MINIX 3.2.0 stable is released. Major features include: Asynchronous, multithreaded virtual file system (VFS) server, Better reliability, Better virtualization support, and much more.
MINIX 3.2.0 stable is released. Major features include: Asynchronous, multithreaded virtual file system (VFS) server, Better reliability, Better virtualization support, and much more.
I tried the latest release today. Couldn’t get it to run at all in VirtualBox, the kernel fails right after the boot menu. Also tried in a QEMU VM and got it to install, but can’t get MINIX to connect to the network. The ifconfig utility keeps crashing with a memory allocation error. Haven’t tried it on physical hardware as MINIX doesn’t support my physical network card.
I checked the documentation and confirmed all my VM settings were what the MINIX devs recommend. The MINIX team may be making some great progress (the installer is certainly straight forward, man pages are well written, etc), but I can’t get the operating system to a point where I can do anything useful with it. No network and no desktop makes for a limiting experience.
Are you using VirtualBox 3.2 or newer? The release notes specifically mention this issue and it’s a bug with VirtualBox (what a surprise).
There is a bug too with VirtualPC 2007 ;-(
I’m using VirtualBox 4.1.
As I mentioned in the above post I also tried qemu and, while MINIX sort of works there, I can’t get a network connection with it. It struck me as odd since I was able to get a previous version (I think 3.1.x) working without any trouble.
I am using Virtual Box 4.1.8 and install seems to be going fine.
how ironic that they use git.
Why is that ironic?
Linus was one of the designers of GIT, apparently him and the creator of Minix do not (or did not) get along when it comes to their views on OSes.
But their views on oses have nothing to do with their views on scms…?
I have never heard or read about them not getting along. Mind pointing me to a source/reference?
Thank you in advance
Edited 2012-03-02 04:48 UTC
Here is the whole thread:
http://groups.google.com/group/comp.os.minix/browse_thread/thread/c…
Wikipedia article about it:
http://en.wikipedia.org/wiki/Tanenbaum%E2%80%93Torvalds…
I agree that Andrew Tanenbaum using Git for MINIX is ironic. In a way I think it proves that professors can be wrong sometimes, too.
Maybe they should learn more about humility and being humble.
Edited 2012-03-02 05:43 UTC
Oh, the great debates?
This doesn’t suggest that they don’t get along. It only proves that they have different opinions regarding the design of os kernels.
Of course professors can be wrong but they usually aren’t .
Professor Andrew seems to be even more humble than Linus.
Using git has nothing to do with the architecture of os kernels so I don’t see the irony.
If Andrew started accepting monothilic kernel designs in his kernel then that would be ironic.
Sometimes a joke is just a joke…
Sometimes a bad joke is just a bad joke…
Considering that VxWorks and QNX are very successful micro kernel OSs, that both Microsoft and Apple have a partial micro kernel approach in their OS, I would say that Andrew was/is quite right.
Remember also, with this particular holy war, such gems as L4 (“one billion L4 kernels” in Qualcomm-based mobiles 1.5 year ago, who knows how much now http://en.wikipedia.org/wiki/L4_microkernel_family#Commercial_deplo… ) or INTEGRITY (flew a modern airliner or… jet fighter lately? http://en.wikipedia.org/wiki/Integrity_(operating_system)#INTEGRITY-178B )
Edited 2012-03-04 21:00 UTC
How does minix using a tool that has nothing to do with the disagreement between Andy and Linus make him wrong in any way?
Perhaps I’ve used the wrong words, I just think there’s some irony in this.
I mean, Linux became bigger than MINIX and Git was created as a result of Linux development and growth.
Andrew Tanenbaum in that post said that Linux would have no future due to its monolithic design, now they’re using Git, which came out from Linux development not a long time ago.
Was Andrew right or not?
Edited 2012-03-02 07:30 UTC
Minix was not always free(as in speech).
Minix was initially just a research operating system. It’s just since recent times that Andrew let known his ambitions for minix.
Too often I see Torvalds’ minions spreading FUD about minix, general misconceptions etc…
Would u suggest that minix use inferior SCM tool just to avoid this perceiced irony? Avoid every and anything related to linux or its creator?
That would be cazy :/.
There is no irony in Andrew’s team using git. Andrew has never publicly disagreed with anything git or related to git.
Was Andrew right or not?
You decide. Linux is widely used though a failure on the desktop. Linux is a huge piece of software, which is difficult to learn and maintain. Many distributions are unstable. Right now I’m using ubuntu which is forcing me to choose between having children in the future and resting my laptop on my lap because it goes well over 80 degrees C whenever I play any media or connect to my desktop over vnc. Sometimes I try to play a mp3 and for some reason the os hangs, then dies. If a microkernel like minix will offer me greater stability than this, then I welcome it.
The linux of yore was nothing like the linux of today. Linux has had to evolve to stay relevant. It will have to continue to evolve. So will minix.
Success for linux doesn’t mean failure for minix or vice versa.
In the near future, minix will not be directly comparable to linux because linux is just a kernel while minix(3) is a full operating system.
Why would you put your laptop directly over your nutsack? Do you have tiny legs or something?
No sir. Please excuse the euphemism.
he just has a massive nutsack.
I don’t think anyone is suggesting that Minix shouldn’t be using GIT, just that as it grows larger its utilization of tools that were born out of necessity to support Linux, who’s popularity the author of Minix severely underestimated, is somewhat ironic.
Linux was born, and grows, by utilization (and kinda stealing the credits) of tools that were born out of necessity to support GNU, a microkernel operating system, oh the irony!!!11111
The overheating issues you mention seems to be more of a hardware issue than anything else.
I’ve tried running several Linux distributions in different laptops, I know my Dell used to overheat like crazy, but my ThinkPad doesn’t.
Perhaps your overheating problems are caused by the hardware itself?
What laptop do you have? I do not experience any overheating issues with my ThinkPad T520, in fact, the laptop remains cool all the time. Heck, I can put the laptop on my bed compiling the full kernel in parallel and it won’t overheat.
Also, read this:
https://lkml.org/lkml/2012/2/22/569
https://lkml.org/lkml/2012/2/24/338
Moving drivers to userspace is not a solution to buggy drivers. The solution is to keep reporting bugs so those bugs can be fixed, microkernel or not.
When drivers are stable, as they should be, microkernel vs monolithic doesn’t matter much anymore IMHO. The microkernel design most likely will just be using more resources and it will feel slower with all the message-passing IPC mechanism.
Just keep reporting bugs if you find any. This is the way of stability, not microkernel.
Edited 2012-03-02 17:41 UTC
I doubt it is a hardware issue as I do not have this problem when I boot into windows 7 or haiku.
I’ve experienced this issue on 3 different laptops while not on my netbook.
I’ve read those already. I would not propose that the drivers in linux be moved to userspace. That would require too much work and the benefits may not justify it.
However, since minix started out as a microkernel, it had the opportunity to offer a different approach to driver development. Writing drivers in urserspace is easier than writing drivers in kernel space and may be much less error prone.
The performance loss experienced with using a microkernel may not be as much as you think.
Experiments running linux on a L4 microkernel shows linux performing less than 10% faster on average.
Copying a file and having it taking 10 seconds rather than 9 seconds is acceptable when the possibility of my system crashing because I’m playing a mp3 while i wait for my copy to finish.
Agreed to the point of your comma. Microkernels offer a valid solution.
You can try comparing the performance of the blackberry playbook with a similarly specced android tablet to see that microkernels aren’t inherently bad for performance.
Agreed: fanboys are annoying but that’s not a reason to spread FUD yourself too..
FUD: micro-kernel improves modularity but performance isn’t modular, so there no reason why performance would be significantly better or worst with a microkernel..
I hope, but for now Minix has to evolve to become relevant not to stay relevant which Linux already is (more or less depending on your needs).
This is obviously not a performance issue, but a stability issue. The performance of my laptop is just fine. However, I was expecting less crashes… And these crashes seems to be caused by a driver issue. If the issue was isolated to a single subsystem e.g sound driver, then the system would be able to restart the sound driver in a microkernel design; not so in the design of linux where the misbehaving driver will most likely crash the system.
When did minix 3 start out? How long did it take for linux to become relevant?
Have you tried to do some bug reporting?
Why don’t you tell us what laptop you have? or what chipset/module/driver you were using, and if you could reproduce this?
I wonder if you’re just trolling…
I did a bug report a while back.
I’m using a Dell Inspiron N4030. I also had the same issue with a Compaq F579WM and My HP at work (I don’t remember the exact model number now but it’s something like nc6120…).
The primary one, the Dell N4030 http://www.villman.com/Product-Detail/Dell_Insp_4030_W7B .
I’m sure I can reproduce this right now.
jayrulez@IN4030:~$ sensors
acpitz-virtual-0
Adapter: Virtual device
temp1: +80.5°C (crit = +87.0°C)
playing a video (note not flash).
If I didn’t kill the player through the system monitor utility then it would have just gotten a little hotter then hang then shut off.
I have windows 7 and ubuntu dual booted. I cannot use ubuntu for skype or google/+ video calls because of this, so I have to log into windows 7 at least once per week.
I’m in the process of backing up the system now to install 12.04 beta hoping that this issue has been resolved in the later kernel version.
In your post you were complaining both of a performance issue and a crash issue.
Maybe, but I remember an issue with a firmware(driver?) where a SoundBlaster hogged the PCI bus, similar bus issues on a telecom equipment, temperature issues causing crashes, DMA issues (IO-MMU are still disabled because they’re too buggy AFAIK): all these issues cannot be solved by microkernels.
The Hurd was started a long time ago, it never became relevant despiste being “micro-kernelish”, I wish Minix3 better luck, we’ll see but remember that historically a better design isn’t enough to be successful.
A lo
Well MINIX IS releven right now. It is the only working open source microkernel *NIX we have and it is the only open source microkernel *NIX with the declared goal of working on embedded hardware. It isn’t there yet but who knows what will come? I’m hopping that ideas from QNX and the HURD will find their way into MINIX. Just imagine having a free *NIX that is actually a RT kernel out of the box!
Apparently the L4 family can be also described as Unix-like (at least in some usages), with open source flavours, and generally already very successful in embedded ( http://en.wikipedia.org/wiki/L4_microkernel_family#Commercial_deplo… )
You are talking about a conversation from 1992!
Tanenbaum thought that Linux had no future because of its architecture and he was right! The Linux architecture back then was a monolithic kernel that was NOT portable at all. Our modern modular and portable Linux is a totally diferent animal.
Linux was right too, MINIX was an incomplete academic teaching toy and NOT free.
Both where right and over time we got a better Linux AND a better MINIX even if MINIX still has some way to before it becomes truly useful.
Consider the situation back then.
MINIX not free and a toy. Linux incomplete, monolithic, x86 only and more or less crap. BSD involved in a lawsuit. We had no good choices at all and now if you want a *NIX your best choices are all free AND we have so many of them that we get to argue about the best choice too!
When it comes to the issue which kernel architecture is better I really don’t give a shit. Linux proved one thing (well Linux AND HURD proved one thing), it is easier to improve what you have then to start over from scratch to build a theoretically better or purer architecture.
No, it’s the same architecture, it’s still monolithic.
Tanenbaum’s criticism was not leveled at the lack of portability, it was at Linux being monolithic. That Linux nowadays only loads the modules into memory that it needs doesn’t change it from being monolithic, they become part of the kernel! So no, Tanenbaum was wrong, Linux is a success, a much bigger success than any micro kernel.
There was a surge of micro-kernel optimism back then in academia which spilled over into commercial/open source space, resulting in failed projects like Apple mkLinux, IBM Workplace OS, Hurd, and of course Mach itself.
Micro-kernels have the benefit of architectural stability at the cost of performance and system simplicity.
If monolithic kernels were indeed prone to crashing then the industry would have moved towards micro-kernel architectures a long time ago. However they are not, which means that monolithic kernels provide both stability and performance. Certainly there are areas in which measures towards stability trumps any performance considerations and in these areas something like a micro-kernel which can keep on running should one component fail would indeed be attractive.
Emphasis on ‘theoretically better’, which is often the reason academia (theoretical) and real world (practical) go different ways.
MacOSX and Windows are both Hybrid Kernels.
This is why when My Nvidia driver crashes it doesn’t take down the whole of Windows (like my S3 Savage Driver used to do in Windows 2000).
There are different measures of success – for example, the INTEGRITY microkernel-based OS is used by avionics computers of quite a few modern military jets, and of more recent airliners like Airbus 380 or Boeing 787, demonstrating those “academic” advantages. Real stability, and also seemingly not a bad performance.
Or L4, apparently running billion+ mobile phone radio stacks ( http://en.wikipedia.org/wiki/L4_microkernel_family#Commercial_deplo… ), many of them presumably Androids (Qualcomm) – so, for this certainly performance-critical part of the handset, choosing a microkernel (probably on very anemic ARM core of radio module) over Linux, and/or even using L4 as a hypervisor for the “user facing” mobile OS (some of the references there would at least suggest this), possibly also underneath Android, down the line.
Edited 2012-03-04 20:39 UTC
This is pure irony. You talk about Linus opponents should be humble? And what about Linus himself? Doesnt he have an attitude problem with his big ego? Stealing all cred from the GNU project. Calling OpenBSD developers “Masturbating monkeys”, for their focus on security. “I have scared away all normal developers, now only the rest is left on the Linux kernel project”. etc etc
So Linus T has no attitude problems, right? He is humble right?
It seems Linus’ original post announcing he was going to do something called Linux was the last time he seemed to be a nice guy.
Since then he has turned to a rude and very easy to dislike bastard.
But hey, the Linux crowd loves this.
Actually Linus just speaks him mind.
I have heard myself saying the same things when being frustrated with people moaning about “browser X” … being a web dev myself.
He is direct and honest … I rather deal with someone like him than someone who smiles to my face and wishes to stab me in the back.
I see this similarly to the “Strict by fair” teachers … these were always the best teachers, because you knew what they expected.
Trust me, you dont mean this. There is no need to insult people when you want to say something to them. It is called respect for others. You are just too weird.
If you are fat, or ugly, would it be ok if everybody told you that you are fat and disgusting? How respectful is that? Even if you dislike some attributes on some people, you do not say that loud. You do not say everything that comes into your mind, only children do. And Linus Torvalds. The worst thing is that some people thinks that such confrontative behavior is ok. No it is not. Where are your manners? Jesus. Have you heard about politeness, diplomacy, respect, etc?
Uh, and this relates to Linus exactly how? Does he personally insult people on a regular basis by calling them fat and disgusting? I’m pretty sure that’s not the case.
The key here is context. Linus made a general statement on his blog, he didn’t post a reply to someone else’s story about their hard life and how bad they felt. If he had done the latter and said “You should kill yourself” that would have been in bad form and despicable.
But…he didn’t.
And, I guess, George Carlin, Richard Pryor and, you know, anyone who doesn’t beat around the bush about things.
I think he does and you know what, I agree with him.
I don’t necessary agree with Linus opinions (mostly I disagree really) but I have no problem with how he voices them. Of course, I have no problem with Theo de Raadt either.
is earned and not automagic.
How about “You are full of shit” in a reply, or “OpenBSD devs are masturbating monkeys” or “I hope this OS dies, because I am in active competition with them”? I bet there are many other quotes, I have not bothered to look. These were just taken from the top of my head.
How about your manager called you “masturbating monkey”? Or “you are full of shit”? Or, you worthless little piece of shit?
You dont really agree that you can reword these formulations slightly? Maybe, “I appreciate the work you have done, but you need to rework this part” instead? What would you prefer?
If you and your friends sit in a car, and you take the wrong turn, would you prefer they insult you? Very respectful, yes?
And, I guess, George Carlin, Richard Pryor and, you know, anyone who doesn’t beat around the bush about things. [/q]
Just because someone behaves bad, it does not justify similar behavior from others. Bad childish behavior is not ok from anyone. Why do you think that? “Because Richard Pryor calls someone fuck head, it is ok to say so to my dad” – huh???
I dont agree with this confrontative insultive managing style. You can choose what words to use, no one forces you to insult people. That only makes them feel bad. If you have a big ego – I understand you think that making people feel bad is ok. I hope you dont treat your (future?) children by calling them fuck face and try to diminish them, just so you can feel superior.
But know that EQ and social skills can be trained and increased.
I think calling BSD developers masturbating monkeys goes a step beyond just being honest and speaking ones’ mind.
It guess it depends what your relationship is when it comes to brutal directness. A teacher or a coach is someone your stuck with and he can push you to extend your limits.
Open Source devs aren’t stuck with Linus, they don’t have to listen to him, he has no authority over them. In that case I think you need to be more political. He can still be critical, even in strong words, but that’s okay if it’s well motivated. It’s not like he has done any coding miracles lately.
If you’re not involved in this world only the nasty bits filter out and paints Mr. Linux as some rude idiot who is in a constant rage.
What I said about humility definitely goes to Linus Torvalds too.
Perhaps he needs to be taught a lesson? Maybe a fork is coming?
Its still funny.
Yes, its different software than they were arguing about, but they were really arguing about philosophies of software development. Pragmatic vs Academic. Linus is the ultimate pragmatist, IMHO. Git was designed with the same pragmatic philosophy ( “screw disk space, I need speed”).
This along with the increased use of NetBSD code, shows that Minux is becoming more pragmatic all the time.
Edited 2012-03-02 15:50 UTC
AFAIK the only thing they “don’t get along with” is which of micro or monolithic kernels are a “better” design.
The whole thing is massively blown out of context by the press and even more so here as Git isn’t even remotely related to kernel design.
It’s a little like saying I’m not going to use any software nor services developed by anyone on OSNews who I’ve ever disagreed on. If everyone acted like that – nothing would ever get done.
Its give and take policy
Nice work for Minix. I hope some day, Plan 9 would also embrace clang or gcc or pcc (modern supported compilers).
The lack of USB stack / peripherals is something that keeps Minix from being used more extensively.Hopefully soon they will have something in this area. Even more important, they could start thinking about DDE kit.
I had just installed 3.1.8 last week.
Now I have to install the new version again, but there lots of new things to play around with.