New Scheduler Available for FreeBSD

flynn_nrg writes "Luigi Rizzo, one of the FreeBSD developers, has just finished the code for a new scheduler. From the announcement: '...as promised, a first version of the Proportional Share scheduler that we developed is available here. These are for a recent -STABLE (i think any version from 4.4 should work; the only 3 files modified are kern_synch.c, kern_switch.c and proc.h, plus a one-line change to kern_exit.c). I have tested it a little bit on a diskless system, and it seems to survive running a full X session with the usual set of xterm, netscape etc. while i do a "renice" of the processes and even switch back and forth between schedulers. But do not trust this yet for a production system!' Read the full post here."

Comments

Why?

[Dratman]

What is the purported advantage of the new scheduler?

Re:Why?

[Anonymous Coward]

"There are compelling reasons to use proportional share scheduling techniques to support multimedia and other soft real-time applications on general-purpose operating systems. First, proportional share (PS) schedulers are a good match for existing infrastructure such as a periodic timer interrupt and mechanisms for assigning priorities to applications -- priorities can be mapped to shares in a proportional-share environment. Second, PS schedulers provide stronger guarantees to applications than do traditional time-sharing schedulers: they allocate a specific fraction of the CPU to each thread, and some schedulers provide error bounds on the allocation rate. Third, PS schedulers have clear semantics during underload: excess CPU time is allocated fairly, in contrast with some reservation-based schedulers that must idle or back off to a secondary scheduling policy once all application budgets are exhausted."

Re:Why?

[Waffle Iron]

What is the purported advantage of the new scheduler?

Doesn't anybody read the submitters' article summaries any more? It says right at the top of this page that the new scheduler is Proportional. That's the advantage. We can therefore infer that the old one is inferior for a related reason: it's not Proportional.

I'm all for good proportions. Kudos to the implementors. Keep up the good work.

Re:Why?

[Dratman]

I was trying to hint, in a (perhaps excessively) subtle way, that the original posting did not offer any easily decipherable clues as to the purpose or strengths of the new scheduler. In other words, I was expressing the opinion that the posting conveyed very little information.

I guess the first moderator picked this up, while the one who called my post "overrated" did not.

This moderation up and then down a point made me laugh quite a bit. I had been faintly praised, then unceremoniously knocked off my molehill-sized pedestal.

Re:Why?

[Dratman]

That's why I found it so funny. Proportionately.

Re:Why?

[Art Deco]

FreeBSD is dead for people who need a bandwagon to jump onto. FreeBSD is dead for people who want a political or trendy OS. For people who simply want the best server OS available for PC hardware FreeBSD lives.

uniprocessor only.

[Anonymous Coward]

what the hell happenned to developing SMP and UP capable schedulers ? or does BSD expect to run well on only UP systems ?

Re:uniprocessor only.

[m0rten]

The scheduler is far from finished, and would likely need a lot more work. Integrating it into 5.0-CURRENT (and bringing in the SMP support along with it) is already under way, but could probably take some time.

Okay, now please tell me...

[RinkSpringer]

Why does this get on the main page, and why does the new Core Team does not?

Either way, FreeBSD does fit the new trend now... more VM's means more freedom :)

Re:Okay, now please tell me...

[Z4rd0Z]

Except this isn't a new VM.

Re:FIFO locks

[Kwikymart]

No, that would decouple the heizenburg compensators and send the flux capacitor into yeager-loop.

No clue what Proportional Share Scheduling is?

[arnoroefs2000]

There's more info here.

Excerpt:

"There are compelling reasons to use proportional share scheduling techniques to support multimedia and other soft real-time applications on general-purpose operating systems. First, proportional share (PS) schedulers are a good match for existing infrastructure such as a periodic timer interrupt and mechanisms for assigning priorities to applications -- priorities can be mapped to shares in a proportional-share environment. Second, PS schedulers provide stronger guarantees to applications than do traditional time-sharing schedulers: they allocate a specific fraction of the CPU to each thread, and some schedulers provide error bounds on the allocation rate. Third, PS schedulers have clear semantics during underload: excess CPU time is allocated fairly, in contrast with some reservation-based schedulers that must idle or back off to a secondary scheduling policy once all application budgets are exhausted."

Re:No clue what Proportional Share Scheduling is?

[ewhac]

Sounds vaguely similar to how BeOS's scheduler worked internally.

Schwab

Low Latency, Pre-Emptive multitasking?

[BrookHarty]

Im currently using linux with the low latency patch and pre-emptive multitasking. Does this help X seem a little smoother on BSD also?

Re:Low Latency, Pre-Emptive multitasking?

[Subcarrier]

Don't forget Linux will soon have a new scheduler, too. The O(1) scheduler by Ingo Molnar kicks some serious ass, especially with SMP.

Re:Low Latency, Pre-Emptive multitasking?

[noselasd]

Actually Linux already has a new scheduler, Ingo Molnars O(1) is already in 2.5. But almost noone(of the kernel developers) thinks it should go in the 2.4. Redhat ships it though.

Re:Low Latency, Pre-Emptive multitasking?

[Dwonis]

Well, my box is running almost that, and I'm not even running very much today:

dwon@zed:~$ ls -d /proc/[0-9]* | wc -l
151
dwon@zed:~$

Re:FreeBSD = Security

[prog-guru]

I don't know about that, considering in the last month we had 2 big exploits (openssh, and libc resolve bug). The advice for the libc bug was to cvsup the whole system, cause lots of stuff depended on that.

Yes I know the openssh bug affected everyone, but only *BSD has it installed and running by default.

Re:FreeBSD = Security

[Anonymous Coward]

Enabled by default? I don't think so!

matt@xena$ uname -sr
FreeBSD 4.6-RELEASE
matt@xena$ grep sshd_enable /etc/defaults/rc.conf
sshd_enable="NO" # Enable sshd
matt@xena$

Re:FreeBSD = Security

[Ashcrow]

In all honesty FreeBSD as exploitable as any Linux distrobution. Sure, it does not use glibc but it does make a regualr appearance on bugtraq and vuln-dev.

And here is some revised code (joke):
if (FreeBSD == "Security") { pray.fortheworld(); }

Re:KARMA WHORE !!!!

[Buck2]

You're so cute when you do that.

0(1) scheduler

[Sivar]

Is FreeBSD's new one a 0(1) scheduler?
0(1) is a "term" from computer science. When applied to schedulers, it basically means that no matter how many processes there are to schedule, a 0(1) scheduler's overhead will not significantly increase.
Of course, with a small number of threads/processes to schedule, the Linux 0(1) scheduler will have greater initial overhead. It isn't until there are quite a few processes that it starts to show its power, and the more processes there are, the more useful it is.
On a busy server with 4+ processors and thousands of processes, a standard scheduler's overhead is so great that it often exceeds the overhead of most of the individual server processes.

Re:0(1) scheduler

[Osty]

Is FreeBSD's new one a 0(1) scheduler?

Just a nitpick. The term is "O(1)", not "0(1)", as in "Big Oh of 1" and not "Zero of 1".

Re:0(1) scheduler

[Sivar]

Yes, I used the number "zero" (0) and not the letter "O" :)

Re:0(1) scheduler

[Sivar]

Oh, I see. Nevermind. Hmm, I could have sworn that the "O" in my books was a zero. I guess i'll have to double-check that. Thanks for the correction. Oops.

Re:0(1) scheduler

[dimator]

I prefer ")1(O" just to throw people off.

Re:0(1) scheduler

[pgpckt]

a 0(1) scheduler's overhead will not significantly increase.

The term you are looking for is probably O(n), but definatly not O(1). O(1) will not increase at all, no matter how much data is put in. O(n) will increase linearly.

Re:0(1) scheduler

[ink]

Actually, it won't increase at all. If computation time increases according to data size, no matter how insignificantly, it is not O(1)

There is a constant in there; big-oh measures the growth rate of an algorithm, not the actual time it takes to run. So, yes, it could take longer as the number of processes go up -- but that function would need to be linear. In other words: a O(1) algroithm isn't guaranteed to ever complete when presented with an infinite set of problems.

Re:0(1) scheduler

[cameldrv]

O(1) means constant time no matter what size input. O(n) is linear with input size.

Re:0(1) scheduler

[PastaAnta]

Yes, you are right about the constant. But for a O(1) algorithm the constant is just that - a constant! So for a data set of arbitrary size the complexity or time to execute will be - constant!

This is like when you look up in an array with a pointer - it will take the same time no matter the data size.

For a linear function the complexity will be O(n) as the previous poster said. Just like when you want to get an arbitrary element in a linked list for example.

For an O(1) scheduler the tradeoff may be larger initial overhead (for few processes) or increased memory usage (for table lookups) or just a plain smarter algorithm :-)

Re:0(1) scheduler

[ldspartan]

I have to take issue with this one...

O(n) is linear time, O(1) is constant time. So, no, it can't take longer as the number of processes increases. An O(1) implementation must have exactly the same complexity regardless of how much data is presented. Thats why this is the holy grail of algorithm design.

--
Phil

Re:0(1) scheduler

[Osty]

Actually, it's not "Big Oh" but Ordo, and the capital 'o' is for show that it's the Ordo value for the worst case.

Yeah yeah yeah, I didn't want to get too pedantic. The common way to pronounce "O(n)" is "Big Oh (of) n", and so that's what I said. Going into why a big 'O' is used, and what it means, and the differences between O, o, Big Theta, Big Omega, etc is out of scope for this topic and my post. Therefore, I didn't get into that and just made the correction from 0(1) to O(1).

Re:0(1) scheduler

[Resistance is futile]

O(n) is linear time, O(1) is constant time. So, no, it can't take longer as the number of processes increases. An O(1) implementation must have exactly the same complexity regardless of how much data is presented. Thats why this is the holy grail of algorithm design.

Actually, O(1) means that it is bounded by a constant so that the time can keep increasing as the input size increases.

Think a-b/n for n->infty

Re:0(1) scheduler

[ZerothAngel]

You're probably thinking about theta, which looks very similar to zero.

Re:0(1) scheduler

[Cryptnotic]

O(1) means that there is some constant C, such that for any problem size N, the execution time of the algorithm is less than C.

O(n) means that there is a constant, C, such that for any problem of size N, the execution time of the algorithm is less than C*N.

Re:0(1) scheduler

[srichman]

The term you are looking for is probably O(n), but definatly not O(1).

No, he means O(1) scheduler, definitely not O(n). Most simple scheduling algorithms (e.g., basic round robin) are constant-time, and a contant-time scheduler patch was recently released for Linux.

Just think about it logically. Why would he expectantly ask if the scheduler was O(n)? Name one super-linear scheduling algorithm in use in modern operating systems.

Re:0(1) scheduler

[ldspartan]

Ok, I concede there. Its still not the common usage, but that doesn't make you any less right :).

--
Phil

Re:0(1) scheduler

[noselasd]

O(1) will not increase - right. That is a good thing, wou want the scheduler to be as efficient as possible - right?

Re:0(1) scheduler

[Anonymous Coward]

The original 4.4BSD scheduler was already O(1) (by use of priority buckets).

It is only Linux that had (for a long time) the O(n) problem (all runnable tasks/threads in a single linked list that had to be fully scanned to find which one to run - ick!).

Re:0(1) scheduler

[platypus]

O(n) means that there is a constant, C, such that for any problem of size N, the execution time of the algorithm is less than C*N.

Small nitpick, the last one should probably be C*N + K, where K is a constant (you already used up the C :) ).

Re:0(1) scheduler

[zdzichu]

Is FreeBSD's new one a 0(1) scheduler?

No, it O(log n) scheduler. Read the mails.

Re:FIFO locks

[Buck2]

Thank god for open source. Could you imagine trying to figure that out in hex?

I mean, wouldn't it be funny if companies did stuff like that, like they sold software that might be "server" software or "client" software depending on one bit. Like NT?

That'd be funny.

Re:FreeBSD = Security

[Peyna]

Just about every *nix distro I've seen now has SSH up and running by default.

Darwin?

[BlackGriffen]

Anybody have any idea when/if Apple will integrate improvements from this scheduler in to Darwin/OSX?

Re:Darwin?

[Sivar]

The scheduler is closely tied with the kernel, and MacOSX does not use the FreeBSD kernel at all. It uses the Mach kernel, which is not only a different kernel entirely but a different core kernel philosophy. Mach is a microkernel whereas FreeBSD's is a monolithic kernel. Both types have their advantages and disadvantages, but microkernels are vastly superior for a commercial OS and for driver installations. Monolithic kernels are theoretically faster and easier to implement.

MacOSX gets its BSD label by using the BSD userland utilities. It is great that Mac's OS is no longer junk. In three months I went from "Macs are toy computers for kiddies and Photoshop pros" to "Wow--I can replace every PC and OS in my house with a single Mac! Great desktop, good server, and all the power of Unix."
I have never been happier with the state of Apple Inc.

Re:Darwin?

[JPriest]

Cool, I always figured OS X was monolithic. That sort of reminds me of the famous usernet debate between Linus and Andy Tanenbaun (MINIX) in 92 covering "microkernel vs monolithic systems". It was an interesting read, in I believe Linus stated "True, linux is monolithic, and I agree that microkernels are nicer."

Re:Darwin?

[Drishmung]

Not quite. This link gives quite bit more background about Darwin. In particular:

Part of the history of Mac OS X goes back to Berkeley Software Distributions (BSD) UNIX of the early seventies. Specifically, it is based in part on BSD 4.4 Lite. On a system level, many of the design decisions are made to align with BSD-style UNIX systems. Many of the libraries are derived from NetBSD (http://www.netbsd.org/), while many of the utilities are from FreeBSD (http://www.freebsd.org/). For future development, Mac OS X has adopted FreeBSD as a reference code base for BSD technology. Work is ongoing to more closely synchronize all BSD tools and libraries with the FreeBSD-stable branch.

Although Mac OS X must credit BSD for most of the underlying levels of the operating system, Mac OS X also owes a major debt to Mach. The kernel is heavily influenced in its design philosophy by Carnegie Mellon's Mach project. The kernel is not a pure microkernel implementation though since the address space is shared with BSD processes.

The Mac OS X kernel (also known as XNU) is a monolithic kernel (unlike Mach, but like Linux and xBSD) with Mach and BSD sitting side-by-side.

Some earlier Apple Unix efforts were true micro kernel implementations. This was also driven by the attraction of a pure hardware abstraction layer. With Darwin this seems to have moved to a more pragmatic recognition that performance matters.

In Darwin, the Mach bits handle memory management, IPC and device drivers. BSD handles users and permissions, the network stack, the virtual file system and POSIX.

So, this won't directly benefit Darwin, though if it is generally useful then someone/anyone can try and put it into Darwin---long live open source! I confess I don't know how the Mach part of Darwin handles scheduling, though I had heard that the Mach VM and scheduling was pretty good.

Re:Darwin?

[Sivar]

That is a very informative link, thanks!
Someone please mod this person up as my comment, currently at +5 (undeservingly), obviously has some flaws that Drishmung has corrected.

Re:FreeBSD = Security

[Anomymous Coward]

I don't know about that, considering in the last month we had 2 big exploits (openssh, and libc resolve bug). The advice for the libc bug was to cvsup the whole system, cause lots of stuff depended on that.


The openssh bug had a one line workaround.

The libc resolver bug has not been successfully exploited yet (so it's not really an exploit). It SEEMS POSSIBLE to exploit it, yes, but it's not trivial (it involves messing up dns replies, so you'd have to have control over an ip block, force the resolver to try to resolve an ip in that block, send the bad response, and then hope it worked). If you know anything about the bsd source code, you know that you can cd /usr/src/lib/libc && make all install. True, it doesn't help statically compiled binaries, but how much do you have on y our system that's statically linked AND answers on remote IPs? certainly not bind, apache, sendmail, or qpopper... or any of the other standard services.

Re:FreeBSD = Security

[quantum bit]

None of the FreeBSD releases, or the -STABLE branch were vulnerable to the openssh bug.

ftp://ftp.FreeBSD.org/pub/FreeBSD/CERT/advisories/ FreeBSD-SA-02:31.openssh.asc

Note the absence of any released version of FreeBSD.

FreeBSD ~= Security

[Sivar]

I am not going to claim that FreeBSD is perfect, but FreeBSD is more secure than the vast majority of Linux-based OSes. It has supported features like the new "GR Security" patch for years, and because it shares a great deal of code with OpenBSD which is audited frequently, it benefits from their work as well.
Of note is that FreeBSD's libc is just over half the size of Linux's Glibc (not that has a thing to do with security)

With FreeBSD, for years, admins have been able to set certain files as "append only" (so even root can only add to, not remove from, log files) and "immutable" (so even root cannot modify or delete the file) and has been able to set firewall rules to the same (immutable) so that creative crackers can't add their personal favorites if they root the system.
This can of course be bypassed by restarting the machine in single-user mode and redusing the kernel security level, but that isn't going to be very easy for your average remote hacker. :)

Furthermore, since 4.0 you can multiple run complete but separate entire copies of FreeBSD on the same system, each with their own FreeBSD system files and such. You can have a single server run an instance of FreeBSD for Apache, one for Postfix, one for BIND, etc. and if any one of them does get compromised (say, BIND since that happens entirely too often) the cracker can not only not effect any of the other instances--he/she cannot even see that they exist! Very interesting stuff.
Of course, IMHO Linux is worlds ahead of FreeBSD on the desktop front, and the new GRsecurity and ACL features will be a real competitor for the *BSD family. It will be most fascinating to see how things turn out. I wish the best to both of them, and I use both of them every day.

Re:FreeBSD ~= Security

[FooBarWidget]

"With FreeBSD, for years, admins have been able to set certain files as "append only" (so even root can only add to, not remove from, log files) and "immutable" (so even root cannot modify or delete the file) and has been able to set firewall rules to the same (immutable) so that creative crackers can't add their personal favorites if they root the system."

Linux can do that too. That is, if you use ext2 or ext3. See chattr(1), options a (append only) and s (secure deletion).

Re:FreeBSD ~= Security

[Fweeky]

if you use ext2 or ext3. See chattr(1), options a (append only) and s (secure deletion).

How do you unset them? In FreeBSD the kernel can have a secure level set which can only be raised without rebooting; setting the kernel into a securelevel stops the various immutable/undeletable/etc flags being unset, among other things.

Check security(7) under "SECURING THE KERNEL CORE, RAW DEVICES, AND FILESYSTEMS".

Re:FreeBSD ~= Security

[FooBarWidget]

chattr +a foo.txt # set Append Only

chattr -a foo.txt # unset

Re:FreeBSD ~= Security

[Fweeky]

Er, yes, so how do you stop someone compromising root and setting -a? In FreeBSD, with a securelevel, the equivilent of chattr -a would fail with a permission denied even for the superuser.

Setting the right things immutable would result in even you being unable to replace the base system or drop the securelevel without local access.

Hmmm

[FrostedWheat]

the only 3 files modified are kern_synch.c, kern_switch.c and proc.h, plus a one-line change to kern_exit.c

I hate to be picky .. but that's 4 files modified. One-line or a thousand, it's still been modified.
</END-RANT>

Re:Hmmm

[Joe Mucchiello]

And modifying a .h file is modifying every file that includes it. How many files include proc.h? Sounds like a popular include file.

Schedulers. (*nix v. win2k)

[Erpo]

I've been poking my newbie nose around in scheduling for a little while, and while I still know very little I've found the field very interesting. It's always neat to see new features and techniques being tried, but there's a feature that exists in the windows nt scheduler that (as far as I can tell) is absent in *nix operating systems. Winnt maintains (I think) four process queues (realtime, high, normal, and idle) into which all processes fall. Every time the scheduler is run, it checks to see if any "realtime" processes can be run, then "high", then "normal", and finally "idle". Processes in "less important" queues are only run if all processes in "more important" queues cannot be run (i.e. they're blocking on input or whatever), or those queues are empty. I find this very useful because I can set a long-running cpu dependent process to "idle" priority and it will be run at nearly 100% cpu usage when the machine is idle, but will instantly get out of the way and not be run at all if I choose to run something else (e.g. a game), no matter how high it's "goodness" value gets from not using any cpu time.

Is there any reason why something like this isn't implemented in Linux or FreeBSD? Low on the developers' feature priority list (har har)? Too difficult? Unnecessary?

Thanks. I'd appreciate any feedback.

Re:Schedulers. (*nix v. win2k)

[trycoon]

eeh? It's allready there, since the dawn of the dinosours. Check "man renice"!
My SETI@home run at 100% only when my CPU is idle...

Re:Schedulers. (*nix v. win2k)

[Sivar]

Not needed. Linux and FreeBSD both handle different priority levels quite nicely, and in fact can handle them in a much more fine-grained fashion. NT actually has additional priority levels in-between each that you described above, but Linux and BSD have a total of 41 possible priority values (from -20 to 20, including zero)
If you set an application to a priority of 20, it isn't going to be bothering any other processes, and if you set an application to -20, it is going to be worshipped like a god by the scheduler.
As far as I know, neither have a real-time scheduling mode like NT, which is actually a good thing in many cases. If a program running at real-time priority goes into an infinite loop, or for any reason uses 100% of hte CPU (SETI@Home, for example) than the system is locked the hell up. Even the mouse will not get any time for cursor movement, and you have to reset the machine.

Read "man nice", "man renice", and probably "man top" (which I use to change priorities of running processes as root)

Re:Schedulers. (*nix v. win2k)

[Ded Bob]

Is there any reason why something like this isn't implemented in Linux or FreeBSD?

FreeBSD does have something like this: idle, normal, and real time. By default it is normal, but you can change it to idle (idprio(1)) or real time (rtprio(1)).

In the man page for rtprio(1) is one relevant bug:
"Under FreeBSD system calls are currently never preempted, therefore non-realtime processes can starve realtime processes, or idletime processes can starve normal priority processes."

Maybe the new scheduler will fix this?

Re:Schedulers. (*nix v. win2k)

[Nevyn]

Not needed. Linux and FreeBSD both handle different priority levels quite nicely, and in fact can handle them in a much more fine-grained fashion. NT actually has additional priority levels in-between each that you described above, but Linux and BSD have a total of 41 possible priority values (from -20 to 20, including zero)

Actually they can have much more than that, but the "user" can only set those 41 values. There isn't even a definition of how they have to map (so for instance you can have the schedular have priorities -20 to 20 including zero but in 0.5 increments. Or just -40 to 40 in 1.0 increments etc.

As far as I know, neither have a real-time scheduling mode like NT, which is actually a good thing in many cases. If a program running at real-time priority goes into an infinite loop, or for any reason uses 100% of hte CPU (SETI@Home, for example) than the system is locked the hell up. Even the mouse will not get any time for cursor movement, and you have to reset the machine.

Do "man sched_setparam" and man "setpriority" to learn about "real time" processes. Linux also had patches for SCHED_IDLE at one point, but that can cause pretty bad reasource deadlocks without priority inversion and so was never integrated.

There can be problems if your "real time" process misbehaves but then don't do that.

NOTE: The term "real time" is used to mean best effort at real time, for true real time you don't use a general purpose OS (Ie. NT doesn't count either).

Re:Schedulers. (*nix v. win2k)

[Burning1]

I disagree to some extent with comments that Linux has a scheduler that is 'good enough.'

I ran some real time multimedia streaming applications under linux a while ago (LiveIce; very nice tool) and found that even using nice, I would occationally get glitches in the stream output during broadcasts.

Switching to FVWM helped somewhat, as did nice... But when it cames right down to it, I don't think this should have been a problem.

Re:Schedulers. (*nix v. win2k)

[Mr. Tuvix]

On Linux and (I believe) FreeBSD, the ability to set processes to different levels of realtime priority has been in place for years. Very recent (last couple weeks?) versions of the new Linux O(1) scheduler can do "idle" priority, and it will probably be in the 2.6 mainstream kernel.

Implementing the idle priority is hard, because if an "idle" task gets a lock on a crucial resource and gets preempted, any non-idle tasks that need that lock must then wait for the system to become idle -- potentially an indefinite amount of time.

Re:Schedulers. (*nix v. win2k)

[mosch]

Is there any reason why something like this isn't implemented in Linux or FreeBSD?

Mostly because it's already implemented on both systems. For freebsd see rtprio, idprio, and of course good old nice.

Re:Schedulers. (*nix v. win2k)

[Birdie-PL]

And there is this patch which allows you to limit the CPU usage if nice is not enough for you.

Re:Schedulers. (*nix v. win2k)

[GypC]

Try it with a low-latency or pre-emptive-scheduling patched kernel. By default the kernel is tuned as a server, not a multimedia workstation. The kernel team seems to be working to correct this by adding compile time options and/or a new scheduler, but for now, on stable kernels, it requires patches.

I've personally found that the patches make an amazing improvement in UI responsiveness and media playback. I'm looking forward to 2.6.

Re:Schedulers. (*nix v. win2k)

[xA40D]

The nice value is merely a weighting factor that is used in the calculation of a task's the priority. The actual priority of a task depends upon what the task is, how much CPU time it's used recently, and which run queue the task has been assigned to. Indeed, each process has two priorities, one for user-mode execution, and one for kernel-mode execution. Renicing a process -20 and expecting it to be treated as god may seem a reasonable inferance, but if you ever delve into the internals of BSD you'll discover how wrong you are - and how brain numbingly complex priority scheduling actual is.

Re:FreeBSD is dieing?

[darqchild]

"Linux-nazi empire. I can't believe it." Wasn't Goldwyns law supposed to stop this thread dead in it's tracks?

Re:FreeBSD = Security

[prog-guru]

you know that you can cd /usr/src/lib/libc && make all install

Right, that's what I did on some of the systems. You do have to restart the programs loading libc too.

Re:FreeBSD = Security

[BrookHarty]

I think you need to re-read that file. +3 indeed.

Affects: FreeBSD-CURRENT between 2002-03-18 and 2002-06-25

Re:FreeBSD = Security

[bovinewasteproduct]

None of the FreeBSD releases, or the -STABLE branch were vulnerable to the openssh bug.

ftp://ftp.FreeBSD.org/pub/FreeBSD/CERT/advisorie s/ FreeBSD-SA-02:31.openssh.asc

Note the absence of any released version of FreeBSD.

I think you need to re-read that file. +3 indeed.

Affects: FreeBSD-CURRENT between 2002-03-18 and 2002-06-25


No, you try re-reading it again. FreeBSD releases (which are based on FreeBSD-Stable) are NOT anywhere near being FreeBSD-Current!

FreeBSD-Current is not a release version. As a matter of fact, people are told NOT to use it unless they know what they are doing, and not just becuase they want one of the new features (which are nice, I've got a FreeBSD-Current SMP box here).

BWP

Re:FreeBSD = Security

[BrookHarty]

None of the FreeBSD releases, or the -STABLE branch were vulnerable to the openssh bug.

-Current isnt a release? I can download it, Seems like a release to me, maybe a beta release, but it IS a release if its been put out. But whatever you call it some version of FreeBSD DID have the bug.

Really get a kick out of you BSD guys trying to use the -Current and -Stable shit to try to say you dont have security holes. So you dont have this 1 bug on your -Stable, Lets check cert and see how many have been on -Stable. Same shit goes with OpenBSD, and "Never A Security bug" bullshit. PHP/SSH/Apache have a security hole, its "LINUX HAS A SECURITY HOLE", well FreeBSD uses the same damn software.
-
Try your Jedi mind tricks on a mircos~1 padawn... Go Go away...

Re:FreeBSD = Security

[generalpf]

-CURRENT isn't a release. It's a cvs repository. It's not even guaranteed to compile.

Re:That's Nothin'!

[timeOday]

Yeah, that was my point, Linux's scheduler upgrade wasn't too graceful. Guess I was a little too "clever" in expressing myself.

Re:FreeBSD is dieing?

[stor]

Sounds like you're after a FreeBSD-centric site. Perhaps you should just jump on the FreeBSD mailing list or something.

Cheers
Stor

p.s. Ahh sheet, IHBT.

Re:FreeBSD = Security

[Arandir]

Really get a kick out of you BSD guys trying to use the -Current and -Stable shit to try to say you dont have security holes.

Go find out what -CURRENT actually is before you comment further. You are in serious danger of choking on your own foot...

Just because -CURRENT is publicly available does not mean that it is released. An analogy is in order. -RELEASE is equivalent to linux-2.4.18. -STABLE is equivalent to linux-2.5.27. -CURRENT is equivalent to whatever is on Linus' harddrive at this very instant.

What are you doing to crash the system?

[axxackall]

It is far more robust and actually more stable.

Hey, people, what are you doing with operating systems that they are crashing like a domino?

I use Linux as well as FreeBSD for awhile with Apache, Sendmail, X11 and lots of other software competing for memory, CPU, exotic devices, network bandwidth and just for disk I/O. No crashes. On linux I use also video, Oracle, Tomcat, JBoss - all works fine, besides own bugs (it's ok for userland, isn't it?).

For my experience, which doesn't meet crashes, it is more important what hardware, filesystems and protocols are supported. And of course I compare systems with only stable manually configured and re-compiled kernels passed "on-the-field" regression tests before I upgrade the production mode with that new kernel.

So, one more time, why should I prefer BSD?

Re:What are you doing to crash the system?

[axxackall]

2.4.7 was not bad in that collection. The rest i did not use. As I told, I do at first regression test by myself before I apporve to myself any upgrades in the production system.

It can vary

[SilentStrike]

The running time can vary for given N, just not for abtrirary N. As demonstrated by the following code. It runs fastest when n = 1, followed by n = 2, but is still O(1) when n is not 1 or 2, because the running time is constant for large n.

int orderOneButVarying(int n) {
if (n == 1) {
return n;
} else if (n == 2) {
for (int i = 0; i 3; i++) {
n*=n;
}
return n;
}

for (int i = 0; i 1000; i++) {
n+= n / 2;
}

return n;
}

Re:FreeBSD = Security

[essdodson]

It was all a ploy on Theo's part. I can't believe anyone with respect for the opensource community would have taken the approach he took to the latest big OpenSSH exploit. It basically went like this : Theo announces that there is some sort of problem, Theo urges everyone to upgrade to upgrade to the latest still vulnerable version, a patch is released, details are provided. I see this as an obvious attempt at bringing other platforms which were not previously vulnerable into the mess with OpenBSD.

Re:This is excellent news

[Dirtside]

...or at least, schedule the funeral ;)

Except Solaris

[moogla]

::shrugs::
Doesn't even come with it.

Re:FreeBSD = Security

[Andrew Francis]

Affects: FreeBSD-CURRENT between 2002-03-18 and 2002-06-25

FreeBSD-CURRENT isn't a distinct release, it's a branch in the CVS tree.

Here is a great reason to prefer BSD

[sawilson]

http://uptime.netcraft.com/up/today/top.avg.html

You'll notice that 45 of those top 50 are BSD
machines. Of those 45, 19 are FreeBSD boxes.
You'll notice 1 Linux box. It's nice to see that
leading industry sites like bongload.com and
twobigirls.com have benefited so much from the
stability of BSD.

Re:Here is a great reason to prefer BSD

[axxackall]

Longest time in that table means for me that people don't have rapidly developed applications there. I guess they host mostly static pages. Maximum what they have is CGI scripts with MySQL. I sure they don't use any Java servlets. Any serious databases servers either. And as far as know the phsychology of most of sysadmins - when they upgrade serious web or database server they use the chance to restart the computer. Either just for a case or to test last changes in boot or monitoring scripts. No matter what OS: Solaris, BSD or Linux - use the chance to reboot your box. So your table doesn't prove anything.

Re:Here is a great reason to prefer BSD

[realdpk]

Rebooting a box isn't necessary when using a database or a "serious web server", at least not on FreeBSD. Maybe it is on other OS's?

I mean, really, what good does rebooting do?

Re:Here is a great reason to prefer BSD

[axxackall]

Technically there is no need in rebooting.

Psychologically many admins prefer to reboot from time to time.

Personally, I like to reboot when the system (including remote monitoring clients) has some scripts related to reboot/shutdown and I'd like to test changes I've recently made. I know, I've already tested it on the testing machines, but ... just to make sure.

Re:Here is a great reason to prefer BSD

[Astastrafal]

It's probably due to that bug whereby the uptime counter wraps around after 497 days and a few timesteps later. From what I gather, there are people who are working to get that bug squashed but I dunno if their patches have been accepted by Linus. See also this page from Netcraft explaining the potential problems about their method of uptime detection. You'll also learn there that some OSes do not report uptime, among them such "enterprise" favorites as AIX, Tru64, VM, OS/390 and others. So Netcraft's stuff, while interesting, is not a definitive comparison of the relative stability of the various server OSes out on the Net.

I use Linux AND FreeBSD

[leereyno]

Right now I've got four systems running Linux (RH-7.3 and 7.2) and one system running FreeBSD 4.6. At times in the past I've run OpenBSD and NetBSD as well.

I can tell you firsthand that in terms of system stability that Linux and FreeBSD are comparable if not indistinguishable. FreeBSD does seem to be more efficient however. The pentium 200 that I have FreeBSD on loads up KDE 3.0 noticably faster than Redhat 7.2 did, and once loaded it is more responsive. On older hardware FreeBSD definitely seems to have an advantage. I consider FreeBSD to be a very fast and well designed operating system. I keep trying to find places where using it instead of Linux would be an advantage.

Not everything about it is all that rosey however. The features and abilities that Linux provides but FreeBSD lacks such as SMP, kernel pre-emption, fast journaling filesystems, certain commerical software packages, 3D acclerated X servers, and generally better device support, make actually using FreeBSD as anything but an interesting toy kind of difficult to justify in many situations.

I worry about FreeBSD. I'd love to see it grow and progress not as a competitor to Linux, but as something of a companion to it. So many people just don't seem to realize that open source isn't about operating systems alone. What Linux and FreeBSD do is provide a foundation, they aren't the whole house. Both provide a powerful and stable platform for running the actual programs that people want to use in the first place. The future of open source development is going to be 90% apps and userland and 10% OS. To have religious and political wars over the OS portion is immature and counterproductive. Linux and FreeBSD aren't genuine competitors from an economic standpoint because it is the applications that both run that make either compelling in the first place.

I want BOTH Linux and FreeBSD to do well, to grow and expand and be the best operating systems anyone has ever seen. I detest the infantile immaturity of those who seek to create division and conflict between FreeBSD and Linux that simply shouldn't be there. I've gotten flames from FreeBSD "advocates" in particular filled with such hatred and obvious zealousness that you'd think they were Mac freaks, all because I described FreeBSD in terms that weren't favorable enough for their religious views. The Linux crowd is full of just as many jackasses, if not more.

Computer enthusiasts are known for generally having high IQ's. Unfortunately our reputation for having low EQ's is equally well earned. There are far too many borderline autistics and asperger's sufferers among us with severely retarded social skills. That is really the only explanation I can come up with when grown men with extensive vocabularies use them to throw a fit on par with that of an eight year old.

Anyway I'm drifting way off from what I wanted to write about. The point that I really want to make is that BOTH Linux and FreeBSD are absolutely fabulous operating systems (save the linux is just a kernel messages for church). The goals and vision behind each are so similar that any ill will between them is manufactured by immature, short sighted assholes. Microsoft is the enemy, not those who prefer another free Unix derivative that runs Mozilla, gnome, kde, etc just as well if not better than what someone else is using.

Lee

Re:I use Linux AND FreeBSD

[Anonymous Coward]

Not everything about it is all that rosey however. The features and abilities that Linux provides but FreeBSD lacks such as SMP, kernel pre-emption, fast journaling filesystems, certain commerical software packages, 3D acclerated X servers, and generally better device support, make actually using FreeBSD as anything but an interesting toy kind of difficult to justify in many situations.

1. FreeBSD does support SMP, albeit not as good as 2.4; but way better than 2.2.

2. Kernel pre-emption is a hack, I'd like it but many developers think it's the wrong path.

3. SoftUpdates are in most ways alot better than journaling, the only time that it isn't is in the event of a crash; as you have to whipe the dangling blocks in the background. But as performance is much better during up-time, it's all worth it.

4. Can't argue with that, but then, I'd run Windows if it was all that important.

5. Accelerated X, the X Server, is available for FreeBSD.

6. BSD was way ahead of Linux when it comes to FireWire, USB and SCSI. In fact, the USB system in linux comes from NetBSD( 2 years later! ); and the SCSI system is way better on FreeBSD than on Linux.

Re:I use Linux AND FreeBSD

[leereyno]

1: My (admittedly limited) understanding of SMP on FreeBSD was that they were having problems finding people to work on it. That they were having to do a rush job on it in order to try and get something marginally workable for 5.0. If SMP is working as well as it did in Linux 2.2 then that would be great.

2: I LOVE kernel pre-emption. On older systems where the kernel can be more or less stuck chewing on something it makes the computer much more responsive. Kernel pre-emption is one of the hallmarks of microkernel architectures (well pre-emption of the userland threads that handle the system calls at least). The nice thing about it is that you don't HAVE to use it.

3: the ext2 filesystem has support for ordered updates just like UFS. It isn't generally used except on servers because of the speed penalty. Ext3 on the other hand uses ordered mode by default. The benefit of jornaling is that your filesystem doesn't get left in and undefined state in the case of a crash. Also when you reboot you don't have to spend who knows how long checking the filesystem. Speed is a secondary concern. I can't make any solid claims about the speed of UFS using soft updates vs. ext3. All I can says that my personal experience with UFS has been that its much, much slower when it comes to creating or destroying files compared to either ext2 in asynchronous mode or ext3. In terms of raw throughput on existing files all I can say is that all three are fast enough that I don't notice any difference between them.

4: It might not be important to you, but I support Unix in the college of engineering at Arizona State and believe me things like Matlab and Maple are a very big deal around here. I know of course that these might very well run on FreeBSD using its ability to emulate Linux (which is a VERY good thing IMHO).

5: It is also commercial software not not cheap. I'm not one of your free software religious zealots, I have zero problems with commercial software. The problem is that if Accelerated X is the only X server with 3d support for FreeBSD, then that tips the scales decidedly in the favor of Linux where such support part of the base system.

6: I know that NetBSD had support for USB back in early '99. I wasn't aware that the support for it under Linux was derived from NetBSD. Even if it was, the comparison here is between Linux and FreeBSD, not NetBSD. I've run NetBSD and I've never been particularly impressed with it. The main problem I see is that there are still people trying to support long dead hardware with it, hardware that NO ONE is going to actually run if they can help it in any way. I got a Quadra 700 for free a year or so ago and decided to try and make it into a real computer. That meant of course putting a real operating system on it. I chose NetBSD and sure enough it would load up and run.....VERY, VERY slowly. The fastest 68k mac ever made wasn't much faster than that Quadra 700 was, meaning that trying to continue to maintain a codebase for the platform line is an utter waste of time. I installed NetBSD for intel shortly thereafter and was less than thrilled by the stability of the system. The kernel didn't crash, but userland stuff I'd compiled from the ports sure did. OpenBSD is much better in that regard, most likely because it doesn't get spread too thin chasing dead hardware. I really like OpenBSD because it is designed to be secure by default, something that other Unix derivatives would do well to emulate. Obviously security is in the hands of the administrator and it is very possible to misconfigure OpenBSD to be insecure. Its just that with it I don't have such a nagging worry that there might be some exploit of some service or another waiting to be discovered by the cracker community. As for SCSI under Linux and FreeBSD I can't really comment one way or the other.

I wasn't really sure how to respond to your post at first. To a large extent I think that for us to debate whether the issues above are real or not is pointless because I'm not out to attack FreeBSD by bringing them up. I WANT FreeBSD to do well, and these are the areas that I think it needs improvement in.

The code base for FreeBSD is VERY mature and the base BSD system itself was created by some of the most brilliant computer scientists who have ever lived, such as Bill Joy. Ever read The Design and Implementation of the 4.4 BSD Operating System? Linux owes as much to BSD as it does to GNU.

Projects like the BSD's are valuable because they provide an alternate view of how to do things from what is available under Linux. If something isn't really being done well in Linux and works better under one of the BSD's, then that provides a valuable contrast so that everyone can see that the problem needs to be fixed. This is a two way street however, which is exactly why I've listed things I thing need to be fixed. By listing them I'm not saying that FreeBSD is "bad," only that there are areas where there is room for improvement.

Lee

Re:I use Linux AND FreeBSD

[gomerbud]

5: It is also commercial software not not cheap. I'm not one of your free software religious zealots, I have zero problems with commercial software. The problem is that if Accelerated X is the only X server with 3d support for FreeBSD, then that tips the scales decidedly in the favor of Linux where such support part of the base system.

freeBSD has had support for hardware graphics acceleration for some time. currently it works for native freebsd binaries. support for linux binaries like quake3 and return to castle wolfenstien can be added with the linux_dri package. take a look for your self

http://people.freebsd.org/~anholt/dri/

this is a cool time for dri under freebsd. anholt is currently working with the dri group at sourceforge. in a little while, XFree86 will be able to 'officially' support DRI under freebsd.

Re:FreeBSD = Security

[Matty_]

He wasn't wrong. The FreeBSD-CURRENT branch was vulnerable, but that is the development branch and few people use it. Developers are the only ones that use it, for the most part.

So, as he said, none of the releases or the -STABLE branch were vulnerable to the OpenSSH vulnerability from a few weeks back.

Please read what the guy says before you start jumping on his bones.

Re:FreeBSD = Security

[Patrick Cable II]

Yes. A Default install doesnt have telnet enabled either...

no remote access == more secure than any remote access.

FreeBSD SMP.

[Patrick Cable II]

(root@oracle)(3/ttyp0)(08:19P:07/23/02)-
(#:/sys/ i386/conf)- grep SMP LINT

# SMP OPTIONS:
# SMP enables building of a Symmetric MultiProcessor Kernel.
# An SMP kernel will ONLY run on an Intel MP spec. qualified motherboard.
# Be sure to disable 'cpu I386_CPU' && 'cpu I486_CPU' for SMP kernels.
# Check the 'Rogue SMP hardware' section to see if additional options
options SMP # Symmetric MultiProcessor Kernel

Re:FreeBSD SMP.

[leereyno]

Yes, but does the SMP actually work well? My understanding having not followed too closely is that SMP is not really working yet. It will detect two CPU's, but only really use one of them. That might have changed recently in which case I'm glad to hear it.

Actually that reminds me of another area where FreeBSD needs improvement, kernel threads. These aren't a make or break feature, but they are nice.

Lee

Re:*BSD is dying

[tooLateNow]

Indeed. If I had only known that BSD was dead I would have avoided having so many happy customers who paid me money.

Re:*BSD is dying

[scalis]

Nail on the head bro' ! Ill probably get my ass kicked by penguin fans but my religion still tells me that FreeBSD kicks the penguin in the nuts.... The deamon gets just in front of the penguin in my book... Solaris and others are way below....

Re:This is excellent news

[scalis]

Yeah! Power to the machines!! ./

Your point makes no sense.

[sawilson]

That's a weak point. It still doesn't explain
why BSD dominates this "table" and some other OS
doesn't. Why isn't it dominated with Linux or
some other OS that isn't being what you perceive as
"rapidly developed". So Linux can't handle the stress
of a non developed environment? You can FUD
around with your philosophies on the psychology
of sysadmins all you want, but this table shows
something very clear. As much as your love your
pet OS, BSD is the clear stability king.