FreeBSD Status Report March-April 2004

Anonymous Coward writes "The FreeBSD project has posted a new status report for March and April of 2004. Work continues on locking down the network stack, ACPI made more great strides, an ARM port appeared in the tree, and the FreeBSD 4.10 release cycle wrapped up."

Comments

An ARM port eh?

[MrIrwin]

x86 life looks ever more limited!

Re:An ARM port eh?

[Anonymous Coward]

Considering the One-chip SMP Multiprocessor Core, things could get very nice.

Re:An ARM port eh?

[killjoe]

Where is the PPC port? I am amazed that I can't install freebsd on my mac.

Re:An ARM port eh?

[MrIrwin]

I thought MACS allready came with a BSD derived OS?

Re:An ARM port eh?

[killjoe]

Yes but it's not the same. I miss the ports.

Re:An ARM port eh?

[evilviper]

Is there some reason you have to have FreeBSD on your PPC, and you can't possibly use NetBSD/OpenBSD?

They aren't the same as FreeBSD, of course, but the differences are quite small.

Re:An ARM port eh?

[Octorian]

Actually, the differences are quite small between NetBSD and OpenBSD because Open is a fork of Net. Free, however, actually has a very different feel to it. Of course they all have a lot of similarities at the source level, and you'll see NetBSD banners all over FreeBSD kernel source files.

Re:An ARM port eh?

[evilviper]

I've used all three and I see very equal differences between them, even in the userland.

OpenBSD has split from NetBSD, but it was quite a long time ago, and much has changed since then. I find that all three are about equally different from one-another, on the user level, administrative level, and at the source-code level.

A matter of opinion of course...

Interesting note from the SMPng status report..

[harikiri]

From here:

Several folks continue to work on the locking the network stack as noted elsewhere in this report. Outside of the network stack, the following items were worked on during the March and April time frame. Giant was pushed down in the fork, exit, and wait system calls as far as possible. Alan Cox (alc@) continues to lock the VM subsystem and push down Giant where appropriate.

Same Alan Cox of Linux kernel hacking fame? Woot! We've attracted him to the dark side... ;)

Re:Interesting note from the SMPng status report..

[Everlone]

Nope, this is a different Alan Cox.

Something tells me I once saw an FAQ list once that involved this same question but I could be wrong ;-)

Re:Interesting note from the SMPng status report..

[R.Caley]

Nope, this is a different Alan Cox.

Something else reimplemented to avoid the evil GPL?

Re:Interesting note from the SMPng status report..

[apocamok]

It's probably like the guy from Fight Club, he's working on Linux during the day, and unconsciously committing FreeBSD patches during the night.

Re:Interesting note from the SMPng status report..

[coolfruit]

Well, These are two different people. Alan L. Cox is a FreeBSD commiter. Here is his homepage : http://www.cs.rice.edu/~alc/

Re:Interesting note from the SMPng status report..

[Anonymous Coward]

No, different guy.

Re:Interesting note from the SMPng status report..

[craig2787]

1st RULE: You do not talk about FREEBSD PATCHES.
2nd RULE: You DO NOT talk about FREEBSD PATCHES.
3rd RULE: If the code says "stop" or goes coredump, the commit bit is over.
4th RULE: Only two comitters to a patch.
5th RULE: One patch at a time.
6th RULE: No GPL, no adware.
7th RULE: Commits will go on as long as they have to.
8th RULE: If this is your first night at FREEBSD.ORG, you HAVE to PATCH.

Re:Misplaced effort

[shlong]

Most of the bugs in your list are marked closed, and one is for a package that has nothing to do with disks or the OS. That leaves two entries that are relevant. Guessing the geometry is a lot harder than it sounds, especially if you already installed Windows or another boot loader and it guessed the geometry differently (as is the case with at least one of the entries in your list). This is a common problem in Linux, too. Windows is 'immune' to it because it'll choose whatever geometry it wants and leave any previously installed OS's stranded.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Good Work!

[Anonymous Coward]

I love FreeBSD! When one of the linux distros will switch over from the Linux effect and adopt one of the BSD??

Anyone interested in starting a bsd company offering a rock solid desktop OS? :)

Re:Good Work!

[Anonymous Coward]

Debian already has something of the sort, and a lone Gentoo developer is working on it.

Re:Good Work!

[Cochonou]

For Debian, there is Debian GNU/NetBSD

Re:PF and ALTQ

[agent dero]

Flip a coin, it doesn't really matter which you choose:

FreeBSD is something i'd put on a critical nfs/http/ftp server or something

OpenBSD is something I'd put on a Pentium 200Mhz box to keep that nfs/http/ftp box safe.

Nothing prevents you from doing either one with either operating system. It's just about preference ;)

Re:PF and ALTQ

[ArbitraryConstant]

OpenBSD will always have the most up to date PF stuff.

What you'll notice with OpenBSD is that you're discouraged from messing with the kernel at all, and ports work better. Theoretically, you may notice it's slower, and you'll probably notice that the software isn't as up to date. Debian-stable should also be in consideration, depending on your needs, but its firewalling capabilities are well behind FreeBSD and OpenBSD.

You're giving something up if you commit to anything period. FreeBSD and OpenBSD have dramatically disjoint sets of stuff they're good at. I've never seen an OS good enough at everything (or even most things) to make it worth commiting to. Not if you can deal with multiple OSes on a day to day basis.

NetBSD strenght

[CaptainPinko]

The one question I always had is when would someone want to use NetBSD except for when they want to t=use *nix on their toaster?

One true ports system?

[killjoe]

When are freebsd, openbsd, and netbsd adopt the one true ports system? Is there any logical reason to have three different source based ports systems?

Re:One true ports system?

[harikiri]

I think the various ports systems emerged as a result of freebsd only supporting x86 (back in the day), and netbsd having a multi-architecture system (thus more effort was required to 'port' something to each arch, and there were fewer ports). Then OpenBSD came along, and imported in the FreeBSD ports system initially, and went on from there.

The reason why FreeBSD's port system has grown so quickly is probably because there's only been one architecture they had to 'port' applications across to. It would be slowed down if they had to unify the ports system to support not only multi-platform architectures, but also the differences between the kernels for each BSD project.

However, this reminded me of this. NetBSD's package collection actually has released their pkgsrc collection to both FreeBSD and OpenBSD.

Re:One true ports system?

[Strog]

Pkgsrc is available for many OSes. It's most matured on BSD/Linux. It would be cool if several of the BSD's and Linux would use it. Check it out www.pkgsrc.org

NetBSD
OpenBSD
FreeBSD
Linux
Solaris
Irix
Darwin (OS X)

Re:One true ports system?

[killjoe]

"However, this reminded me of this [netbsd.org]. NetBSD's package collection actually has released their pkgsrc collection to both FreeBSD and OpenBSD."

The problem is that pkgsrcs ports collection is not as complete as the freebsd ports collection. It has some extra features but the number of ports is lacking.

If freebsd was to adopt pkgsrc then the development effort would be unified and a more uniform set of ports would be available to everybody.

Re:One true ports system?

[DashEvil]

I can think of one good reason.

On OpenBSD, apache and perl are part of the base system. bzip2 is not.

On FreeBSD, bzip2 is part of the base system. Apache and perl are not.

There are probably more examples.

PowerPC port

[IRLQBall]

Where is the PPC port? I am amazed that I can't install freebsd on my mac.

The current status of the FreeBSD on PowerPC is here

Short version: It's a Tier 2 architecture which means it's not quite there yet. According to the project page it's "on the verge of booting to single-user mode".

Re:PowerPC port

[Anonymous Coward]

OS X is based on Next's Unix which was one of the hybrid Mach/Unix jobs from the early/mid 90s. The only real difference was that Next used BSD instead of a sysV Unix. What happens is OS X is actually a Mach microkernel with several layers on top of it. The BSD networking code and other parts of the kernel were made to run on top of Mach to give it a Unix layer that a lot of other things build on. This unix layer is augmented by a freebsd 5.1 userland (at least that's what I have on my new powerbook).

Re:PowerPC port

[Anonymous Coward]

Actually, if you visit the mailing list archives, they're probably a lot further along than that. The developers working on porting it are a bit lazy in updating their status page.

dead trees!

[MavEtJu]

Book: The Design and Implementation of the FreeBSD Operating System

I know a birthday present for this year!

Re:Dragonfly

[Anonymous Coward]

I fail to understand your logic somewhere.

You are either saying that FreeBSD developers *don't* try to improve things; Or do try to improve things but fail miserably.

I'll assume the second, because if they didn't even try to improve FreeBSD then they wouldn't really be developers. So that is saying that the single really active Dragonfly kernel developer outclasses all the FreeBSD kernel developers put together?

If he doesn't, then I don't see how Dragonfly is going to become "nice". If he does, then I don't see how you would call the loss of those outclassed developers a "serious brain drain".

I don't know why you get off making these wild predictions and then publically stating them as fact when it is just a wild statement based on your personal prejudice and very likely to be wrong. Seriously, 5 years ago when FreeBSD-5 and Linux 2.3 development first opened, you were going around saying how FreeBSD-5 will be more scalable, stable, better-under-load, better networking, faster, etc than Linux, weren't you?

Re:PF and ALTQ

[pkplex]

#include "imo.h"

I think the advantages of FreeBSD are drivers ( for newer toys ), speed, and that jail thing ( which I have not actually used ) which AFAIK lets you run a virtual machine chroot thing. Also, freebsd ( and netbsd ) have automagical update the ports/packages tools and things. On openbsd you need to pkg_delete them yourself.

Other than that, I think OpenBSD is the ticket. Lots of people seem to think OpenBsd is only a firewall OS... which is unfortunate. OpenBSD works fine as a standard server ( eg, web, dns, mail, ftp, samba, etc ). The security effort which goes into obsd is also a deeper than just things disabled by default, too.

Re:PF and ALTQ

[runderwo]

Debian-stable should also be in consideration, depending on your needs, but its firewalling capabilities are well behind FreeBSD and OpenBSD.

Nice troll. How about an actual comparison?

Why is this at Score 5?

[Anonymous Coward]

Seriously?

Re:PF and ALTQ

[Octorian]

Personally, I see the different BSDs as all good OSes, but for different purposes...

FreeBSD - Good general purpose server OS, and my 'nix of choice on x86

OpenBSD - Good firewall/network-device OS, runs wonderfully on good hardware (like old SPARCs) Though it can often be behind the ball in places you'd least expect it until you run into them head-on, out of nowhere. (Like when I tried the sparc64 port on my Netra, and it ran slowly, and didn't like more than 2 hme network interfaces)

NetBSD - Tinkerer's OS of choice. Runs on anything, and is easy to manually install. Perfect for figuring out how to net-boot a VAXstation, SPARC, or simply to put that Sun3 through its paces. (and anyone who thinks you can't have gcc and X in a 200MB full system, has never used NetBSD)

Re:PF and ALTQ

[ArbitraryConstant]

Firewall Failover

"In test environments, we have run up to 4 pfsync+carp hosts (all different architectures: i386, sparc, sparc64, and amd64!), randomly rebooting them. TCP sessions were not interrupted through over two days of such torture testing."

Linux has UCARP, but has no way for the stateful firewall to do transparent failover.

Don't really feel like researching enough detail for a comprehensive summary of the other stuff.