Wednesday, 2023-08-02

*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC03:00
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc03:00
ghostmansd[m]Hi folks, yesterday I finally became a dad for the second time!06:20
ghostmansd[m]Sorry, I didn't manage to participate in the meeting, but the absence reason is respectful :-)06:21
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC07:30
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc07:30
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC07:35
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc07:35
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC07:41
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc07:42
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC08:29
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc08:29
*** guest_ <guest_!> has joined #libre-soc10:48
*** guest_ <guest_!> has quit IRC10:55
*** octavius <octavius!> has joined #libre-soc11:34
octaviusCongrats ghostman[m], great to hear :)11:39
octaviusoops ghostmansd[m]11:39
markos_ghostmansd[m], congratulations!11:52
sadoon[m]Congrats ghostmansd[m]  !! :D12:20
sadoon[m]<lkcl> "sadoon, yes, the ppc emulation..." <- Not much harm, it does mean we have to patch qemu but I have an incremental plan we can discuss tonight so that a- things don't take too long and b- we can start testing ASAP12:21
sadoon[m]The short version is to use qemu-user-static to chroot, patched with whatever needs to be done to disable or (this might be easier) print an error on any instruction not valid for sffs12:22
sadoon[m]It means we can never emulate a full running system but should be more than enough to test sffs compliance in user-space12:23
sadoon[m]Because making the kernel boot without VSX and AltiVec is more difficult than toggling them12:23
lkclghostmansd[m], wha-heeey! :)13:59
lkclsadoon[m], ahhh yeah that would work well, and yes the userspace is the important bit here. we *know* the kernel can be built (microwatt 5.7 by joel shenki)14:00
lkclbtw do look up this discussion and put the link into the bugreport14:01
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC14:06
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has joined #libre-soc14:12
lkclghostmansd[m], we'll take care of the MoU for Ongoing NLnet tonight14:25
sadoon[m]<lkcl> "btw do look up this discussion..." <- Will do now14:32
sadoon[m]re: The 5.7 microwatt kernel, I can look into that afterwards because it will be immensely useful later14:33
lkclit's on the bugreport - i leave it with you to take care of the project management and tracking for audit purposes (strong hint: without which i'm not signing-off on the RFPs!)14:41
sadoon[m]Don't worry about it14:42
octaviussadoon[m], I was looking over IRC logs and managed to find several conversations about your sffs porting. May I link them in bug 999 for you?14:48
sadoon[m]Sure would greatly appreciate it :D14:48
sadoon[m]Linked today's chatlog, made some typo fixes, and now creating the subtasks14:49
sadoon[m]I still enjoy that it's task 99914:53
octaviusGreat. I've also posted the chats. Found some chats from my ls2 verilator work I need to link to bug 1086 as well. Thanks lkcl, sadoon[m] for encouraging me to do a cleanup :)14:53
octaviusIndeed, nice number14:53
sadoon[m]I think we should break it into 3 subtasks in this heirarchy (typo unintended)14:59
sadoon[m]1- task 999 (sffs porting)15:00
sadoon[m]   a- task 1128 (qemu-user patch for testing)15:00
sadoon[m]        a1- task x (qemu-system + linux-microwatt-5.7)15:00
sadoon[m]   b- task x (wiki pages for gentoo and debian reproduction)15:00
sadoon[m]Aww it didn't indent them, you get the point though :P15:00
sadoon[m]I could feasibly merge b into 1 if it seems unnecessary15:01
sadoon[m]I didn't assign 1128 to a milestone since I presume it's going under this year's right?15:02
octaviusIt's probably better to have separate a and b tasks15:04
octaviusbecause b may become pretty substantial (and should have its own budget at least015:04
sadoon[m]I don't think it will be too difficult, probably two days if I don't do anything else15:06
sadoon[m]Unless debian proves too difficult to port with.. normal means15:07
octaviusIt's not like we're going to run out of bugs any time soon, may as well split them15:07
sadoon[m]Like I said gentoo is already done, I'm considering uploading the rootfs and config files tonight15:07
sadoon[m]When I say config files15:08
sadoon[m]It's one file xD15:08
sadoon[m]But it took weeks to get right so15:08
sadoon[m]A total of 141 packages are necessary for a debian chroot to run, with about 30 packages built for all archs. Another reason why I think mini-buildd will bring more headaches than we need right now, manual scripted building should be 100% fine and take less than two days once I get the build flags sorted15:34
markos_a trick I used long ago is to have the builder use the gentoo chroot to build the initial packages15:42
markos_once the main set of packages was built, I replaced the gentoo chroot with the actual debian chroot and rebuilt everything15:43
sadoon[m]That would be useful if we were doing a cross build, I'm very thankful that's not necessary since it's still power9 :)15:43
markos_but indeed, mini-buildd assumes a proper debian environment I think you're better off doing it manually for the first set of packages15:44
sadoon[m]crap, need to reboot because I had /tmp on noexec, need it for a bit15:44
markos_mini-buildd can be used later15:44
sadoon[m]Yeah definitely should use mini-buildd once we need the full debian repo15:44
sadoon[m]The chroot packages along with a few useful utilities should be enough for now15:44
sadoon[m]Back, that was one long reboot16:00
sadoon[m]Nice to see the debian team redoing their sbuild wiki page, it now includes automated ccache, eatmydata, and package mirroring :D17:06
sadoon[m]no need for manual setup17:06
sadoon[m]lkcl  so should I rewrite task 1128 so that it is only objdumping? I was under the impression this task had separate funding at first17:10
sadoon[m]And then create an independant task for qemu stuff17:10
sadoon[m]successfully forced sbuild to respect CFLAGs18:12
sadoon[m]Doing a tiny test with hello.deb to see if things are looking ok18:19
sadoon[m]Aha! this looks like VSX no? lxvd2x  vs33,r29,r1018:22
sadoon[m]A simple comparison showed that with the correct flags, no VSX instructions are there18:23
sadoon[m]I'll stop for today, huge progress18:25
programmerjakeone thing to watch out for is vsx/vmx code that won't get used because it checks for cpu capabilities before trying to run it18:30
programmerjakein particular glibc is likely to have code like that18:31
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC18:46
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc18:47
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC19:11
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc19:12
octaviuslkcl, I created a new bug for tracking the overall tasks I've been working on for ls, bug #1129. Although there is already bug #1037, I figured it's better to keep it more general19:26
*** ghostmansd[m] <ghostmansd[m]!> has quit IRC19:31
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has joined #libre-soc19:32
ghostmansd[m]Sorry folks, I was extremely busy for the recent days. I might be 5-10 mins late for meeting, still finishing the preparations for the tomorrow day when wife and kid go out of perinatal house.19:34
ghostmansd[m]Thanks for congratulations! I couldn't reply to them yesterday, so much calls and tasks...19:34
sadoon[m]<programmerjake> "in particular glibc is likely to..." <- Oh yeah and I expect things like nss and ffmpeg to also have them19:43
sadoon[m]But I don't think we'll be using ffmpeg for a while now lol19:43
sadoon[m]perhaps nss for some stuff19:43
sadoon[m]I can't hear anyone in the meeting20:02
octaviuslkcl, meeting20:03
sadoon[m]Tried different browsers and the app20:04
octaviusAnother try sadoon[m]?20:12
toshywoshysadoon: can you try one more time20:12
toshywoshyuse firefox, that should work20:12
sadoon[m]ok downloading firefox on my phone20:12
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@> has quit IRC20:26
*** ghostmansd[m] <ghostmansd[m]!> has joined #libre-soc20:27
*** ghostmansd <ghostmansd!> has joined #libre-soc20:28
programmerjakelkcl: what do you think about skipping 0 budget tasks in json? feel free to respond later21:27
programmerjakei copied the description changes from #1026 to #961, i'll rerun budget sync after checking the rest of the descriptions are synced22:04
*** octavius <octavius!> has quit IRC22:05
programmerjakesome of the bug's comment #0 is very clearly not intended to go in the MOU, so I only copied some of them to the MOU. lkcl, please check #1026 when you have time. I'll generate the json now22:20
programmerjakein the future it'd be nice to have comment #0 match what's in the MOU, so it's much easier to keep them in sync22:21
programmerjakeso, please keep that in mind for future nlnet grants22:21
programmerjakecomment #0 of all the top-level subtasks22:22
programmerjakeyes, budget-sync can be optimized, it's sending one https request per bug when requesting comments, probably because luke wasn't thinking about that when he wrote that part22:45
programmerjakethe part I wrote groups them into 100 bugs per request22:45
programmerjakefor now, I added a print() for each JSON file written, so it acts as a sort of progress bar, can do additional work on budget-sync later22:47
programmerjakelkcl: i attached the json as requested, since I attempted to sync the top comment with the other tasks, it needs to be checked before submitting to nlnet22:54
programmerjakethe mou needs to be checked, not the json22:54
*** choozy <choozy!> has joined #libre-soc23:11

Generated by 2.17.1 by Marius Gedminas - find it at!