Mariusz@fbits.ttank.com (Mariusz Stanczak) (05/12/91)
In article <75230@brunix.UUCP>, cgy@cs.brown.edu (Curtis Yarvin) writes: > In article <110@fbits.ttank.com> Mariusz@fbits.ttank.com (Mariusz Stanczak) writes: > >the differences between Mach kernel, and the future microkernel [I think in > >CTR]) mentioned a 1000 line C code for the whole thing. > > Your decimals need some work. > > We have Mach 3.0 code here. Note that this is NOT the version with BSD > wedged into the kernel. BTW, not intending to nit-pick, but maybe things could be wedged into a kernel, but they are layered around a microkernel... one is rather monolithic, the other modular. > Kernel source alone is about 100,000 lines of code. This does not include > any of the machine-specific sections, which can be quite large; for example, > the i386 machine-specific code is 40,000 lines. I wonder... are we talking about the same thing?! Mach MICROkernel, not kernel. Big difference (though, admitedly, I had no idea of the actual size of the kernel) in fuctionality. Mach kernel includes most of the kernel of the *nix OS as we know it, microkernel isn't much more then task/context switcher, and a process cop, with "everything" else modula- rised over it. Because of the modularity of all the functionality in Mach, one can mix and match what is implemented. A far cry from the monolith bloat of the current kernels. So what does the decimal-point-corrected number of lines of source contain? > The idea is about ten years old. How long have you known about? The idea yes, but the code has been taken out of the labs barely four/five years ago, and even then it contained enough of AT&T code not to be freely distributable. The Mach microkernel is, and it's been so for about a year (if that!). > >and OSF's implementation (with all its resources, and commitment) is a > >couple years away. > > A prerelease version of OSF/1 is running quite stably on a Decstation 3100 > here at Brown. Note that OSF/1 is BSD wedged into the microkernel, not > built around it. Again, are we talking about the same thing... the current OSF/1 release is build around the Mach KERNEL, with late 1992/early 1993 date for the new version that WILL be built around the Mach MICROkernel. > Vaporware is any software or hardware which has been announced but not yet > completed. Ours seem to be a difference of understanding... there's a distinction you overlook, and that's between a preannounced PRODUCT, and an intent to work on "some" (as in "any") software. Do you see Intel's press announcment of a 100 MIPS CICS (some months back, and prompted be the wave of RISC chip releases), and sombody's on the net word of intent to work on some problem as the same? ... I hope not, and the usage of the word "vaporware" has a rather more specific meaning to the former of the two cases. -Mariusz -- INET: Mariusz@fbits.ttank.com CIS : 71601.2430@compuserve.com UUCP: ..!uunet!zardoz!ttank!fbits!Mariusz
Mariusz@fbits.ttank.com (Mariusz Stanczak) (05/12/91)
> to work on "some" (as in "any") software. Do you see Intel's press > announcment of a 100 MIPS CICS (some months back, and prompted be the Maybe I have a problem with placing the decimal mark... Intel predicts a 2,000 MIPS x86 chip -Mariusz -- INET: Mariusz@fbits.ttank.com CIS : 71601.2430@compuserve.com UUCP: ..!uunet!zardoz!ttank!fbits!Mariusz
john@chance.UUCP (John R. MacMillan) (05/13/91)
|> Kernel source alone is about 100,000 lines of code. This does not include |> any of the machine-specific sections, which can be quite large; for example, |> the i386 machine-specific code is 40,000 lines. | |I wonder... are we talking about the same thing?! Mach MICROkernel, not |kernel. Big difference (though, admitedly, I had no idea of the actual |size of the kernel) in fuctionality. Mach kernel includes most of the |kernel of the *nix OS as we know it, microkernel isn't much more then |task/context switcher, and a process cop, with "everything" else modula- |rised over it. Actually all the references to the Mach kernel I have _don't_ include the UNIX emulation stuff, which can be run at the user level, outside the kernel. Most commercial OSes based on Mach put it into the kernel for performance reasons. As for what really is in the Mach kernel, it's a little more than you've mentioned. It includes a rather neat virtual memory management system designed to be readily portable to a wide range of different architectures, including huge multiprocessors, task/thread management, again on multiprocessors, inter-task communication that's transparent over a network, and resource management (memory, processors), again on mulitprocessors. Considering these design goals, it's not so surprising that it's more than a few K big. I remember hearing Rob Pike at Usenix picking on billing Mach as a microkernel, and telling how much of Plan 9 you could get for the same number of lines of source. It included the entire OS, the compiler and related tools, the window system, I think the UNIX compatibility library, and I forget what else.
Mariusz@fbits.ttank.com (Mariusz Stanczak) (05/17/91)
In article <1991May13.014449.15581@chance.UUCP>, john@chance.UUCP (John R. MacMillan) writes: > As for what really is in the Mach kernel, it's a little more than [a list of impressive features deleted] > mulitprocessors. Considering these design goals, it's not so > surprising that it's more than a few K big. Still, (;-)) I'll be thrilled to have a chance to see the Mach MICROkernel, and to think about "what it is", and maybe ^^^^^ emphasis added ("Mach", and "micro" are not synano- mous(sp)) even about "what it would take". [...] (an aside) > I remember hearing Rob Pike at Usenix picking on billing Mach as a > microkernel, and telling how much of Plan 9 you could get for the same > number of lines of source. It included the entire OS, the compiler [a list causing extensive salivation deleted] (and even further aside) hey, here's a truely futuristic desire... one we can have for a safe time to come, BTW. And it should unfold afront of our eyes in a somewhat familiar fashion... news of it first, then educa- tional institutions, then... I don't know, wouldn't once be enough? [...] > library, and I forget what else. And so do I [would like to], no matter how nice it is. As long as the mentality to sell ideas as products persists. (and the whole of the above completely not on subject, which should not matter either way ;-)) -Mariusz -- INET: Mariusz@fbits.ttank.com CIS : 71601.2430@compuserve.com UUCP: ..!uunet!zardoz!ttank!fbits!Mariusz