Linux Couldn't Have Happened Without This Other Man

When you think of Linux you first think of Linus Torvalds and then maybe Richard Stallman for all the GNU stuff but don't forget about what came before Linux, a project called Minix created by Andrew S Tanenbaum
==========Support The Channel==========
► Patreon: brodierobertson.xyz/patreon
► Paypal: brodierobertson.xyz/paypal
► Liberapay: brodierobertson.xyz/liberapay
► Amazon USA: brodierobertson.xyz/amazonusa
==========Resources==========
Linux is Obsolete Video: • Linux Was Obsolete 30 ...
Linux Announcement: groups.google.com/g/comp.os.m...
Linux Is Obsolete: groups.google.com/g/comp.os.m...
2017 Minix 1: www.techpowerup.com/238514/in...
2017 Minix 2: liveatpc.com/widely-used-os-w...
Tanenbaum Open Letter: www.cs.vu.nl/~ast/intel/
=========Video Platforms==========
🎥 Odysee: brodierobertson.xyz/odysee
🎥 Podcast: techovertea.xyz/youtube
🎮 Gaming: brodierobertson.xyz/gaming
==========Social Media==========
🎤 Discord: brodierobertson.xyz/discord
🐦 Twitter: brodierobertson.xyz/twitter
🌐 Mastodon: brodierobertson.xyz/mastodon
🖥️ GitHub: brodierobertson.xyz/github
==========Credits==========
🎨 Channel Art:
Profile Picture:
/ supercozman_draws
🎵 Ending music
Track: Debris & Jonth - Game Time [NCS Release]
Music provided by NoCopyrightSounds.
Watch: • Debris & Jonth - Game ...
Free Download / Stream: ncs.io/GameTime
#Linux #OpenSource #FOSS #LinuxDesktop #Minix
DISCLOSURE: Wherever possible I use referral links, which means if you click one of the links in this video or description and make a purchase I may receive a small commission or other compensation.

[@aquilafasciata5781]

I love the beginning from Linus “just a hobby, won’t be big and professional like gnu” Oh boy he didn’t know what would happen

[@denoww9261]

"... and it probably never will support anything other that AT-harddisks" fast forward 30-something years and it has the widest OOTB hardware support of any other OS

[@htpc002Weirdhouse]

That's the same Tanenbaum that reminds you not to ignore the bandwidth of a station wagon full of tapes hurtling down the highway.

[@potatoes5829]

honestly... reasonable take imo

[@blindsniper35]

I'm pretty sure that's how NASA moved around the black hole research data that was in the news a while back. Well I think it was planes with luggage filled with hard drives. If I remember correctly there are some companies that will essentially pull up with an 18-wheeler sized NAS for you to migrate lots of data. I will also admit to have mailing hard drives before because it was faster. (Although that round trip latency is horrible) When you need to move an awful lot of data around the internet starts becoming very expensive and impractical. When you need to move massive amounts of data. The fact that sending a guy on an airplane is the fastest and cheapest solution is very amusing.(well at least compared to the internet)

[@MaDmanEXE]

And with 18TBs in each LTO-9 tape, this still applies.

[@netbotcl586]

this still applies. Cloud providers use trucks to move customer data to their servers.

[@angeldude101]

I'm pretty sure there are some network protocols that are actually designed with this in mind.

[@wannafedor4]

there are two types of brodie videos: the ones where he rants about something stupid, and the compressed movies.

[@BrodieRobertson]

Aren't those the same thing

[@yep596]

The two types of Brodie videos are wayland content and non-wayland content

[@aintnochange]

​@@yep596Lmao

[@Megalomaniakaal]

@@BrodieRobertson That's the joke

[@NJ-wb1cz]

The ones where he fingers Tux in the thumbnail, and the ones where he doesn't

[@Problematist]

The thumbnail makes it seem like it was MX Linux. With it being #1 on DistroWatch so much you'd think it was developed alongside Linux.

[@vitordelima]

Did you mean MkLinux?

[@user-ks1oh2wx6o]

@@vitordelima No MX Linux. For some reason it's always number 1 on distro watch.

[@arunabraham9382]

Minix

[@AbteilungsleiterinBeiAntifaEV]

​@@user-ks1oh2wx6othat reason is bots

[@apina4]

Big thanks to M...x and A....w T......um!

[@salazar1554]

I feel that the way Linux shoves all the drivers for CPUs, printers and stupid 30 year old peripherals into the kernel makes it more portable rather than less. It's nice that you can just shove an old linux hard drive into a new computer and mostly expect it to work.

[@mathman0569]

That's not what was meant, portable accross x86/ARM/RiscV, obviously now, that's an issue of the past

[@vidal9747]

The surprising part is that the kernel is only 70 MB. You can also strip most of the stuff you aren't using, so it is even smaller if you like to tinker with your kernel.

[@rangeramg]

​@@vidal9747my custom one is 12

[@nobodyimportant7804]

Microkernels simply don't work for the majority of use cases. That is why they have gone nowhere.

[@mathman0569]

@@nobodyimportant7804 yeah, though fun fact: the Nintendo switch uses a micro kernel

[@kxuydhj]

the lack of immature jokes displeases me greatly and as such i will point out the the "ACM Software System Award" can be abbreviated to "A.S.S. Award". i'll see myself out.

[@NJ-wb1cz]

Watch where does Brodie's finger go in the thumbnail. It seems he's way ahead of you

[@zzco]

Windows NT didn't even start out as a microkernel. It was highly modular, and has always been designed to be its own OS-with-a-kernel. You have system services that sit on top of the NT kernel, which provide what you think of as Windows (e.g., the "NT Executive" services). You never directly interact with the NT kernel itself, and this is by design.

[@framegrace1]

Well, Windows NT started, basically, as a version of VMS for mach microkernel. So the microkernel intention was there. But as all times it has been tried (Just like MacOS), it devolved to an hybrid kernel with just two "microkernels", the kernel and the Executive. They are not really isolated as they use the same memory space, so, well... so much for being "kernels".

[@SianaGearz]

@@framegrace1 All times? There's QNX.

[@zzco]

Nope, Cutler despised UNIX. He thought it was an asinine system design.

[@framegrace1]

@@zzco Don't understand, no one has mentioned UNIX. Of course he didn't like UNIX , that's why he was on the team that created VMS. We are talking about kernels here

[@zzco]

@@framegrace1 Yes, I am specifically referencing the NT kernel (ntoskrnl.exe). It was engineered as an operating system-within-a-kernel.

[@horusfalcon]

Glad to hear Dr. Tannebaum got an ACM award. He deserves it.

[@azmi3333]

I am an old guy who was around during these times. Absolutely Tanenbaum was very influential, now maybe this was because I worked in an academic environment from the time I was 18. Two of my earliest books as I moved on from computer repair technician ( in those days this was component level repair) to systems and networks. Tanenbaum's, Computer Networks and Operating Systems: Design and Implementation. Unfortunately my originals were lost in a move between Arizona and California. I suspect a box of my books got mixed up with Goodwill donations. I remember receiving the first edition of Linux Journal at my office in San Jose of course the other engineers and myself were having similar discussions on kernel architecture etc.. However it became clear Linux had legs and regardless of the philosophy we had a free Unix like OS. Good times.

[@mskiptr]

Well, the GPL isn't about "giving your work away to the competition". It's more like, "if you want to use this conveniently-available code, you have to let all the future users see the source as well".

[@user-vv8vz9iv1e]

It's litterally the same thing.

[@mskiptr]

@@user-vv8vz9iv1e The result might very well be similar, but the purpose is quite different. Whether that matter is up to you I guess.

[@user-vv8vz9iv1e]

@@mskiptr On that we both agree.

[@mskiptr]

@@user-vv8vz9iv1e I guess I was just annoyed by the tone there. It's like feeling entitled to the work of others and getting angry because there are some conditions attached. "Oh no, those pesky developers released that nice code for free, but won't let us use it in this proprietary product." If your fine with 'all rights reserved' copyright, it's just plain hypocrisy to be angry at sharealike licenses.

[@user-vv8vz9iv1e]

@@mskiptr I hope you were not annoyed towards me. I really believe that a truly free source code is behind an ISC license and not a GPL one. On this "issue" I completely follow the OpenBSD guys, the people who contributed the most to the OpenBSD guys were companies and not developers or users. My personal arguments: 1. Even Stallman never found a viable way to monetize GPL projects. 2. Your code being behind a GPL won't necessarily protect it from being stolen. 3. The only thing the GPL does is to spread like a virus to every project that uses even just 0.001% of GPL code and make everyone re-invent the wheel for no good reason that just "you should not make proprietary code". I rather let companies do what they want to do because GPL or not they will continue to monetize their code, but maybe, just maybe, if companies love your project, they will want you to keep your project up to date because they actually rely on it.

[@_DATA_EXPUNGED_]

And generally: please don't forget greg. Everybody needs a greg.

[@vidal9747]

This radicalized me against using non GPL licenses in projects that could be used for something other than their original purpose. Unless there is a clause that states: "Privacy is a human right. If there is any reasonable suspicion of using any of this code for surveillance purposes, a third party under NDA is to have access to an unobfuscated copy of the code to verify if it violates the non-human rights violation clause." Together with what reasonable suspicion entails in an addendum. Imagine writing an innocent code that is used for mass surveillance.

[@damouze]

One interesting fact: from those early flamefests you would think the two men have little respect and a great dislike for eachother. In fact, nothing could be farther from the truth. You can see from later discussions that they do in fact have great respect for eachother. Tanenbaum actually came up in Linux's (and Torvald's) defense in the days of the early copyright infringement lawsuits made against Linux, especially when someone had written a book that brought up "proof" against Linux that most of its code was copied directly from the Minix source code (a fact that the investigator this person had hired to find that "proof" was able to debunk pretty easily).

[@SheeplessNW6]

I think it's important to remember that Linus started Linux when still a university student, and Tanenbaum was a CS professor (at a different university, of course). So Tanenbaum almost inevitably displayed a didactic attitude when writing about Linux. When a CS professor criticises a student's code, that doesn't come from a place of disrespect, it's just part of the job.

[@damouze]

@@SheeplessNW6 So true.

[@xard64]

5:18 Linus "my first, and hopefully last flamefest" Torvalds I wonder how long that hopeful bliss actually lasted? PS. That thumbnail is not the type of image I'd use to depict phrase "Linux at your fingertips".

[@SnowTheParrot]

Alan Cox? Greg Kroah-Hartman? Those are the second and third names i think of after Linus...

[@1DwtEaUn]

I think of Ted Ts'o before Greg, being he's behind ext2, ext3, ext4 ...

[@NJ-wb1cz]

​@@1DwtEaUn they weren't essential. Other filesystems would've worked just fine

[@1DwtEaUn]

@@NJ-wb1cz that may be logically true, but still doesn't change the order I think of folks

[@CptJistuce]

Hans Reiser.

[@forzatoro89]

So basically NSA or whatever foreign spy agency can search security issues in older version of minix and then test on Intel cpus if intel fixed them

[@Raspredval1337]

somebody would've tried that already, idk. We had some 300iq stuff, like spectre. If there was some sloppy backdoor, it would've been exploited instead

[@CptJistuce]

​ @Raspredval1337 It's been done, actually. There've been publicly-disclosed Management Engine security vulnerabilities before. And the fix was "buy a new computer lol"... okay, it was actually "download a firmware update from your motherboard manufacturer, but that didn't happen for most systems because motherboard manufacturers do not support boards indefinitely and it hit a LOT of parts. Tthis is the first line of Intel's affected processor list when they disclosed the 2017 ME CVEs(which here is the first link on a search for "management engine vulnerabilities"): "1st, 2nd, 3rd, 4th, 5th, 6th, 7th, and 8th generation Intel Core Processor Families" That line covers everything in the Core family from 2009's Nehalem on through to the then-brand-new Coffee Lake(I believe those launched two months before Intel published their acknowledgement in the end of 2017). The third line in the list was "Intel Xeon Processor Scalable Family". No bracketing by generations, or years, or models. Just "Xeon". (Several lines were various specific Atom models, because the Atom brand was WEIRD) The only bright spot in that one was that the ME is far less powerful on processors intended for home use. A lot of the most devastating abilities(like the ability to image a hard drive without ever turning the system on(yes, really)) are only available to enterprise customers.

[@ingenarelitems]

damn brodie these chicks really love your content eh?

[@user-in2cs1vp6o]

Sometimes i feel like this kind of bot is actually from a bad actor, but they're just learning/honing their programming skills for controlling the bots. Where they're less likely to get banned in the process of learning.

[@ingenarelitems]

@@user-in2cs1vp6o true. even ph bots are better than this

[@softwarelivre2389]

RedoxOS is also trying a microkernel approach, so let's see where this goes.

[@bookle5829]

From what I heard, microkernel is just too slow but you can easily maintain it. I'm sure if they're focus on making a productive os, they would go for monolithic.

[@softwarelivre2389]

@@bookle5829 it can be slow, but it can also not be slow. It is more secure, because there's less things running as root, and if something crashes, it doesn't kernel panic. It creates better separation of attributions and prevents the need of bundling drivers in the kernel, as Linux currently does. It is overall much better, and speed really depends on how you write it.

[@GeorgeGzirishvili]

@@bookle5829, yes, microkernels usually have a higher overhead, but it's not as bad as it once was. L4 microkernels are almost as fast as monolithic kernels. They still suck at virtualization, unfortunately.

[@CharlesGriswold]

@@bookle5829 Microkernels have had a fundamental design flaw that made them very slow compared to macrokernels. GNU HURD has that flaw. Redox does not.

[@framegrace1]

​@@CharlesGriswold All this new microkernel O.S. appear now because CPU's are faster and the overhead is less noticeable on context switching speed. They all still need message copying, and that will always be a burden. It is a very hard to solve issue, because is intrinsic in the architecture. The advantages are not that big for big players to pay the penalty. They are perfect for embedded systems, although they also have competence there.

[@mnoxman]

Well.... Not quite. There was *BSD out there and Mark Williams coherent. GnuHerd has had the ability to "crib" from both *BSD and others and ++30 years later.... we are still waiting. tick tock stallman....

[@c128stuff]

Heh, yes... 12000 lines of code is tiny for an OS. I'm writing a multithreaded OS for an old 8 bit machine.. it is not complete enough for release, and currently at just over 15000 lines of code. Ok, that is all handcoded assembler, so it is a bit less code dense than a more high-level language like C, but this also is for a tiny system in comparison (128k system ram, 2mhz 6502 cpu).. Ah before someone asks... why? because I can 🙂

[@FrankHarwald]

(the entirety of the original L4 micro kernel consists of about 5000 lines of C source TOTAL! But it really is just a kernel & just that: no drivers, no plumbing layer, no user space to speak of, which is also exactly what caused its disadvantage is: L4 never had any consistently let alone stable eco-system to begin with)

[@ahettinger525]

MINIX, at the time, _wasn't_ limited to just one machine type. That was one of Andrew's complaints with Linux. With that considered, think about how much that 12k LoC gets you. The performance problem with microkernels isn't inherent, Liedtke (and his acolytes) proved that. The problem with microkernels is most of them (including mach and GNU) aren't micro enough. In a microkernel the kernel's _only_ job should be inter-process communication, and the goal should be doing that as fast as possible.

[@fuseteam]

Soooooo just dbus/binder

[@SianaGearz]

But the cost of TLB flush on context switch? Where does it go?

[@IsmaelLuceno]

@@SianaGearz normally you counter that with syscall chaining and bytecode boilerplate, think eBPF, but for chaining syscalls together, that way programs only go back to busy work loops, and the rest is computed before a context switch. It ends being better than monolithic kernels performance-wise.

[@ahettinger525]

@@fuseteam not exactly. A microkernel is running a *lot* more in usermode. All drivers are usermode. Want to write something to the HD? that NVMe driver is in usermode. It gets pretty extreme, the seL4 guys have a usermode daemon for requesting memory allocations. (the kernel takes what it needs, which is a small fixed amount, because the kernel is so small, and hands the rest to the usermode allocater). I think they ended up doing a basic scheduler in Ring 0, but that's about it, other then IPC.

[@FrankHarwald]

L4 & Liedtke mentioned! I did an intern at a company which did produce embedded systems in the early 2000s, & they were using a custom driver & userspace on top of the L4 kernel running on a smaller & older 32-bit ARM cpu (some kind of OMAP from STM). L4 was an academic project, but one of the few academic projects which actually worked, but which became obsolete nevertheless simply because it lacked & still lacks maintenance big time - it's abandon ware nowadays.

[@slebetman]

MacOS is technically a microkernel OS with Mach being the microkernel and Darwin being the macro OS running on top of it. There are versions of Linux that run on top of microkernels (linux itself is not the microkernel but it runs on top of it). I remember booting MkLinux on my old PowerPC Mac and it was one of the distros that ran well on old Apple hardware. And yes, like MacOS, MkLinux also ran on top of Mach.

[@a7i3n93]

Odd history this. I had Minix running on a Mac SE but switched to Linux to run on my new Bondi colored iMac when it became available as "Linux PPC". Minix's role in Intel ME is just sad.

[@MW-mn1el]

I still have the Minix book wrote by Tanenbaum from 2006 in my bookshelf, that's 17 years ago as part of my software engineering curriculum. Micro kernel design hasn't been prove and disprove in practice in larger scale as desktop/mobile/server/cloud os. Monolithic kernel and hybrid kernel dominances doesn't exactly prove micro kernel is inferior. OS design have been very conservative last 30-40 years. Maybe Redox and new breed of micro kernel os will put myth to the test.

[@johnr3936]

Awesome video, would love more like this.

[@israelayaladiaz9410]

I'm a simple man, if I hear "Tanenbaum" I just think of Computer Networks.

[@joshallen128]

oh Tanenbaum oh Tanenbaum you gave us linux (oops i mean minix)

[@CFWhitman]

It turns out that nobody wants to live with the overhead of microkernels in a production environment. Popular operating systems are usually based on "hybrid" kernels, which are more like monolithic kernels than microkernels, though are supposed to be a cross between the two.

[@joshallen128]

windows is hybrid, linux is monolithic, hurd is micro,

[@framegrace1]

All the "hybrid" kernels, are microkernel projects that failed. Notably, NT, and MacOS. NT was build in the mid of the microkernel fad, and tried very hard to be a "Microkernel version of VMS". But as overhead problems arised, they ended up just bundling everything they had until then on a single module and run everything on the same memory space. But it was very close to be one. MacOS is hybrid microkernel by accident. They just used NeXT kernel which was already theirs and fully developed. NeXT kernel was also made in the middle of the "Microkernel" fad but it used the "Mach" architecture, which allowed for a gradual "microkernelization". Something that was never done. (And doesn't seem it will never be).

[@SianaGearz]

QNX works well. I don't know what black magic it is doing but filesystems drivers etc are all processes.

[@framegrace1]

@@SianaGearz At least for filesystems, they seem to use some sort of Batch Copy (DMA like) procedure to copy buffers from kernel to userspace. But that's one more copy monolitic OS doesn't need to do. It has to use more CPU and add some latency, that's for sure. As I say, hardware got to a point that it may be acceptable for the most part.

[@CFWhitman]

@@framegrace1 Yes, though I believe when they based OSX (later renamed to MacOS, like the old system) on a modified Mach kernel, they had the choice of using the newer version of it which was a true microkernel, but didn't because it wasn't performant enough. Even the Linux kernel is more of a modified monolithic kernel so that it can have different modules for various drivers, though it can still be compiled as completely monolithic which is often done for custom embedded systems. Linus Torvalds has stated that kernels called "hybrid" are really still monolithic in all the ways that matter.

[@MW-mn1el]

Maybe Redox would be the new hope at some point 😄

[@stefanalecu9532]

Yes, like all the other OSs... BeOS and OS/2 were also the new hope in the 90s, and they had actual reasons to exist, not just "oooh, Rust OS"

[@MW-mn1el]

@@stefanalecu9532 Competing options and new ideas are always good for innovations. But it's 30 years ago, we can always use a "new" hope and a "new" hope take 2 spin after that, otherwise Disney won't keep make money from nostalgia with Star Wars universe.😄

[@LostieTrekieTechie]

For a moment I thought the ellipsis was replacing "AR", rather than "INI", lmao

[@ottolehikoinen6193]

Yes I heard of Minix, it was about the times I was not into computers, the it-guy said something like "can't be really good since it's free". Two years afterwards i heard of a student who was building an OS as a hobby. 1991 and 1993 I think it was.

[@jack_irl]

Hey nice vid Brodie, I read a lot about this in Rebel Code, its a good book you will find interesting and related to Linux / Hacking history.

[@SheeplessNW6]

Before Minix, there was another resource widely used by CS students studying operating system internals: "A Commentary on the Unix Operating System" by John Lions, more commonly known as just "the Lions book". This came in two parts: the source code of Bell Labs 6th Edition Unix for the PDP-11 (less than 10k lines of C and assembler), and the commentary itself. It was an excellent resource for studying operating systems. It wasn't formally published, and not available in the usual way for textbooks, but was distributed to source licensees: mainly universities. It was so popular with students that it was widely photocopied, and samizdat copies abounded. (Don't ask me how I know, and definitely don't check my bookshelves). When 7th edition Unix was announced, the new license no longer allowed the source to be used in teaching, and this was one of the things which spurred Tanenbaum to write Minix. The Lions book was finally properly published in 1996, with a cover depicting two nervous students at a photocopier. Anyone with an interest in the history of operating systems ought to find a copy (though I think it's out of print).

[@PhilBoswell]

I picked up a copy of this on the charity shelf in our local Tesco this last weekend, for the princely sum of 50p donated in the tin 🤯 It's not the easiest read, I' think it'll take me a while 🤣 I was surprised to see the date of 1987 for the Tanenbaum book because I had that for my OS course at UCL and 1987 was my last year: we had a PC (clone? I forget the exact model ;-) in the basement with MINIX set up for people to gawp at.

[@SheeplessNW6]

@@PhilBoswell see lines 2230-2239. I was at UCL three years before you. Spent many a happy hour at a Newbury Lab terminal in the basement, using 6th edition on one of the 11/44 machines.

[@13thravenpurple94]

Excellent video 👍 Thank you 💜

[@FrankHarwald]

Yes! Andrew S. Tanenbaum - one of the OG of OS design. Along with Abraham Silberschatz, Peter Galvin, Dennis Ritchie, Ken Thompson, Gene Amdahl, Jochen Liedtke...

[@peterjansen4826]

I have known for a few years that Minix is used to boot a system with an Intel CPU, I suspect that AMD also uses Minix or something very similar, they also have an engine like the IME and they also need to get the POST done without the main OS running so they need some OS for that. Fun little fact: Ryzen-CPU's contain a few ARM-cores, for the POST-process and as I understand it also for security.

[@oxyacetylene_]

If it were me, this would have been my Oppenheimer moment

[@groos3449]

Time is a loop. Microkernels are once again the future.

[@SnowTheParrot]

@BrodieRobertson can you link the video you mentioned about the back and forth between Andrew and Linus? Thanks!

[@BrodieRobertson]

It's in the description now

[@SnowTheParrot]

@@BrodieRobertson appreciate it!

[@capability-snob]

If you had him on Tech over Tea I'd be down for that. He's not very prolific on youtube.

[@jktolford8272]

Thanks for the history.

[@pcartisan2721]

Well written. I actually stayed awake during this entire history lesson. Good job!

[@peterjansen4826]

Let me guess: Minix. This guy worked in the Netherlands (where I live) when he worked on Minix. His book on operating systems is the standard book for computer scientists.

[@lul202]

Tannenbaum is a total legend. Both Stallman and Torvald are simply unable to compete with him in terms of actual influence and insight. There strengths have always been in showmanship and out ass-ing the other assholes in the room. Tannenbaum's real weakness is that he constantly under-estimates how the market will select "it works well enough and is gaining traction" over correctness. The really funny thing is just how portable Linux has become.

[@knghtbrd]

I think if Linus had never created Linux, GNU HURD would've gotten more developers … but it probably would've been FreeBSD that most of us were running today. A big part of that is as much as Stallman hates to admit it, a big part of Linux's wider adoption has been the companies, and HURD would've been actively hostile to them. Then again, we might all be living in a brave GNU world if Linux hadn't come along, I dunno. I guarantee you wouldn't have nvidia GPU support on a HURD system. That said, MINIX is obsolete. 😉

[@IsmaelLuceno]

there's plenty of companies profiting from GPLed code, even GPLv3, the main reason BSD would have triumphed is it's community, but we would be nowhere near where we are now, because there's less incentives for companies to properly support a BSD-licensed OS on the desktop. For smaller companies selling services instead of products, the GPL makes a lot of sense, for a product by-the-unit sales model it's a lot more complicated.

[@MonochromeWench]

IME exploits seem like they tend to be hardware attacks, not attacks at Minix. IME is so locked down you can't even talk to IME without first finding a hardware exploit, you can't just make a syscall into IME. The aim of the exploits is usually to get write access to IME memory. If you get write access you can rewrite the IME page tables and interrupt vector table and execute arbitrary code. The problem is it's just too hard to get a copy of all the IIME code without first being able to do arbitrary code execution. And if you've already got arbitrary code execution, you don't need the Minix code to find exploits. Getting the Minix code is more useful to see if Intel is doing anything that is a serious immediate privacy/security concern

[@Sqwert-g6h]

The Tannenbaum guy sounds a lot more based in his later writings.

[@wonderings8973]

Linux would not have made it past the early years if it hadn't been for the incredible work & coding prowess of Alan Cox

[@polinskitom2277]

Imagine a world where andrew didn't licence minix as BSD and intel couldn't create the IME. that's one bad scathing mark that will haunt andrew way past his grave unfortunately

[@nou712]

The guy who singlehandedly backdoored every computer. Amd copied intel with their psp.

[@TrolleyMC]

In all fairness Intel could've made their own OS for the ME regardless if Minix was GPL or not. The reason they used Minix was just to save time and money so that they can integrate ME quicker.

[@joshallen128]

@@TrolleyMC they could have used NetOpenFreeBSD

[@fish3977]

Naah, cucks get off on that

[@No-mq5lw]

I've been told that Intel now uses their own OS for IME

[@kuhluhOG]

Since from what we know, the Intel Management Engine is ibased on MINIX, Let's say "IME = MINUX". So the question which one, MINIX or Linux, is more used is actually pretty hard to answer. On the one hand, MINIX is pretty much only used there. But also a good chunk of Intel CPUs run Linux. So they have quite a big overlap. So now we need to get the number of Intel CPUs with IME but without Linux (but instead Windows or macOS or something else) and we need to get the number of Linux-based OS which don't run on top of an Intel CPU. Intel has a considerable majority when it comes desktop and server CPUs (currently) and we also need to count at least a share of these servers as non-Linux, because a lot of corporations depend on Active Directory. But there's also Android and embedded Linux. My guess would be that Linux is used more, but that's pretty much it, a guess. But I would argue that MINIX has more deployments than Windows (be it Desktop or Server).

[@BUDA20]

yes, I read the book back in the day, and I also believe at the time that microkernels where the future or at least modularity at runtime

[@IsmaelLuceno]

without Linux, probably we would be using some of the mainstream BSDs now (386BSD was released in 1992)

[@DeeaA.-qu2bn]

Cheeky comment section today

[@TenTypekMatus]

13:23 NT and XNU are microkernels, btw.

[@human-V-oid]

His tome on Operating Systems is recommended by my college.

[@jamesyoung151]

I remember hearing about Andrew Tanenbaum often when I was studying Computer Science and Computer Information Systems back in the (dating myself) 1990s. The ironic part of all is that I had his textbook in college. I got rid of it for the cash of course. Too bad I didn't keep the book then.

[@sussteve226]

Bro what are these bots this is a tech channel 😢

[@tranthien3932]

Brodie going mainstream...?

[@deadfry42]

you know they wouldn’t be doing it if it didn’t work

[@vaisakhkm783]

Ig guy making this bots might watch brodie

[@cameronbosch1213]

The bots making it 👏 or claiming "2 views in 1 minute, bro fell off"? Either way, 🙄

[@johnwick1877]

It's a channel and that's all that matters but I do wonder, what is their purpose? What do they achieve?

[@seanscon]

YES MINIX We strated using it to run a management engine to drive a set of neuromorphic computer chips.

[@siljrath]

thankyou

[@esra_erimez]

My dad *still* agrees with Tanenbaum

[@SianaGearz]

He's an OK boomer.

[@TheLonelyMoon]

MX Linux sets the stage, huge respect

[@galen__]

MINIX Inside 😅

[@paulstubbs7678]

I would have thought if we didn't get LInux, we'd be all on Windows - as minix was buried in coding hell

[@IgorGuerrero]

Oh Professor Tanenbaum? EVERYONE read his book and Minix back in the day, even when I went to university, I read it ~2003, kinda sucks people didn't experience that time / researched it at school, even OS like Amoeba were interesting back in the days when Ubuntu didn't even existed.

[@IgorGuerrero]

I remember, for example, TYPING Minix, just to learn some algorithms :)

[@simonscott1121]

Alan Cox?

[@TilmanBaumann]

To be fair, Linux is a bit of a sunk cost fallacy at this point. A Microkernel would totally be the right solution on todays hardware.

[@yep596]

That usenet thread between Tanenbaum and Torvalds is still getting posts 30 years later, lol

[@BrodieRobertson]

Google Archived things weirdly, those were not actually being sent on usenet. They were being sent on Google's thing

[@yep596]

@@BrodieRobertson Ah right, thanks for clearing that up. It was also interesting to see Ken Thompson participating in the thread as well

[@mtxn]

if linux wasn't a thing and the hurd would be the OS of our choice. companies like nvidia would have a hard time developing software for hurd. Stallman would probably add paragraph to the license about restrictions to use proprietary software under gnu hurd based Distros. I bet that in described previous timeline the 'VMS/BSD/Minix' would be installed on every server. something like black mirror scenario

[@stefanalecu9532]

Or BSD would become more popular, if not for the lawsuit in our current timeline

[@dozerd42]

Now I am curious if i can get MINIX running on an RPi Pico. Honestly, i have not heard of anyone running MINIX in embedded. Anyone know of any pros/cons over FreeRTOS?

[@FrankHarwald]

I don't know about Minix on RPi Pico, but I know for sure that you can pretty much run bare metal assembly on any Raspberry Pi if you know how their firmware bootloader works. config.txt is very well documented, shouldn't be a problem if you also have hardware documentation or the OS you're trying to boot has basic support for the Raspberry Pi platform hardware.

[@rogerdahl0]

What are you doing to the butt of that penguin in the thumbnail?

[@supremesonicbrazil]

> "If I had suspected they might be building a spy engine, I certainly wouldn't have cooperated" Oh the fucking irony. The guy got an award for indirectly enabling the biggest and most prominent backdoor in all of modern computing out of his own inertia and stubbornness about "being right about microkernels", and then tries to save face by acting all pissy whiny about not being told about it in a future where his POV was actually wrong. If that's not the literal embodiment of a "pedantic hypocrite" I don't know what is.

[@nobodyimportant7804]

You are correct, if not for the incompetence of Tanenbaum, there would be no Linux. Minix(and Hurd) was a bad idea that needed to be stomped on and Linus was the perfect person to do that. Great ideas are usually started in opposition to brain-dead ideas. It is a shame that Linus kept the shortbus Minix(well Unix, I suppose) FS structure as it makes it so difficult to find files as they can be in one of about a dozen places. His kernel architecture ideas were wrong and the OS book that he wrote that subpar CS programs inflict on their students is a crime against humanity. Unlike Torvalds, Tannenbaum didn't have the intellect to back up his ego. Corporations can't stop lying about the GPL. You only have to make changes available if you distribute it and only those who you distribute it to - of course they have distribution rights also. Keep changes in house and they stay in house. I am not a fan of the GPL or Stallman and even I know that.

[@Matt2010]

I actually think a Monolith Kernel like Linux is way better. and In-Kernel-Drivers. Does it have to be totally open source, well how about this NVIDIA Didn't you hire someone that comes from the Linux side of things?? Have him see the source code, then work with him to apply a way for a in-kernel way and have semi-open source only the one you hired can truly know, to be sure there's no intentional bad code, and have him and any other future person always from the Linux side of things way. It's an idea. No, This is for me the first time hearing of this guy. Know of the others though.

[@user-vv8vz9iv1e]

The only reason why the world was not taken over by microkernels is because it is much simpler to write a shitty monolithic kernel than a shitty microkernel and it's much easier to just reuse the same kernel base for decades for compatibility purposes for example than taking a stand. Microsoft will never take a stand unless forced to, it's not in their interest for example.

[@tecoberg]

👏

[@andymann6061]

Perhaps it would all be BSD.

[@jagagemo8141]

Going to comment just to offset the bot to human content. 🙃

[@shardzkaylar]

I wish I had half the brain power Linus has. Hell, I wish I had 5% of the brain power Linus has. Thanks for the history lesson, I'm new to Linux.

[@ichauch110]

nope. GNU Hurd still is not ready.

[@HeCodes2Much]

Wow made it here in the first 30 seconds lol

[@justacat3639]

Congrats Also you're first

[@ParticularSky]

Strange horni bot

[@HeCodes2Much]

Bot ya right lol

[@HeCodes2Much]

Ya speak for yourself lol

[@ustrucx]

The thumbnail looks like your finger is.....nvm...will you talk about nixos or ladybird news?

[@CRYPTiCEXiLE]

yes im well aware of mnix and yes i well aware of this story and yes Linux is the best :)

[@valerianmp]

Or you'd probably know that name from his textbook on computer network lol

[@genstian]

Haven't we had over 60 CVEs regarding this spyware OS on intel chips? Why does on OS meant to do system initialization needs a network stack - isn't it just a big potensial security vulnerability? Don't trust closed source OSes, they are most probably spyware, like Windows.

[@stefanalecu9532]

That's like saying Linux is a spyware OS because Red Star OS and Deepin exist, this is so dumb

[@zxuiji]

GNU Hu? :D

[@Supervisor360]

why is your finger in tux...

[@BrodieRobertson]

Why isn't yours

[@Supervisor360]

@@BrodieRobertson Reserved for Xenia

[@BrodieRobertson]

@@Supervisor360 do you only have 1 finger?

[@Supervisor360]

@@BrodieRobertson nah

[@Supervisor360]

@@BrodieRobertson no i have more

[@Vantaz]

I swear to god you made basically yhe same video like 2 yeara ago

[@Sim-rh4tj]

The difference between Linux and the better ways of doing it is that Linux got written!

[@stefanalecu9532]

There are still better ways than Linux, but sadly those were forgotten (look at the other OSs available at that time). But then, Linux's existence is a huge factor for why OS research is barely a thing anymore, because they think "Unix is good" and don't bother doing anything else

[@know_how_games]

snd i thought malcom x couldnt get better

[@sarkybugger5009]

What the hell is Linix?

[@BrodieRobertson]

What indeed

[@philipjohansson3949]

Lennart Poettering?

[@SnowTheParrot]

Stallman? I dont think Stallman when i think of Linux. I think of Stallman when I think about people eating the dry skin off of their feet.

[@BrodieRobertson]

He does more than 1 thing lol

[@SnowTheParrot]

@@BrodieRobertson lol yeah i was just being funny, hope i didnt come off disrespectful

[@SianaGearz]

Is that a category, how many of such people have you got?

[@SnowTheParrot]

@@SianaGearz just Stallman so far.

[@TheRealFallenDemon]

My friend lent me the raccoon book in uni. I learned a lot from it

[@U1TR4F0RCE]

I probably heard of Tanenbaum in my OS class but I first remember hearing about him from your KZbin video about the infamous debate. Funnily enough when talking to a co-worker he remembered being there at the time that there was that debate and overall felt that micro-kernel and some versions of a hybrid kernel are technically superior but since they aren't deployed for desktop it doesn't necessarily matter. Though, he did mention that Windows NT took the worst of both worlds.

[@ENJERUNEbutLIVE]

woah, i didnt knew women were into this kind of tech content

[@felixfourcolor]

"women"

[@No-mq5lw]

Is it OK to YWNBAW pron bots?

[@Artoooooor]

Tanenbaum! And f*** Intel.

[@TheCurtisnixon]

yet another reason to use amd...

[@mskiptr]

AMD has its Platform Security Processor though. We have to abandon x86 really…

[@TheCurtisnixon]

and lol.....we could all stop using x86 tomorrow, and use arm, but we'll still be dealing with x86 for the next 50 years. eg, we're still using cobal, every day. every single one of us....

[@SianaGearz]

@@TheCurtisnixon Hold on. We might not abandon x86 ISA, but abandoning the x86 processors in particular... is not that hard. Apple changed the processor architecture... 68k to PPC, PPC to x86, x86 to ARM, that's 3 times now, every time aided by software emulation, just-in-time translation. It's difficult to abandon and obsolete software, but hardware can be abandoned by adding one more piece of software. Adding things is easier than removing things.

[@Alioth-sd5rz]

M..x Marx?

[@BrodieRobertson]

1 more dot

[@Alioth-sd5rz]

@@BrodieRobertson maarx

[@nikbl4k]

Wow, "Donald Trump" really?.. Thanks donald for that hefty donation of 3trillion in 93'