User: Password:
|
|
Subscribe / Log in / New account

MINIX 3.2.1 released

MINIX 3.2.1 released

Posted Feb 23, 2013 12:26 UTC (Sat) by hadrons123 (guest, #72126)
In reply to: MINIX 3.2.1 released by mabshoff
Parent article: MINIX 3.2.1 released

Well it is possible one might assume my post as trolling. When you look deeper, things are not so bright for this OS as per your post.

But what has this minix have really achieved?

I dusted my old P4 to see if this really works, but then I got kernel panic and the screen went blank. Didn't want to try anymore beyond that. Happened the same last year with the previous release.
So practically pointless release with 2 CD wasted.


(Log in to post comments)

MINIX 3.2.1 released

Posted Feb 23, 2013 12:57 UTC (Sat) by Zack (guest, #37335) [Link]

>But what has this minix have really achieved?

Well, presumably, AST is managing to scratch his itch by it, and the rest , I imagine, are having fun hacking on it and using it, which is all the justification a free software project would ever need when it comes to achievement.

MINIX 3.2.1 released

Posted Feb 23, 2013 19:12 UTC (Sat) by patrick_g (subscriber, #44470) [Link]

Except AST received a 2.5 Million euros grant from European Union (see this interview). What has been achieved with this money ?

MINIX 3.2.1 released

Posted Feb 23, 2013 19:32 UTC (Sat) by Zack (guest, #37335) [Link]

>What has been achieved with this money ?

Ostensibly, at least the release 3.2.1 of MINIX.

MINIX 3.2.1 released

Posted Mar 12, 2013 22:08 UTC (Tue) by Baylink (guest, #755) [Link]

Well, I thought we got a lot of great political commentary on the last election out of it. :-)

MINIX 3.2.1 released

Posted Feb 23, 2013 14:15 UTC (Sat) by mabshoff (guest, #86444) [Link]

> Well it is possible one might assume my post as trolling. When you look deeper, things are not so bright for this OS as per your post.

I would hardly see how anyone could interpret my reply in this way. I wrote about FreeBSD specifically and I thought made it clear that there is a difference. The Hurd as well as Minix are research systems IMHO. That is usually a nice way of saying that they tend not to go anywhere.

> But what has this minix have really achieved?

Not much TBH, you have for example actual implementations of network drivers that can be restarted upon crash and have its state restored, but I somehow have my doubts that this is really useful in the real world. It is neat, but unless you show that working on a storage driver I am less than impressed.

AST is 68, so I would be surprised if he keeps working on Minix much longer. He has gotten EU funding for resilient OS research, but it all looks like it will not really help the commercialization of Minix since I see no reason to pick it over Linux or a BSD even if that project bears fruit beyond nice demos, because its bad points (hardware support, performance, available talent) outweigh its potential plus points. The same applies to its ARM port which has been talked about for years and most people around here will know that ''it boots'' is quite different than ''it is stable and works well for real workload''. And it does not even boot on any ARM platform yet, much less on a wide variety of ARM platforms.

> I dusted my old P4 to see if this really works, but then I got kernel panic and the screen went blank. Didn't want to try anymore beyond that. Happened the same last year with the previous release.

Yeah, I am not surprised. Today's main problem for toy and research OSes is hardware support, so I see the potential tendency of them running well on hypervisors just like on the IBM 390 decades ago. But once you do that you kind of miss the point of running an alternative OS in the first place, i.e. if you do RT most of the potential goes away if you run it on top of say Xen.

> So practically pointless release with 2 CD wasted.

Maybe for you and most other people, but I am always surprised to meet people hacking on ''pointless'' code. It is really not my prerogative to come to that conclusion since it is not my time.

In the end I agree with you that Minix is pointless, but the brevity of your comment and the lumping in of the BSDs just made it look trollish. Had you written the second comment as your first I would have pretty much agreed with you.

Cheers,

Michael

MINIX 3.2.1 released

Posted Feb 23, 2013 15:47 UTC (Sat) by ibukanov (subscriber, #3942) [Link]

> Today's main problem for toy and research OSes is hardware support, so I see the potential tendency of them running well on hypervisors just like on the IBM 390 decades ago. But once you do that you kind of miss the point of running an alternative OS in the first place,

With IOMMU visualization like Intel's Vt-d one can run a toy OS against a single piece of real hardware like a network card while the rest will be provided by a hypervisor. That can bear very useful results like isolation of increasingly complex network drivers and protocols behind a hardened special-purpose OS. This reduces the attack surface against other software running in the hypervisor.

MINIX 3.2.1 released

Posted Feb 23, 2013 22:56 UTC (Sat) by mabshoff (guest, #86444) [Link]

> With IOMMU visualization like Intel's Vt-d one can run a toy OS against a single piece of real hardware like a network card while the rest will be provided by a hypervisor.

Absolutely, but given that the Hurd does currently not have any USB support (At least it did not have it toward the end of 2012 even though I think that a USB DDEKit is being worked on also by the Minix folks) the IOMMU support for something like Hurd or Minix seems unlikely.

> That can bear very useful results like isolation of increasingly complex network drivers and protocols behind a hardened special-purpose OS. This reduces the attack surface against other software running in the hypervisor.

Yeah, I still think that if you took some of the ideas/goals from the Hurd and tried to implement them on top of the Linux kernel they would have gotten much further along, but then you would have had to compromise. These days there are plenty of userspace driver infrastructure bits in the Linux kernel. I cannot imagine that the theoretical advantage of the Hurd microkernel design will even pay off because most of the interesting bits can likely be done with the Linux kernel and no one should care about the boring driver bits, but the cool stuff.

Cheers,

Michael

MINIX 3.2.1 released

Posted Feb 23, 2013 23:17 UTC (Sat) by ibukanov (subscriber, #3942) [Link]

> IOMMU support for something like Hurd or Minix seems unlikely.

I meant running Minix or other toy/research OS under a hypervisor like XEN or KVM that supports IOMMU so Minix could manage a piece of the real hardware like a network card. Such OS can implement complex network protocols or WIFI drivers isolating the rest of the system from bugs there.

I hope such setups would be more widespread allowing once again small teams or even a single person to try new OS ideas against latest hardware.

MINIX 3.2.1 released

Posted Feb 23, 2013 23:41 UTC (Sat) by mabshoff (guest, #86444) [Link]

> I meant running Minix or other toy/research OS under a hypervisor like XEN or KVM that supports IOMMU so Minix could manage a piece of the real hardware like a network card. Such OS can implement complex network protocols or WIFI drivers isolating the rest of the system from bugs there.

Ok, got your point. That certainly makes sense and if for example you think about VFIO coming from the Cisco folks it does not take much imagination why those folks were motivated to do that work since instead of porting their various routing OSes to various hardware platforms just take Linux with KVM and hand control of the networking hardware to the routing OS. That sidesteps the whole GPL issue and isolates the routing OS from the boring hardware bits.

> I hope such setups would be more widespread allowing once again small teams or even a single person to try new OS ideas against latest hardware.

I think it is already happening. I would be hard pressed to name a OS that does not run on VMWare, i.e. Haiku, the Hurd and Minix all run on top of it. Even OS/2 Warp and later is a supported configuration, but I might have thought about some earlier OS/2 releases which IIRC did some strange things in ring 2, but I am too tired to research it at this time.

I am not sure about the quality of those OSes running on top of say VMWare since I recall strange stability issues with FreeBSD 8.3 on some ESXi targets for example, but that is a different problem. Jump five years ahead and I cannot imagine anything but the various hypervisors being a mandatory target platform for any research OS out there. IIRC last year's linux.conf.au had a session about using Linux as the L4sec boot loader for example for some ARM target. That just sounds like an insane thing to do unless you think about what it would take to write all those drivers for L4sec I assume :p.

Cheers,

Michael

MINIX 3.2.1 released

Posted Feb 23, 2013 17:45 UTC (Sat) by andreasb (subscriber, #80258) [Link]

> The Hurd as well as Minix are research systems IMHO. That is usually a nice way of saying that they tend not to go anywhere.

Minix used to be a simple OS for teaching OS concepts (not researching, AFAIK). It tries to be usable as a general purpose (embedded) OS now, so that's not really an excuse anymore.

The Hurd is GNU's replacement for Unix, not a research project.

Research OSs may not go anywhere in most cases, however that does not make OSs that are not going anywhere research OSs.

MINIX 3.2.1 released

Posted Feb 23, 2013 23:00 UTC (Sat) by mabshoff (guest, #86444) [Link]

> Minix used to be a simple OS for teaching OS concepts (not researching, AFAIK).

True IMHO for Minix before the 3.0 release.

> It tries to be usable as a general purpose (embedded) OS now, so that's not really an excuse anymore.

Yeah, but I would consider the resilience work done via the EU grant mentioned above in a comment puts it into the research OS space. It certainly tries to be embedded, but I think AST is kidding himself if he believes that he can outcompete the BSDs, much less Linux or commercial options like QNX if one desires a pure RTOS.

Cheers,

Michael

MINIX 3.2.1 released

Posted Feb 23, 2013 23:10 UTC (Sat) by mabshoff (guest, #86444) [Link]

Oops, forgot about this one:

> The Hurd is GNU's replacement for Unix, not a research project.

Well, it certainly started out as a intended Unix replacement to complete the GNU ecosystem since a GPLed kernel was the last missing piece. But looking at its history and detours with the attempted replacement of Mach with L4 and Coyotos I think they definitely strayed into the research space. It was probably never intended that way, but things tend to change a bit over 20 years :).

> Research OSs may not go anywhere in most cases, however that does not make OSs that are not going anywhere research OSs.

Yep. I still think it applies to both the current Hurd as well as Minix 3.0 to some extend, but the discussion about what is a research OS and what not is about as decisive as talking about hybrid kernels (see [1]), i.e. the NT as well as OSX kernels are prime example where nebulous claims just cloud up the whole discussion.

Cheers,

Michael

[1] http://en.wikipedia.org/wiki/Hybrid_kernel

MINIX 3.2.1 released

Posted Feb 24, 2013 16:33 UTC (Sun) by keesj (guest, #55221) [Link]

Hi,

I have started working for the MINIX 3 team about a year ago. I have mostly been busy with the ARM port. Having done a lot of Linux work in the past I think I might be able to answer some questions.

The design MINIX 3 allows drivers to crash just like other userland programs under Linux. If you do nothing special at best your driver will be restated. To make transparent restarability a fact
you need to do some additional work like splitting your drivers to split the program state and the driver itself. This splitting has been done for at least network *and* block drivers. There is more work going on to allow hot replacement of components (The Linux analog would be something like ksplice but better).

Next week, at embedded world, we will be giving a restartability demo (running on ARM) of the crash and recovery of a graphics driver. This is quite unique and unseen feature in the Linux world. I think there is a market for MINIX 3 on ARM. The system is small and simple enough for people to tweak and modify to their own needs.

Hope this helps.

As last tip. If you have problems running MINIX 3 try interacting with the community. The only reason I still have CD's it probably because of MINIX :p

MINIX 3.2.1 released

Posted Feb 24, 2013 17:22 UTC (Sun) by Cyberax (✭ supporter ✭, #52523) [Link]

> Next week, at embedded world, we will be giving a restartability demo (running on ARM) of the crash and recovery of a graphics driver. This is quite unique and unseen feature in the Linux world.
Uhm...

Linux GPU drivers have had support for hang detection and reset for _years_.

MINIX 3.2.1 released

Posted Feb 24, 2013 21:02 UTC (Sun) by drag (subscriber, #31333) [Link]

hehe.

It seems that the biggest problem with all of this is that when the driver crashes it puts the hardware into a bad state were recovery is just not going to happen.

That is why I suppose people don't notice that hang detection and reset exists.

MINIX 3.2.1 released

Posted Feb 25, 2013 10:09 UTC (Mon) by adobriyan (guest, #30858) [Link]

> Linux GPU drivers have had support for hang detection and reset for _years_.

Oops inside kernel driver was always either-or event.
If you're lucky, kernel continues to run, no restart needed!

MINIX 3.2.1 released

Posted Feb 25, 2013 13:23 UTC (Mon) by ibukanov (subscriber, #3942) [Link]

> MINIX 3 allows drivers to crash just like other userland programs under Linux

Does MINIX support IOMMU when available to really prevent the driver from affecting the rest of the system?

MINIX 3.2.1 released

Posted Mar 1, 2013 13:45 UTC (Fri) by renox (subscriber, #23785) [Link]

> Does MINIX support IOMMU when available to really prevent the driver from affecting the rest of the system?

I doubt it, a googling showed this "Not assigned yet":
http://wiki.minix3.org/en/StudentProjects/DependabilityIn...

But Genode/NOVA seems to have it:
http://www.osnews.com/story/26819/Genode_13_02_supports_I...

MINIX 3.2.1 released

Posted Mar 2, 2013 11:42 UTC (Sat) by keesj (guest, #55221) [Link]

>> MINIX 3 allows drivers to crash just like other userland programs under Linux
>Does MINIX support IOMMU when available to really prevent the driver from affecting the rest of the system?

No, this is not supported but would certainly fit design goals of MINIX.


Copyright © 2017, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds