Friday, 2023-02-17

*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc00:07
sadoon[m]It's a start00:16
programmerjakeafaict if you need to modify qemu to remove vsx/vmx support, you'd modify it here:
programmerjakefor experimentation you can just modify the power9 definition, for upstreaming you'd want to figure out how to copy it as a new cpu definition and modify that00:48
sadoon[m]That's awesome, thanks00:48
programmerjakei haven't tested it so that may not be the only place that needs to be modified00:49
sadoon[m]It's a good start00:49
programmerjakenote MSR_VR is referred to as msr.vec in the powerisa v3.1b pdf00:50
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc01:07
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC01:33
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc01:41
*** Ritish <Ritish!~Ritish@> has joined #libre-soc05:29
*** ghostmansd <ghostmansd!> has joined #libre-soc05:46
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC06:21
*** ghostmansd <ghostmansd!> has quit IRC06:23
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC06:32
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc06:33
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc06:33
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC06:58
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc06:59
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC07:05
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc07:08
Ritishlkcl, I have a tiny issue07:13
RitishSo, we have this power server at office, and Im trying to run the whole thing on that07:13
Ritishbut, when I am trying to use: schroot -c libresoc07:14
RitishIt says Im not in the passwd database07:14
Ritishoh wait, that error occured when I used sudo bash, inside the libresoc envt07:15
Ritishthough I am able to use sudo bash out of the libresoc envt07:15
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC07:16
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc07:17
programmerjakehmm, schroot by default will copy or bind mount /etc/passwd and related files, did we disable that?07:17
programmerjakethat only works if /etc/passwd actually contains your user's info though, if you're on a computer using LDAP, your account might not be in /etc/passwd (i haven't read LDAP's docs, so idk)07:20
Ritishoh :007:24
RitishI'll try checking that with the admin, thanks! :)07:24
programmerjakeapparently schroot 1.3 added LDAP support...07:25
programmerjakeyou should have 1.6 so it should work if we didn't disable it07:26
programmerjakeassuming your host system is debian07:27
Ritishyep, I have version 1.6.1007:27
Ritishoh no, the host system is unfortunately ubuntu TwT07:27
programmerjakeubuntu should work fine as the host...07:29
programmerjakecurrently checking if we do something funny with schroot's config...07:30
Ritishokii thanks ^^07:30
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC07:32
programmerjakeoh, discovered schroot is EOL,
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc07:34
programmerjakelkcl: thought you might find that interesting07:34
programmerjakeapparently we intentionally make schroot not copy the host's passwd or other auth files:07:43
programmerjakeah, found the bug: if your user id is not 1000, our script will assign you the wrong uid/gid in /etc/passwd and related files in the chroot:07:48
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC07:48
programmerjakeapparently you're the first one who bothered to run the script on a computer with multiple users and then try to run sudo07:49
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc07:50
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc07:50
Ritishwas in a meeting x.x07:50
RitishOh :D07:50
RitishSo, can I consider that I've found a bug xD07:51
programmerjakethough, otoh our talos server has multiple users, so maybe everyone else just ignored that quirk...07:52
programmerjakeyup, you found the bug :)07:52
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC07:54
Ritishthank you!! hehe *dances*07:55
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc07:55
RitishI am kind of relieved honestly07:58
RitishI actually first got that error I think 2 weeks back07:59
RitishBut I've some time making a couple of presentations and a little for my capstone project (which is on Litex)07:59
*** Knar <Knar!~Knar@> has joined #libre-soc08:00
KnarI am part of the team developing the transpiler under Abhinandan. just wanted to introduce myself08:02
Knarwe met up with @lkcl online last time08:03
Knarfunny thing was i had some difficulties finding this channel. lost track of IRC news. found out Libre-soc had moved on from freenode08:05
programmerjakenice to meet you!08:05
programmerjakeit should be linked on the main page at
programmerjakeit is, but may be easy to miss...08:07
Knaryea was scouring the site for updates found it later on08:08
Knari followed this
programmerjakeyeah, that page needs to be fixed, i'll create a bug for it08:10
programmerjakeRitish: i don't have time to fix it right now (since it's nearly 1am here), so i created
programmerjakemaybe luke will fix it08:46
programmerjakeotherwise i can try to fix it another day08:46
Ritishno problem!08:46
RitishThanks for the help :)08:47
*** Knar <Knar!~Knar@> has quit IRC08:49
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC08:49
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc08:49
Ritishrest well!08:52
*** knar <knar!~kinar@2401:4900:4a65:96ef:8f6d:a706:573c:5c45> has joined #libre-soc08:57
*** blind_grandma <blind_grandma!~kinar@2401:4900:4a65:96ef:fd78:f82f:16be:da1> has joined #libre-soc09:20
*** knar <knar!~kinar@2401:4900:4a65:96ef:8f6d:a706:573c:5c45> has quit IRC09:25
*** blind_grandma <blind_grandma!~kinar@2401:4900:4a65:96ef:fd78:f82f:16be:da1> has quit IRC09:31
lkclRitish: a workaround is to su to root on the host, then schroot, then su to the user *in* the chroot.09:33
lkclgiven that it is absolutely essential to have root access on the host in order to create the chroots, this should not in the least bit be a problem09:33
Ritishoh awesome09:40
RitishI do have root access09:40
Ritishsince I had to run devscripts09:40
Ritishthanks for the hack :)09:41
Ritishworks :D09:51
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC09:54
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc09:58
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC10:51
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc10:52
*** knar <knar!~kinar@2401:4900:4a65:96ef:fd78:f82f:16be:da1> has joined #libre-soc11:09
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC11:22
lkclyyep. programmerjake i closed the bugreport as it is invalid.  the solution is as above, which i used successfully on the University of Oregon Server named "silicon".11:23
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC11:23
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc11:23
*** knar <knar!~kinar@2401:4900:4a65:96ef:fd78:f82f:16be:da1> has quit IRC11:36
*** knar <knar!~kinar@2401:4900:4a65:96ef:fd78:f82f:16be:da1> has joined #libre-soc11:36
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc11:37
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC11:38
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc11:39
*** Ritish <Ritish!~Ritish@> has quit IRC11:54
*** blind_grandma <blind_grandma!~kinar@2405:201:d032:9bf4:24f8:69ef:b5a3:b632> has joined #libre-soc11:58
*** blind_grandma <blind_grandma!~kinar@2405:201:d032:9bf4:24f8:69ef:b5a3:b632> has joined #libre-soc12:00
Ryuno-KiAndrJaenIs it worth to add the steps as comment on the issue?12:02
Ryuno-KiAndrJaen(Haven't clicked the link)12:02
*** knar <knar!~kinar@2401:4900:4a65:96ef:fd78:f82f:16be:da1> has quit IRC12:02
*** blind_grandma <blind_grandma!~kinar@2405:201:d032:9bf4:24f8:69ef:b5a3:b632> has quit IRC12:02
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC12:17
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc12:19
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC12:24
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc12:29
sadoon[m]Gentoo running xfce4 now, time for firefox and webkit12:44
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC13:19
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc13:20
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC13:37
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc14:32
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC14:40
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC14:41
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc14:49
sadoon[m]firefox working well14:49
sadoon[m]esr 102.714:50
sadoon[m]oops, meant to type that in the command line lol14:50
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc15:17
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC15:49
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc15:51
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC16:14
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc16:15
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC16:21
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc17:15
sadoon[m]Epiphany (webkit-gtk based) is also working fine17:18
sadoon[m]I'll build a few more random applications I use and then proceed to the rebuild sffs17:18
lkclsadoon[m], webkit-gtk is a pretty high bar, there's 30+ complex package dependencies/libraries involved there, that's a really good sign17:26
sadoon[m]Oh no keep in mind this is still p917:26
sadoon[m]And it is by far easier to build cross platform than firefox17:26
sadoon[m]But I have a feeling we won't have trouble building it on sffs17:27
sadoon[m]Since it uses C code for its javascript on POWER anyways, no JIT17:27
lkclRyuno-KiAndrJaen, yes - and usually (because of the public logs) we put the IRClog link in the bugtracker.17:27
sadoon[m]It seems webkit-gtk also works on risc-v which corroborates my assumption that it is easier to port, mostly C/C++ code in it17:28
sadoon[m]I've previously run it successfully on G4 and G5 machines too17:29
sadoon[m]Extremely usable there17:29
lkclmarkos, you'll need this;a=commitdiff;h=b112cab083345d283ebd02126d30afc3280b1ed1;hp=182a9f262d887a7a306a4f81ce95318b991c5f5017:29
lkclstarts here;a=blob;f=src/openpower/decoder/isa/;hb=HEAD#l11117:30
lkclyou'll need a routine in assembler (ppc64 or svp64 it matters not) which loads the indices into the required registers17:30
lkclwhere they come from (start) you must *DOUBLE* the first argument of svindex17:31
lkcl 115             'svindex 11, 0, 1, 3, 0, 1, 0',  # SVSHAPE0, a17:31
lkcl 147             set_masked_reg(initial_regs, 22, i, ew_bits=8, value=a)17:31
lkclsvindex 11 --> register **22** is used as the start-point for reading the indices17:32
lkclexactly how those numbers get there is entirely up to you. you can store 8 indices per 64-bit register, but "svindex ...... ->>>3<<<- ..." has to be used to indiate "read indices as 8-bit values"17:35
lkcl 145         # offsets for a b c17:35
lkcl 146         for i, (a, b, c, d) in enumerate(schedule):17:35
lkcljust run the unit test, run a print() statement of what a,b,c,d is, then replicate that in sequence, as bytes, at the registers you choose to start from.17:36
lkclit looks like a lot of registers for the indices but actually it's not17:42
lkcl  74 def chacha_idx_schedule(x, fn=quarter_round_schedule):17:42
lkcl  75     fn(x, 0, 4,  8, 12)17:42
lkclqty 8 times 4 because it's a quarter-round17:42
lkcl*divided* by 8 because you can get qty8 indices per 64-bit register17:43
lkclso actually only QTY4 64-bit registers needed for each of a, b, c and d for a total of only 16 64-bit registers for the entire Indices Schedule normally done as massive loop-unrolling.17:43
lkcland it would be perfectly fine to do a sv.ld of pre-computed qty16 64-bit values17:44
lkclbut, ironically, it may be quicker (CPU/memory-bandwidth-wise) to calculate them. they're pretty regular.17:46
programmerjakesadoon, for firefox, did you remember to change the rust target as well? otherwise rust will use vsx/vmx instructions17:53
sadoon[m]I still did not start the rebuild but thanks for telling me17:54
programmerjakeyou need to change it when you build rust, not just when you build firefox, otherwise rust's standard library could have vsx/vmx instructions17:54
sadoon[m]Will do!17:54
programmerjakei'll create a patch for you17:58
sadoon[m]Awesome, thanks17:58
programmerjakewhich compile flags/target are you using for clang?17:59
sadoon[m]So far the plan is to change the COMMON_FLAGS18:00
sadoon[m]Does clang require specific changes as well?18:00
programmerjakeidk, i was just wondering if you were using a new target name, e.g. powerpc64lesffs-unknown-linux-gnu18:01
sadoon[m]Not yet, that's the next step to build it into gcc clang etc18:01
programmerjakei'd guess clang just needs the same config changes as gcc18:01
sadoon[m]-mno-altivec -mno-svx should be fine18:02
sadoon[m]On an unrelated note I desperately need a way to build the unity DE into debian instead of using ubuntu just for unity18:04
sadoon[m]It is available on gentoo as an overlay but that sounds like a can of worms18:09
programmerjakeafter some experimentation with clang, it turns out that, to have the proper target features set/unset, you need:18:38
programmerjake -mcpu=power9 -mno-altivec -mno-crypto -mno-htm18:38
programmerjakepower9 is needed to get isa v3.0 instructions enabled18:38
programmerjakeclang has a bug where -mno-altivec doesn't imply -mno-crypto, hence why both are needed. -mno-vsx *is* implied however18:39
sadoon[m]Good thing you told me, I'm starting the rebuild now18:44
sadoon[m]I added -mno-vsx just to be safe too18:45
programmerjakewhich version of rust do you have?18:53
programmerjakek, thx18:55
programmerjakei think i figured out the rust target modifications, building rust to test it...19:37
programmerjakeok, i have a working patch. lkcl: should i push the patched rust branch to a new repo on libre-soc or just put the patch somewhere, maybe utils.git?21:22
programmerjakelkcl: ( case your name has to be in front to be notified)21:25
programmerjakeif you do make a git repo, can you name it rust-ppc-sffs.git?21:27
programmerjakeoh, sorry sadoon i just realized i gave you flags that assume isa v3.0, but iirc we also want to support the E5500 for awilfox, lemme fix that21:41
sadoon[m]No harm in targeting v3.0 for now21:46
sadoon[m]And it's actually the other way around, awilfox is working on a distro already and wants to support libre-soc potentially21:47
programmerjakeposted the patch to bugzilla, i'll also push it to git after luke decides where it should go and possibly makes a new git repo22:05
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc22:22
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC22:34
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc23:02
sadoon[m]Awesome thanks23:28
sadoon[m]I'm having some technical difficulties, 3 GPUs have failed me xD23:29
sadoon[m]HD 7970, R9 290X, and HD 4890, all with various issues, and the undying HD 5450 chugs along23:29
sadoon[m]Luckily this is not my talos machine having issues so no worries there23:29
sadoon[m]I would keep using HD 5450s forever if it weren't for their slow performance on heavy desktops and their 3rd port being VGA.. Perhaps something higher up in the 5000 series would be a good upgrade23:30
programmerjakei'm using a rx 580 and haven't had any issues that i recall...23:47
sadoon[m]Yeah that's too modern to have any issues on xD23:47
sadoon[m]Mine were all bought used23:47
programmerjakethe rx 580 should be pretty cheap used too...23:47
sadoon[m]100$.. hmm23:48
sadoon[m]I do have a preference for fanless cards so even at that price I'd pass23:48
sadoon[m]Less things to break and worry about lol23:49
sadoon[m]I'd go for an AMD firepro single slot even if it wasn't fanless though, being single slot is nice23:49
programmerjakemine's 5-6 yr old and the fans are still fine even though i mined on it for 1 1/2 yr23:50
sadoon[m]Anyways, I got really frustrated with unity not being available on anything but ubuntu basically (there's an arch port and a gentoo port but meh), currently looking into switching my desktop and then switching to debian bookworm, I'm fed up23:51
sadoon[m]I tried to build unity on debian but it's too much of a hassle23:51
sadoon[m]At least this way I'll know the quirks of deb12 for when we do need to rebuild it23:52
programmerjakei wonder if it would work to use the ubuntu packages on debian...23:52
sadoon[m]Not quite, already tried xD23:52
sadoon[m]The choice now is between KDE, Cinnamon, Xfce, and Xfce + dwm which I made a while ago and still should work23:53
sadoon[m]Still too many choices to be comfortable and it's 3AM here and I'm sleepless lmao23:53
programmerjakewell, sleep well :)23:54
sadoon[m]The major advantage of xfce + dwm is that I tried it even on a PB G4 and still worked flawlessly, so having a unified DE would be nice23:54

Generated by 2.17.1 by Marius Gedminas - find it at!