Is Minix3 useful yet? - Minix

This is a discussion on Is Minix3 useful yet? - Minix ; Hi... I've been using FreeBSD for some 6 years now, although not to the exclusion of some sort of winblows box. Recently there was some chat in the BSD community on the state of the health of some BSD monolithic ...

+ Reply to Thread
Results 1 to 13 of 13

Thread: Is Minix3 useful yet?

  1. Is Minix3 useful yet?

    Hi...

    I've been using FreeBSD for some 6 years now, although not to the
    exclusion of some sort of winblows box. Recently there was some chat in
    the BSD community on the state of the health of some BSD monolithic
    kernels. That got me thinking about the monolithic-vs-micro kernel
    issue. So I'm lurking on this newsgroup and have poured through the
    Minix website, in an effort to acquaint myself with Minix.

    Question: Is Minix3 mature and stable enough enough to be used as an
    X-Windows workstation at this very moment?

    Question: Could I use Minix3 to run Apache/mysql/perl?

    TIA....
    --
    duke

  2. Re: Is Minix3 useful yet?


    > Question: Is Minix3 mature and stable enough enough to be used as an
    > X-Windows workstation at this very moment?

    Well, I think that the choices of what DE you can have is rather
    limited (there is, to my knowledge, only wmii and one other available
    for Minix).

    Further, the hard disk driver <-> File System interface is limiting the
    partition of the hard drive; for a 4 KB block, only a 4 GB partition is
    supported. I am not sure if this has been fixed in the experimental
    versions of Minix, but if you were bored enough you could solve the
    problem.

    > Question: Could I use Minix3 to run Apache/mysql/perl?


    Good question. I think that if you can get the GCC to work properly on
    Minix (there isn't virtual memory support so you have to assign the RAM
    manually, etc.) then you could theoretically compile Apache/MySQL/Perl
    source codes then have them run on Minix...IN THEORY (that's my
    disclaimer!).

    Although I am fairly certain there is a port of Perl to Minix, don't
    take my word for it. It may be in its beta stage.

    But for what you are looking for, a workstation microkernel, I would
    suggest L4Linux (http://l4linux.org/) or k42
    (http://www.research.ibm.com/K42/); Minix is, in my humble opinion
    (which I am guarenteed to get flamed for), a hobbyist's and a hacker's
    OS. It's done enough so that you can start working on it, and tinker
    around with the pieces in isolation without fear of destroying the
    hardware or crashing the system...but not done enough to be a fully
    functional, "runs-on-every machine" type system.


  3. Re: Is Minix3 useful yet?

    On 2006-10-14, Pablo Rodriguez wrote:
    >
    >> Question: Is Minix3 mature and stable enough enough to be used as an
    >> X-Windows workstation at this very moment?

    > Well, I think that the choices of what DE you can have is rather
    > limited (there is, to my knowledge, only wmii and one other available
    > for Minix).


    I'm not picky -- i'd even settle for some good tools using CLI.

    [snip]

    >> Question: Could I use Minix3 to run Apache/mysql/perl?

    >
    > Good question. I think that if you can get the GCC to work properly on
    > Minix (there isn't virtual memory support so you have to assign the RAM
    > manually, etc.) then you could theoretically compile Apache/MySQL/Perl
    > source codes then have them run on Minix...IN THEORY (that's my
    > disclaimer!).


    You mean `gcc' the compiler -- *it* doesn't even work? I'm pooched ....

    > Although I am fairly certain there is a port of Perl to Minix, don't
    > take my word for it. It may be in its beta stage.


    I'll have to look hard...

    > But for what you are looking for, a workstation microkernel, I would
    > suggest L4Linux (http://l4linux.org/) or k42
    > (http://www.research.ibm.com/K42/);


    I'll have to go snoop...

    >Minix is, in my humble opinion
    > (which I am guarenteed to get flamed for), a hobbyist's and a hacker's
    > OS. It's done enough so that you can start working on it, and tinker
    > around with the pieces in isolation without fear of destroying the
    > hardware or crashing the system...but not done enough to be a fully
    > functional, "runs-on-every machine" type system.


    Nothing wrong with a kernel being a work-in-progress, or should I say a
    labor-of-love in progress. I can hack Perl but not C(++) -- yet. There may
    be hope for me yet. Thanks for your input.
    --
    duke


  4. Re: Is Minix3 useful yet?


    Merrilee Larson wrote:
    > >> Question: Could I use Minix3 to run Apache/mysql/perl?

    > >
    > > Good question. I think that if you can get the GCC to work properly on
    > > Minix (there isn't virtual memory support so you have to assign the RAM
    > > manually, etc.) then you could theoretically compile Apache/MySQL/Perl
    > > source codes then have them run on Minix...IN THEORY (that's my
    > > disclaimer!).

    >
    > You mean `gcc' the compiler -- *it* doesn't even work? I'm pooched ....
    >


    No, I just haven't figured out how to use it. There are instructions
    online at the minixtips.com blog; I just have too much work to do it
    right now

    Here are some relevant pages:
    When configure fails on some GNU Tools:
    http://www.minixtips.com/2006/06/con...ling-some.html

    Minix Tips homepage: http://www.minixtips.com/

    > > Although I am fairly certain there is a port of Perl to Minix, don't
    > > take my word for it. It may be in its beta stage.

    >
    > I'll have to look hard...


    I'm sorry, I made a mistake, perl-5.8.7 is supported. And both the
    apache portable runtime and the apache portable runtime utils are in
    the beta stages.

    With MySQL, you'd have to port it yourself. Though I am not sure how
    much space you would have on your computer (your 4 GB partition) when
    you have all of these things ported...I've never used Apache/MySQL/Perl
    nor been involved with server stuff.

    You can tell what software is and is not supported here:
    http://www.minix3.org/software/

    > > But for what you are looking for, a workstation microkernel, I would
    > > suggest L4Linux (http://l4linux.org/) or k42
    > > (http://www.research.ibm.com/K42/);

    >
    > I'll have to go snoop...


    k42 is great for multiple processor machines, I'm not sure if it has
    been tested for single core machines. L4Linux is a great alternative if
    you are lazy like me

    > >Minix is, in my humble opinion
    > > (which I am guarenteed to get flamed for), a hobbyist's and a hacker's
    > > OS. It's done enough so that you can start working on it, and tinker
    > > around with the pieces in isolation without fear of destroying the
    > > hardware or crashing the system...but not done enough to be a fully
    > > functional, "runs-on-every machine" type system.

    >
    > Nothing wrong with a kernel being a work-in-progress, or should I say a
    > labor-of-love in progress. I can hack Perl but not C(++) -- yet. There may
    > be hope for me yet. Thanks for your input.


    Don't take my word for it, as there are undoubtedly people who will
    violently disagree with me

    In my honest opinion, Minix will be a lot better when it finishes the
    work being done on virtual memory and expanding the maximum partition
    size beyond 4 GB...or you could simply increase the block size during
    installation


  5. Re: Is Minix3 useful yet?

    On 2006-10-15, Pablo Rodriguez wrote:
    >
    > Merrilee Larson wrote:
    >> >> Question: Could I use Minix3 to run Apache/mysql/perl?
    >> >
    >> > Good question. I think that if you can get the GCC to work properly on
    >> > Minix (there isn't virtual memory support so you have to assign the RAM
    >> > manually, etc.) then you could theoretically compile Apache/MySQL/Perl
    >> > source codes then have them run on Minix...IN THEORY (that's my
    >> > disclaimer!).

    >>
    >> You mean `gcc' the compiler -- *it* doesn't even work? I'm pooched ....
    >>

    >
    > No, I just haven't figured out how to use it. There are instructions
    > online at the minixtips.com blog; I just have too much work to do it
    > right now


    The extent of my knowledge of gcc is: make / make install /make clean

    >
    > Here are some relevant pages:
    > When configure fails on some GNU Tools:
    > http://www.minixtips.com/2006/06/con...ling-some.html
    >
    > Minix Tips homepage: http://www.minixtips.com/


    I'll read them for sure.


    > I'm sorry, I made a mistake, perl-5.8.7 is supported. And both the
    > apache portable runtime and the apache portable runtime utils are in
    > the beta stages.


    2 out of three --- not bad!

    > You can tell what software is and is not supported here:
    > http://www.minix3.org/software/


    More reading for me ...

    >
    >> > But for what you are looking for, a workstation microkernel, I would
    >> > suggest L4Linux (http://l4linux.org/) or k42
    >> > (http://www.research.ibm.com/K42/);

    >>
    >> I'll have to go snoop...

    >
    > k42 is great for multiple processor machines, I'm not sure if it has
    > been tested for single core machines. L4Linux is a great alternative if
    > you are lazy like me


    So exactly what is L4Linux? I visited both the above sites -- and I left not
    quite sure what I had been reading. Is L4Linux a standalone microkernel O/S,
    which is capable of running Linux binaries? Or does it install itself on top
    of an existing Linux installation? I'm a bit confused.

    > In my honest opinion, Minix will be a lot better when it finishes the
    > work being done on virtual memory and expanding the maximum partition
    > size beyond 4 GB...or you could simply increase the block size during
    > installation


    Minix does sound promising, all right! What we need to do, is to convince all
    the core developers for free/Net/Open BSD, to drop what there doing and
    volunteer for a month or so on Minix. and if we could get the Linux crowd
    involved....

    I guess I should get busy learning C so I too can help.
    --
    duke

  6. Re: Is Minix3 useful yet?


    Merrilee Larson wrote:
    > So exactly what is L4Linux? I visited both the above sites -- and I left not
    > quite sure what I had been reading. Is L4Linux a standalone microkernel O/S,
    > which is capable of running Linux binaries? Or does it install itself on top
    > of an existing Linux installation? I'm a bit confused.


    Well, you know how Mac OS X's Darwin is essentially BSD running on the
    Mach Kernel? Well, L4Linux is basically Linux running on the L4 kernel
    (similiar idear of monolithic OS running on microkernel).

    It may be promising, it may be useless...I don't know! I think that it
    is a good microkernel for those that don't want to do a lot to make it
    work.


  7. Re: Is Minix3 useful yet?

    Merrilee Larson wrote:
    >
    > Hi...
    >
    > I've been using FreeBSD for some 6 years now, although not to the
    > exclusion of some sort of winblows box. Recently there was some chat in
    > the BSD community on the state of the health of some BSD monolithic
    > kernels. That got me thinking about the monolithic-vs-micro kernel
    > issue. So I'm lurking on this newsgroup and have poured through the
    > Minix website, in an effort to acquaint myself with Minix.
    >
    > Question: Is Minix3 mature and stable enough enough to be used as an
    > X-Windows workstation at this very moment?


    Actually yes but Minix3 as any other Minix as well (but some forks)
    does not support swapping.

    >
    > Question: Could I use Minix3 to run Apache/mysql/perl?
    >


    Apache compiles even under Windows,
    I did not attempt to do that, but I hope it must be compiled.

    The beauty of Minix 3 architecture it may be extended without kernel rebuild,
    only you need to obtain new functionality build respective driver,
    and/or respective library.

    However, some kinds of functionality may not be reached in such way.

    There are such kinds of computer users:
    (1) those who may not configure anything but may use certain computer
    applications.
    (2) those who may configure computer solve collisions.
    (3) those who may develop software.
    (4) those who design new types of software and new ways of computer usage.

    Most of users are kind (1).
    Minix seemed to be found as OS for educating kinds of (3) and (4)
    more rarely (2). However with project 100$ computer,
    it will be addressed for educating also (1) users.

    I'm all four kinds user, that's why Fedora Linux is ugly for me,
    even with its flexibility with some special rarely used API.

    Minix3 is a new beginning with little abilities,
    but promises to be extendable.

    --Quas.co.ua


    > TIA....
    > --
    > duke




  8. Re: Is Minix3 useful yet?

    Pablo Rodriguez wrote:
    >
    > > Question: Is Minix3 mature and stable enough enough to be used as an
    > > X-Windows workstation at this very moment?

    > Well, I think that the choices of what DE you can have is rather
    > limited (there is, to my knowledge, only wmii and one other available
    > for Minix).
    >
    > Further, the hard disk driver <-> File System interface is limiting the
    > partition of the hard drive; for a 4 KB block, only a 4 GB partition is
    > supported. I am not sure if this has been fixed in the experimental
    > versions of Minix, but if you were bored enough you could solve the
    > problem.
    >
    > > Question: Could I use Minix3 to run Apache/mysql/perl?

    >
    > Good question. I think that if you can get the GCC to work properly on
    > Minix (there isn't virtual memory support so you have to assign the RAM
    > manually, etc.) then you could theoretically compile Apache/MySQL/Perl
    > source codes then have them run on Minix...IN THEORY (that's my
    > disclaimer!).


    Look, Apache works such: first it is loaded in memory,
    after when you connect it from user agent
    (no difference which one, even telnet on port 80)
    it makes fork() and a respective child process serves your request,
    opens and loads files, gets memory, runs interpreters
    (external or built-in), and after terminates with all reserved memory,
    and descriptors of opened files when request is complete
    or when the time out for request is occurred.

    Such schema itself does not need swapping,
    but when in some rare cases when you need memory occupied by apache.

    >
    > Although I am fairly certain there is a port of Perl to Minix, don't
    > take my word for it. It may be in its beta stage.
    >
    > But for what you are looking for, a workstation microkernel, I would
    > suggest L4Linux (http://l4linux.org/) or k42
    > (http://www.research.ibm.com/K42/); Minix is, in my humble opinion
    > (which I am guarenteed to get flamed for), a hobbyist's and a hacker's
    > OS. It's done enough so that you can start working on it, and tinker
    > around with the pieces in isolation without fear of destroying the
    > hardware or crashing the system...but not done enough to be a fully
    > functional, "runs-on-every machine" type system.



  9. Re: Is Minix3 useful yet?

    Merrilee Larson wrote:
    > Hi...
    >
    > I've been using FreeBSD for some 6 years now, although not to the
    > exclusion of some sort of winblows box. Recently there was some chat in
    > the BSD community on the state of the health of some BSD monolithic
    > kernels. That got me thinking about the monolithic-vs-micro kernel
    > issue. So I'm lurking on this newsgroup and have poured through the
    > Minix website, in an effort to acquaint myself with Minix.
    >
    > Question: Is Minix3 mature and stable enough enough to be used as an
    > X-Windows workstation at this very moment?
    >
    > Question: Could I use Minix3 to run Apache/mysql/perl?

    If you built Apache as a static binary with the MySQL and Perl modules
    linked statically, it is possible, but you'd end up with a 40MB apache
    binary. Also, you can't have a threaded httpd, either.

    This is all because of the lack of pthreads and shared libraries. I have
    been looking into the feasibility of using ELF shared libraries, but it
    doesn't seem like an easy overnight task.
    >
    > TIA....
    > --
    > duke




    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.1 (GNU/Linux)
    Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

    iD8DBQFFMrIdsjeOFtd+nycRAhqyAJ4vNXLZBRjOPQQKfb8hNr bozgBrrgCfY9I6
    1/2LDdEcS7C/AlvLhE9uibs=
    =9CmX
    -----END PGP SIGNATURE-----


  10. Re: Is Minix3 useful yet?

    On 2006-10-15, Pablo Rodriguez wrote:
    >
    > Merrilee Larson wrote:
    >> So exactly what is L4Linux? I visited both the above sites -- and I left not
    >> quite sure what I had been reading. Is L4Linux a standalone microkernel O/S,
    >> which is capable of running Linux binaries? Or does it install itself on top
    >> of an existing Linux installation? I'm a bit confused.

    >
    > Well, you know how Mac OS X's Darwin is essentially BSD running on the
    > Mach Kernel? Well, L4Linux is basically Linux running on the L4 kernel
    > (similiar idear of monolithic OS running on microkernel).
    >
    > It may be promising, it may be useless...I don't know! I think that it
    > is a good microkernel for those that don't want to do a lot to make it
    > work.


    OK! So, if Linux is running on the L4 kernel, do you automagically loose the
    negative aspects of a monolithic kernel? Or is this merely a work-around hack
    for the time being -- sort of thing. It seems a little "hybrid" to me.

    For sure I'm waaaay over my head in this newsgroup. You guys are kernel
    hackers. I'm a 6-7 year old newbie to Unix -- but I'm curious, and open to
    suggestions. What I'd really want to see is for Unix to take its rightful
    share of the desktop PCs around the world -- be it with a monolithic or
    microkernel. It just seems to me that using a microkernel as a foundation is
    the way to go.

    So to all that have replied -- thank you for your input, but I'm afraid that
    I'm out of my league here. I'll continue lurking and reading.
    --
    duke

  11. Re: Is Minix3 useful yet?


    Merrilee Larson wrote:
    > On 2006-10-15, Pablo Rodriguez wrote:
    > >
    > > Merrilee Larson wrote:
    > >> So exactly what is L4Linux? I visited both the above sites -- and I left not
    > >> quite sure what I had been reading. Is L4Linux a standalone microkernel O/S,
    > >> which is capable of running Linux binaries? Or does it install itself on top
    > >> of an existing Linux installation? I'm a bit confused.

    > >
    > > Well, you know how Mac OS X's Darwin is essentially BSD running on the
    > > Mach Kernel? Well, L4Linux is basically Linux running on the L4 kernel
    > > (similiar idear of monolithic OS running on microkernel).
    > >
    > > It may be promising, it may be useless...I don't know! I think that it
    > > is a good microkernel for those that don't want to do a lot to make it
    > > work.

    >
    > OK! So, if Linux is running on the L4 kernel, do you automagically loose the
    > negative aspects of a monolithic kernel? Or is this merely a work-around hack
    > for the time being -- sort of thing. It seems a little "hybrid" to me.


    Well, yes and no. Yes in the sense that if Linux crashes, the system is
    still fine.

    No in the sense that it's really more of a hybrid kernel, in my honest
    opinion. My fuzzy reasoning behind this is such: (Monolithic Kernel) +
    (Microkernel) = Hybrid Kernel. Others are more likely than not to
    disagree with my fuzzy reasoning

    > For sure I'm waaaay over my head in this newsgroup. You guys are kernel
    > hackers.


    Most people who contribute to Minix are (in my opinion) hackers
    (disclaimer: I have not contributed to minix, and therefore am not a
    hacker).

    But same for anyone who contributes to any OS.

    > I'm a 6-7 year old newbie to Unix -- but I'm curious, and open to
    > suggestions. What I'd really want to see is for Unix to take its rightful
    > share of the desktop PCs around the world -- be it with a monolithic or
    > microkernel. It just seems to me that using a microkernel as a foundation is
    > the way to go.


    Well don't be afraid of nasty algorithms

    Monolithic kernels are easier to program, but more prone to fail; as
    opposed to microkernels, which are harder to program yet also less
    prone to fail. I'm not sure if you know how a microkernel works, but it
    takes system calls and turns them into messages. It has a stack of
    messages which are then passed off on to the drivers/servers/whatever
    and invokes something. Programmig that is a lot harder than it sounds.

    I think that as soon as some Microkernel becomes really user friendly,
    and doesn't crash (ever), Micro-unix will go like hotcakes. But that
    will be way down the line, and that supposes that the information
    necessary to make drivers is still freely available...nVidia and a few
    others are not giving such information out freely

    There are ways around that however (where there is a will, there is a
    way).

    > So to all that have replied -- thank you for your input, but I'm afraid that
    > I'm out of my league here. I'll continue lurking and reading.


    Well, just to let you know, I'm more out of my league than you. I took
    one course in programming, and it was Java programming with a two week
    historical focus on C/C++ programming.

    The only knowledge that I have from the Minix kernel's inner workings
    is through the book "Operating Systems: Design and Implementation" by
    Andrew S. Tanenbaum and Andrew Woodhall; so I am not pretending that I
    know how the entire system works. I know how the filesystem works, and
    I know a little bit about the kernel's workings; all you really have to
    do is download and study the source code failing that, reading the
    book and studying the source code.

    (Here is the source code from the appendix of Operating Systems Design
    and Implementation:
    http://www.minix3.org/download/minix..._3_1_1.tar.bz2
    and here is the complete source code for Minix 3:
    http://www.minix3.org/download/minix...mplete.tar.bz2
    I would however suggest studying OSDI religiously, as well as the
    source code if you would like to contribute something; that's what I'm
    doing )


  12. Re: Is Minix3 useful yet?

    On 2006-10-16, Pablo Rodriguez wrote:
    >
    >
    > I think that as soon as some Microkernel becomes really user friendly,
    > and doesn't crash (ever), Micro-unix will go like hotcakes. But that
    > will be way down the line, and that supposes that the information
    > necessary to make drivers is still freely available...nVidia and a few
    > others are not giving such information out freely


    So there's hope for the future -- of Unix that is! Presently, I try to convince
    all I come in contact with to support Mac OS boxes over that Winblows stuff.
    I love the new Mac commercials! For me , it;s not about which flavor of
    Unix is better -- it all about doing away with a system that perpetuates
    the notion that it's normal to reboot if something goes wrong.

    In your opinion, what would it take to move the Minix/et al Project(s)
    along? I suppose manpower (hackers) are in short supply! Is there any way
    that grunt work could be farmed out to willing spirits like myself, in
    an effort to free up hacker time and resources?

    >
    > The only knowledge that I have from the Minix kernel's inner workings
    > is through the book "Operating Systems: Design and Implementation" by
    > Andrew S. Tanenbaum and Andrew Woodhall; so I am not pretending that I
    > know how the entire system works. I know how the filesystem works, and
    > I know a little bit about the kernel's workings; all you really have to
    > do is download and study the source code failing that, reading the
    > book and studying the source code.
    >
    > (Here is the source code from the appendix of Operating Systems Design
    > and Implementation:
    > http://www.minix3.org/download/minix..._3_1_1.tar.bz2
    > and here is the complete source code for Minix 3:
    > http://www.minix3.org/download/minix...mplete.tar.bz2
    > I would however suggest studying OSDI religiously, as well as the
    > source code if you would like to contribute something; that's what I'm
    > doing )
    >


    Thanks for the references. I'll follow up on them!
    --
    duke

  13. Re: Is Minix3 useful yet?


    Merrilee Larson wrote:

    > In your opinion, what would it take to move the Minix/et al Project(s)
    > along? I suppose manpower (hackers) are in short supply! Is there any way
    > that grunt work could be farmed out to willing spirits like myself, in
    > an effort to free up hacker time and resources?


    Well, to be honest, I don't really know what is in dire need of help
    right now. I know that probably the following need serious help:
    * Virtual Memory
    * Kernel Threading and the ability to handle multiple threads
    * Fixing the memory address pointer from 32 bits to 64 bits (hackers
    far more experienced than me inform me that this will allow larger
    partitions to be handled, that is why I mentioned: supporting MUCH
    BIGGER partitions than 4 Gigs while using 4 kb block size)
    * Porting Packages
    * Writing Drivers

    That's just a check list of my personal wishes of Minix, THOUGH you may
    want to ask someone in charge of a project (see the who's in charge of
    what page: http://www.minix3.org/who_doing_what.html). As you can tell,
    the stuff that needs to be done is quite a bit. Minix 3 is a great
    operating system, don't get me wrong, but there is plenty of work to be
    done.


+ Reply to Thread