marcan changed the topic of #asahi to: Asahi Linux: porting Linux to Apple Silicon macs | General project discussion | Topics: #asahi-dev #asahi-re #asahi-gpu | Keep things on topic | https://github.com/AsahiLinux | Logs: https://freenode.irclog.whitequark.org/asahi
jamadazi has quit [Ping timeout: 268 seconds]
mha has joined #asahi
<mha> \o/
<JTL> Welcome
comex has joined #asahi
_whitelogger has joined #asahi
_whitelogger has joined #asahi
<TheJollyRoger> Woah.
<marcan> so, apparently, what marketing called "icestorm" and "thunderstorm"
<marcan> are actually "vortex" and "tempest"
<marcan> #define MIDR_ARUBA_VORTEX (0x010 << MIDR_EL1_PNUM_SHIFT)
<marcan> #define MIDR_ARUBA_TEMPEST (0x011 << MIDR_EL1_PNUM_SHIFT)
<marcan> I think we finally have a codename for the M1: "Aruba"
<davidrysk[m]> then what's A12? it's documented that A12 is Vortex and Tempest
<davidrysk[m]> (in LLVM sources and other places)
<marcan> #define MIDR_CYPRUS_VORTEX (0x00B << MIDR_EL1_PNUM_SHIFT)
<marcan> #define MIDR_CYPRUS_TEMPEST (0x00C << MIDR_EL1_PNUM_SHIFT)
<marcan> one of those would be A12, the other M1
<marcan> there's this though too
<marcan> #define CPU_PART_LIGHTNING 0x12
<marcan> /* H12 e-Core (ARMv8 architecture) */
<marcan> #define CPU_PART_THUNDER 0x13
<marcan> wait no, this isn't right
<marcan> #define CPU_PART_VORTEX 0xB
<marcan> there is mention of "M9" around here
<davidrysk[m]> prior documentation states that Vortex is A12 big and Tempest is A12 little, while Lightning is A13 big and Thunder is A13 little
<marcan> yeah this makes more sense
<davidrysk[m]> Interestingly though, LLVM has been using the same scheduler model since Cyclone (A7)
<davidrysk[m]> <davidrysk[m] "Interestingly though, LLVM has b"> Since I thought Apple may have been pulling closed-source-forked-LLVM shenanigans, I perused their clang/llvm binaries as well and could not find a different scheduler model for the newer cores
<marcan> so: H11=Cyprus[Vortex/Tempest], H11G=Aruba[Vortex/Tempest], M9=unknown[Tempest], H12=[Lightning/Thunder]
<marcan> M9 sounds like a Watch core
JTL has quit [Quit: WeeChat 1.7.1]
<davidrysk[m]> the S4 and S5 use a dual core Tempest
<marcan> so if H12 = A13, then the M1 isn't in here
<davidrysk[m]> does this mean we need to write a kext and read the registers? :)
<marcan> I just wanted to know the codename tbh :p
<marcan> but I intend to do all such hardware probing from mini once I get that ported
<davidrysk[m]> ugh
<marcan> #elif defined (BCM2837) /* Raspberry Pi 3 */
<marcan> I like how this just casually supports the rPi
<davidrysk[m]> xnu-6153.141.1 has more files in pexpert/pexpert/arm64
<davidrysk[m]> but that might not be meaningful
<davidrysk[m]> no mention of T8101 here
JTL has joined #asahi
<davidrysk[m]> see the boot log that's posted here: https://eclecticlight.co/2020/12/09/what-happens-when-an-m1-mac-starts-up/
<marcan> I like the RC_HIDE_XNU_LIGHTNING in the old one
<davidrysk[m]> yeah, that's the "fun" part
<davidrysk[m]> download all the different xnu source bundles and look for stuff they forgot to redact
<marcan> #if __arm64__
<marcan> need_wa_rdar_55577508 = cpuid_get_cpufamily() == CPUFAMILY_ARM_LIGHTNING_THUNDER;
<marcan> #endif
<marcan> ah yes I love it when people put bug numbers into variable names
<davidrysk[m]> lol at least they say what the bug is
<marcan> yeah, thankfully
<marcan> we're going to need that one quite possibly
<davidrysk[m]> if it's not fixed in A14/M1
<marcan> also:
<marcan> // Disable clock dithering
<marcan> // <rdar://problem/29022199> [Tunable] Skye A0: Linux: LLC PIO Errors
<marcan> they use Linux for hardware validation, clearly :p
<marcan> I'm not amused as to how their source references stuff like pexpert/arm64/H12.h which does not exist
<davidrysk[m]> yes.
<marcan> I guess to them these xnu source dumps are just garbage not intended to be buildable...
<davidrysk[m]> they USED to be buildable
<davidrysk[m]> that's the frustrating part
<davidrysk[m]> the KDKs with debug symbols might end up being more useful
<davidrysk[m]> (those can be found under https://developer.apple.com/download/more/)
<marcan> yeah...
<davidrysk[m]> I'm partly hoping that this is because there's too much broken crap in 11.0.1 that they weren't clear for release and that we'll see more in 11.1 and 11.2 releases
<davidrysk[m]> I updated my feedback entry though
<marcan> quite possible
<davidrysk[m]> from the KDK we can see the build line though
<davidrysk[m]> (note, long lines)
<davidrysk[m]> that's the t8020 one though — A12, not M1
JTL has quit [Quit: WeeChat 2.9]
<davidrysk[m]> I don't think we have an M1 KDK yet, though lemme see
JTL has joined #asahi
<davidrysk[m]> yeah not even the 11.2 KDK has the t8101 kernel
tarzeau has joined #asahi
JTL has quit [Client Quit]
JTL has joined #asahi
<davidrysk[m]> I filed a feedback for that
JTL has quit [Quit: WeeChat 2.9]
JTL has joined #asahi
JTL has quit [Quit: WeeChat 2.9]
JTL has joined #asahi
jamadazi has joined #asahi
roxfan2 has joined #asahi
roxfan has quit [Ping timeout: 260 seconds]
svenpeter[m]1 is now known as svenpeter[m]
roxfan2 is now known as roxfan
artemist has joined #asahi
dyniec[m] has joined #asahi
minicom has joined #asahi
sferrini has joined #asahi
eichin has joined #asahi
_ghost_ has joined #asahi
karlyeurl has joined #asahi
ransom has joined #asahi
sven has quit [Quit: ZNC 1.7.4 - https://znc.in]
dhewg has quit [Ping timeout: 256 seconds]
mha has quit [Ping timeout: 264 seconds]
jcs has joined #asahi
LeonardJanisRobe has joined #asahi
Jasper[m] has joined #asahi
eta has joined #asahi
marcan has quit [Quit: Now where's my screwdriver...]
maximus64 has joined #asahi
leah2 has joined #asahi
rjzak has joined #asahi
Shiz has joined #asahi
gruetzkopf has joined #asahi
FFY00 has joined #asahi
Raqbit has joined #asahi
awesomebing1 has joined #asahi
GaveUp has joined #asahi
numa[m] has joined #asahi
<numa[m]> Hi, just testing the matrix bridge
josiahmendes[m] has joined #asahi
rcombs has joined #asahi
djb has joined #asahi
jn__ has joined #asahi
chuskywalker has joined #asahi
Fanfwe has joined #asahi
<TheJollyRoger> Woah, this channel's grooooownnn!
<Fanfwe> Hello, world!
spikebike has joined #asahi
<jn__> o hi TheJollyRoger! :)
<Shiz> lotta names here
<djb> This irc thing...I don't think I've done it in like 15 years now.
DrWhax has joined #asahi
austriancoder has joined #asahi
<spikebike> djb: your nick looks familiar
tpw_rules has joined #asahi
<djb> I don't recognize any of you from the old linuxnet days, but that might be because many of you weren't born yet
<LeonardJanisRobe> we use(d) irc on irc and it simply scales so nicely
<djb> @spikebike yours too
<spikebike> ha, yeah, on linuxnet
<rcombs> that escalated quickly
<numa[m]> <djb "This irc thing...I don't think I"> I think it's the better option, everything has a irc bridge
<LeonardJanisRobe> but here i simply use it via element-bridge
jcs has left #asahi [#asahi]
<djb> @numa[m] I've got no complaints about it!
flokk has joined #asahi
neobrain has joined #asahi
<djb> I'm no kernel hacker, but I used to enjoy hanging out with them so I figured I'd check in here.
x56 has joined #asahi
jonmason has joined #asahi
Yuzu has joined #asahi
<jn__> djb = Dan Bernstein?
<djb> @jn__ OH DEAR GOD NO
<j`ey> lol
<jn__> alright, ok
hspak has joined #asahi
<djb> Donnie Barnes, formerly of Red Hat
<Shiz> hah
<djb> Bernstein and I had some epic flame wars bitd
<Shiz> i had the same thought, but a quick IP check proved otherwise
<Shiz> unless bernstein suddenly moved to NJ
<LeonardJanisRobe> well, with that nick, i also thought first of bernstein :)
vlad-k has joined #asahi
<djb> He and I both go back a long ways
<djb> I've used this nick since 1990, no idea about him but it was probably close
<djb> I'm way less of a dick
<djb> :)
jld has joined #asahi
<spikebike> heh, mentioning bernstein was like an evoking some summoning spell and he'd appear and rant at you about the much better qmail is
<djb> until he saw me, when he'd start ranting at me about why Red Hat wouldn't ship qmail
<LeonardJanisRobe> we shouldn't press our luck, better shut up about him :D
titanous has joined #asahi
DarthCloud has joined #asahi
jamadazi has quit [Ping timeout: 260 seconds]
<djb> he didn't like when I told him to make it open source and we would
<spikebike> ha
<Shiz> on a more constructive note, how many people here actually own an M1 device
<Shiz> (m1ni here)
<spikebike> I'm pondering a M1, but figure I'll wait till it's at least a little bit useful with Linux.
<jonmason> macbook air should be delivered this week
<djb> I do not, but I'll get one before there's anything useful to test
<numa[m]> I feel like a 5 year old hearing titans talk
<spikebike> got a mbp i9 which is just terrible CPU wise
* jn__ doesn't own an m1, already has enough random computeroo to hack on
<Shiz> i also got the i9 mbp for $dayjob, the compilation times were rather nice ime
<LeonardJanisRobe> i hope to be able to afford one as soon as i've finished my thesis and perhaps then contribute as well
<Shiz> but my baseline is a 2013 mbp :)
<numa[m]> <Shiz "on a more constructive note, how"> Nope, possibly I will get one used years from now when the linux experience is stable.
<djb> When it's stable? That's missing out on all the fun!
<Shiz> :)
<spikebike> Shiz: yeah compiling is fast, but a video conf often makes the fans go into turbo mode
<Shiz> my m1ni is currently being ~~occupied~~ used by an ffmpeg developer, so all for a good cause :p
<djb> Some of my best times at Red Hat were helping with ports to the Alpha and the SPARC
Avion[m] has joined #asahi
<djb> I contributed zero code, mind you
<LeonardJanisRobe> precisely, I'm just in the midst of migrating from Arch to Fedora Silverblue "Rawhide", since Arch has been just faaar to stable for being fun
<LeonardJanisRobe> oh, SPARC <3
<artemist> I'm probably going to get one when there's a kernel booting / I find one on sale, I'll make a NixOS overlay for it
<spikebike> djb: heh, yeah, I bought a alpha PC (or whatever it was called) when they went on fire sale
<davidrysk[m]> even with ARCH -testing?
<LeonardJanisRobe> yes, testing
<spikebike> alpha udb or somesuch
<jn__> djb: infrastructure and bug reports are just as important
<LeonardJanisRobe> Last year I still set up some SPARC compute clusters running debian in domX with Solaris in dom0, that was fun!
<djb> @jn__ indeed
nufflee[m] has joined #asahi
<numa[m]> <djb "When it's stable? That's missing"> The flesh is willing but the wallet is empty and third world-ly
<Shiz> if other people bugreported my software i'm sure it would be significantly more stable :p
<djb> @numa[m] Understood, understood.
<Shiz> yay for bugreporting+community
<LeonardJanisRobe> david.rysk: I've been running -testing on my workstation for, well testing purposes, for some time and forgot to disable it but didn't notice for a year. so i figured i keep it.
<davidrysk[m]> since there are more people here now: preferred matrix client that isn't element for macOS?
<cdesai> I just keep element.im open in a browser
<LeonardJanisRobe> ditto
WhyNotHugo has joined #asahi
<nufflee[m]> I'm using element.io for the first time but it seems nice so far
<x56> now that parallels is available on M1, my MBA is getting more use than either of the maxed out 2018 Mini or 2020 MBP x). MBP is just on the shelf, Mini still gets a bit of use for now because it fits better on my workbench and still need Catalina for some things
f4n4tix[m] has joined #asahi
Foxboron has joined #asahi
<x56> M1 MBA is just categorically better than the other machines for 95% of my use cases
<q3k|m> (for new element users: settings -> appearance -> advanced -> Enable experimental, compact IRC style layout )
<davidrysk[m]> Parallels is great, but the ARM64 emulation in Windows stinks
<davidrysk[m]> Then again it's really new so
<q3k|m> (also smol font size and fixed width font go a long way)
konstanter has joined #asahi
<davidrysk[m]> also hi x56
marcan has joined #asahi
<Avion[m]> <davidrysk[m] "since there are more people here"> [Seaglass](https://github.com/neilalexander/seaglass) was really promising, but there hasn't been any development on it since 2019. Element client is the best we have for now.
<x56> hey davidrysk[m]! I'm curious whay issues you've had with the emulation, I spent a bunch of time playing x86 steam games over the holidays in a Windows VM without issue
<davidrysk[m]> I tried installing Autodesk Inventor and Solidworks
<davidrysk[m]> both failed during install
<x56> ah damn :/
<davidrysk[m]> both require x86_64
<x56> ah gotcha, I guess I was using 32-bit stuff primarily
sven has joined #asahi
dhewg has joined #asahi
<davidrysk[m]> what fails? various MS-redistributable Windows dependencies. My guess is that there are still MS bugs in handling dependency/library installation\
<Shiz> marcan: wb :p
<numa[m]> <davidrysk[m] "since there are more people here"> In the same spirit, any feature complete and lightweight matrix client in Linux?
<numa[m]> besides element ofc
<LeonardJanisRobe> <numa[m] "In the same spirit, any feature "> the only feature complete irc client that's not irssi or one of the other classics is Konversations. lightweight, if you have a KDE desktop :-)
<LeonardJanisRobe> Most others don't support things like custom certificates and login with cert
jjanzic has joined #asahi
<artemist> A lot of that isn't necessary but weechat works well enough
akemin_dayo has joined #asahi
<davidrysk[m]> marcan: regarding copyright policy: how is code developed for other projects that require a CLA handled?
snowlake[m] has joined #asahi
<davidrysk[m]> that require a CLA for upstream inclusion **
nkaretnikov has joined #asahi
nemith has joined #asahi
burra has joined #asahi
<marcan> davidrysk[m]: I'm not aware that we'll have to contribute to any such projects, if we do we'll figure it out at that point
Axenntio has joined #asahi
flokk has quit [Quit: Leaving]
chivay has joined #asahi
Axenntio has quit [Remote host closed the connection]
<q3k|m> what's the plan for tracking 'things to do' as a community? i know i would like to at some point chip away at something in this project (that's not strictly in a blocking path for larger efforts), but i'm not sure how much time i would have for reliably following IRC/project updates. i'm not even sure that sort of help will ever be needed.
<davidrysk[m]> regarding firmware blobs: I'd expect one would start by documenting them, then using an extractor (e.g. broadcom drivers), meanwhile attempting to obtain redistribution permission from Apple so that they can be included in the linux firmware distribution
<marcan> my understanding is that firmware is largely loaded and locked by iBoot before we are called, for security reasons
siriusfox has joined #asahi
<Fanfwe> q3k|m: good point, let's setup a Jira
<marcan> so with luck we won't have to deal with any blobs
siriusfox has left #asahi ["Leaving"]
<Foxboron> marcan: Is packaging going to be discusse in -dev or this channel? Are you plan still to use ALARM?
<q3k|m> :( not what i meant
<Fanfwe> this was sarcasm, just to make things clear :P
<Foxboron> Is your plan* :p
<q3k|m> Fanfwe: yeah but i'm sorry i only got a sarcastic response
<q3k|m> anyway, i'll just assume i'll continue checking in as much as i can
blu42 has joined #asahi
<davidrysk[m]> Yeah I think you're right about that, which imo is a nice (though not FSF-approved) way of doing firmware
<Shiz> doon't get started on FSF-approvied firmwares
pg12 has joined #asahi
<blu42> heya. just dropping by to say asahi.org is dope
ky0ko has joined #asahi
<jn__> blu42: wrong domain :)
snowlake[m] has quit [Quit: authenticating]
marcan has quit [Ping timeout: 246 seconds]
snowlake[m]1 has joined #asahi
sven has quit [Read error: Connection reset by peer]
marcan has joined #asahi
<marcan> q3k|m: good point that we should set something up. at least a priori a wiki page would be good
<q3k|m> i also don't want this to be, you know, unnecessary overhead for people that actually do the damn work - but if there's something that would be useful for everyone, then that would be nice to set up early
m42uko has joined #asahi
Crisco has joined #asahi
sven has joined #asahi
<marcan> my IRC bouncer box is flapping and it's 5AM here, so I'm afraid I need to call it a day; I'll catch up on IRC logs (good thing we have those now :)) and reply to folks later.
<marcan> good night all! and thanks for joining to everyone new!
<jn__> thanks for you administrative work, etc
bublik_ has joined #asahi
Hexagon has joined #asahi
<TheJollyRoger> hi jn__! Eee you're here too!
<TheJollyRoger> holy cow! djb, are.... are... are you THE DJB?
<jn__> yep yep :)
<TheJollyRoger> goodnight marcan! Sleep well!
<TheJollyRoger> I'm excited~~
<jn__> i already asked. not THE djb of qmail etc
<TheJollyRoger> Ohhh, ahahahaha
<djb> oh ffs
<GaveUp> a djb, not *that* djb
<TheJollyRoger> Hehe, sorry about that.
<djb> I'm way cooler than that guy!
<jn__> thanks for clarifying :)
* djb grumbles
<TheJollyRoger> Hehehe. And in case anyone asks, I'm not the Jolly Roger who wrote the darknet Opsec Guide either :P.
<davidrysk[m]> heh, I lost a lot of respect for bernstein after the whole Appelbaum situation and his handling of it
<djb> I'm the Red Hat djb
<TheJollyRoger> Oh wow.
* JTL thinks "the" djb is busy with academic students
<JTL> and being a professor
<JTL> djb: I mean, could be still interesting :p
<djb> I lost all respect for that djb when he started cursing me with f'bombs in a public forum because he didn't like my responses to why Red Hat wouldn't ship qmail
<JTL> Lemme guess, license concerns?
<djb> but hey, that was the 90's and now it's a gagillion years later
<Foxboron> (and qmail hasn't improved)
<JTL> Foxboron: because djb thinks it's perfect, duh
<TheJollyRoger> *grumble*
<djb> @JTL Yes, as in his license didn't allow forks or any changes he didn't approve
<TheJollyRoger> Oh jeez.
<TheJollyRoger> Now I see why.
<JTL> figures
mogery has joined #asahi
<djb> The funny part was we *used* qmail at Red Hat for the support mailing lists
<djb> it was the only thing that would actually scale to our mailing list needs
<davidrysk[m]> oh I remember that
<spikebike> I was tempted by qmail, up till I took a close look at postfix.
snowlake[m]1 is now known as keystone[m]
<davidrysk[m]> the license mess
<djb> It was seriously the only thing that could handle our load at the time in a reasonable way
<JTL> if I was djb and had concerns about forks "tarnishing" my image I'd try and license with a license condition that forks neeed to be named differently from the upstream project and have a disclaimer. Mozilla does similar for Firefox.
<djb> it was good in that way, if you didn't need anything fancy and just needed raw speed
<djb> but it wasn't compatible with our philosophy on what we could ship since customers weren't free to fix it and neither were we, actually
Brnocrist has joined #asahi
mah has joined #asahi
<djb> and for pointing that out and standing my ground I was a "f'n asshole" or some such nonsense
<djb> many many times, too
<JTL> ...
<davidrysk[m]> some people don't like being told that they're wrong
jannau has joined #asahi
<JTL> Kind of reminds me of the attitude of the main Signal developer, oops too soon :|
<djb> the funny part was how he'd say he didn't care if we shipped yet, but yet got SO worked up that we wouldn't ship it
<djb> *it, not yet
<JTL> It's 2020 and secure IM is still "meh" :P
<TheJollyRoger> Sheesh.
<JTL> er 2021
<JTL> My brain is stuck on 2020
<JTL> AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
* TheJollyRoger unsticks the 2020-2021 lever.
<mogery> by the time i get used to the new year it's a new new year
<JTL> Haha fair
<JTL> I didn't have that problem with 2019->2020, but new decade and all
TheJollyRoger has quit [Quit: TheJollyRoger]
<ky0ko> just follow the lead of 16-bit-era games developers and say "it's 202X"
prusnak has joined #asahi
<prusnak> hi all
<numa[m]> helloo
<LeonardJanisRobe> yeah, djb did the "I'm Knuth and you don't change TeX" move, but just with a more dick-ish license and while also not being, actually, Don Knuth
jchillerup has joined #asahi
blu42 has quit [Ping timeout: 245 seconds]
<jchillerup> Just dropping in to say I'm super psyched about this :3
pdrew has joined #asahi
<spikebike> heh, same here, M1 is an impressive CPU, seems a shame to limit it to OSX.
mha has joined #asahi
<numa[m]> <LeonardJanisRobe "yeah, djb did the "I'm Knuth and"> As far I'm aware TeX is public domain-ish, you just can't call it TeX if doesn't pass his stress test
<numa[m]> the first line of TRIP (the test) "% This is a diabolical test file for TeX82. Watch your step."
<numa[m]> lmao
<LeonardJanisRobe> numa: yes
dshep has joined #asahi
<LeonardJanisRobe> that's why i said "more dick-ish license"
<davidrysk[m]> it makes the difference between free and nonfree, though
<LeonardJanisRobe> the spirit is the same "don't taint the name". But while Knuth was about protecting his project with reasonable means, djb is about protecting his ego with unreasonable memes
<LeonardJanisRobe> anyway, after djb pulled the applebaum shit off, he was more then dead
<LeonardJanisRobe> also, actually, he became quite a joke in the security world
<davidrysk[m]> I haven't been following the latest on the joke in the security world part, can you give me a link? :)
<Foxboron> I don't think he is a joke? His ECC talks at CCC has been popular since then
<davidrysk[m]> oh, software security, but how about cryptography?
<davidrysk[m]> I thought he was still respected, in the very narrow field of cryptography
<LeonardJanisRobe> Noooo, my code ain't broken, you should just increase ulimit for my special code! That exploit is completely the users fault >.<
<LeonardJanisRobe> Fair enough, yes, but I don't think he did do much new stuff there, so mostly respected for what he did... ages ago
<spikebike> AFAICT DJB finally let go of gmail, moved on to crypto, and has done quite well there.
<LeonardJanisRobe> I mean, he covers newer developments as well, but he's far from the leading force AFAIK
<Foxboron> I think he does novel work in the quantrum crypto field these days.
<Foxboron> quantum*
<LeonardJanisRobe> anyway, i'm mostly in the software security bubble rn, yes
<spikebike> yeah, and seems like many trust his crypto curves more than NIST's
<LeonardJanisRobe> well, that's not that surprising :)
orbot has joined #asahi
<davidrysk[m]> well, people stopped trusting the NIST curves after the whole Dual-EC mess
<LeonardJanisRobe> I mean, even the developers of libdalek use curve 25519 which was proposed by DJB
<LeonardJanisRobe> But you usually still wanna kinda stay away from him :D
<numa[m]> I love hearing drama that's older than me :D
<spikebike> all the unreasonableness I associate with him was in his qmail phase. Heard generally positive things since.
<davidrysk[m]> Again, the Appelbaum situation.
<LeonardJanisRobe> the unreasonableness is also regarding how to act around other people
<LeonardJanisRobe> precisely
<LeonardJanisRobe> he's had a notoriously hard time with "don't be an asshole"
<Crisco> I feel like half the Internet has that issue
david1888 has joined #asahi
<numa[m]> <Crisco "I feel like half the Internet ha"> That's being optimistic
matthewrizaldi has joined #asahi
<LeonardJanisRobe> Fair, but given the time he's had exposure to people one could have thunk he'd learn
Necrosporus has quit [Read error: Connection reset by peer]
Necrosporus has joined #asahi
matthewrizaldi has quit [Client Quit]
<LeonardJanisRobe> He's a great mind, sure, but great minds usually don't like to get criticized and told that they should read up on X. That was the issue with qmail, and that's still his issue. it just happens less in crypto :)
<q3k|m> djb's crypto is fine
<q3k|m> but the way it's been released as code is questionable
<q3k|m> ed25519 key/message format is a mess
<q3k|m> because every ed25519 implementation does this differently, and that format was mostly implementation-defined
<q3k|m> so even as he's moved from doing questionable sweng to crypto, the questionable sweng practices still linger
<LeonardJanisRobe> I know meany people doing questionable [design] choices, but most of them accept criticism and try to get better :)
* djb is glad his initials could cause such spirited discussion
dgb has joined #asahi
<LeonardJanisRobe> :D
<Shiz> there'def people who mark his criticism about other curves as FUD, but i dont know enough about the matter to say if there's any truth to it
Ismas has joined #asahi
Ismas has quit [Remote host closed the connection]
Axenntio has joined #asahi
<Crisco> Crypto is one of those things that I wish I knew more about but feel overwhelmed every time I try to dive into it.
Ismas has joined #asahi
Axenntio has quit [Remote host closed the connection]
mah has quit [Ping timeout: 260 seconds]
TheJollyRoger has joined #asahi
<LeonardJanisRobe> I've dabbled in crypto and it's definitely fun, at least if you like maths! However, the best crypto often doesn't hold much against rather simple attacks, either against the user or the software around it, in my experience.
<LeonardJanisRobe> Most security issues I've found related to the usage of crypto, rather than the crypto itself. E.g. different cli APIs between GPGv1 and GPGv2 which resulted in the private keys being publicly accessible
<LeonardJanisRobe> [insert relevant xkcd here]
<TheJollyRoger> Heh. Spafford's joke about using encryption on the internet being like arranging an armoured truck to pick up a package left on a park bench comes readily to mind.
<LeonardJanisRobe> precisely
brosco has joined #asahi
<TheJollyRoger> I have near-complete faith that modern, thoroughly-tested, academically studied ciphers are secure. But I feel a lot of the time endpoint security is in a terribly sorry state :(
<LeonardJanisRobe> Yup
* TheJollyRoger starts drowning his sorrows.
<Crisco> I've written implementations AES, RSA, and other stuff like Diffie-Hellman implementations to learn how it works, but I wouldn't know where to start to try to design new crypto algorithms or really how to evaluate already created ones for security.
<LeonardJanisRobe> There's a reason why many security people kind of try to replace GPG as the swiss-army-knife. As it's more alike to a swiss-army-knife with many sharp edges s.t., you mostly cut urself
<Foxboron> Crisco: seen cryptopals?
<Crisco> No, is that a movie?
<TheJollyRoger> Crisco: well, my programming experience pretty much ends at 10 PRINT HELLO WORLD ; 20 END ; so you're way ahead of me!
<LeonardJanisRobe> Crisco: Yeah, the implementations of these are astoundlingy easy-ish.
<Foxboron> Crisco: It's a set of challenges that builds up the fundamentals of crypto by helping you implement the fundamentals, how to break them, and how to improve them.
<Crisco> yeah, I was surprised at how quickly I could get RSA working
alimj has joined #asahi
<TheJollyRoger> I was working on writing a manual for GnuPG, and I got about 200 pages deep, but then I realized, I hadn't even covered the entirety of key management in that space, and who the heck is going to read a >200 page document to send an E-mail?
<Foxboron> Like why AES has different modes :)
mrasmus has joined #asahi
<LeonardJanisRobe> Oh, AES modes
<LeonardJanisRobe> And why are the defaults so shitty
<TheJollyRoger> Then I just... I sort of just got overwhelmed by it all and sort of set that aside and went to go get drunk.
<LeonardJanisRobe> reasonable course of action, i guess :D
<TheJollyRoger> I want to finish it someday, just... I feel like getting motivated to do so is really difficult.
<spikebike> TheJollyRoger: heh, yeah, I think of GPG as a trap for geeks.
<jn__> TheJollyRoger: wow, that's a lot of work you did there
<Crisco> We're talking about block modes, right? there's not some other AES mode I don't remember is there?
<spikebike> (even more so than crypto)
<TheJollyRoger> *hic* It'sh shertainly been a long *hic* ri-ride...
<LeonardJanisRobe> I think the best way to start is analysing crypto algorithms, i.e., cryptology
crabbedhaloablut has joined #asahi
<LeonardJanisRobe> Here's some material, I think most of it is bilignual, in EN and DE: https://www.staff.uni-mainz.de/pommeren/Cryptology/
<Crisco> I've been wanting to learn German for a while anyways
<TheJollyRoger> jn__: hehehe, yeah, I did put a lot of work into it. I wanted to cover not only the what to do, but more importantly cover the rationale of why because I realized that GnuPG makes the assumption that the user already knows why and won't try to stop them from doing anything stupid... so I had to delve into all that and that's what made things balloon so much in scale and scope.
<LeonardJanisRobe> <Crisco "We're talking about block modes,"> yup, and basically, the default is basically encrypting each block separately, making all kinds of easy-peasy attacks possible
<TheJollyRoger> But the unfortunate thing was that... that's a lot of stuff to cover.
mah has joined #asahi
<Crisco> Yeah, I'm familiar with all the block modes and their different features
<LeonardJanisRobe> The course linked is basically pure maths though, it does only barely touch on precise ciphers but mostly looks at specific concepts of cryptography
<TheJollyRoger> There's something that upsets me about block cipher modes and GPG, and that's sort of...
<LeonardJanisRobe> e.g., shift, stream ciphers, block ciphers, etc.
<LeonardJanisRobe> TheJollyRoger: If you need inspiration (and can read German): https://www.hauke-laging.de/sicherheit/openpgp.html
<Foxboron> that's fairly old
benjamindc has joined #asahi
jamadazi has joined #asahi
<TheJollyRoger> One thing that kind of upsets me about PGP is that we threw every cipher including CAMELLIA256 and TWOFISH into the mixture, but it uses all those ciphers in CFB mode. I find that sort of upsetting because it feels like GCM would be great to have and prevent a whole bunch of ciphertext modification attacks that the MDC doesn't stop. But since it's not in the spec, we can't simply switch it.
<TheJollyRoger> LeonardJanisRobe: oh cool! Uh... I might have to find someone who can read it to me, I don't know German, so I'd have to find some help.
alimj has quit [Remote host closed the connection]
david1888 has quit [Quit: Leaving]
<Crisco> best way to learn a language is by using it :P
* TheJollyRoger gulps audibly.
<LeonardJanisRobe> Foxboron it is, but so is GPG :D
<Crisco> that's true for both programming and human languages
<Foxboron> LeonardJanisRobe: Stuff changes over 10 years.
<LeonardJanisRobe> oh, I've had too many of my students who rather read or watch 5h of videos before writing 1 line of code
<Foxboron> Also recommends 2048 bits RSA (I don't read german)
<LeonardJanisRobe> Foxboron sure, but that's why I said he could use it as inspiration, not just translate it ^^
<LeonardJanisRobe> yeah, 2k was the fancy shit :D
<Foxboron> There is probably better material these days.
<Foxboron> In english, even.
devinvs[m] has joined #asahi
<TheJollyRoger> Once I used OpenSSL and pem2pgp to mint myself a 32 kilobyte RSA key.
<LeonardJanisRobe> It's the only I found
<LeonardJanisRobe> or rather, I didn't really find much better
<TheJollyRoger> But using it was kind of difficult because the digital signature files it produced were longer than the E-mails I was sending!
<LeonardJanisRobe> haha
<TheJollyRoger> I also couldn't fit it onto any of my homemade smartcards (they cap out at 2048, although I've located a supplier that claims they initialize their cards with 4096 capabilities), so I just had to put it in my ~
<LeonardJanisRobe> but I'm all open for better gpg ressources honestly, despite trying to phase it out
<TheJollyRoger> well er...
<TheJollyRoger> I still have some of my work posted. I suppose I could get back to that sometime...
* TheJollyRoger shudders apprehensively.
<LeonardJanisRobe> good luck!
<Foxboron> gnupg isn't going anywhere. Distribution and FOSS is going to use it for decades to come
* TheJollyRoger gulps
<Foxboron> Distributions*I
<TheJollyRoger> I'm... I'm going to need it.
Ismas has quit [Remote host closed the connection]
MeLoveCookies has joined #asahi
mogery has quit [Read error: Connection reset by peer]
bent_ has joined #asahi
<Crisco> Sometimes I wonder if it isn't a good idea to just get a cabin up in the mountains and disconnect completely from the conveniences of modern technology.
MeLoveCookies is now known as jamesboehmer
<pdrew> Unabomber
<TheJollyRoger> Crisco: I'd argue a pirate ship is better! That way you can go anywhere you like on the high seas!
<TheJollyRoger> And spend all your time drinking rum and singing sea shanties~~
<Crisco> You can sing sea shanties on dry land
bent_ is now known as RicBent
<Crisco> but yeah, maybe I should do that
RicBent is now known as RicBent_
<Crisco> I lived in the Caribbean before. I can go back.
<TheJollyRoger> ~What shall we do with a drunken sailor; what shall we do with a drunken sailor; what shall we do with a drunken sailor, eaaarlayy in the moooorn-ning~
fridtjof[m] has joined #asahi
<TheJollyRoger> I wonder if there are any signing utilities that can utilize HSMs, smartcards, and such... that'd be ideal. I know that shifting the root of trust into a blobby, opaque little chip from the 1990's that the vendors are curiously secretive about isn't ideal, but it's a whole lot less attack surface than keeping it on a desktop with a lot of extra attack surface.
<artemist> TheJollyRoger: I frequently just use openssl with the pkcs#11 module
<TheJollyRoger> I heard that Trezor-Agent was hoping to add support for Signify to their Trezors, which would be fantastic?
<TheJollyRoger> artemist: hey, maybe I should check that out.
<Foxboron> There are many tools that support it, yes.
<TheJollyRoger> woah.
kido has joined #asahi
<TheJollyRoger> I'm going to have to figure that out sometime!
<TheJollyRoger> Gonna need to stock up on JCards again, heh.
ml has joined #asahi
OatmealDome has joined #asahi
sharpiro has joined #asahi
Simonx22 has joined #asahi
HeN has joined #asahi
stemnic has joined #asahi
sharpiro_ has joined #asahi
sharpiro has quit [Ping timeout: 246 seconds]
jamesboehmer has quit [Quit: Textual IRC Client: www.textualapp.com]
sharpiro__ has joined #asahi
mah has quit [Ping timeout: 240 seconds]
sharpiro_ has quit [Ping timeout: 240 seconds]
jabamin has joined #asahi
linkmauve has joined #asahi
<LeonardJanisRobe> I always say, Computers were a mistake
sharpiro__ has quit [Quit: Leaving]
brosco has quit [Remote host closed the connection]
<LeonardJanisRobe> Jared Diamond calls agriculture the worst mistake of humanity I'd say Computers are the second worst.
<LeonardJanisRobe> Speaking of, we should have probably never left the trees, or maybe eben the sea. As once the great philosopher [sic] Douglas Adams wrote.
<mrasmus> Teaching sand to think is definitely the greatest example of humanity's hubris that *I* can think of.
blazra[m] has joined #asahi
jabamin has quit [Quit: Textual IRC Client: www.textualapp.com]
<LeonardJanisRobe> well, know your enemy, they said. so i wound up studying CS
<Crisco> Arizona is the greatest example of humanity's hubris, but teaching sand to think is a close second
jchillerup has quit [Ping timeout: 246 seconds]
sharpiro has joined #asahi
grumble has joined #asahi
os[m] has joined #asahi
<numa[m]> obligatory "In the beginning the Universe was created. This has made a lot of people very angry and has been widely regarded as a bad move."
dibas has joined #asahi
henni has joined #asahi
jannau has left #asahi [#asahi]
kakobrekla has joined #asahi
<davidrysk[m]> "Boot Camp" isn't CSM, specifically — "Boot Camp" is "entire set of tools and drivers that allows Windows to boot on an Intel Mac"
<davidrysk[m]> it's as much bootloader as drivers
mah has joined #asahi
noneucat has joined #asahi
dstzd has joined #asahi
riker77 has joined #asahi
larbob has joined #asahi
<JTL> davidrysk[m]: heh
<JTL> and I think it was the 2015 that deprecated/removed CSM from the UEFI firmware
<JTL> 2015 MBP that is.
<davidrysk[m]> do we know if the OS does the RAM training?
undvasistas[m] has joined #asahi
<Shiz> i would be very surprised if it did
<Shiz> how would it go anywhere without RAM training :p
talisein has joined #asahi
<jn__> re-training perhaps
<jn__> to adjust for temperature variations or something
<davidrysk[m]> I mean — as opposed to the bootloader/bootchain doing it
TonyRedondo has joined #asahi
<tpw_rules> doesn't that imply the bootloader couldn't use main RAM? where is it supposed to put the OS?
mah has quit [Ping timeout: 260 seconds]
<davidrysk[m]> tpw_rules: the M1 has some undetermined amount of shared cache
damia has joined #asahi
awordnot has joined #asahi
<tpw_rules> that doesn't seem like it would necessarily be enough. and i assume the M1 still supports internet recovery and other heavyweight bootloader features
<davidrysk[m]> that's implemented with a stripped-down macOS
<tpw_rules> in ROM?
<davidrysk[m]> "the bootloader" is iBoot, which is fairly small
<davidrysk[m]> no, on disk
<tpw_rules> intel macs can do that without disk
<davidrysk[m]> if you wipe that, you need to use Apple Configurator 2 and another Mac to restore it
<davidrysk[m]> yeah, M1 Macs don't have that in ROM
<damia> The project has been announced on ycombinator some discussions there https://news.ycombinator.com/item?id=25649719
<Shiz> tpw_rules: it got changed in M1
<Shiz> if you have a completely empty SSD you need to use DFU
<Shiz> no more internet recovery :(
<tpw_rules> i guess that's a lot less likely of a scenario
<Crisco> I wonder why they made the downgrade...
mah has joined #asahi
<Crisco> also hi tpw
<tpw_rules> hello
<comex> Crisco: because that was based on EFI and there's no more EFI
<Crisco> Ah makes sense
<Shiz> based in EFI in the sense that it used TcpDxe etc drivers?
<Crisco> I'm not very familiar with what Apple has changed with the M1
<Shiz> (aside from obviously well, being a part of the EFI firmware)
<comex> dunno
mah has quit [Ping timeout: 246 seconds]
<comex> but iBoot is very simple
<comex> for example, it doesn't even have USB host support
scubasteve1 has joined #asahi
brinly has joined #asahi
max_thrun has joined #asahi
max_thrun is now known as bear24rw
benjamindc has quit [Remote host closed the connection]
benjamindc has joined #asahi
scubasteve has joined #asahi
<davidrysk[m]> I expect Apple to implement an on-device ROM-DFU eventually, but it's gonna be a while
<davidrysk[m]> WiFi-ROM-DFU
<davidrysk[m]> A major thing is that it would mean extra flash, or a protected area of flash that's completely separate from the regular SSD
ransom has quit [Read error: Connection reset by peer]
burra has quit [Quit: WeeChat 3.0]
chuskywalker has quit [Remote host closed the connection]
kevincox[m] has joined #asahi
Namidairo has joined #asahi
jamadazi has quit [Read error: Connection reset by peer]
jamadazi has joined #asahi
doof has joined #asahi
veyron has joined #asahi
Stary has joined #asahi
ransom has joined #asahi
hthh_ has joined #asahi
damia has quit [Quit: damia]
damia has joined #asahi
mofux[m] has joined #asahi
milek7 has joined #asahi
jamadazi has quit [Ping timeout: 264 seconds]
mcnight[m] has joined #asahi
saaam has joined #asahi