Software Tweak Makes Linux Boot In Under 200 ms

An anonymous reader writes "A version of Linux has been created that radically speeds up system boot time -- to less than 200 milliseconds (ms) from power-up to application code startup. The techniques, created by Real-time Linux vendor FSMLabs, are processor independent, and boot times of under 100 mS are expected in the future." Update: 09/30 01:04 GMT by T : Yep -- both headline and post should have read "ms" (milliseconds) rather than "mS" (milli Siemens); thanks to all the alert readers.

Comments

Only for embedded devices

[Professor_Quail]

This isn't for desktop linux, only for embedded devices.

Re:Only for embedded devices

[herrvinny]

yeah, I saw that too.. Slightly off topic, but what's the fastest boot time you can get with a 300mhz laptop? I have a compaq armada currently running Win 98, and if Linux can boot up faster than win 98 (I disabled a lot of stuff that autolaunches, like virus scanners, firewall, etc) I can use the laptop to take notes in lecture.

Re:Only for embedded devices

[JesseL]

The article says this work was done to improve the boot time of embedded devices, but I don't see anywhere why these changes couldn't be applied to any other computers running linux.

Re:Only for embedded devices

[Fareq]

ahh... makes much more sense...

I had something resembling
"My BIOS takes at least 10-20 times that, how can my OS speed that up..."

On topic, it is very cool to have an operating system that can start that quickly -- assuming stability, which is essential in embedded stuff (what happens when your Microwave has a kernel panic while cooking, for example)

Embedded devices really should start very fast, so boot time is really important. That said, this thing about 100ms being available eventually doesn't matter. Most people can't really do much with 100ms. In fact, most people can't tell if audio synchronization is off unless its at least about 75-100ms off... and only then when looking at people's faces. My point is: going from a large number of seconds to a large number of milliseconds is a great feat. But, I think that I can wait the full 1/5 of a second for my system to boot... I won't hold my breath (or waste my $) making it 100ms faster...

Re:Only for embedded devices

[Anonymous Coward]

you could look into BeOS or QNX. They both have impressive boot times.

Of course, DOS with edit.com, or an early version of wordperfect or wordstar would put everything else to shame

Re:Only for embedded devices

[wo1verin3]

I'm all for that last minute getting to class thing, but you could always boot it a minute before class starts?

Or leave it in suspend?

Re:Only for embedded devices

[HidingMyName]

I'm not privy to his techniques, but he may be hardwiring the compiled kernel for the target architecture to get more speed (recall that when programming, using early binding times trade off flexibility for speed). Yodaiken's a smart guy, so I may not have guessed his tricks.

Embedded devices may not need to do things like hardware discovery, plug and play configuration, etc. since their hardware configuration may be constant (so this stuff could be compiled into the kernel). Additionally, booting the kernel is different than doing various daemon startups and file system initializations, network configuration, etc. that one typically wants for non-desktop devices.

Re:Only for embedded devices

[Dahamma]

> "My BIOS takes at least 10-20 times that, how can my OS speed that up..."

Hopefully we'll see this project combined with LinuxBIOS (which can load & boot the kernel in 500ms - from compactflash at least. With IDE, you have to wait a few seconds for the disk to spin up). 5 seconds (or 1 with CF) to get to init on an x86 box would be impressive. Now to get a parallel service startup into a major distribution...

Re:Only for embedded devices

[yomegaman]

He's talking about suspend mode (on mine it's called standby). You know, you close the lid, the computer goes into a low-power mode, then when you open it you're right back where you left off. After there really laptops that don't have this feature?

Re:Only for embedded devices

[LilJimbo]

and put it in the autoexec.bat so you don't even have to type it in when it boots :)

Re:Only for embedded devices

[ihopMaintenance]

I disabled a lot of stuff that autolaunches, like virus scanners, firewall, etc

Good idea. Virus scanners and firewalls use WAY too much overhead. Heck I used to run them and they interfered with my screen savers and stuff. Who needs em.

Re:Only for embedded devices

[Anonymous Coward]

Or how about this, mod the kernel to detect a successful boot and then immediately take a snapshot of the kernel memory.

Next time on boot, just copy the memory image into ram, set the registers up right and then jump to the instruction pointer after where the snapshot was taken and just continue.

I think windows does something like that.

And I know for a fact that emacs does this to help speed up its startup time.

Re:Only for embedded devices

[wo1verin3]

>> Perhaps the instant-on concept confuses you?

No, it's not a matter of confusion, the parent seems to be deciding to use Windows vs Linux based on the bootup time which is silly. There are tools for every job. Sometimes it is Windows, sometimes it is Linux.

The alternative I suggested was to use suspend, a better option would be hibernate perhaps where the laptop is completely powered down. The Ram contents are written to harddisk. On an ACPI system windows begins loading and sees that it was in suspend and loads the hibernation file into ram, or APM where this function is controlled more at the bios level.

Re:Only for embedded devices

[danheskett]

Windows does exactly that for its "hibernate" feature. Essentially hardware is returned to a safe state (write operation suspended, etc) and the contents of memory written sequentially to a block of disk space. On boot, a small stub is loaded, which copies the data from disk to memory. Execution resumes at that point. The last thing that Windows does before hibernating is to schedule hardware re-initilization in a short period of time - essentially ensuring that things like sound cards, video cards, USB devices, etc get a "soft" reset when out of the hibernation period.

On a decent reasonable recent PC this takes a few seconds. On my fast AMD box I hit the power button and about 2 seconds later the box is off. Hit the power button again and its back on in about 4 seconds exactly where I left off. If I lost my DHCP lease in the interim it takes another 3-4 seconds for network connections to resume.

But whats best about the changes MS made to Windows XP is the multi-threaded boot process. Other than the kernel load the userland stuff is mostly all multi-threaded even on boot. This means services and whatnot are starting concurrently without waiting for other things to finish. Its somewhat less safe - a failed service could potentially cause others to hang and lock the system and/or cause a reset - however, I've yet to see that problem in the field. The net result is that on a typical AMD/P4 Windows XP box you can get to a login prompt/welcome screen in under 30 seconds, often under 20.

Users I've switched from XP to various Linuxes have all complained about the boot times - Red Hat 8 on similiar hardware takes eons to get a graphical login prompt. SuSE is just as bad. Mandrake is somewhat better, but all are in the minutes not seconds metric.

Re:Only for embedded devices

[mirko]

RiscOS could get from a cold boot to the GUI in less than 4 seconds on an ARM3/8MHz.
I can't even measure it now on a StrongARM/202MHz, also consider the newest RiscOS powered computer : Iyonix (Xscale/600MHz)...

Re:Only for embedded devices

[Zog The Undeniable]

I was wondering, as the article says "Normally Linux takes a leisurely 5 seconds or more to complete the boot process" and my PC takes a postively indolent 90 seconds!

Re:Only for embedded devices

[-brazil-]

But whats best about the changes MS made to Windows XP is the multi-threaded boot process. Other than the kernel load the userland stuff is mostly all multi-threaded even on boot. This means services and whatnot are starting concurrently without waiting for other things to finish. Its somewhat less safe - a failed service could potentially cause others to hang and lock the system and/or cause a reset - however, I've yet to see that problem in the field. The net result is that on a typical AMD/P4 Windows XP box you can get to a login prompt/welcome screen in under 30 seconds, often under 20.
Yup, I absolutely love XP's lightning-bootup. But the main thing that made it possible was the implementation of non-exclusive read locking of the registry, without which the multithreaded approach would have been useless. I don't think it's any less safe - there are well-known techniques for avoiding deadlocks that should be applicable.

Re:Only for embedded devices

[Anonymous Coward]

Why would an embedded device need a firewall or virusscanner if it takes no input but from a certified source?

Re:Only for embedded devices

[blibbleblobble]

"Windows does exactly that for its "hibernate" feature. Essentially hardware is returned to a safe state (write operation suspended, etc) and the contents of memory written sequentially to a block of disk space. On boot, a small stub is loaded, which copies the data from disk to memory. Execution resumes at that point."

Yeah, tried using that mode once on Windows98: Restarting after the hibernate didn't go as smoothly as one hoped:

IExplore has caused illegal operation and will be shut down
IExplore has caused illegal operation and will be shut down
Kernel32.exe has caused illegal operation and will be shut down
Kernel32 is not responding, press space to try again or Ctrl-Alt-Del to reboot

Re:Only for embedded devices

[RAMMS EIN]

``Users I've switched from XP to various Linuxes have all complained about the boot times - Red Hat 8 on similiar hardware takes eons to get a graphical login prompt. SuSE is just as bad. Mandrake is somewhat better, but all are in the minutes not seconds metric.''

IMHO, this is due to the Linux boot process having been inherited from Unices run on servers. Waiting a couple of minutes for the system to boot is annoying, but boot time is not really an issue on a system that is normally always up.

I have an old 486 that boots Linux in seconds (not counting the power on self test). The trick? Don't load anything you don't need. The kernel is a aggressively stripped modular kernel (with support for IDE harddisks and ext2fs built-in). Other things (drivers for the network card, floppy drive, iso9660fs, ...) are provided as modules that are only loaded when needed. The init scripts mostly just remount the root filesystem read-write, and spawn a few ttys and allow users to log in on them. Pity I can't get decent colors on the gaphic card, otherwise it would make a great terminal.

Re:Only for embedded devices

[dublin]

Yeah, tried using that mode once on Windows98: Restarting after the hibernate didn't go as smoothly as one hoped:

Oh, come on now... I'm no MS apologist, but Win98's power management support was notoriously bug infested. (I know - I was Program Manager for Dell's laptops when it was introduced!)

Comparing this to W2K or XP is like comparing DOS to VMS. There are similarities, but they are only superficial. Power managment in XP is flawless, and implemented FAR better than in even the latest mainstream Linux distros, which always seem to be two years behind the times in hardware support. (

Sadly, PM has never been very good in Linux, I think mostly because of the anti-MS bias - If you don't hang out at the Windows Hardware Developer Conference, how are you going to know enough about the PCxx standards (which, like it or not, *define* what a PC is and how it works) to write good PM code? Answer: You can't!

I have to say though, despite the fact that I really dislike some of Microsoft's business practices, I recently upgraded my primary desktop to XP, and it's *by far* the best desktop environment I've ever used (after expunging IE/OE for Mozilla). Like it or not, XP is a real OS, and particularly as a desktop (still 100% BSD/Linux for servers), it's the most stable and functional setup I've seen. (And this is with the low-budget XP Home, since I didn't really need the few extra features of Pro they charge another $100+ for.)

If I sound surprised, it's because I am - W2K was the first "real OS" from Redmond, and XP is a much bigger improvement on it than I expected. Now if they could just secure it, and would quit intentionally breaking things... (like for instance *every* (older) version of Visio - Grrrr)

Re:Only for embedded devices

[danheskett]

It isnt a big issue for "always up" servers, but for desktops that real users use - in offices and at home - people feel like its going back to "Windows 98". Its a minor issue, but RedHat 9 on a certain machine I've supported takes about 6 minutes to get to a graphical login prompt (and then of course the user logins and everything else loads etc) compared with 45 seconds on XP. If they use the hibernate feature instead of fully powering the machine off that time is reduced almost in half to about 25 seconds.

The bottom line is that for a user who boots the machine 7 times a week they will save nearly 30 minutes of waiting time during that week. Its a small thing but its the reason I get calls from my Linux users asking why "their PCs are way slower than the other users machines".

Re:Only for embedded devices

[Bert64]

But a multithreaded startup can also be better in some respects, if something hangs, then the startup continues anyway, instead of it hanging the whole machine and leaving you with no input methods

Re:Only for embedded devices

[Bert64]

AmigaOS was similarly fast, and for the same reasons, most of the OS is stored in ROM and it`s a fairly efficient and simplistic system. I guess RiscOS is still faster because its entirely rom based, whereas AmigaOS required a small program to be loaded from disk, which took a few seconds on a floppy.. but was instant from a ramdrive and very fast from a decent HD...
My A4000 boots almost instantly from a scsi hd..

Re:Only for embedded devices

[vasqzr]

Doesn't your laptop have a 'sleep' feature?

You really don't need to turn the thing off and then on every time.

Re:Only for embedded devices

[herrvinny]

I hear what you guys are saying, but the last time I tried to suspend the armada, Windows crashed :-(. Another time, the screen went all weird, it's hard to describe, but imagine the screen in front of you was divided into about 20-30 vertical bars, and every other bar was a lot lighter than it should be, or somehow "fuzzed up" the screen (wow, there's a technical word for you). I don't know whether it's a hardware or software conflict, and personally I don't want to waste the money to buy a newer Windows version, and neither do I want to switch to Linux unless it can offer me faster bootups, because Win 98 is holding so many settings hostage, and it would be a pain to reconfigure Linux with the same settings, similar programs, etc unless it can offer me faster bootups.

Re:Only for embedded devices

[Mr Z]

Are there really laptops for which this feature doesn't start mysteriously crapping out on you after the third time you use it?

Seriously. Every time I've gotten a new laptop (3 so far), I try out the suspend feature. It works the first 3 or 4 times I try it, and after that, it's a crapshoot. And, no, I don't load down my laptops with all sorts of useless crap. I do have corporate-mandated backup software and antivirus on it, though.

--Joe

Re:Only for embedded devices

[hesiod]

> Are there really laptops for which this feature doesn't start mysteriously crapping out on you

I have a Dell C610 laptop which has no problems like the one mentioned.

Re:Only for embedded devices

[Mr Z]

Power managment in XP is flawless,

That might be true for some desktop machines. However, as the annoyed user of a recent Dell laptop running WinXP, I disagree. I tried using suspend and hibernate on this laptop a few times. It worked the first couple times I tried it. Then, I started having weird problems, such as "no keyboard after resume", or "mouse and touchpad mysteriously die after unlocking" or "spontaneous reboot on resume."

(Another unrelated annoying things include all the crashes-at-shutdown in the ATA driver, and the fact that the laptop no longer sees the DVD-ROM/CD-RW drive anymore.)

and implemented FAR better than in even the latest mainstream Linux distros

No arguments there. :-) Honestly, I've never seen anyone get power management (at least 'suspend/restore') truly 'right.'

--Joe

Re:Only for embedded devices

[hesiod]

> the last time I tried to suspend the armada

The last time we tried to suspend the Armada, all our ships sank.

Re:Only for embedded devices

[Mr Z]

I have a Dell C640 running WindowsXP Professional that does.

Specifically, I encountered each of these problems before I gave up on hibernating my laptop and resuming later.

• Laptop hangs during suspend/hibernate, with a pretty, blank blue screen. (I use dark blue for my backdrop, and that's what it displays.)
• Laptop resumes, but I have no keyboard. I have mouse, though. Kinda hard to hit Ctrl-Alt-Del to log in.
• Laptop resumes, I log in, and then mouse/trackpad "go away" after a second. Gotta go to the Synaptics mouse control panel and initiate a "reset" to get them back.
• Laptop freezes during resume.
• Laptop starts to resume, decides to reboot instead. Too bad for that Word document I had open.

Of course, I use a lot of command prompts. Maybe that's it? (No, I'm not running old DOS apps. I'm using command-line utilities compiled with MinGW and Cygwin.)

--Joe

Re:Only for embedded devices

[ovit]

What I want to know is how Windows detected here that "Kernel32 is not responding"... Maybe Kernel32 is not the REAL kernel? :) Maybe their is a shodowy overlord kernel who's really running the show (perhaps a linux kernel?) :)

ovit

Re:Only for embedded devices

[Anonymous Coward]

There is no way a linux desktop machine should take anywhere near 7 minutes to boot up.

Take a look at what services are starting up. You should be able to disable most of them. Also check for any network connections you aren't using. Redhat has an annoying habit of hanging while waiting for a dhcp response for all network devices unless otherwise configured. After a little while it will give up and continue. Disable those unused network devices.

Sendmail can sometimes take a long time to load too. Most likely you can disable it.

After you trim off all of the unused services your startup should be back down to a minute or two.

As a nice side benefit, not running sendmail, ssh, rpc, apache, telnet, ftp and such will remove nearly all possible remote root exploits.

Re:Only for embedded devices

[blibbleblobble]

"Maybe Kernel32 is not the REAL kernel? :) Maybe their is a shodowy overlord kernel who's really running the show"

Ask the Master Controller Program

Re:Only for embedded devices

[ihopMaintenance]

Why would an embedded device need a firewall or virusscanner if it takes no input but from a certified source?

Read the parent post. The post I responded to was about 'something else' unless Compaq Armadas (a laptop) come with embedded Win98

Re:Only for embedded devices

[edxwelch]

But whats best about the changes MS made to Windows XP is the multi-threaded boot process

It's the *only* thing good about Windows XP. I see no other _useful_ features that weren't in Windows 2000... and if you are unfortunate enough to be running a SCSI disk under WinXP, the file system crawl, making that speed gain on the boot-up time largely redundant.

Re:Only for embedded devices

[ebricca]

you cant be really serious to think ms came up with their xp for good .. i think you're first quite soon in a lock in position if you follow their lead .. second they copied quite massive stuff as i know from bsd as ex in network handling and third as they have damn much money so it isnt remarkable they know more about the proprietary suspend mechanisms yet .. it seems as ms is on cullumination point and while with linux all directs upwards - while they are somewhat on the same height ..

Re:Only for embedded devices

[jesup]

XP is definitely better than say 98 in boot speed.

Back in '91 or so, an Amiga 3000 (25MHz 68030) could boot to gui (Workbench) and bring up TCP/IP & NFS and do NFS mounts in 7 seconds (warm boot; cold boot was ~12 seconds). And that was with 40MB 3600RPM slow-seeking drives, and a processor that was (even in raw MHz) 1/100 the speed of modern ones. Admittedly, it wasn't starting anywhere near as much, though it was bringing up the GUI, TCP, NFS, etc.

Re:Only for embedded devices

[dublin]

That might be true for some desktop machines. However, as the annoyed user of a recent Dell laptop running WinXP, I disagree. I tried using suspend and hibernate on this laptop a few times. It worked the first couple times I tried it. Then, I started having weird problems, such as "no keyboard after resume", or "mouse and touchpad mysteriously die after unlocking" or "spontaneous reboot on resume."

I don't know whether your Dell is an Inspiron (just a badge-engineered Compal or Quanta box, usually using a fairly standard Phoenix BIOS) or a Latitude (Dell's attempt to justify a 2-3X cost differential for the "enterprise" market.)

The Latitude BIOS is truly weird, and does things no BIOS should really *ever* do. If they are still using that same code, I expect they're having fits with it, since the people that wrote it and really understood it (mostly one guy) left Dell some time ago.

Here's how weird it is: I was Program Manager in charge of getting Win98 to run on Dell's laptops when it came out. ACPI support in W98 was completely broken, but it worked perfectly in NT, which unlike 9x, was developed in a reasonably professional way at MS. This problem was so severe that it threatened to keep us from shipping Win98 at launch, something that *absolutely must never happen* in Dell's world. (Whether it works for the customer is almost a secondary concern to letting Michael keep Bill happy.)

Our brilliant BIOS programmer (actually a Unix/Linux/BSD guy, by inclination and interest) came up with an idea after noticing that each OS made the initial ACPI call slightly differently: Hold off on setting up the BIOS for power management until that call is made, then re-write the BIOS tables for whichever OS is actually running. Brilliant in one way, doomed in another: we were able to ship at launch (and it actually worked!) and were the only OEM to do so for several months. This was dirtectly responsible for establishing Dell as a much more serious laptop player, as corporate customers *had to have* the latest and greatest MS OS, regardless of warts. But sooner or later, you have to pay the piper - If Microsofft has *ever* changed those bizarre little differences in the way 9x and NT-derived OSes make that initial ACPI call, ugly failure can be the only result. (I don't even want to think about the confusion something like VMware could cause...)

Check to make sure you're running the recommended BIOS for your machine and OS combination and DO NOT run other versions.

There's a reason I prefer ThinkPads and PowerBooks...

Re:Only for embedded devices

[dublin]

When you have enough consideration for others here to use correct grammar and capitalization (so that your post is at least readable), I'll respond to your arguments (which I are pretty weak, BTW.)

Really, though, couldn't you take another minute or two to make your post readable? Isn't the idea here to communicate? Or is your goal just to restate a politically correct position poplular among some \. readers?

Re:Only for embedded devices

[Mr Z]

Ok... ok... I'll rephrase: Are there really x86-based laptops for which his feature doesn't start mysteriously crapping out on you after the third time you use it? ;-)

It sounds like at least Apple's got it right, which is good. Unfortunately, the software I run for work on my laptop is all Windows based.

--Joe

Re:Only for embedded devices

[yomegaman]

On my Acer Travelmate LCi800 both standby and hibernate work pretty much every time. I think there's been fewer than a half-dozen times I had problems getting out of standby in the many months I've owned it. This is with XP Home.

Re:Only for embedded devices

[ebricca]

well i personally like to write everything in lowercase :) .. though you were right it was a bit unstructured (but arent we all a bit chaotic :)) as for ms (or m$) .. i think they are quite interested in a lock in situation as they stated (in)officially in the haloween documents (btw they act as quite every company would like) .. so a good reason to not use or pragate their products is to not get and let them get you in such a situation (on a small scale) as for your signiture .. is this a quotation of your own thought .. "GPL is a greater long-term threat" .. and then what threat ? from making money, freedom, "prosperty of the western world" ? and why .. i for myself consider the gpl as a good buisness model also in respect to exploitation of any kind btw my mother tongue is not english .. so i may make spelling mistakes :)

Re:Only for embedded devices

[ebricca]

(the comment box ate my return formatting before :| )

well i personally like to write everything in lowercase :) .. though you were right it was a bit unstructured (but arent we all a bit chaotic :))

as for ms (or m$ as you'd like) .. i think they are quite interested in a lock in situation as they stated (in)officially in the haloween documents (btw they act as quite every company would like)
.. so a good reason to not use or pragate their products is to not get and let them get you in such a situation (on a small scale)

as for your signiture .. is this a quotation of your own thought .. "GPL is a greater long-term threat"
.. and then what threat ? from making money, freedom, "prosperty of the western world" ? and why
.. i for myself consider the gpl as a good buisness model also in respect to exploitation of any kind

btw my mother tongue is not english .. so i may make spelling mistakes :)

as if

[commodoresloat]

any linux user wants to sacrifice their uptime to boot faster

Re:as if

[jellomizer]

Well just remember these are for imbedded devices. Where Uptime is not much of an issue. A lot of these things are running on battery, so if you get 3 days of uptime that is absolutely grate because the wonderful battery life. For these devices boot speed is very important imagine a Palm pilot having to boot up normal linux when you turn it on. Oh your batteries would be dead before sendmail realized that there is no network configured on the system.

Re:as if

[Ymerej]

Don't you hate it when you get moderated up. But there are no responses to your post.
There, there.

Re:as if

[robberbarron]

Uh, a Palm pilot doesn't boot when you turn it on. It merely comes out of sleep. The only time it boots is when you hit the reset button on the back. Then it takes about 5-10 seconds to boot.

However, there are a lot of embedded devices that do need to boot quickly. Automotive electronics like your radio, Nav-system, etc.. do boot up when you turn on the car, at least today they do.

Re:as if

[Erik Hensema]

Real men measure uptime as a percentage, not as an absolute value.

Re:as if

[Overly Critical Guy]

Laptops.

incredible

[alienhazard]

this is certianly incredible, but it is not yet available for x86 platforms. Do note, that this is not the boot sequence up till you get the login prompt, but just the initial loading of the kernel.

Re:incredible

[ysachlandil]

Why do you say this is not available for x86?

check out www.linuxbios.org for x86 details...

Re:incredible

[alienhazard]

according to the article, the developments that this is about are stated as having not been ported to x86 yet. RTFM

Re:incredible

[ysachlandil]

But then according to the website of linuxbios, they boot linux in 3 seconds, and that is because they have to wait for the harddisks to spin up.

Linuxbios is not related to this project, but then they also get the quick boot times.

RTFWebsite

NOT Only for embedded devices RTFA

[Anonymous Coward]

RTFA
"Yodaiken said the new fast boot technology also supports many Intel x86 boards"

Re:NOT Only for embedded devices RTFA

[mcspock]

Did you read the article? This was only in the context of EMBEDDED devices, and there are MANY Intel x86 embedded boards. These are very different from desktop linux, with the myriad of devices your standard PC has.

Re:NOT Only for embedded devices RTFA

[jonadab]

Nevertheless, it's clear in the article that we're only talking about
kernel boot time here (which is usually about five seconds). The
_other_ three hundred seconds your system spends booting (starting
all the services and stuff, then X, then your desktop environment,
then any apps) are unaffected.

Re:NOT Only for embedded devices RTFA

[linwoes]

There is another article at IBMs technolgy center that describes a method of parallelizing the init startup services. They don't give concrete numbers but it is possible to reduce this time substantialy also.

They (re)invented XIP?

[Dancin_Santa]

It works by loading the OS to RAMDisk from Flash... Sounds like eXecute In Place.

Not the most original thing in the world, but definitely necessary to keep Linux in step with other heavy embedded operating systems like WinCE and VxWorks.

Re:They (re)invented XIP?

[Vancouverite]

Whoops. Please try again. The article states:

That version boots in less than 200 mS, or 500 mS if it includes the loading of a RAMDisk from Flash memory.

Out of cheese error!! Please reboot universe!

Re:They (re)invented XIP?

[Dancin_Santa]

Read that marketspeak again.

They are only including the time from bootloader to kernel jump in the 200ms number. The actual boot time is 500ms. They still need to initialize the Flash as a bootable device and that is what takes up the 300ms. It certainly isn't anything to sniff at, but it's a little disingenuous to say 200ms when real world usage will show the lower speed.

Still, getting a system to kernel startup in under half a second is pretty decent for any bootloader.

kernel boot != complete system initialization

[Ankh]

Note that for embedded systems the main interest is how long it takes for the kernel to load, not how long it is before a multi-user server or workstation has a prompt that says "login" on a pretty X display.

So, this is a good improvement it seems, but shaves away 4.5 seconds or so out of maybe 30 sconds or over a minute for many people. Combined with the parallel init scripts work mentioned a few days ago,though, I'm guessing that Linux systems will be booting a lot faster with the major releases in 6 months to a year.

Re:kernel boot != complete system initialization

[Weaselmancer]

So, this is a good improvement it seems, but shaves away 4.5 seconds or so out of maybe 30 sconds or over a minute for many people.

True, but this is still great.

The article says, "less than 200 milliseconds (mS) from power-up to application code startup." The thing that makes this great is that not every device is going to go through the entire *nix init sequence.

How about a device like a Linux embedded router, or something like that? Just a kernel running and that's all. Or how about a dashboard mp3 player that only needs to run one application?

This makes Linux much more like customized firmware, and there are plenty of places to use that.

Granted, this'll be great when it makes it to the desktop, too. =)

Weaselmancer

200 mS?

[Shaklee39]

and boot times of under 100 mS 200 is too slow for me, I think I will wait for the 100 to be released.

Re:200 mS?

[Seehund]

Is 200 millisiemens (mS) fast or slow?

Oh, Timothy and the anonymous story submitter meant milliseconds, ms.

Why is it that not even alleged nerds can use units and prefixes properly these days? Measuring disk space in "mb", CPUs operate at clock frequencies of a certain number of "mhz", and their modems are of the "56 KBPS" kind...

Or is "S" another US-centric medieval unit, for measuring time? ;)

Re:200 mS?

[Russ Steffen]

Actually, the "S" unit refers to the Imperial Second, which is defined as the interval of time between now... and now.

Re:200 mS?

[frisket]

> "mS" (milli Siemens)

Surely this is what all the current spam is about curing?

The best cure for sea-sickness is to go and si under a tree -- Spike Milligan (RIP)

Re:200 mS?

[Thelxepeia]

At least the typo wasn't 200 MS (Microshafts). That would take most of the afternoon.

Re:200 mS?

[Surt]

Booting with a low conductance is neither fast nor slow but safer from viruses, of course.

Now I can get my pr0n in 200ms!!!!!!

[scosol]

Happy day!

Re:Now I can get my pr0n in 200ms!!!!!!

[Technician]

I hope you enjoy it in ASCII. This does not include a graphical interface. ;-)

Slightly off topic but about *nix boot times

[Zelet]

I have noticed that *nix boot times are noticibly longer than Windows XP boot times. I have never been able to figure out why this is - does anybody know?

Thanks
John

Re:Slightly off topic but about *nix boot times

[gordyf]

Linux starts its services before it brings up the password prompt. Windows loads, displays the login prompt and continues starting services in the background.

Re:Slightly off topic but about *nix boot times

[Zelet]

Could Linux theoretically do this? If so - why doesn't it?

Re:Slightly off topic but about *nix boot times

[Libor Vanek]

Windows XP are usually nothing more then OS+GUI. But Linux usually boots all kind of services also.

Re:Slightly off topic but about *nix boot times

[Libor Vanek]

There are some projects which are trying to modify rc scripts to boot on background (now it's sequential - after one is completely started up next one is started). Problem is dependency but I've seen some quite good solution to this. But I don't remember the name any more :-(

Re:Slightly off topic but about *nix boot times

[benjamindees]

You could try moving XDM/GDM/KDM to an earlier position in the init scripts. On my system, KDM is the last thing to start. I don't know if this is because it is dependent on other things being started first or what.

Re:Slightly off topic but about *nix boot times

[madcow_ucsb]

Dunno, but at least it's consistant. Unlike windows where the boot time seems to double every couple months, even when you keep unused apps uninstalled, defrag and run registry cleaners and spyware detectors often and kill all unneeded services...

I'll be damned if I can figure out how to fix that without doing a monthly reinstall...

Re:Slightly off topic but about *nix boot times

[TD_3G]

Windows XP doesn't finish loading everything before you actually see the screen. Not sure about your system, but my XP get me to the welcome screen in maybe 15 second... the welcome screen then sits there for about another 5 seconds, then I finally get my login... after logging in it continue to load a lot of my device drivers which freeze the system up (not completely but make it slow to the point I can't use it) for about another 30 seconds. You could probably modify init to provide you with a login prompt before it starts running all the services... this way here you'd be able to login and use the system and the services would load up in the background while you were doing this. That's basically what XP does, realistically the time it takes me to get from loading to services start up in Linux is probably under 10 seconds... the services are what seem to take the most amount of time. They take about 20 seconds all around to load -- Depending on how many I have. My Server system boots up in about 15 seconds total from loading to login prompt... all services loaded.

Re:Slightly off topic but about *nix boot times

[localghost]

There are probably a lot of services starting that you don't need. KDE has a good SysV Init editor you can use that has an interface kind of like the Windows service editor. Just get rid of all the stuff you don't need, and you should find that your bootup time is reduced drastically.

Or if you want to start a lot of services, you could do it the way I do it. I've got a script I run as soon as I log in to start most services. While that's initializing all that stuff, I go to tty2 and start my X session. This, as another poster has previously pointed out, is how Windows does it.

Re:Slightly off topic but about *nix boot times

[exhilaration]

Could Linux theoretically do this? If so - why doesn't it?

cuz you want you want your server to be useful without having to log in

Re:Slightly off topic but about *nix boot times

[LucidityZero]

I have noticed that *nix boot times are noticibly longer than Windows XP boot times. I have never been able to figure out why this is - does anybody know?

Sure. Very simple.

The number of services normally started up in someone's standard Linux install are many more. My friends and I noticed this actually, and once making the services similiar to a Windows boot, I was able to boot quicker on my 1700+ laptop than my friend's 2.0ghz Centrino, which was running XP.

Re:Slightly off topic but about *nix boot times

[croddy]

that's true of any interpreted language (though lisp, and to some degree python, are pretty zippy), but I'm more than happy to trade 30 seconds of boot time for the ability to completely reprogram the init sequence. far better to trust system initialization to something I can read and change -- without even having to recompile it -- than to some black box binary :-)

Re:Slightly off topic but about *nix boot times

[be-fan]

Nope. WinXP boots tons of sevices in the background as well, like RPC servies, etc. Its just that WinXP is pretty optimized in the bootup department.

Re:Slightly off topic but about *nix boot times

[dspeyer]

With a few exceptions (such as starting the network) Linux services could be started in arbitrary parrellel order. I think that someone, somewhere, has dones this. The question is, would you really want it?

On windows, it boots up and quickly presents the creen in a finished-looking way, but the daemons launching take up so much CPU that the widgets are unresponsive. With intermittant splash screens disrupting focus. I've gotten in the habit of listening carefully for the hard drive to stop growling, and then a few more seconds just to be sure. I'd much rather it get it over with first, and show me the desktop when its actually ready for it.

Maybe if all daemons were +20 nice this could work tolerably, but you'd think windows would have tried that?

Re:Slightly off topic but about *nix boot times

[iabervon]

In my case just now, "/dev/hdb1 has gone 180 days without being checked; check forced". But normally my machine (which only starts things I want) takes less time than it takes Windows XP to wake up if you let it go to sleep (I don't think I've ever seen Windows XP boot).

Many distributions are slow because they check for new hardware, involving probing for a ton of stuff that requires I/O bus cycles. But if you happen to have Oracle, that's what will take all the time.

Re:Slightly off topic but about *nix boot times

[motorsabbath]

It's easy. Install something else. Anything else.

Re:Slightly off topic but about *nix boot times

[UniverseIsADoughnut]

There is still a world of differance. WinXP is fully usefull for me in about 30-35seconds from the time i hit the power button. This is on any of my computers. Linux for me and any other computer I have seen it on is in the minutes range to boot. I'm sure people have it boot fast, but that in no way seams to be normal. Also if your using some sort of tweak to get faster boots that doesn't count either since when you are compairing you have to work in terms of out of the box. Since if you tell a person linux boots in X seconds and they go to install it, thats what they will have.

Re:Slightly off topic but about *nix boot times

[gerardrj]

Yes, GNU/Linux could theoretically do this. But we're mostly missing/off the point of the article:

The article is talking about Linux boot times. As they mention, Linux boots in 5 seconds on most cases (embedded or on your system) from the boot loader to kernel initialization is quite fast.
The kernel (Linux) then starts loading application code (network services, security daemons, x windows, etc).

To your question: Parallelizing services initialization can dramatically improve time to log on screen (LOS). Apple did this around OS X 10.2 IIRC (went from sequential to concurrent service loading) and cut down the time to LOS dramatically.

You can probably (for example) load your network stack and local modular drivers at the same time. Then you can start all the services that depend on those: NTP, name server and mount any remote file systems. After that, most dependencies should be handled and the rest (smtp, http, ftp, telnet , ssh, etc) could all be started in the background while the system started in to multi-user mode and perhaps loaded X.

In an ideal world all those startups would be completed before any LOS is presented, but the LOS should probaly be delayed until the system has completed the load-in process.

Re:Slightly off topic but about *nix boot times

[drinkypoo]

My windows xp was like the above with 512MB of ram. When I upgraded to a gigabyte of ram, my boot time was cut in half, and it was ready for me to log in just as soon as the screen appeared. I loaded a bunch of crap which starts up taskbar icons and makes the system effectively unusable for about 15 seconds while all that crap starts up, but that's my own fault. Most widnows users don't feel a need to load three IMs, waste, a livejournal post agent, an antivirus program, a transparent window manager, a raid utility, a mixer app, a dvd manager, a cdrom emulator, and an X server.

Re:Slightly off topic but about *nix boot times

[swillden]

Unlike windows where the boot time seems to double every couple months, even when you keep unused apps uninstalled, defrag and run registry cleaners and spyware detectors often and kill all unneeded services...

This is true; any Windows gurus out there want to take a stab at explaining why this happens -- and maybe what can be done about it?

Why is it that Windows machines seem to get slower and slower over time? It's most pronounced on boot time, but it seems to me that there's some run-time performance as well. They also seem to get gradually less stable, although I only see that on desktop machines where lots of software gets installed and removed. Server stability seems to be more consistent, although boot time still degrades.

What gives? Most people I know just assume that you *must* reinstall your workstation every couple of years, and it's a good idea to reinstall annually. Just part of system maintenance, like rebooting weekly, or even daily. There's got to be an underlying reason, though, and if there is, it should be possible to fix it.

Re:Slightly off topic but about *nix boot times

[Kashif Shaikh]

In other words, it loads the kitchen sink AFTER you login...

Re:Slightly off topic but about *nix boot times

[yerricde]

Then present a login prompt on the first console while starting all the daemons on the second console.

Re:Slightly off topic but about *nix boot times

[Zelet]

I've never built a kernel. I wasn't complaining either - I was asking a technical question and got a smart-ass answer from you. I wish I could mod you down... damn.

Re:Slightly off topic but about *nix boot times

[binary paladin]

Eh. I disagree. Win2k and Linux boot about as fast on this machine including loading X. Both are liveable but neither are BeOS... that was a fast booting OS.

Re:Slightly off topic but about *nix boot times

[_Splat]

It could, but did you ever notice that windows is basically useless for a while after you're given 'control' of the desktop after startup or login. That's why.

Re:Slightly off topic but about *nix boot times

[koali]

The IBM solution used make. You sure know about make, don't you? :-b

Re:Slightly off topic but about *nix boot times

[swillden]

It was supposed to be funny. I thought it was, anyway. Don't be so touchy.

Re:Slightly off topic but about *nix boot times

[GundyRage]

Now that you mention it, I realize that I subconsciously draw a marquee (I think that's what it's called) on the desktop after I'm presented with a desktop just to test if the system is really ready to go. When I get good redraws on the marquee, I know the system is ready for business.

Also, did you ever notice how the desktop gets drawn once and then again after a few more seconds? You press start - the menu comes up and then it disappears while being drawn for the second time - now you click and it works. I've always hated that. Let's not copy that implementation. - G

Re:Slightly off topic but about *nix boot times

[cptgrudge]

This is true; any Windows gurus out there want to take a stab at explaining why this happens -- and maybe what can be done about it?

Hmm. I guess I don't consider myself a guru, but I know more than any of my personal friends.

Although this does still seem to exist somewhat, I think it's taken a real drop since Win2k and XP. I'm a sysadmin for a large Win2k network, and from what I can see, the servers stay happy until they are taken down.

We had the most problems when people were on 95/98. With the ability to install every type of software you can imagine (not my call, I was a grunt back then) the machines grind to an agonizing crawl. After a wipe and reinstall, they become speedy quick. Perhaps it would still happen on Win2k and XP, but I don't let users install software now. The machines do fine.

I really think that the root of all those problems is shitty software. Seems that most commercial software you install nowadays wants a bit of your screen and processor real-estate by installing all sorts of icons and background programs. Quicktime? Realplayer? Winzip?!? Wtf?

Check where programs run on startup. And that means registry digging too. Check HKLM\Software\Microsoft\Windows\CurrentVersion\Run . Check for hooks into explorer.exe in the registry, though that usually means your box is compromised. Legacy programs can still put stuff in win.ini and system.ini. Check the Windows Start Menu startup folder.

Check running services in the MMC snap-in. Run services.msc to see them. See something you don't recognize? Track it down. Look at what processes are running for clues. See a process that you don't recognize? Find out what it is. I use Process Explorer from Sysinternals, a nice little free app. That running process that doesn't need to be there is stealing from you. Stealing your cycles and time.

I probably forgot some things, but there are lots of places to "startup" things in Windows. And be careful. One wrong keystroke with this stuff and you can kill your system.

My machine that I use for work is solid, has been for years. I will usually get uptimes measured in weeks with no problems. Stupid security patches.

Which prompts the question:

[Atario]

...what about a utility to set your uptime to an arbitrary value?

Re:Which prompts the question:

[aardvarkjoe]

-- /usr/local/bin/uptime --

#!/bin/sh
echo "42 years!"

There ya go.

Re:Which prompts the question:

[Anonymous Crowhead]

alias uptime="echo '5:33pm up 22342352324 days, 6:28, 2124315623 users, load average: 2432.40, 12312.31, 123123.19'"

Re:Which prompts the question:

[Wakkow]

Load average of 123123.19?! Holy crap. I'd rather it look like this:

5:33pm up 22342352324 days, 6:28, 2124315623 users, load average: 0.01, 0.00, 0.00

The uber-server.

Re:Which prompts the question:

[YOU ARE SO SUED!]

For the record, your load averages won't go above 1024.

It's been 17 seconds since you hit 'reply'! It's been 16 seconds since you hit 'reply'! It's been 15 seconds since you hit 'reply'! It's been 14 seconds since you hit 'reply'! It's been 13 seconds since you hit 'reply'!

Re:Which prompts the question:

[elvum]

5:33pm up 22342352324 days, 6:28, 0 users, load average: 0.01, 0.00, 0.00

users are bad, mmkay?

Seems pretty simple to me

[MerlynEmrys67]

Rather than keeping a kernel that gets loaded at boot time...

Keep the image that the kernel creates AFTER boot - simply load that into memory and restart.

That said - you still need the long boot the first time, and after any hardware changes. Also, I am guessing to get it into the sub second range - hard drives are right out as well - and all of the silly boot managers. But for an embedded device - who cares

Re:Seems pretty simple to me

[Dahamma]

> Keep the image that the kernel creates AFTER boot - simply load that into memory and restart.

The kernel also does a lot of hardware initialization, etc. This might work to some degree on a soft reset, but you still need to EXECUTE a lot of the kernel startup code when powering on.

Re:Seems pretty simple to me

[MerlynEmrys67]

The kernel also does a lot of hardware initialization, etc. This might work to some degree on a soft reset, but you still need to EXECUTE a lot of the kernel startup code when powering on.

The joys of hot plugging and module insertion... Driver init code actually doesn't take much. Perform a hot swap event on the hardware when you restore... probably where most of the 200 mS gets spent

Re:Time To Abandon Windows?

[spectecjr]

Sure looks like it. XP takes 30 seconds to start up, and another 30 before you can run any apps without them locking up the system.

You go girl! Abandon your XP Desktop and have fun checking your email on your... toaster.

This, if you hadn't noticed, is for Embedded Linux.

Re:Time To Abandon Windows?

[MerlynEmrys67]

Funny, I have a functioning login within 10 seconds of pushing the ON button - and never had problems after logging in... So I guess you must be running on an old Pentium or something (compare and contrast your startup numbers numbers with Linux on the same system). Oh, and I am not talking about Standby either... Don't like the battery drain on my laptop

Booting Linux Faster through Blocking

[Eponymous Cowboy]

Although this article refers to embedded systems, the earlier Booting Linux Faster article contained an overlooked post by TornSheetMetal, who had a great idea on how to make Linux, or any operating system start up faster on any system.

Simply run every startup script simultaneously, but have each script block until its dependencies have started. Nothing waits longer than it needs to, and there is no need for additional complex systems to check and manage dependencies.

This is VERY easy to do with daemontools and svok (both written by D.J. Bernstein, the author of qmail). Switch over and you'll never go back.

Re:Booting Linux Faster through Blocking

[spacefrog]

Hehehehehehe. We used to use a trick very similar on the Amiga way back when. "LoadWB" in your startup-sequence would actually load and display the Amiga WorkBench (similar to Finder, Explorer, whatever).

You would do your basic initialization (establish your path, your assigns, etc.) then run "LoadWB" and *then* load all your other crap.

End result....System would boot and be "useable" in a couple of seconds, even on an 8Mhz system.

Re:Booting Linux Faster through Blocking

[Anonymous Coward]

This is VERY easy to do

Um...yeah.

I visited your links and it all immediately became blindingly clear to me...

daemontools and svok...

Yep.

Re:Booting Linux Faster through Blocking

[targo]

imply run every startup script simultaneously, but have each script block until its dependencies have started.

Btw, this is pretty much what Windows XP does, that's how it achieves a much better boot time compared to earlier Windows versions.

Re:Booting Linux Faster through Blocking

[FooBarWidget]

Havoc Pennington said he already tried that before, but the end result was *slower*, not faster. Basically, it's because of disk seeking time. Harddisks are slow. Simultanous loading will result in a seek storm which makes everything load noticably slower. The same is true for applications: if I try to load Galeon and Mozilla Mail at the same time, the total time is higher than when I load Galeon, and then Mozilla Mail.

Re:Booting Linux Faster through Blocking

[Anonymous Coward]

That's why it is necessary to do what Windows XP does: each time it boots, it records the order of sector accesses. Then, when it is later idle, it reorganizes those sectors to optimize them for the next boot--a much smarter form of the traditional "defragmentation." The result is a recursive iteration to the optimal sector organization, and the fastest possible boot. (Incidentally, it also does this while loading each individual application, which is why applications start so much faster under XP than 2000.)

Re:Booting Linux Faster through Blocking

[Ziviyr]

That was a very naturally occuring "trick". Amiga doesn't have much beyond the Workbench to load on startup. And its not wholly smart to autorun stuff before that.

Its just a simpler system.

It'd be interesting to see if Linux can reach that sort of tightness someday. Right now it seems to be moving the other way though.

Re:Booting Linux Faster through Blocking

[yerricde]

parallel startup scripts isn't that of an improvement on single processor systems.

So you really think it's fun waiting fifteen seconds to get an IP address? Do you think that modern network card drivers have to busy-wait to get IP addresses? No, they block until the NIC sends an interrupt when receiving a packet from the DHCP server.

If you 'make' the daemons start in parallel, you 'make' the computer boot faster.

Re:Booting Linux Faster through Blocking

[Bert64]

95 used to take ages on a p233, but 3.11 started almost instantly

Re:Booting Linux Faster through Blocking

[Bert64]

My Amiga used to take ages to boot, because of all the crap i had running.. i ended up using a multithreaded boot process, and it seemed slower, so eventually i cached all the boot files before loading them, this was much faster, since the machine had a lot of ram, by amiga standards anyway, and the disk was pretty slow.

Re:Booting Linux Faster through Blocking

[michaelhood]

I use daemon tools. It lets me mount up the warez ISOs..^W^W

Re:Booting Linux Faster through Blocking

[Bert64]

Spacefrog ? spacefrog.demon.co.uk user from ircnet #worms/#amiga ?

Re:Booting Linux Faster through Blocking

[Bert64]

I liked the original sourcemage, aka sourceror, which had just a simple single script for startup, no fancy init stuff...
all i had in mine, was:
ifconfig eth0 bla bla bla /usr/bin/iptables-script
route add default bla /usr/sbin/sshd &

and it was to a login prompt pretty much instantly once the kernel had initialized, the system this story talks about could cut the kernel init time down, so now all we need is to speed up the bios post.

Re:Booting Linux Faster through Blocking

[mst76]

> 95 used to take ages on a p233, but 3.11 started almost instantly

A cleanly installed win95 system was very fast on a P233 with sufficient memory. You may have forgotten it but the P233-mmx wasn't even introduced until 1996, the Pentium II-233 followed in 1997. Booting was (and still is) slow if you installed virus scanners, ms-office, and other autostart items, and if you used networking. Sadly, more and more application writers feel that their application is so important that is should be started on every boot.

Another Windows optimization

[MerlynEmrys67]

Windows does some very intersting things with both optimizing the location of sectors on the hard drive and loading drivers

For the hard drive - rather than put executables down 1-n on the hard drive - Windows (for many years) figures out the load order of sectors of the executable - and fragments them across the sectors in that order - net effect +10-50% load time boost from using the hard drive effectively

For drivers - there is this really interesting way that windows is now initializing driver loading by putting them into the kernel image itself... Kind of like taking modules in Linux - and rather than having the overhead of loading the module each time you boot - insert it into the kernel - and letting the kernel load (with a "static" module in now) - This one is a little trickier to put into a Linux environment... what does the GPL say if I have a loadable module - yet the kernel now statically links it in as an optimization... I don't even want to go there

Re:Another Windows optimization

[iabervon]

If you have a binary module and you statically link it to the kernel, you can't distribute the result. In this situation, though, you don't want to distribute it, so you're fine. Of course, you could also arrange so that the module is not linked to the kernel but just contained in the kernel image (and linked at runtime, as if it had just been loaded into memory off of the disk), at which point it's "mere aggregation" and permitted for distribution by the GPL.

For that matter, if you're exclusively an end user, you don't even have to accept the terms of the GPL, so it doesn't matter what it says.

As for whether the vendor can distribute the module that you would do this with, it's fine as long as you don't have to statically link it. They can't be blamed for how you link it, and you can't be blamed under the GPL if you're only an end user, so everybody's happy.

Re:Another Windows optimization

[swillden]

If you have a binary module and you statically link it to the kernel, you can't distribute the result. In this situation, though, you don't want to distribute it, so you're fine.

In the case of developers of embedded systems, they generally do want to distribute the result, inside a device.

However, in most embedded devices, the time required to load a module from media isn't relevant, since all of the code is probably in flash RAM anyway, so they just run the code from where it is. As a result, the cost of loading a module is insignificant.

It's also worth pointing out that the GPL does not distinguish between static and dynamic linking; it just talks about "linking". This means that distribution of binary-only modules probably technically infringes on the kernel developers' copyrights. However, Linus has taken the pragmatic stance that allowing binary-only modules is a reasonable compromise that serves to widen the range of hardware that can be supported. So, since it's good for Linux, no one makes an issue of it. It would be interesting to see what would happen if some kernel contributor decided to get sticky about this point. Odds are that unless he had contributed really significant amounts of code, his code would simply be rewritten to make the issue go away again.

OTOH, when Linux achieves sufficient market presence that hardware manufacturers *must* have Linux support to be competitive, we may see that policy decision change. Kernel developers accept binary-only modules, but they don't really like them (c.f. the whole kernel tainting thing).

Some might worry that Linus might suddenly change his mind and start suing vendors who've provided binary-only drivers, but given that Linus has publicly stated that binary-only drivers are okay (and no one else in the kernel development community has objected), it's unlikely that a court would do anything more than just order the vendors to stop distribution (or start providing source, their choice).

Re:Another Windows optimization

[Scarblac]

You mean if Linux is changed to hijack a binary module and swallow it into the kernel without the driver's permission?

And enforce the GPL on it?

No, just use it. That's allowed (that's always allowed). If you do that, you're not allowed to distribute the result anymore though, by the GPL, and probably by the license of whatever proprietary module we're talking about.

Remember, copyright law always allows using things, it just restricts copying and distribution.

Re:Another Windows optimization

[scrytch]

> what does the GPL say if I have a loadable module - yet the kernel now statically links it in as an optimization... I don't even want to go there

If it writes out a static kernel with a non-GPL driver linked in, you simply have a kernel you can't distribute under the GPL, not a "GPL-tainted" kernel.

Secondly, Linux has a specific exemption to the GPL for closed modules.

Re:Oh well

[binarybum]

Huh? please show your work.

ps. the sticker is 4000 not 40000.

I'll be impressed

[commodoresloat]

when linux boots 200 mS before I turn the computer on.

Great.

[Executive Override]

Not only it almost never crashes, but when it does you barely even notice it! :-]

Re:Congratulations

[MerlynEmrys67]

Actually what you are talking about is not very effective in terms of daily usage. The time necessary to save off the memory may be longer than acceptable.

Then you go off and talk about Hibernate. Suspend simply stops the CPU and keeps the RAM hot... Allowing the CPU to come back where it left off. Hibernate actually writes it off to hard drive.

Now in an embedded environment - I can have a HOT OS on a fast flash that I can execute from. Why do I store the kernel there - just store a suspended kernel - and restart it

100 mS?

[pesc]

But what does electrical conductance have to do with boot times? 100 mS is 100 milliSiemens. Milliseconds is abbreviated ms.

And for the desktop, there is always LinuxBIOS

[drfreak]

check out www.linuxbios.org. Too bad I don't have any motherboards with a supported BIOS. It sounds way cool. Kind of like turning a legacy PC into a modern embedded device. heh

Re:Time To Abandon Windows?

[Penguinshit]

But do you have a functioning SYSTEM?

XP-login != fully-booted

Re:Wait a second, or rather, 200ms

[Penguinshit]

Linux doesn't crash. It "creatively parks".

They removed the offending code:

[cybermace5]

// Code to make Linux startup look like it's doing something
// Insert in a few hundred random places
long foo = 0;
while (foo < 10000000)
{
foo++;
}

Re:They removed the offending code:

[Anonymous Coward]

And this is why Linux requires a 32 bit processor. It just hangs here on 16 bit procs.

Re:They removed the offending code:

[praxim]

Hey, hold up there! You don't want a nasty letter from SCO's lawyers, do you?

Re:They removed the offending code:

[Soul-Burn666]

That will just make compile time longer.
Since foo isn't volatile, the compiler will notice that code does nothing and will remove it ;)

Re:They removed the offending code:

[maxmg]

That's only because somebody got scared that Microsoft could sue them for IP violations.

Re:They removed the offending code:

[thunderbird46]

mod this guy up, my sides are splitting from laughing!

Re:They removed the offending code:

[IainHere]

foo++;

OMG! That's just disgusting; imagine using a post-increment operator when there's no need for a temporary.

I'm off to cleanse my mind with another read of Sutter's GOTW.

eh?

[Anonymous Coward]

"...speeds up system boot time to less than 200ms... The techniques are processor independent"

Could anyone think about the 386SX/20 please???

Re:eh?

[TeknoHog]

Could anyone think about the 386SX/20 please???

I couldn't think of it, but my Beowulf AI of 386SX/16s just imagined it.

No, it is not that simple at all.

[seanadams.com]

Go RTFA - several points: You don't just simply "load an image into memory" and have a running system. This is why properly supporting APM is difficult on any machine. All your hardware needs to be reinitialized. Network connections need to be reestablished (getting IP and so on), file systems need to be remounted, there are all kinds of timer-driven things that need special handling, and so on and so forth.

What these guys are doing is optimizing for embedded systems - where the kernel is hardwired for exactly the same hardware every time. You don't need to probe, and you don't need to guess what state the hardware is in - it's a closed system and it's the same at every power-on. Furthermore there are all kinds of things you can initialize simultaneously when you can optimimize for a deterministic environment - if your video system wants a moment to do a POST, you can spend that time initializing a network interface, for example.

Also, the definition of "boot time" for this dicussion is the time until the first application-level code runs. That's something like only 1/3 to 1/2 of the boot time for a typical linux server or desktop that you're thinking of. Most of the time is spent bringing up userland services and loading the graphical environment. There's a big savings on big workstation in flushing RAM to disk, but not so much for small embedded systems, where application state is very minimal (eg a Tivo, or a wireless router).

Re:No, it is not that simple at all.

[Mattsson]

Something that's been bugging me for years is that computers and OS's that's capable of saving state to disk and then turn off won't let me create a state-image of my newly booted environment and then always load it at power on instead. (Wah! That was a long one.)
I've been wondering about this since I got my first IBM Thinkpad around 1998.
It could hibernate to disk and then power off.
This was done to a separate partition and was independent of OS.
To "boot" again took the time to read 24 MB from the harddrive. (It had 24 MB ram)
But even though I can now do exactly the same thing in windows, I still can't keep a hibernationfile for fast booting.
Is this possible in Linux? I've searched for it, but haven't been able to find anything.

Misleading...

[DroopyStonx]

What a misleading article.

Could've at least put non x86 or embedded device in the title.

Re:Misleading...

[Krach42]

Not nearly as misleading as "College Freshman builds Fusion Reactor" :P Here I thought it was a scientific breakthrough from some kid thinking outside the box, and I find out it's just a reduplication of something someone else has already done, and documented.

Maybe "College Freshman replicates Fusion Reactor" or something like that, at least then I wouldn't think he invented it.

Would make a great commercial

[niko9]

Scene one, Act One

Adrian Lamo in a orange prison jump suit conferencing wit his attorney.

Ring Ring

Attorney: Hold on Adrian, I have to take this call. (talking into cellphone) Yeah.. ok... Great! ok..Thanks!

Great news Adrian!

Adrian: (curious/happy): What?!

Attorney: I just saved 4.8 seconds on my Linux boot time with FSMLabs!

Adrian: :(

Re:Time To Abandon Windows?

[MerlynEmrys67]

Yes... my system works very nicely after I log in as well. Quite fully booted.

Are you making the HUGE mistake of shutting down and rebooting the system when you hit the power button ?

What a waste - hibernate works well on laptops - suspend (which is even faster) works well for desktops

why is it ...

[sl0ppy]

... that when a company doesn't put its kernel changes out immediately, there's calls for hanging them for violating the GPL, but when a linux company optimizes boot-up routines in the kernel, nobody is asking when the patches are going to be making it into the mainline kernel?

from the looks of the article, FSMLabs has been basically profiling the kernel, looking for sticking points, and optimizing them.

why wouldn't at least some of this work be contributed back to the mainline kernel? it is modifications on a GPL'd kernel, after all.

Re:why is it ...

[seanadams.com]

They only have to distribute their source to whomever they distribute a binary, if and when they do so. Under the GPL you do not bear the burden of publishing, distributing, and supporting source changes that you made for your own use, or changes which you have not yet distributed.

Re:why is it ...

[sl0ppy]

They only have to distribute their source to whomever they distribute a binary

and this is different than linksys how?

this isn't meant to be a troll, but i'm really not seeing how this is different.

FSMLabs sells RTLinux, a real-time version of linux. i'm sure that they are doing their duty and distributing the source of the RTLinux kernel. wouldn't their additions to the kernel become GPL, and be able to be integrated into the main-line kernel?

Re:why is it ...

[seanadams.com]

and this is different than linksys how?

When I buy a linksys router I am gettting linux binaries running on the system.

FSMLabs sells RTLinux, a real-time version of linux. i'm sure that they are doing their duty and distributing the source of the RTLinux kernel. wouldn't their additions to the kernel become GPL, and be able to be integrated into the main-line kernel?

At this point there is no reason to doubt that when they release these changes in the form of a product, we'll get source.

Re:why is it ...

[sl0ppy]

At this point there is no reason to doubt that when they release these changes in the form of a product, we'll get source.

exactly! there seems to be a lot of confusion as to what this entails. reading the article, it looks as if this is pointed optimizations.

the company releases its run-time for free, under the GPL license, and probably (i haven't looked, i'm giving them the benefit of the doubt, i'm lazy) the source to their changes as well.

at the time of my post, there was a plethora of "this is only for embedded systems" posts. my points were as follows:

o the article talks about finding places to fix, and fixing them

o these are modifications to the kernel, and should make it to mainline

Re:why is it ...

[seanadams.com]

exactly!

Exactly what? Do you agree with me or are you just very confused?

I would suggest that you read the GPL. It makes no distinction between software for PCs versus software for embedded systems. It just says you have to ditribute the source to whomever you distribute a binary. What is confusing about that?

Re:why is it ...

[sl0ppy]

i state again:

at the time of my post, there was a plethora of "this is only for embedded systems"

at that point, there were three reasons why people seemed to think that this would not help the desktop:

o skipping hardware initialization
o the rtlinux kernel is somehow radically different than the mainline kernel
o FSMLabs wouldn't give back the changes

i pointed out that the article stated that many of the optimizations were from what appeared to be profiling, and that people seemed to be missing that this was a modification to the kernel, that the changes would (or should) go back into the mainline kernel, or at least become patches against the mainline kernel by someone who gets the GPL'd source.

i see no point in my original post where i state that these wouldn't/shouldn't make it into the mainline kernel, are you sure you read it correctly?

Re:why is it ...

[TheLink]

What if they don't distribute the binary but distribute a saved memory image of a running kernel?

That's not distributing binaries in many people's dictionaries.

Re:why is it ...

[seanadams.com]

Really you had two questions: "why haven't they released this", and "why won't this go into the mainline kernel". Two totally different questions. I addressed the first. We can discuss the second if/when the code is released. :)

Re:why is it ...

[sl0ppy]

nah, i know why they haven't released it (or at least i have a pretty good idea). my proof-of-concept code is usually tersely written and not tested under all conditions, and i'm assuming theirs is as well. i'm sure they'll release it when it's all cleaned up :)

my first part wasn't "why haven't they released this", but more a jibe at people that seem to think that releasing code is super-fast.

i've tried to go thru legal to get supposedly "proprietary" code released to GPL (even if it's a requirement of the license that legal already accepted). my thought is that we should be more patient, rather than attacking linksys, or apply that rage evenly, and complain in this case as well.

Re:why is it ...

[sl0ppy]

if it runs as a native, compiled image, it's a binary.

even if it's a saved memory image, it still has an entry and exit point, and is still in a form that the machine executes, it's still a binary :)

Re:why is it ...

[nathanh]

... that when a company doesn't put its kernel changes out immediately, there's calls for hanging them for violating the GPL, but when a linux company optimizes boot-up routines in the kernel, nobody is asking when the patches are going to be making it into the mainline kernel?

Because the GPL only guarantees your right to source code if you first receive the modified binaries. You can't demand the source code "immediately" once the change has been made. FSM could conceivably keep this project in-house forever without releasing the source code. However as soon as they distribute binaries they must provide the source code.

Re:No, it is not that simple at all. - but it is

[MerlynEmrys67]

I did read TFA - And guess what - you have a few edge cases to consider... But reinitializing hardware in a running system is rather easy - especially in a closed system environment. Would this work for the desktop - probably not with a rearchitecture of the driver interfaces... but nothing stops it. I mean what is the difference between this and the Windows suspend feature, especially when you are using Flash to load your running image off of.

Not for me.

[MarcQuadra]

I just booted my 500MHz laptop in 20 seconds in Linux, that's from GRUB to login. Booting Windows 2000 on it takes about 80 seconds from 'press F8' to 'Login'

Also, when the linux box says I'm good to login, it is, everything is already loaded. The windows box sits there for a while AFTER I login to sort itself out and let me to the desktop. I think it's because *nix INIT process wants to load services first and let users in later, Windows is the other way around.

Good news. Now my software can be more buggy.

[stienman]

I had a professor tell us this story of one of his previous coworkers:

She had designed and implemented a simple service on top of unix which was accessed by a moderate number of users. When the time to put it into production came, she looked at her remaining few crashing bugs and determined to put in a monitoring loop that would reboot the server if such a situation happened. She also determined that no data loss would occur.

Why did she do this workaround, and how did she determine what bugs she could leave in?

She had a 5 digit company phone extension. She determined that someone could call her, if she let her phone ring twice, in a short period of time. During this time the server would have finished rebooting and start serving again. She could answer the call and simply say, "Try it again", whereupon the user would find that his operation worked this time.

So remember - if your server can reboot itself (and does so automatically and safely) before they can finish dialing tech support, you have no worries!

-Adam

With XP, outer rim.

[phorm]

Try instead setting an "automatic login" to a default user... count your time period up to the point where everything has stopped loading (your hard-drive will stop ticking, heh) and all your tray icons etc etc are in place. Significant difference between that and booting just to the initial user-selection/login screen

For linux, you can try a few hard-disk tricks. Also, your filesystem might be slow... try reiser as it seems to run quite nicely. For swap partitions, put them at the end of your hard disk. Just before the swap partitions, if you have any always-loading speed-required files, try making a partition there where you can stick 'em.

There are lots of things you could tweak... being able to stick "/lib/modules" or "/usr/bin" at the end of a hard-disk could actually save you some time if you're loading a lot of stuff from there on bootup, as files on the outer rim of the HDD would load faster...

Re:With XP, outer rim.

[gordyf]

The significant difference between booting to login-screen and booting to the desktop comes from all the apps set to load for a user, such as aim, icq, msn, some virus scanners (most should load as a service), volume control apps, video driver helpers, not to mention explorer itself! Ever used a person's computer and seen 95371 systray apps? They (well, most) load on login, not on boot.

Re:With XP, outer rim.

[Nogami_Saeko]

Part of the windows problem seems to be that startup applications and services all try and execute in parallel, thus thrashing the hell out of the drive and slowing the machine a whole lot.

Why not force (or allow the user to set) a limit on the number of startup applications that execute simultaneously to avoid thrashing the drive and making the startup sluggish.

Re:With XP, outer rim.

[nivedita]

The outer rim of the hard drive (the faster part) is usually at the logical beginning of the drive, not the end. Also, a large portion of the time required to load things, especially small executables, is seek time, not read. Putting things at the beginning of the drive improves disk throughput for those files, but if they are small, that is irrelevant. Frequently accessed files should be closer to the center of the drive to minimize seeks (of course, if they're small and frequently accessed, they're probably sitting in RAM anyway, so this won't apply). For the same reason, it may be better to keep your swap (if you actually ever hit it) in the middle of the drive.

Re:With XP, outer rim.

[phorm]

The beginning of a drive is the RIM? So this would be pretty much the reverse of a CD-ROM then?

Seems I've been somewhat mislead previously.

Re:Wait a second, or rather, 200ms

[caillon]

Don't forget about users of laptops and other portable devices who occasionally do turn their machines off.

I think this needs to be updated

[spir0]

the heading paragraph should be updated to point out that this is for embedded devices. now we have morons thinking an ide drive can boot an OS in 200ms.

Re:Summary of preceding posts

[WindBourne]

Linus Torvalds' dad can beat up Bill Gates' dad.
While I am not sure about Linus's father, I am quite certain that Linus's wife could :)

My stupid question for XP or Linux desktop...

[grolschie]

Ok, I am not trying to be a retard in asking this. I am curious. Why the normal boot process, start-up many different services, wait, wait, wait, etc? Why not take a memory dump/snapshot of system after the first boot and load that into memory mega fast at every boot? Boot time would be as quick as loading the image into RAM. Or has this been done?

Re:My stupid question for XP or Linux desktop...

[vadim_t]

Just wouldn't be much faster. I have a dual system. I'm pretty sure the CPU is not the bottleneck. It's the hard disk.

Loading my RAM (1GB of it) from disk is probably going to take longer than booting the old fashioned way. In any case, whatever you read from the disk can't be much smaller than the sum of all the executables

Re:My stupid question for XP or Linux desktop...

[kulpinator]

See the previous similar question. The answer by seanadams.com is very helpful.

And it's not a stupid question...very reasonable, if you assume it can be done.

Re:My stupid question for XP or Linux desktop...

[grolschie]

wrong. I don't want to restore to previous state, but initial state of a clean boot. Otherwise system could be restored with memory leaks and the aftermath of program crashes.

WTF?!?!

[monkeyboy87]

Why do we care how long it takes to boot something that can have an uptime measured in months and years??!?!?

oh wait. these are embbeded systems for things security, monitoring equipment etc. yeah i can see reboot times being critical.

Linux != init.d

[Voivod]

This is a great example of where nit picking "Linux the kernel" vs "Linux the operating system" is right on.

"Linux" does NOT "start its services before it brings up the password prompt". Redhat does that. SuSE does that. That is a limitation of the init.d system used by some distributions, and has nothing to do with this work that FSMLabs has announced.

If you slim down a 2.4.x kernel a bit and run it with a minimal setup like BusyBox it is very easy to get a login prompt within 5 seconds of the BIOS screen beginning to load Linux. Getting a 2.4.x kernel to boot much faster than this involves kernel hacking. Fortunately, the GPL requires that FSMLabs release this source code to their customers. As an embedded systems developer, I hope that some day we'll get a "boot quickly" option in the generic kernel as well. ;-)

200ms?! That's forever!!

[Ignorant Aardvark]

I can boot up DR-DOS in an emulator in a few milliseconds, not 200 of 'em! And keep in mind, the emulator just adds overhead. If I somehow managed to install DR-DOS on my tripped-out Athlon XP 3000+ system, it'd boot in even LESS time!

Re:Wait a second, or rather, 200ms

[pixas]

sometimes accidents do happend.

(yes the picture is fake but you all know what I mean by it)

Login Prompt vs. Useful System

[billstewart]

Yes, Windows does a login prompt earlier in its boot sequence than Linux usually does, but just because you've typed in your login and password doesn't mean you've got a usable system, especially on Windows. After you do that, it spends a while loading lots of System Tray applications, running whatever it runs to get pretty icons on your desktop, and also running whatever equivalents to .profile are set up on your network login as well as running your Startup folder apps. Moving the request for a password earlier doesn't really help much.

Also, with Linux, you could log in to a shell or console window before starting X, rather than waiting for X and XDCMP to put up a graphical login window, but you'd then have to wait for X to start after you boot, and most people don't seem to want that for the main user interface.

That's "milliSlashdottings"

[billstewart]

100ms is when your machine is 10% of the way to totally hammered by network connections.

Windows can do something else though...

[fmaxwell]

Sure, Linux can now boot in 200ms, but Windows can blue-screen 200ms before you can save your work. So it sounds like the Linux folks put all of their efforts into quickly beginning productive work while Microsoft focused on very quickly ending it.

Re:No, it is not that simple at all. - but it is

[damiam]

But reinitializing hardware in a running system is rather easy

Go tell that to the kernel swsusp developers. Or, better yet, since it's so trivial, why don't you just go implement it?

Re:Wait a second, or rather, 200ms

[niko9]

Linux doesn't crash. It "creatively parks".

Yeah! You sure showed him you rocket scient.. I mean, aerospace propulsion engineer!

XP = Newer, Faster Hardware

[billstewart]

One difference is that you may still be running Linux on that "cutting edge when you bought it" P2-233, but you're unlikely to be running XP on anything less than 800 MHz or so, plus you've got a bigger faster disk drive and more memory. Sure, you could do an apples-to-apples comparison, but the big reason that Win2k Pro boots so fast on my new work laptop is that it's an 1100MHz box. (Another reason is that Suspend/Resume finally works reliably enough that I don't need to reboot more than every week or so :-)

Massive Potential

[555-5555]

THis has the potential for a Linux based PVR that is the problem using a windows machine for a pvr is the mad boot times

Re:Massive Potential

[beerman2k]

Linux basd PVR? Oh, you must be reffering to the TiVo.

Re:Time To Abandon Windows?

[TheLink]

What does suspend for desktops do in event of power failure?

I'd rather leave it on, on UPS and when there's a power failure it shuts down systematically. When the power resumes, it starts back up.

Re:Why bother anyway?

[trippinonbsd]

For servers a big draw back is that if you have to patch the kernel (or update it) you must reboot, but if the acual reboot only takes less than a second, users might hardly notice the downtime (you could reboot and keep a 99.99% uptime).

Oh, you're killing me

[The Tyro]

but I must say my geek hat is off to you, sir... I would have never come up with that one.

Kudos.

And everyone knows...

[Kashif Shaikh]

...Windows might boot faster, but as we all know windows has D.S.S. capabilities which means "Delayed Service Startup":)

In other words, it loads everything AFTER you login, no joke;)

Re:And everyone knows...

[Des Herriott]

That matches my experience. I get a Windows 2000 login dialog much faster than I get a Red Hat 9 login dialog on my laptop, but the first 2 or 3 minutes of using Windows involves blinking hourglasses and a constantly churning hard drive. I'd say the total time from power-on to usability is roughly the same between Windows and Linux.

Re:And everyone knows...

[matvei]

That's pretty much what you get if you use KDE/GNOME. It seems to take about the same time to log into a KDE environment than it takes for the computer to boot up to the display manager's login screen.

Re:And everyone knows...

[Chester K]

...Windows might boot faster, but as we all know windows has D.S.S. capabilities which means "Delayed Service Startup":)

In other words, it loads everything AFTER you login, no joke;)

Let's see.... I've never heard of this before, I know it to be false, and The Hive Mind has absolutely no references to it.

Care to try to salvage your credibility?

Re:And everyone knows...

[RupW]

In other words, it loads everything AFTER you login, no joke;)

No, it doesn't - it offers you the login prompt whilst it's still loading services. But it does finish loading most / all the services before it'll give you a desktop.

If you left it for a minute or so and *then* logged in it'll go straight in.

Re:And everyone knows...

[RupW]

the first 2 or 3 minutes of using Windows involves blinking hourglasses and a constantly churning hard drive

You almost certainly need more RAM. One of our servers did this and we found it was loading ~1Gb of services into swap. Now we have a motherboard that takes more RAM it gets a desktop in 20 secs or so.

Re:And everyone knows...

[TrippyZ]

He was probably thinking of 'delayed group policy application' which effectively shortens 'time to desktop' but screws up the admin of our network.

TZ

Re:And everyone knows...

[Kashif Shaikh]

There was a smiley at the end of my sentence after my phrase...

But if you look at it,

- It executes the all the programs in the "RunOnce" registry key
- Windows connects to all your SMB shares after you login
- It loads all the common internet apps after you login(i.e. messenger takes a long time to startup and connect). There should be a common backend server spawning a new messenger interface for each user instance
- it reads a 2-5megabyte user profile file at startup
- when you logout: all your processes are destroyed(when you logout in linux the background processes or processes not attached to a tty are left intact).

BTW, for what its worth I have a stock Windows 2000 Server on my Desktop and it takes a really loooooong time to boot and startup. Moreso than RH9.0 with "Everything Installed".

Kashif

Re:And everyone knows...

[Chester K]

- It executes the all the programs in the "RunOnce" registry key
- Windows connects to all your SMB shares after you login
- It loads all the common internet apps after you login(i.e. messenger takes a long time to startup and connect). There should be a common backend server spawning a new messenger interface for each user instance
- it reads a 2-5megabyte user profile file at startup

That's effectively your .profile under Windows, though. Those aren't "services", they're "applications" -- they can't load prior to login because which ones load depends on who's logging in. My account may have different SMB shares than your login; and I might have MSN Messenger disabled on my account, while you don't, for instance.

- when you logout: all your processes are destroyed(when you logout in linux the background processes or processes not attached to a tty are left intact).

Windows doesn't kill a process if you start it unattached from your session with the "at" command -- so its functionality is strikingly similar to Linux in this regard. If you set /interactive, the app may terminate when you log out if it hasn't been written to correctly handle the logout message -- Windows doesn't kill it, it kills itself incorrectly.

BTW, for what its worth I have a stock Windows 2000 Server on my Desktop and it takes a really loooooong time to boot and startup. Moreso than RH9.0 with "Everything Installed".

I may be mistaken on this point, but I don't think Microsoft started touting shorter bootup times until Windows XP.

The Power Companies

[Kujah]

It's too bad the damn power companies havn't spent some time improving their boot times... it took them like a week to restore my power *grumble*

Want a faster boot? "Make" it so.

[yerricde]

Problem is dependency but I've seen some quite good solution to this.

If you want to make some daemons run, and some daemons depend on others, why not use a GNU tool specially designed to handle dependencies? Make the initscript into a makefile, and GNU Make should handle starting daemons quite nicely.

RTF?!?!

[Sebby]

Did you?

Re:aSlightly off topic but about *nix boot times

[zambuka]

I don't see this problem at all. about 30 seconds power on to useable system (with all systray items loaded).

But then I have done some tweeking with the system.

The biggest killer is web browser caches, hundreds of small files loaded and deleted on a regular basis. Also the temp directory tends to get a lot of workout, particularly when you install and remove programs regularly.

This causes a lot of fragmentation which then causes the swap file to fragment which then cascades into worse fragmentation for any new files. (at least this is my personal experience).

the solution is simple, well... kinda.

1. Put your swap file on a seperate partition or make sure it is a fixed size and located at the begining of the drive. An alternative is to put the swap file on a partition on a different drive.

2. Redirect the temp directory and browser cache to a seperate partition or better yet a seperate drive. Even if this is the only thing you do you will notice a long term performance boost.

In other words pretty much the same kind of thing you do when seting up a linux system, dividing the drive into different partitions for different purposes.

occasionally cleaning out the registry with regclean can't hurt but as yet I have seen no real benifit from doing so.

With these steps I can now happily run my win box for as long as I like, well almost. In the last three months the only thing that has taken down my system was a confilct between my tv tuner software(a little dodgy to start with) and Simcity4 that took out the sound driver.

My Linux box only takes about 20 second but it is an old AMD K6 400 and takes about 10 seconds just to get through the bios and I don't load X, just use it for an experimental web server. Something went wrong with the hardware on this box so it won't run windows at all, not a real loss since It forced me into upgrading to a decent system and gave me something to experiment with.

Cheers

Zambuka

Faster booting! I have to...

[glenebob]

I for one welcome our new, more rapidly booting overlords.

Re:Faster booting! I have to...

[binary paladin]

In Soviet Russia the Linux box rapidly reboots you!

I wish I could attach a picture of a burly Russian dude with a furry hat and a red star pointing at out at whoever is reading this. Kind of like a communist version of Uncle Sam with some cool 50s era propoganda look to it. Any artists out there want to immortalize the much overused joke?

Oh yea?

[waspleg]

how about those guys at NASA; i bet they'd be REAL interested in getting a failed shuttle/sattelite system to reboot in less than a second

one obvious example; i'm sure there are others

hello! This doesn't affect you!

[GISGEOLOGYGEEK]

Won't someone here get a clue! This has nothing to do with a server or desktop computer! ... so give up all the bragging about how this beats windows, and all the boasting that you haven't booted in 4 years so it doesnt matter...

This doesn't affect you.

Re:I want the opposite

[bockman]

You would need something to preload the most heavy KDE shared libraries at boot time and keep them in memory.

Some slow-to-start program (e.g. galeon) can be started in 'daemon mode' to speed up the GUI start-up response (most of the initialization time is done by the daemon at boot time).
But it should not be difficult to make a program that just ldopen() a bunch of shared libraries and then stays alive (dunno if it will be swapped out, however).

If there is some KDE program that does not display anything but still uses most of the KDE libraries, it could be started as a service with a fake account, just to keep those libraries in memory.

make -j

[Julian Morrison]

Seeks storms aren't inevitable doing stuff in parrallel. The Kernel's quite capable of handling a modest amount of simultaneous disk accessing. It's only doing too much at once in parralel that leads to "thrashing".

Perhaps a solution would be the equivalent to "make -j", where you can tune how many simultaneous things to run. In fact "make" is a good model for this whole approach, since the control mechanism will also need to do dependency-blocking.

Other refinements that occur to me:

- Things could be marked as "light", "medium", or "heavy", and the "weight" of simultaneous running processes kept constant.

- The control process could be adaptive, tracking the "load average" and altering the "weight" of individual processes to seek a best aggregate run time.

Re:make -j

[eyeye]

interesting comment, thanks.
I dont get mod points anymore so this will have to do :-)

Re:make -j

[julesh]

It's only doing too much at once in parralel that leads to "thrashing".

Sorry to nitpick, but thrashing isn't really what we're talking about here. Thrashing is the state a system gets into when in order to do any useful work more stuff needs to be in memory than will actually fit there, so the system is instead almost constantly waiting for pages to be fetched from swap space.

We're just talking about bad optimisation of disk accesses.

You have a good point, though, and that kind of system could certainly work fairly well.

Other suggestions that come to mind:

- the system might want to try out different start orders on different runs and time things to see what works better, so startups would gradually improve (eg maybe starting a process which does a lot of disk access at the same time as mounting NFS filesystems which uses the network connection is better than doing two disk-intensive startups at once).
- Other parameters could be tuned by an automatic feedback process too, like the amount of parallelism.
- A kernel option could be provided that gives a process lower priority access to the disk system for the first few seconds of operation, meaning that the higher priority process would effectively have to finish using the disk before the second one would be allowed to. This might or might not require substantial changes; I don't know the Linux disk scheduling code at all, I only know about how such things work in theoretical terms.

Where is it?

[Kynde]

Where's the source?

In my perspective Yodaiken comes with a rather questionable track record of toying around with linux only commercially.

Sort of depends the details...

[macshit]

This is a timed version of my kernel boot sequence on an RTE-V850E/ME2-CB board (a rather pokey processor -- 80MHz); the first column is seconds:

[ 0.002619 ] Linux version 2.4.21-uc0 (miles@mcspd15) (gcc version 2.9-v850ice-000414-nmit-20010327) #62 Wed Jul 16 16:03:57 JST 2003
[ 0.009299 ] On node 0 totalpages: 8192
[ 0.019597 ] zone(0): 8192 pages.
[ 0.030390 ] zone(1): 0 pages.
[ 0.030635 ] zone(2): 0 pages.
[ 0.030891 ] CPU: NEC V850E/ME2
[ 0.031065 ] Platform: Midas lab RTE-V850E/ME2-CB
[ 0.031322 ] Kernel command line:
[ 0.031869 ] 50 BogoMIPS (precomputed)
[ 0.067024 ] Memory: 24388K/32768K available (291K kernel code, 150K data)
[ 0.068884 ] Dentry cache hash table entries: 4096 (order: 3, 32768 bytes)
[ 0.069639 ] Inode cache hash table entries: 2048 (order: 2, 16384 bytes)
[ 0.070279 ] Mount cache hash table entries: 512 (order: 0, 4096 bytes)
[ 0.071467 ] Buffer-cache hash table entries: 1024 (order: 0, 4096 bytes)
[ 0.072234 ] Page-cache hash table entries: 8192 (order: 3, 32768 bytes)
[ 0.073991 ] POSIX conformance testing by UNIFIX
[ 0.074414 ] Linux NET4.0 for Linux 2.4
[ 0.074663 ] Based upon Swansea University Computer Society NET3.039
[ 0.075648 ] Starting kswapd
[ 0.078020 ] Serial driver version 5.05c (2001-07-08) with no serial options enabled
[ 0.078538 ] ttyS00 at 0xfe08000 (irq = 90) is a 16550A
[ 0.079150 ] Blkmem copyright 1998,1999 D. Jeff Dionne
[ 0.079349 ] Blkmem copyright 1998 Kenneth Albanowski
[ 0.079544 ] Blkmem 1 disk images:
[ 0.079889 ] 0: 876000-FCE7FF [VIRTUAL 876000-FCE7FF] (RW)
[ 0.084282 ] VFS: Mounted root (romfs filesystem) readonly.
[ 0.085781 ] Freeing unused kernel memory: 20K freed

Whoo, 80ms!

Not that useful though (no network devices; network devices seem to take forever to start)...

Re:Why bother anyway?

[BrokenHalo]

My kernel load is a lot quicker than the POST and BIOS loading, which takes > 45 seconds.

Re:Good news. Now my software can be more buggy.

[Spit]

So remember - if your server can reboot itself (and does so automatically and safely) before they can finish dialing tech support, you have no worries!

Why reboot the whole server when you can restart the service? Only morons resort to rebooting in the first instance, and only an ass dragging incompetent would go to so much trouble to work around implementing a real solution.

Re:Congratulations

[vidarh]

"fast flash" is an oxymoron. Flash is slow slow slow. You most certainly don't want to execute code straight from flash unless your system is ridiculously slow anyway.

Just copying a kernel or a suspend image from flash will give a quite noticeable delay.

And take a look at swsusp - restarting a suspended kernel is NOT trivial. You need to reinitialise hardware, some of which may not allow you to read back their state (graphics cards being a common culprit) so that you need to know what state they were in on suspend.

Re:aSlightly off topic but about *nix boot times

[stray]

yo zambuka

thanks for the suggestions.. i usually stick to linux workstations, but i do have windows installed for graphics apps and games, so hints to tweak the system and install stuff properly are very welcome...

just a question.. how can i "redirect" the temp directory? i sometimes tried to make those link-thingies windows has, but obviously they don't behave the same way in windows than softlink in linux do.. i had applications compain that the target path (temp = link to other directory on other partition) was not a directory. and indeed, if you examine them closely, they just seem to be regular files with a string pointing to the target..

how do you do it?

thanks
stray

Re:Good news. Now my software can be more buggy.

[myom]

My guess is that this story has changed its exact form when passing between people, and that it restarts the service. I think few servers would restart in the short time where her two ring signal wait would make a difference. Welcome to the real world; sometimes workarounds are necessary, when you are working funded by someone elses money - not in your parents' basement.

As OS for embedded devices...

[SharpFang]

...that are usually shut down by pressing the power button, it doesn't have th have any exit point :) As for entry point - just have all the bits in CPU, RAM and the rest set as frozen at arbitrary operation point and start ticking the clock. Not a normal entry point either. But I agree, it's still a binary. Even a core dump is a binary :)

Re:MilliSECONDS

[dosius]

Well one good thing, I think they're the only company that released an IBM-compatible CP/M 4.1 (Personal CP/M-86 .2.0/4) that actually felt like CP/M. (DOSPLUS 1.2, which came with the Amstrad PC1512, felt like MS-DOS. But you could run a CP/M 4.1 CCP on it.)

-uso.

Uhhh, milli Siemens?

[sdijkstra]

"milli Siemens"?

Is that some sort of miniturized German electronics manufacturer?

I always thought that mS stood for "milli Sievert", a unit for ionizing radiation...

Re:Uhhh, milli Siemens?

[haschka]

Siemens is the european equivalent for mho

Re:Uhhh, milli Siemens?

[julesh]

Siemens is the european equivalent for mho

Huh? Siemens are SI (that is, international standard) units, there's nothing European about them other than their name :-)

I'd never heard of 'mhos' before today, although having looked it up in a suitable reference it is defined as an 'archaic' term for siemens...

More relaxed approach to writing software

[tgt]

Before: let's make it a separate process, and if it crashes - who cares, start another one !
After: come on, just reboot the box, nobody even notices !

Re: pr0n in 200 mS

[TeknoHog]

BTW, mS stands for milli-Siemens. Siemens is the European equivalent of mho (inverse ohm). Goes quite well with pr0n I guess *splat*

Exit point in embedded device software

[yerricde]

that are usually shut down by pressing the power button, it doesn't have th have any exit point

If the power is software controlled, such as on any modern TV, VCR, or computer, the "exit point" is where the program writes to the register that shuts off the power supply. This happens a few milliseconds after the user presses the power key on the front panel.

Some flash looks more like disk than like ROM

[yerricde]

since all of the code is probably in flash RAM anyway, so they just run the code from where it is.

Not always. Hooking the flash device up directly to the memory controller works well with fast parallel flash devices such as Game Boy Advance Game Paks, but some flash devices, such as MMC flash, are slower and serial and are treated more like disk than like ROM. (In those devices, bootstrap is performed by a small mask or OTP ROM device.) Even CF looks like an ATA disk.

Re:Some flash looks more like disk than like ROM

[swillden]

Thanks.

I was speaking based on the flash-using embedded devices I've developed for, but I wouldn't claim to have extensive knowledge of the thechnology.

Is the main difference the I/O channel?

Re:Some flash looks more like disk than like ROM

[yerricde]

Is the main difference the I/O channel?

Yes, that, and the speed of the memory itself. If the flash memory is accessed only at boot time through a copy to RAM, the flash chip doesn't need to be as fast, and slow flash memory is less expensive.

Excellent, now I have even more time to...

[gatkinso]

...sit around and wait for driver support.

Re:Why bother anyway?

[bluGill]

Microsoft has fixed that problem. A PC Must show the windows XP logo in 6 seconds, or something like that. Of course that is nothing coompared to linux booting completely in 200ms, but a good start compared to the full minute BIOS takes on my PC. Details can be downloaded at http://www.microsoft.com/whdc/winlogo/downloads.ms px. (I didn't like it cause microsoft changes links often, and I can only find a word document there, which I can't open so I'm not ever positive I have the details right)

I'm told that linux bios can get your system up fast too, but I don't know much about that project.

Key difference between PCs and embedded

[yerricde]

It makes no distinction between software for PCs versus software for embedded systems. It just says you have to ditribute the source to whomever you distribute a binary.

The essential difference here is that unlike embedded systems, PCs almost always store their programs on an easily rewritable magnetic disk. But say I am writing software for an embedded device, and the software is stored in ROM. How would the GPL apply to this case?

Or say I am writing software for an embedded device that cannot run binaries that have not been signed by the device maker. The GPL specifies that "source code" includes linker scripts, and in the case of signed-only devices, these linker scripts reference a private key, which is the subject of a non-disclosure agreement with the device maker. I'm guessing that it would be impossible to distribute GPL software in this case, right?

Bernstein needs to release a UNIX distribution

[emil]

This guy has written so much stuff, and all of it is unusable by most major Linux/BSD distributions because of his licensing.

His software has about the same reputation as OpenBSD for security, but it is extremely difficult to configure.

A preconfigured distribution, maintained by Bernstein himself, would get his software in the hands of many more people.

Re:Bernstein needs to release a UNIX distribution

[Bob Uhl]

Rather, he needs to get his act together and release his software under the GPL. Then we users can take the good and toss out the bad (slashpackage, anyone?).

Re:Bernstein needs to release a UNIX distribution

[RobbieW]

Bob, with all due respect, have even read Dan's short explanation of Software User's Rights? You have the right to modify Dan Bernstein's software as you see fit. If you don't want to use a particular feature, don't. If you want to install qmail into the /etc directory, go right ahead. If you want to use /daemons instead of /service you are free to do so.

What you don't have permission to do is distribute changed versions of Dan's software without his express permission. You can distribute a patch that makes the software act differently or install differently, but you can't modify the software and then distribute the modified version.

Personally, I'm extremely thankful I have DJBDNS, QMAIL, etc. because they make my life easier, not harder like you seem to imply.

Pray for the peace of Jerusalem!

Re:Bernstein needs to release a UNIX distribution

[Bob Uhl]

Yes, and I've read the many materials at the FSF--and I find the latter far more reasonable. I trust software which I can modify, and others can modify, without having to fall back on kludges like distributing patches far more than software which is completely controlled by one man.

I like freedom. There's also the fact that I disagree with several of djb's technical decisions--and the fact that his software is not truly free means that if I wished to use it as a base for my own, I'd have to maintain a parallel set of patches and update them every time he updated the code. The GPL is far more user-friendly, as is the BSDL.

Re:Why bother anyway?

[BrokenHalo]

I've often wondered about Microsoft's OSs booting so quickly. I can put up with it by simply not running Windows, and never rebooting Linux except when I build a new kernel.

But I guess if it's a choice of being able to boot in no time at all while being aware that the machine can blue-screen in about the same interval, I'll take the longer boot-time...

What have we become ...

[fygment]

... when a wait of 4-5 seconds is too long?

Re:What have we become ...

[Mike Buddha]

Impatient?

BeOS

[Potent]

Anyone remember how fast the early releases of BeOS booted (around the non-BeBox PPC era)?

That was awesome!

--

How it is different

[nuggz]

...and this is different than linksys how?

Apparently nobody in possession of the binaries has been refused the full source code.

Pfft ... this is easy:

[Greedo]

Edit lilo.conf, and change

delay = 50

to

delay = 0

What about

[bplipschitz]

putting / on a solid-state HD? Granted, it has to have constant power [even when the machine is off], but might be one way for things to start more quickly.

Kinda pricey, though. . .

Suspend vs Hibernate

[arete]

First of all, most people in this thread. are talking about suspend/standby, not hibernate. Hibernate is a no-power mode ('cept the clock) which writes all RAM data to disk, while standby is a low-power mode (no display, disk, etc.) that just basically waits. Hibernate is historically somewhat less reliable, although either will destroy open network connections. Both modes work quite well in Win2k+, although I'd certainly be scared in 95/98/ME.

On most laptops I've seen, you can set the bios to send standby, hibernate, or nothing to the OS when you close the case. Any effect can be used with the case open by selecting it from the Start-> Shutdown menu.

On my 233Mhz Thinkpad 600 (which is a wonderful if aged machine, but which isn't technically supported by Windows 2000 which is running it) I have case-close mapped to standby and the power-button mapped to hibernate. I also have autohibernate on low battery. So if I close it it goes to standby and from standby it'll shutdown if it gets in danger of running out of power to stay going. (Standby does _not_ last indefinitely on battery)

I've had Windows 2000 fail to recover on hibernate, but not very often and that machine had other issues. I've _thought_ standby was failing on this laptop because the response time to the wakeup button-press varies and isn't particularly fast. It's occasionally woken up without the mouse, but the mouse is finicky to start with (again, W2k doesn't properly support the onboard mouse) Sleeping and waking again has always fixed the problem, all applications still running.

All of that said, however, I certainly recommend saving everything important before sleeping, hibernating, changing power sources, or transporting the machine (even in case-closed but staying on mode) All of those things are somewhat more risky than doing nothing, but less risky than some applications.

Re:Suspend vs Hibernate

[Mr Z]

You're right, I've conflated suspend with hibernate.

I've had problems with both suspend and hibernate. I've got my laptop set to suspend-on-close, and I've tried the old "close the lid, walk from meeting room A to meeting room B, open the lid" trick a couple times, and have met with mixed success. I've lost keyboard, had spontaneous reboots during resume, etc.

Hibernate is monotonically worse than suspend. Unfortunately, hibernate is what I really want most of the time.

Now I forget: Is there a suspend->hibernate transition that kicks in at some point just based on time (not battery)? If so, then maybe all my problems really are with hibernate.

--Joe

Re:Suspend vs Hibernate

[arete]

In Windows 2000, at least, you must first enable hibernate. This all happens in Start->Settings-> Control Panels-> Power Options. With hibernate enabled, you can select separate idle times for:
monitor off, disk off, standby, hibernate (separate for battery or AC, and also arrangeable in profiles) You can ALSO select what to do on a battery alarm (for which you set the battery left) a critical battery alarm (again configurable) and laptop close, power button, and sleep button.

So, in short, yes, you can make it do pretty much whatever you want. Mine, at least, will hibernate on critical battery alarm, even if it's already in standby.

Right

[greygent]

Like X11... Gnome... the GNOME apps, no wait, that's Linux...

MacOS X

[arete]

MacOS X is by far the best desktop environment _I've_ ever used, and it's far better than XP. It does tend to be more expensive, of course.

It does almost everything you want it to, and it does it automatically, and they're constantly building new features in smooth ways.

It also has one of the largest available software bases around: It runs MacOSX software, MacOS7/8/9 software, linux/BSD software compiled for PPC (X11 isn't installed by default, but the OpenOffice.org installer smoothly includes it, for instance, and it's on the OSX CD) and has an available emulator (VPC) that allows you to run x86 in a window, including Windows or Linux.

Even they aren't perfect, but they're closer than anybody else. Oh, and "Mail" rocks, hard.

Re:MacOS X

[dublin]

MacOS X is by far the best desktop environment _I've_ ever used, and it's far better than XP. It does tend to be more expensive, of course.


Like they say, you get what you pay for. I love OS X, too, but I don't think it's markedly superior to XP for most desktop use. It's a better gee-whiz media machine, but as far as getting real work out the door, no, it's really no better, and although the software situation is getting much better, there are still at least two orders of magnitude more quality software for XP than for OS X. Visio alone is enough reason for me to stay with a Windows desktop, and has been the main thing keeping me in MS-land since 1999. (And you can't share local Mac printers with PCs in OS X, either. (At least not without stupid and very un-Mac-like Samba hacking. Grrrrr.)


Even they aren't perfect, but they're closer than anybody else. Oh, and "Mail" rocks, hard.

Ugh. The Jaguar Mail program is a helicopter crash (those are *always* ugly...) It doesn't have half the features Netscape/Mozilla Mail has, and it seems riddled with bugs that prevent correct interoperablity with Windows users. My Dad uses Jaguar, and I'm about to move him off Mail and onto Mozilla Mail just to get Microsoft attachments working cleanly with PC mail users. (AppleCare's answer to Mail's refusal to properly tag an attachment as say, application/ms-word (or whatever it is...) instead of application/octet-stream was, "Use another Mail program - that's just the way Mail is for some users..." This is not an acceptable answer in today's world, where MS is dominant, whether Apple likes it or not.)

I've also spent several hours rebuilding (by hand, the sed/awk way) mailboxes botched by that stupid program. (At least it gets points for using a standard and fixable mailbox format, unlike the Microsoft mail tools.) I'm glad you're happy with it, but IMO, Mail is among the worst MUAs I've ever seen, and a huge dark smudge on the image of the Mac as a "just works" computer...

So complicated..

[MasTRE]

Yep -- both headline and post should have read "ms" (milliseconds) rather than "mS" (milli Siemens); thanks to all the alert readers.

Wouldn't it have been simpler to just say 1/5th and 1/10th of a second? Try to drive the point home in simpler terms when possible - otherwise general geekness will start to be compared to political correctness. This is not the same as Intel switching to 90 nanometers instead of 0.09 micron - that's actually a good idea, so we don't end up with 0.9e-7 micron. But anything over 100 ms can/should be expressed as a good ol' easy-to-understand FRACTION of a second.

Think, therefore be!

P.S. When I first tried to post this on 030929 11:11 PM EDT, /. was broken and wouldn't accept my post!

Re:POST

[julesh]

POST checks are for wimps! ;-)