29
ghostmansd[m]I think you can even omit host.16:29
ghostmansd[m]configure guesses this anyway.16:29
ghostmansd[m]So only --target16:29
ghostmansd[m]That is, simply drop --host, and check if it works both on Talos and x86.16:30
Veera[m]That I can make out. Are you doing development now in talos and x86_64 simultaneously.16:31
ghostmansd[m]Kinda. I run tests there. Plus its PPC, so native.16:34
Veera[m]ok16:34
ghostmansd[m]And most devs use it I think.16:34
Veera[m]H.J.Lu releases testing releases(!) (tag) for binutils(perhaps for x86_64). So did you forked from there(should?). svp64 branch shows some random fork.16:36
ghostmansd[m]Nope, I forked from upstream. Perhaps I should update it, it's outdated a bit.16:38
ghostmansd[m]I think you can rely on branch svp64 w/o tags16:39
ghostmansd[m]They likely won't appear soon, not unless we're merged.16:39
ghostmansd[m]This is WiP.16:39
Veera[m]ghostmansd[m]: ye16:39
lkclVeera[m], only one script is needed because the target is the same regardless of the host18:22
lkclas i mentioned yesterday here on IRC the only "problem" is to ensure that the target *always* has the prefix "powerpc64le-linux-gnu-" even when host==target==powerpc64le18:24
lkclautoconf *may* detect this and go "oh, you are doing native.  let me just overwrite /usr/bin/ld for you"18:24
lkclyou need to SPECIFICALLY ensure that does NOT happen18:24
lkcland it's probably as simple as always setting "--target=powerpc64le-linux-gnu"18:26
lkclwith that done i would *not* expect you to have any problems or in fact any differences in the script of any kind.18:26
lkclbtw please don't use --prefix=/xyz in the script18:27
lkclthe default is /usr/local and that's where the programs need to be installed18:27
lkclnowhere else18:27
ghostmansdyeah totally agree with lkcl18:29
lkclghostmansd, how did you build it?18:30
ghostmansdfrom my point of view, the best'd be to have `mkdir build && cd build && ../configure --target=powerpc64le-linux-gnu`18:30
ghostmansdlkcl, reading my mind again ^^^18:30
ghostmansdof course `make -j$(nproc)` afterwards18:31
ghostmansdsimple and sweet; you _can_ reinvent autoconf magic and detect host automatically... but _why_?18:31
ghostmansdor, rather, _what_for_?18:32
ghostmansdreminds me of an old demotivator meme: http://apikabu.ru/img_n/2012-10_4/jn2.jpg18:34
ghostmansdthat's in Russian, an approximate translation is...18:34
ghostmansd"So with some simple tools you can make trolleybus from bread.. but what for? (Linux at home... what for?)"18:34
ghostmansdVeera[m], I've synced with the recent binutils-gdb upstream19:13
lkcllol19:53
tinybronca[m]Hi all!  This just caught my attention, they claim to have a company "Open Silicon" https://www.sifive.com/custom-soc-solutions23:52
tinybronca[m]what exactly is "open" about this?23:52

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