channel logs for 2004 - 2010 are archived at http://tunes.org/~nef/logs/old/ ·· can't be searched
#osdev2 = #osdev @ Libera from 23may2021 to present
#osdev @ OPN/FreeNode from 3apr2001 to 23may2021
all other channels are on OPN/FreeNode from 2004 to present
http://bespin.org/~qz/search/?view=1&c=osdev2&y=21&m=11&d=9
00:43:00 <geist> probably no reason to, but i wonder what a modern equivalent of a merlin would look like
00:43:00 <zid> like a merlin, tbh
00:43:00 <zid> big engine looks like a big engine
00:43:00 <geist> if you started off with a lighter engine block and used more modern computerized injection, might be substantially crap hanging off it
00:43:00 <zid> just it'd have a few wires going to a loom for some management and monitoring
00:44:00 <zid> you might be able to make the same horsepower smaller etc
00:44:00 <geist> may be you can get a lot more horsepower out of a smaller block though
00:44:00 <zid> with better block design etc
00:44:00 <zid> exactly
00:44:00 <geist> yeah, not saying it'd be as impressive, etc, just thinking
00:44:00 <dh`> good question, I don't think anyone builds radial engines for light planes these days
00:45:00 <geist> a merlin is a V12, it's just a bigass block though
00:46:00 <dh`> huh, I thought they were radial
00:46:00 <dh`> shows how much I know
00:46:00 <geist> looks like.... 27L displacement
00:46:00 <geist> it's entirely possible there are light plane, non turboprops that have similar performance modern ICE engines
00:47:00 <kazinsal> I imagine if someone was going to cook up a four digit cubic inch displacement V12 that sounds like the devil and sucks gas like a mad max bandit with a siphoning tube it would be mopar
00:47:00 <geist> but maybe not. may be there's no point building something like that now, since you're better off going turbine
00:48:00 <geist> nice, discord is classy. someone learned i worked on zircon and then went on a rant about how he dispises everything i do and wishes i'd fail
00:48:00 <geist> well, okay then. keep it classy guy
00:48:00 <kazinsal> yeah I'm not in any really public discords because honestly fuck most people
00:49:00 <kazinsal> it's amazing how the internet is just full of the most atrocious people that the universe saw fit to unleash upon the world
00:49:00 <zid> geist: is he mad it isn't a real microkernel
00:49:00 <geist> zid: actually no, going on about how microkernels are a 'meme' etc
00:49:00 <zid> kazinsal: internet has a lack of punch in the mouth problem is all
00:49:00 <geist> and CISC, mono, etc. and etc etc
00:49:00 <geist> like, okay man.
00:49:00 <zid> I mean, microkernels are a meme, but that's for playfully ribbing microkernel devs with
00:49:00 <zid> not going on hysterical rants about
00:49:00 <geist> yeah i mean sure, it's fun to playfully rib
00:49:00 <geist> exactly
00:50:00 <geist> you know, rant for a few lines and put a smiley face.
00:50:00 <geist> or dont. and maybe not someone you literally just met
00:50:00 <zid> zircon cares too much about performance to be a real microkernel so you're fine anyway
00:50:00 <zid> (Do you see what I did there, I said microkernels are slow, hahahahaha)
00:50:00 <kazinsal> the problem with the idea of an osdev discord is that it's going to be full of the twelve year olds
00:50:00 <geist> noice. glad you explained it :)
00:51:00 <zid> np babe
00:51:00 <zid> kazinsal: It used to be astrophysics
00:51:00 <geist> 12 year olds are one thing, but i also think there's just a bit more of a meanness to that style of comms
00:51:00 <kazinsal> little timmy's first python script was too simple for the helicopter-parent-induced god complex he has so now he's trying to write an operating system in bash
00:51:00 <zid> The crazies would write hand written letters to professors about their theories of the universe and perpetual motion etc
00:51:00 <zid> now they just join ircs and discords to do it
00:51:00 <geist> much how lots of internet forums go instantly to flamewars, etc (also a think i never participate in)
00:52:00 <zid> I've yet to see a discord that was actually moderated, either
00:52:00 <zid> this channel isn't really moderated either but we managed to luck into some people that don't fight
00:52:00 <geist> you can, it's just very time consuming, presumably because people assume they can just be asshats
00:52:00 <zid> ..much
00:53:00 <geist> but maybe it's simply beause we've whittled it down to a handful of people, and so statiscally speaking there's less BS
00:53:00 <geist> and if we had 5 or 10x as many active users here it'd turn into a disaster
00:53:00 <geist> like some sort of non linear relationship of active users to BS
00:53:00 <geist> and we're in or below the sweet spot, maybe
00:54:00 <zid> yea that also helps
00:54:00 <zid> maybe it's the birthday paradox
00:54:00 <dh`> based on experience with similar issues in other domains I'd say there's a certain quality control filter applied by the fact that your basic bozo doesn't know how to operate an irc client
00:54:00 <zid> We all fight with 1/365ths of the users
00:54:00 <geist> dh`: yah that's been my suspicion
00:55:00 <geist> hmm, anyone here born on jan 2nd?
00:55:00 <geist> cause we have to fight, highlander style
00:56:00 <geist> speaking of bday paradox i actually just a few years ago ended up sitting next to two people in my office that had the same bday
00:57:00 <kazinsal> we have fewer drive bys now than we did on freenode I've noticed
00:58:00 <geist> yah reminds me i still need to get my dad set up on here again
00:58:00 <geist> he had a client pointing at freenode, but never re-set it up when it switched
00:58:00 <kazinsal> at this point I think if someone is aiming to seek out IRC directly and go through the hoops of nickserving up etc. there's a fair chance they're going to hang around and contribute
00:58:00 <geist> also funny about the time we switched it said screw it and finally went in on irccloud
00:59:00 <geist> end of an era, dont really need a shell box with irssi anymore
00:59:00 <geist> irccloud works well enough, and is better at doing simple formatting too
00:59:00 <kazinsal> yeah, I'm considering switching over as well. would make checking in on IRC on my phone easier
01:00:00 <geist> yeah it's pretty decent, honestly
01:00:00 <kaichiuchu> i've considered irccloud, but our past bretheren once banned irccloud on freenode
01:00:00 <kaichiuchu> i'm concerned they'll do the same thing here
01:01:00 <geist> well, i was just about to point out that irccloud also sealed ever going back to freenode since Dude banned it
01:01:00 <geist> nah, he banned it because irccloud tweeted something that hurt his inner child
01:01:00 <geist> no real technical reason
01:01:00 <kaichiuchu> ah
03:54:00 <junon> geist: I won't say which day but recently I shared my birthday with a guy I dated, a coworker, and a mutual friend - all without knowing until about 48 hours before. I do not have a large social circle.
04:40:00 <klange> It's no 2.0 yet, but: 🎉 https://github.com/klange/toaruos/releases/tag/v1.99.9
04:42:00 <zid> 1.99.999999 in a few weeks?
04:43:00 <junon> klange in january marks 11 years of working on ToaruOS, did you work on it completely through those 11 years or did you take any significant breaks?
04:43:00 <junon> (11 years if my idiot brain math is correct :D)
04:43:00 <klange> 11 years is correct
04:44:00 <junon> That's incredible.
04:45:00 <klange> It's been on-and-off. The last year has been very consistent, but 2020 I did next to nothing, 2019 I was mostly working on my editor...
04:46:00 <junon> How many hours a week on average do you think you devoted to it over the years? I'm sure that number varies a lot but less than 10? More than 20? 40?
04:49:00 <klange> Hard to say. I would bet the number is pretty low overall. I'm very sporadic. It's not like I set aside dedicated time.
04:49:00 <junon> Sounds about right. It is, unfortunately, very much a hobby for most :\
04:49:00 <klange> Used to be a lot of days I would be up late into the night trying to hack on something; don't do that much anymore - the meds help :)
04:50:00 <junon> I can certainly relate. (:
05:54:00 <ZetItUp> it's snowing outside :(
05:57:00 <ZetItUp> btw i have design issues with mountpoints, since im using FAT32 as fs, should i just go over each partition and say "Hey partition 0, you are now known as IDx", or should i save some table in the kernel where i go Partition 0 = IDx? where IDx is some ID
05:57:00 <kazinsal> mount points are often a userspace problem -- do whatever the sysadmin tells you
05:58:00 <Mutabah> Probably a table?
05:58:00 <ZetItUp> hmm true
05:58:00 <Mutabah> Not sure I understand where you issue is?
05:59:00 <zid> not sure how you'd tell a partition anything
05:59:00 <zid> or for why
05:59:00 <Mutabah> Usually you'd have a map that determines which filesystem is used for a given path
06:00:00 <kazinsal> yeah I'm a bit curious as to what the issue is, how to lay it out in memory, or how to bolt mount points into a VFS
06:00:00 <kazinsal> or how to lay it out as a config, etc
06:00:00 <zid> what's the "it" here though
06:00:00 <ZetItUp> bolt mount points into the VFS
06:01:00 <zid> vfs layer is typically only giving a shit about resolving paths to LBA + device pointer
06:05:00 <ZetItUp> im thinking around the windows style of A/B/C/D: etc
06:06:00 <kazinsal> then you just need to keep a mapping of what volume label maps to what logical filesystem on what device
06:07:00 <ZetItUp> yeah, but where would be a good place to store that mapping? kernel space? some file on the disk?
06:08:00 <ZetItUp> can't really decide hehe
06:08:00 <kazinsal> some file on the boot disk
06:08:00 <kazinsal> then load it into your VFS at boot t ime
06:08:00 <kazinsal> you can even do that part in userspace, just expose a mount() call
06:09:00 <ZetItUp> that seems like a good way
06:09:00 <kazinsal> kernel automatically mounts the device it boots from to eg. boot:, then your userspace init loads boot:config/filesystems.cfg and for each valid entry it does mount(stuff based on entry)
06:09:00 <kazinsal> this is functionally how /etc/fstab works
06:11:00 <ZetItUp> ah that makes sense
06:12:00 <zid> windows has 'device paths' for this reason btw
06:12:00 <zid> the //Device0/?/C/kaz/ stuff
06:12:00 <zid> \\.\Volume{b738473-348939483-34893\test\kaz.txt
06:13:00 <ZetItUp> yeah i guess the issue was mostly, where to store it hehe
06:13:00 <ZetItUp> or how to handle it
06:14:00 <kazinsal> the NT VFS is really complex
06:14:00 <zid> (and slow)
06:14:00 <ZetItUp> i guess there really isn't any "right or wrong" way of doing it
06:14:00 <kazinsal> you can do stuff like \\?\GLOBALROOT\Device\WebDavRedirector\10.20.30.40\foo\bar
06:25:00 <zid> ZetItUp: Do you plan to have an 'internal' name that userspace can even translate to?
06:26:00 <ZetItUp> yeah
06:27:00 <zid> so it's up to you I guess then, if it's in userspace you can use it as like, userspace aliasing
06:27:00 <zid> to allow drive letters to change to representing different devices
06:27:00 <zid> idk how useful that is though
06:27:00 <ZetItUp> yeah it's probably just gonna be used on bootup
06:28:00 <zid> so like how I only use (root0,0) for grub? :P
06:30:00 <ZetItUp> or i just do hda, hdb, hdc or something like that :P
06:30:00 <zid> sounds like it then yea
06:39:00 <ZetItUp> good thing i started to think about this now, cause i think i would have to rewrite alot of stuff later if i dont :P only a couple of lines now thou :D
06:40:00 <zid> yea I spent most of my time staring at notepad making notes
06:40:00 <zid> instead of actually coding
06:41:00 <ZetItUp> right now for this i just read straight from the disk without a vfs, like this https://i.gyazo.com/c85ee9f0d1716e438f6f38b55725adab.png
06:42:00 <ZetItUp> sure it works, but it was kinda annoying to write it like that :P
06:45:00 <ZetItUp> speaking of something unrelated, i hate my neighbours, i live with my windows to the parking lot, there is like 5 meters to the parkings, EVERYONE fucking slams the doors on their cars when they comes or leaves
06:45:00 <ZetItUp> during this hour (6AM-9:30AM) all i hear is SLAM SLAM SLAM WROOM WROOM
06:46:00 <ZetItUp> and eventually someone forgetting to put in the gear and just revs his engine WROOOOOM
06:46:00 <ZetItUp> Rant Over
06:53:00 <wleslie> sound-activated water balloon launcher
06:54:00 <zid> is it actually for 386
07:18:00 <geist> huh interesting. never noticed this before but the generator .ctor section from gcc at least seems to mark it as writable
07:19:00 <geist> i wonder why? seems like it'd just be a RO section. maybe it's a relocation thing where it clusters it in a RW section of the binary?
07:36:00 <zid> what if you wanna mark em off as done as you go? :P
07:43:00 <geist> possible. maybe there are some implementations that do that
07:43:00 <geist> set an atomic that it's already been run orsomething
07:45:00 <zid> idk how resolveable the order you need to run things is and whether you need to track whether a certain constructor has already ran when you're parsing it all
07:45:00 <zid> or if the linker figures it all out and puts them in the right order
07:50:00 <klange> aw did i miss a discussion on paths and vfs
07:54:00 <zid> not especially
07:58:00 <geist> hmm, that's always alarming. just heard a tree come down in the woods outside my house
07:58:00 <geist> went out and can't see it, so it must have been fairly big and farther away
08:00:00 <klange> I probably should have built a new VFS for Misaka but I just ported over the one from toaru32. It has some... unique quirks.
08:00:00 <klange> But it works okay, gives the impression of a functioning system at least.
08:00:00 <zid> close enough then
08:02:00 <klange> https://klange.dev/s/vfs-mounts.png fun times
08:06:00 <zid> The internet has informed me that there is no time to die
08:07:00 <zid> how do those enp names get generated btw
08:07:00 <zid> I take it you're using the 'proper' schema seeing as it ends in 17?
08:07:00 <zid> I always use the old style, never learned wtf the new ones mean
08:08:00 <zid> ethernet's always gunna be funny though, there isn't really anything identifying about them other than MAC and that's after drivers are loaded, and is still software configurable
08:11:00 <klange> pXsY → p for PCI, bus X, s for slot Y
08:12:00 <zid> ah I see
08:12:00 <zid> Enjoy.. slot 17!?
08:12:00 <klange> 00:11
08:13:00 <klange> I confused myself for a minute there looking at lspci output but then I remembered that uses hexdecimal
08:13:00 <zid> oh your machine just has a bunch of early stuff like bridges and vga and stuff I take it?
08:14:00 <zid> not physical slot, bus childo
08:14:00 <klange> This screenshot is from VirtualBox which places things a bit sparsely
08:14:00 <klange> My laptop has enp0s25 ;)
08:15:00 <zid> laptops are full of bullshit
08:18:00 <klange> little more straightforward in qemu :) https://klange.dev/s/qemu-ifconfig.png
08:18:00 <zid> looks like vmware goes up to like 20, then starts a new bus for my ethernet
16:12:00 <chibill[m]> I am very annoyed right now. Can't get at my OSDev stuff because my home internet is out, so I can't connect into my server.
19:16:00 <Bitweasil> That's why you dev locally.
20:39:00 <Qubasa> Does someone know how to get the memory map if I boot over PXE with ipxe and then into my x86 bootloader? I checked the ipxe source but they don't place a pointer into any register when jumping into the loaded image
20:42:00 <kazinsal> PXE should drop you in real mode, use the BIOS E820 memory map functions to build a map
20:42:00 <kazinsal> http://www.uruk.org/orig-grub/mem64mb.html
20:42:00 <bslsk05> ​www.uruk.org: INT 15h, AX=E820h - Query System Address Map
20:45:00 <Qubasa> ah hmm i don't boot with PXE directly but PXE into IPXE
20:45:00 <Qubasa> and IPXE drops me into protected mode
20:46:00 <Qubasa> pxe -> ipxe -> bootloader
20:48:00 <Bitweasil-> Scan for ACPI tables? :/
20:57:00 <Qubasa> Found it out. You need to set some special make flags to make this wor
20:57:00 <Qubasa> k
22:06:00 <jjuran> If a tree falls down in the woods and nobody sees it, does it make a scene?
22:22:00 <vin> I am little confused, in ext4 if two threads are writing to the same inode/file at different offsets in parallel, would a lock be acquired on the file making the writes sequential?
22:25:00 <moon-child> atomicity is not generally guaranteed in such situations, no. You will of course have to maintain valid kernel/fs state
22:30:00 <vin> So parallel writes to a single fiile are sequential to make it crash consistent?
22:31:00 <immibis> jjuran: a scene requires at least one camera and at least one light