I have to select one of the micro-kernels for further development. I need answers to following questions to go forward:
- What are the limitations/advantages of Fiasco, L4Ka: Pistachio and OKL4. Which one should I select? - Is Fiasco still alive? When I visit Fiasco http://os.inf.tu-dresden.de/fiasco/ site, last updated date is 26 Sep 2005 *Fiasco 1.2 released!* Do we have further releases or is it stopped? - Does anyone have performance figures of Fiasco, L4Ka: Pistachio and OKL4, based on which i can decide which micro-kernel to go for? - Does L4Linux work on OkL4, I read L4Linux runs of L4Ka: Pistachio, will it run as it is or we need to make some changes to it?
Thanks Sarita
Hi,
sarita kale wrote:
I have to select one of the micro-kernels for further development. I need answers to following questions to go forward:
- What are the limitations/advantages of Fiasco, L4Ka: Pistachio and OKL4.
In a nutshell, OKL4 is a commercially suppored derivative of L4Ka:Pistachio, Fiasco and Pistachio are both research kernels. For application developers the feature sets of these kernels and of the Nova Microhypervisor differ only at a very detailled level. The low-level OS layers (e.g., L4RE for Fiasco) typically abstract from most kernel details. This leads to the answer of your next question:
Which one should I select?
It depends on what you want to do with this kernel.
As a small guidance: - Real-Time: Fiasco / Jan's version of Pistachio - Virtualization: Fiasco, Pistachio, Nova, (OKL4?) - Verification: seL4 - Product: OKL4 or P4 (from sysgo)
- Is Fiasco still alive? When I visit Fiasco
http://os.inf.tu-dresden.de/fiasco/ site, last updated date is 26 Sep 2005 *Fiasco 1.2 released!* Do we have further releases or is it stopped?
Yes, although the webpage has become a little outdated. We are currently preparing a release of our new Fiasco version which will also include capabilities, kernel memory management and full virtualization support.
- Does anyone have performance figures of Fiasco, L4Ka: Pistachio and
OKL4, based on which i can decide which micro-kernel to go for?
Since you are asking about virtualization below, please have a look at: http://os.inf.tu-dresden.de/papers_ps/schildlackorwarg09_rtlws_faithful_virt...
- Does L4Linux work on OkL4, I read L4Linux runs of L4Ka: Pistachio,
will it run as it is or we need to make some changes to it?
You didn't ask that but L4Linux runs on L4RE on Fiasco
Best regards
Marcus
On 100129 at 19:45, Marcus Voelp wrote:
I have to select one of the micro-kernels for further development. I need answers to following questions to go forward:
- What are the limitations/advantages of Fiasco, L4Ka: Pistachio and OKL4.
In a nutshell, OKL4 is a commercially suppored derivative of L4Ka:Pistachio, Fiasco and Pistachio are both research kernels. For application developers the feature sets of these kernels and of the Nova Microhypervisor differ only at a very detailled level. The low-level OS layers (e.g., L4RE for Fiasco) typically abstract from most kernel details. This leads to the answer of your next question:
Which one should I select?
It depends on what you want to do with this kernel.
As a small guidance:
- Real-Time: Fiasco / Jan's version of Pistachio
- Virtualization: Fiasco, Pistachio, Nova, (OKL4?)
- Verification: seL4
- Product: OKL4 or P4 (from sysgo)
It might also be worth noting that TUDOS ships quite a few tools, libraries and frameworks to get you started. Genode also provides some infrastructure for OKL4 and Pistachio, but in general it appears you start mostly from scratch when developing for those others.
- Is Fiasco still alive? When I visit Fiasco
http://os.inf.tu-dresden.de/fiasco/ site, last updated date is 26 Sep 2005 *Fiasco 1.2 released!* Do we have further releases or is it stopped?
Yes, although the webpage has become a little outdated. We are currently preparing a release of our new Fiasco version which will also include capabilities, kernel memory management and full virtualization support.
That is GREAT news! Are we talking weeks or months?
/steffen
- Is Fiasco still alive? When I visit Fiasco
http://os.inf.tu-dresden.de/fiasco/ site, last updated date is 26 Sep 2005 *Fiasco 1.2 released!* Do we have further releases or is it stopped?
Yes, although the webpage has become a little outdated. We are currently preparing a release of our new Fiasco version which will also include capabilities, kernel memory management and full virtualization support.
To me, capabilities are nothing but a buzzword that a tiny fraction of a minority of the academic computing community have latched on to, fetishized, and have turned into a cult religion. Now they are starting to mercilessly ram it down the throats of every kernel project that isn't already cast in stone such as unix.
To me, any security mechanism is nothing more than something I have to hack around to make the computer do anything useful at all. =( That's why I still love DOS, it does **ANYTHING** you ask it to.
I've tried to read articles about capabilities. They were just jibberish to me when read straight through. When I crossed out all instances of the word "capabilitiy", they document made sense but seemed hopelessly pointless, as in the concept served no conceivable >> net << benefit to anyone except the NSA.
In the real world, people tend to ignore security issues, even to the point of intentionally using stupid passwords for the sake of making sure they can still get things done and to minimize the risk of forgetting the password, which could be catastrophic to their business!!! (I work at such a company, the password to the most heavily used account on their main database is an easily memorized variation of the username. =P, the root password is equally stupid.)
At my previous place of work, a geophysical laboratory, the old guy there insisted that when I upgraded a windows 98 machine to 2000 (in 2007), that I make sure I never entered a password so that the machine would always boot to the desktop. If I failed to do this, I had to install it again. Once again, the key motivation was to bypass security to *make it possible to use the computer*.
The root and user passwords for my own home PC and the computers on my network are exceptionally weak too because I need to be able to get into them.
They know that security is not a benefit, it's a potential obstacle for them staying in business!
If I were dictator, I would banish these capabilities nutcases from academic institutions for five years and give them guaranteed employment as an IT pro or a programmer for a small company and give them an education in how hard security makes life in general.
Alan Grimes agrimes@speakeasy.net writes:
To me, capabilities are nothing but a buzzword that a tiny fraction of a minority of the academic computing community have latched on to, fetishized, and have turned into a cult religion. Now they are starting to mercilessly ram it down the throats of every kernel project that isn't already cast in stone such as unix.
To me, any security mechanism is nothing more than something I have to hack around to make the computer do anything useful at all. =( That's why I still love DOS, it does **ANYTHING** you ask it to.
+----------+ | PLEASE | | DO NOT | | FEED THE | | TROLLS | +----------+ | | | | .|.||/..
Hello Sarita,
I have to select one of the micro-kernels for further development. I need answers to following questions to go forward:
- What are the limitations/advantages of Fiasco, L4Ka: Pistachio and
OKL4. Which one should I select?
- Is Fiasco still alive? When I visit Fiasco
http://os.inf.tu-dresden.de/fiasco/ site, last updated date is 26 Sep 2005 *Fiasco 1.2 released!* Do we have further releases or is it stopped?
- Does anyone have performance figures of Fiasco, L4Ka: Pistachio and
OKL4, based on which i can decide which micro-kernel to go for?
- Does L4Linux work on OkL4, I read L4Linux runs of L4Ka: Pistachio,
will it run as it is or we need to make some changes to it?
if you are interested in L4 but don't want to make your development depend on one particular kernel implementation, the Genode OS Framework may be of help. This is a user-land infrastructure that works on all those kernels and also on Linux. It enables you to develop a microkernel application once and port it to another kernel by a simple recompile. The project website is:
The project is actively maintained and plans to broaden the number of supported base platforms even beyond the already supported kernels. To see what is possible now, you can take a look at the release notes from the last version released in November 2009 or try out the last live CD released in September:
http://genode.org/documentation/release-notes/9.11 http://genode.org/download/live-cds
Regards Norman
l4-hackers@os.inf.tu-dresden.de