Linus Says No To Annoying Boot Messages

Pants Ripper writes: "In a victory for all Linux users, Linus Torvalds declared jihad on annoying 'informational' kernel boot messages. I'm sure we'll all miss the inspirational 'spewtron driver 0.09 installed (C)2000 by Wardwick Extrusion' messages in our dmesgs." I've always thought those messages looked pretty interestingly verbose, but people want pretty boot-ups. And this Linus guy seems to know a lot about this "Lee-nuks," too.

Comments

Re:So let me get this straight

[Anonymous Coward]

To be honest, the need for all those messages is useless anyway. Like the gent said, you need to go though your logs to see them anyway. They go by way too fast. If they really was useful, they'd stick an alert after you logged on, saying "Hey! Something goofed! - Check your logs"

If we need pretty logon screens, I hope (and probably will be vindicated) that we have a) JPG login screens, instead of that bloated BMP format, and b) we can change them *easly* instead of having to change registers, or other dangerious things so MS can have their ugly face on every computer.

Hey, who wouldn't want a hot burnette in a skimpy bikini (or a muscular stud in a g-string) saying "Welcome back " when they reboot after 2 years? :)

Way to go, Linus!!!

Why not make money off of boot messages?

[Anonymous Coward]

Okay, it's obvious that the Great Ship Linux is going down. Between VA's decision to *ahem* get out of the hardware business and all the IPOs tanking, it's time to expand the paradigm. Why not keep the boot messages, and sell ads?

For instance, instead of getting messages in your inbox about mortgage, it would display while the kernel boots up! We could also make it fsck each time, to increase the impact of the display.

Also, with SVGA mode, we could have graphics.

And therefore make money off of the only profitable internet business--porn.

Each time you connect to the internet, Linux could unobtrusively download a new set of ads for various services and websites. It's about the only way Redhat and VA are going to stick around.

Re:Why not make money off of boot messages?

[codejnki]

Advertisers might not be too keen on that idea. I reboot my linux box maybe once a month. Therefore the ad impact would be minimal.
----
"War doesn't determine who's right, just who's left"

Re:Why not make money off of boot messages?

[Da Web Guru]

So, would that mean I would have to sit and wait the same amount of time it takes for Windows to boot? It's like DVD's that force you to view trailers before the actual movie...

Besides, as usual, we would try to find a way around the ads while still getting paid for them.

Anyway, I doubt that they would be real interested in buying ads on my workstation...

admin@sys3:~$ uptime
3:53pm up 213 days, 7:56, 1 user, load average: 0.06, 0.02, 0.00

Re:Why not make money off of boot messages?

[JonWan]

>And therefore make money off of the only profitable internet business--porn.
Man talk about Uptimes Going Down. ;-)

Re:Good == Good Bad == Failed

[Ranger Rick]

Those aren't kernel messages, those are init messages. Linus has no (direct :) control over those.

What a great idea

[mosch]

This idea is almost as good as when Linus decided that Linux shouldn't have a kernel debugger, that real men use printf. Now if your machine is booting and it just hangs, you'll know that it happened during the initialization of the... oh wait, you'll have no idea when it happened unless you're a "real kernel hacker" thus making it harder for normal people to figure out what to search the mailing lists for.

If the suggestion was to make the statements display only display if in a special verbose booting mode, that might be reasonable. Killing all messages, always, that's just dumb.

--

Re:What a great idea

[mosch]

No, normal linux users don't use kernel debuggers. Normal linux users are almost identical to windows "power users", which is to say they're fairly competant at using other people's software. However, there's a large body of professional programmers who are extremely comfortable with debuggers, and know how to quickly diagnose problems with them, so that they can either fix the problems, or provide meaningful bug reports.

Secondly, removing informational messages provide an important function, in that they give you status updates as to the progress of the boot process. Thus, if the machine crashes due to a bug, where there is no error message, you have a clue where to start looking.

--

Re:What a great idea

[mattdm]

If you read the article, you'd know which it was.

But in any case, Linus doesn't think that real men use printf() (or printk(), as the case may be). Real men grok the code.

Re:What a great idea

[binner]

The best solution in my opinion, is to just make it a compile time option!

If you like (I do) leave them on, if you don't like them, grab The Linux Progress Patch, and forget that boot up messages ever existed!

It's a win/win!

-Ben

printk

[mosch]

ah yes, printk, which internally uses a 1K buffer and doesn't check for overruns. Apparently dynamic allocation is a new concept to Linus.

The whole "real men grok the code" idea is just arrogant and stupid. If your goal is to hack on the kernel, then yes you need to understand it deeply. However, if you just want to figure out why your linux 2.4.5 machine just panic'd after loopback mounting an iso image, then all you really need is a kernel debugger and some basic software engineering skills.

--

Re:printk

[mosch]

what the fuck are you talking about? haven't you ever heard of DE-allocating memory?

my point about the limitations of printk is simply that it's NOT a debugging tool, and linus shouldn't pretend that it is.

the linux kernel is far from intuitive, it's kind of amazing that it works when you read the code, let alone that it works well. telling the developers that they should be allowed access to NO debugging tools other than a 1024 character print buffer just makes people's lives unneccessarily difficult.

Sure, it's fine if all you want to do is hack kernel code, but it sucks ass if you want to hack userland code, but need to fix a glitch in the kernel first.

--

Re:printk

[ttfkam]

FYI: Allocating on the stack is usually more efficient in time and space than allocating memory on the heap. The speed difference in allocating a static 1K buffer and a dynamic, variable size buffer outweighs the size or flexibility issues while in the kernel.

If a kernel coder can't keep there messages under 1K, you've got worse problems than a rigid API. If you want type safety, you should be using C++ and compile-time type traits. If you want raw speed for an internal kernel API in C, you expect a tradeoff in the lack of bounds checking (which has a non-trivial overhead in large quantities).

And his reasons for resisting the kernel debugger were not so cavalier as "real men grok the code." His assertion was that a proper design should not allow random errors to occur. His assertion was that the debugger encouraged a quick hack/variable test instead of looking at the overall picture of whether or not the implementation was inherently flawed.

Re:printk

[da groundhog]

apparently good os design practices are a new concept to you. prink is a RING BUFFER meaning it will over-run the begining when it's full -- ie to SAVE MEMORY

if printk was dynamically allocated and you were running ip aftables after about a week you would have used up all your available memory for on filtered packet messages from iptables

so yes a static rung buffer is a good idea

ps - I also agree with linus's feelings on kernel debuggers

Re:saftey

[mosch]

printk is properly designed, for a routine which is designed to print out informational messages, and short errors. it is not properly designed for a debugging tool, despite the fact that it's the only officially acceptable kernel debugging tool in existance.

There's no way to put a large amount of text on screen atomically, thus if you have code with a bug that's getting called repeatedly you have a choice of:

1. limiting yourself to 1024 characters of information which is probably enough, but maybe not
2. calling printk multiple times, which depending on where it happens in the code, might make it possible to have a race between multiple printk's, making it difficult to pull the status information from each invocation, as they're mixed
3. or just install the kdb patch, and act like a programmer, not a cs100 student

I stand by my assertion that linux makes kernel debugging unneccessarily difficult, asserting the arrogant notion that only total kernel hackers know how to find or fix bugs.

--

Re:This is a good thing

[dair]

No "cryptic messages" -- but it does display extensions as they load (those little icons that pop up along the bottom of the screen). For better or worse, Apple does seem to believe that the user should be notified of what drivers are being loaded.

Actually, there has never been a public API for displaying these icons - they're shown by the extensions themselves rather than displayed by the OS. A couple of spare bytes in low-memory are used to hold the coordinates of the last displayed icon - each extension reads the coordinates when it gets invoked, draws its icon, and then increments the values for the next extension.

Quite a neat hack, considering that the bytes used to hold this state were appropriated from the end of the space normally used to hold the application name once the system boots.

On Mac OS X, where there are no extensions, the only visible messages are a single-line caption in the middle of the boot screen that flicks through a couple of high-level tasks ("Starting AppleTalk", etc).

-dair

Re:how many people here actually *read* Linus's ms

[jandrese]

If you read all the way to the bottom, he talks about removing the "good status" messages as well. This means if a driver loads fine it won't show up in the boot sequence. Some people were expressing concern that they might leave old cruft in their kernel (drivers for hardware they don't have, and pseudodrivers that aren't useful anymore) and not notice it. Some drivers report if they can't find the piece of hardware they're looking for, but I belive some are silent other than the informational messages (version foo copyright so and so messages).

Down that path lies madness. On the other hand, the road to hell is paved with melting snowballs.

How about the way FreeBSD does it?

[jandrese]

FreeBSD has an option (although it is disabled by default.) to display a boot screen (similar to windows). IIRC, the boot screen can be dumped for the regular boot message by a flag to the loader, or by simply pressing escape during the boot (unless your error is: atkbd0: Error failed to initalize keyboard...

Really, what Linus seems to be annoyed at here are the excessivly verbose messages that some drivers like to print out (like I need to see the algorigthm benchmark each time I boot) that might drown out an important message by scrolling it off of the screen before you see it (although it should still be available through dmesg, just like the FreeBSD boot messages are still are even when you have the "graphical" boot.). The Linux boot sequence is getting a bit heavy on the pointless informational messages these days, so a bit of a pruning won't hurt too much.

Down that path lies madness. On the other hand, the road to hell is paved with melting snowballs.

This is a good thing

[Have Blue]

Attention, We-Want-Linux-On-The-Desktop crowd: Support this and help out, it is a big step in the right direction. To a consumer, diagnostic messages are confusing and pointless. Admit that Apple did something right, for once (interpret that as you will): the Mac OS, up to and including X, will never show cryptic messages or break out of the GUI unless you give it a direct order to do so (launch Terminal or Console, hold down key sequences during boot) or a fatal error occurs. This is a good thing, it makes the experience seamless and friendly. Remember that consumers don't care about what drivers got loaded when (and isn't improving the drivers themselves a much more important goal than improving the error messages?) and similar arcane knowledge of the computer's internals, and saying "learn it anyway because it's good for you" will not win you any friends or customers.

Re:This is a good thing

[cpt kangarooski]

Speak for yourself - half the lights in my car (US Ford Taurus) don't come up when it's started, but I've seen a number of them light up for various reasons. (e.g. the low coolant light)

I'm not against having a diagnostic mode so that if someone is interested in seeing status messages they can, but generally no one gives a crap. I mean, do you really need the computer to tell you how much ram it has whenever you boot if it's the same every time? Just give the stupid box enough of a memory of its prior state so that it can compare it to what it finds now, and it'll likely be fine.

Re:This is a good thing

[cpt kangarooski]

Precisely- the messages are not informative at all.

Imagine you're driving in your car, and all of the sudden, buzzers and warning lights start going off, announcing EVERYTHING'S FINE. Do this enough, and you'll not only be upset by it until you can manage to ignore it (if possible) but you'll have trouble distinguishing it from NOT EVERYTHING'S FINE.

Let the status messages come up if there's something new, or different, or wrong. But as long as things are running smoothly, there's no need to hurl that in our faces. (and when warnings are displayed, also include some fricking suggestions as to how to fix it or who to ask, dammit!)

Re:This is a good thing

[Ben Hutchings]

That's disgusting.

Re:This is a good thing

[Lumpy]

Agreed, there is alot of crap that is printed during boot that is pure fodder. although if you want a pretty boot screen you can use the linux boot logo patch (up to Kernel 2.4.2 the author hasn't released anything for the newer dev kernels.) Now.. putting all that crap in a bmesg file... that would satisfy the "gotta see everything crowd" (you know the ones that would like the kernel to display the name of each person and a Loaded... after every 4 lines of code )

We really could quiet down the kernel on boot.

Re:This is a good thing

[Simon Brooke]

Attention, We-Want-Linux-On-The-Desktop crowd: Support this and help out, it is a big step in the right direction.

While I agree that short, informative messages are a step in the right direction, a splash screen or other sweeping under the carpet is not. People who want to make money out of Linux products may want Linux on Joe Ordinary's desktop, but those of us who use Linux now should not - or at least, should not be prepared to make compromises in the features we value to do so.

Linux is high quality, very stable, rapidly developing system because it is a hackers operating system. When it doesn't work, hackers are motivated to fix it, and have the ability to fix it. Joe Ordinary will never have the ability to fix an operating system. If you compromise Linux to suit Joe Ordinary in a away that doesn't suit the hacker community, the hacker community will drift off to the next cool operating system, and Linux will start to suffer from bit-rot.

Re:This is a good thing

[Old Wolf]

If you've ever written a driver, you will know that it is -not possible- to make it show an error message iff it fails. The best you can do is display an error for a number of common conditions. In fact, the larger part of driver testing is in trying to run it on different machines and see if a situation comes up that the programmer has failed to anticipate.
On the other hand, it is easy to display a message upon success, since all possible success situations (namely, everything goes according to the coder's plan) are known beforehand.

Re:This is a good thing

[kimihia]

It is *normal* for commodity PC hardware to randomly blow up and do weird things, and being able to handle it without training people to press special key combinations is *good*.

I'd hide the messages unless the person watching can do something about it.

If Alan is unable to understand what via-rhine.o is, then being able to press a magic key that prints 'via-rhine.o' to the screen will do him no good.

However Bob understands what via-rhine.o means, and he knows the key to print it onto the screen.

Take for example 'F8' on old MSDOS bootups. I never knew about pressing that key until I understood enough to know what I was being asked. If I knew the key before understanding the questions MSDOS was asking me, I would have been lost.

If you can do something with the message, then you'll know how to find the message.

Re:This is a good thing

[Trepalium]

The point isn't to have a pretty bootup, but to remove the useless information that's obtainable via other, more reliable, means. Do I really care about copyrights and versions of all the components on bootup? Not really. It might be handy having everything in one place with dmesg, but it's still too cluttered. I'd rather have information only on drivers probing hardware and the results it gets, and so on... You know, the stuff that's useful for diagnosing problems.

If nothing else, announcing module load/init should be output at debug priority, not informational priority. Debugging a non-working system would be the only real reason most people would ever need to see that.

Re:This is a good thing

[sigwinch]

Admit that Apple did something right, for once: the Mac OS, up to and including X, will never show cryptic messages or break out of the GUI unless you give it a direct order to do so or a fatal error occurs.

You can get away with poor brittle engineering if you are willing to tightly restrict the hardware that is used. OS X runs on, what, a single CPU architecture and half-a-dozen machine variants? Linux runs on half-a-dozen architectures and tens of thousands of machine variants. It is *normal* for commodity PC hardware to randomly blow up and do weird things, and being able to handle it without training people to press special key combinations is *good*.

This is a good thing, it makes the experience seamless and friendly. Remember that consumers don't care about what drivers got loaded when...

Until some random driver wedges for 5 minutes during boot up or shut down, as it eventually will on commodity no-name hardware. If you can see a message for which thing is wedging, you are 99% of the way towards solving the problem. OTOH, with the Windows/Mac approach, the poor end-user will need a trip to the technician for a reformatting. Transparency and diagnosibility are Good Things.

Re:This is a good thing

[Adam Jenkins]

It's strange but I thought that Unix was meant to follow the "no news is good news" philosophy. When Linux spews out a line or two about every device it finds, it really makes you wonder which Linux is trying to be, Unix or just some you-beaut "look how technical I can look" OS.
--
Never try to teach a pig to sing. It wastes your time and annoys the pig.

Re:This is a good thing

[friscolr]

when the obscure server in the 3rd computer room (the one 2 buildings away) needs to be rebooted after a good year of uptime, all those messages are real nice like.

i deal with a slew of different machines for my job and even my home network is composed of a wide variety of hardware on various machines. all those ugly messages really help remind me of what i'm dealing with whenever i reboot a machine.

Having no boot messages should be an option in lilo. Plenty of messages should be the default.

-f

Re:This is a good thing

[enneff]

I agree.

I can't believe it's been going on for so long.

Re:This is a good thing

[fredrik70]

oh, wish I had mod point, 'cause then I modded you up!
As mentioned though, the boos msgs can be handy while checking a crashing kernel. A -v switch should do it...

tada

Re:This is a good thing

[update()]

Admit that Apple did something right, for once (interpret that as you will): the Mac OS, up to and including X, will never show cryptic messages or break out of the GUI unless you give it a direct order to do so (launch Terminal or Console, hold down key sequences during boot) or a fatal error occurs.

No "cryptic messages" -- but it does display extensions as they load (those little icons that pop up along the bottom of the screen). For better or worse, Apple does seem to believe that the user should be notified of what drivers are being loaded.

Unsettling MOTD at my ISP.

Re:This is a good thing

[jrockway]

Thanks for that! I always wondered why some extensions didn't show an icon (not to menton the ones that fscked up the rest of the icons:). Neat hack.

Re:This is a good thing

[Muad'Dave]

Correct me if I'm wrong in your part of the world, but every car I've ever driven does illuminate every idiot light when you turn the key to start the car. They go off once the car is up and running. Everyone I know knows to ignore them (in fact, that's when you check for burned-out bulbs!) So what's wrong with messages from the kernel as it boots? They're easy enough to ignore, and are very handy when things get scrogged up.

Re:This is a good thing

[Muad'Dave]

Interesting. Please try this: turn the key to 'on' (the step right before 'start'). Do the lights come on then? If not, I guess I need to buy a new car to keep up with things!

Re:This is a good thing

[bay43270]

Attention, We-Want-Linux-On-The-Desktop crowd: Support this and help out, it is a big step in the right direction. To a consumer, diagnostic messages are confusing and pointless. Good point. The first time my brother saw my linux machine boot, he thought I had a problem. It never occured to me that Linux running correctly look just like the windows blue screen of death.
-----

Re:I did and I still don't like it

[cduffy]

Whadayathink 'cat /proc/interrupts' is there for?

Re:Ummmm...

[DunbarTheInept]

There are two ways a driver could "fail". One is to try to do something and fail, the other is to never get called in the first place because the kernel didn't recognize that the hardware was of the appropriate type for that driver to be called. It would be handy to tell the difference between the two. Today you can because today if the driver is being called you will get either a good or a bad message, but you know you will get *something*. If you see no message, then you know it didn't even *try* the driver in the first place, and that helps you figure out the exact type of error you are dealing with (probably a driver that's older than the hardware, so it doesn't recognize its ID string.)

Maybe you should.

[DunbarTheInept]

Read all the way to the bottom. He also rants about not wanting "yup, I worked" messages to show up either. I agree about the copyright and version numbers, but NOT about pruning the "it worked" messages. Take them out and now "This driver worked just fine" looks exactly the same as "The kernel never loaded this driver in the first place." When you are trying to figure out why some piece of hardware doesn't seem to be working, that is a relevant difference.

"good status" messages ARE relevant.

[DunbarTheInept]

I agree with getting rid of ads and versions, but NOT with getting rid of good status messages. The difference between a driver being silent because it isn't even being loaded in the first place and a driver being silent because it was working okay is a HUGE difference. It's something I want to know when I'm trying to figure out why I can't get my new WidgetMasterProBlaster device to work.

Re:Ummmm...

[DunbarTheInept]

Why not just use /sbin/lsmod to see which drivers are loaded? If you do that and see that a driver is not loaded, can you tell the difference between that driver not being loaded because it failed at start-up (didn't see the right hardware), vs that driver not being loaded because the attempt was never made to do so?

Re:Please read the real story before replying...

[drsoran]

That's all Linus wants to remove from the boot messages, and I hope he gains support on this, because I want to use my computer, not read the full credits for the kernel 5 times a day.

Are you a kernel developer? If so, perhaps it would make more sense to just include some option to shut the messages off but leave them on by default. I don't mind people's names being credited on the modules and bits of code of the kernel they worked on when it loads. They're certainly not getting rich writing kernel code so they might as well get some fame out of it. Besides, like I said, unless you're a kernel developer, why are you rebooting 5 times a day?? This isn't Windows95. :-)

Re:I did and I still don't like it

[Xenophon Fenderson,]

What if you can't boot to a working system? Thank God NetBSD would desplay interrupt and Ethernet MAC ID, because it would crash hard soon after and that was the info I needed to make things work.

Rev. Dr. Xenophon Fenderson, the Carbon(d)ated, KSC, DEATH, SubGenius, mhm21x16

I like those boot messages

[Malc]

Why would we want to get rid of them? I have never been annoyed by them. Please, make it configurable! I like to see what's going on whilst my machine is booting: it reminds me what modules I have forgotten to remove when I change me hardware; it let's me see what is having problems if boot-up takes a long time. I hate the way Windows boots... when it sits there for 60 secs with no hard drive activity, I would really like to know what is going on and whether there is something I can do to fix it (e.g. when searching for a non-existent DHCP server). Otherwise it's just time-consuming trial and error.

Re:I like those boot messages

[Rendus]

I have never, ever seen the option "Attempt to reach DHCP server? (Will timeout after 60 seconds) [Y/N]: "

Nor have I seen the option

"Load video drivers? [Y/N]"
"Load mouse drivers? [Y/N]"
"Load Aureal Vortex Wave Device? [Y/N]"
"Load VIA UDMA ATA/100 drivers? [Y/N]"
"Load VFW Interface for ATI Radeon All-In-Wonder? [Y/N]"

Re:I like those boot messages

[garcia]

I would not understand why this wouldn't be an option. I find it very annoying to try and fix a Windows machine and not know what the hell is being loaded at startup.

I think it is great that people don't want to see that shit. Hell, we are trying to make a move towards having more "dummy friendly" OS. "Dummies" don't like watching kernel messages. They would rather stare at the pretty pictures :)

I am fine w/all of that, in fact I believe it is a great idea and it should be implemented! Just make it optional. At least have a keystroke bring up the kernel messages.

dmesg is nice to have but I just like to know that there is something really going on other than just watching that stupid little bar at the bottom of Win bootups :)

Re:I like those boot messages

[snubber1]

You want to know what windows does? Hold F8 *before* you see the first windows splash screen and you will get a menu. Pick step-by-step confirmation.
You will get to watch each step and see how long things are taking. You also have the option of logging it all.

Re:Hope we don't lose our verbose logs with this

[Fnord]

No he wasn't talking about removing usefull stuff. Specifically, he's talking about removing anything not usefull. All the driver version and "written by Joe Shmoe 2001" messages. Also the "blah.c driver loaded OK" are being considered for removal because, really, they serve no purpose. Error messages and actually usefull status updates (like the ide driver spitting out the device ids of the drives attached) will be left in. And for all those of you who really value the ego boost of having your name in the kernel boot, he also specifically said to leave this stuff in module load outputs. Just get it out of compiled in drivers.

Message uniformity would be better

[iabervon]

I think what would be nicer is if everything used a common way of presenting information. Each driver that gets an IRQ or something has a different wording for telling you which one it has, and some drivers' messages sound like they might indicate errors even when the situation is pretty normal and supported.

I'd much rather see a list of:

driverx.c Driver for X on IRQ n DMA m
drivery.c Driver for Y device not detected

so you could see, at a glance, what stuff is in your kernel, what is actually active, and how it autoconfigured itself. Currently, you have to look through the messages for the one that applies to a misbehaving device, and then sometimes look at the driver source to figure out what the message means.

It's been so long...

[Coppit]

I forget what the boot screen looks like. :)
---------------------------------------------- ---------

Can't you people read?

[Chang]

Torvalds is just doing everybody a favor.

He's only getting rid of annoying messages, i.e., non-informational crap.

He's not getting rid of all boot messages.

Re:no version numbers?

[Tet]

cat: /proc/scsi/driver_name/0: No such file or directory

that's why.

Pah! Out-pedanted -- how embarassing :-) Substitute "driver_name" with the name of the driver for your particular SCSI card. In my case, that's "aic7xxx". Note that if your SCSI driver is compiled as a module, you won't see it unless the module is loaded -- but then you wouldn't have seen it at boot time either...

Re:no version numbers?

[Tet]

it's tough to load the scsi driver as a module when all my disks are scsi

initrd is your friend -- this is the technique most of the main distributions use for dealing with SCSI install disks. However, if you're compiling your own kernel, then there's no point in having modular SCSI for an all-SCSI system. You might as well just compile it straight in.

Re:no version numbers?

[Tet]

I knew I had done it properly when I saw the correct version number. leave that in!

cat /proc/scsi/driver_name/0

In my case, that gives:

Adaptec AIC7xxx driver version: 6.1.13

Now tell me again exactly why you need the version printed at boot time?

Re:how many people here actually *read* Linus's ms

[evilandi]

Lally Singh: Linus was just saying that *NON-INFORMATIONAL* messages are just a waste, and that they shouldn't be done. Information related to DEBUGGING is logged. Error messages are still printed. Relax and get over it.

Well, Linus is wrong.

Error messages only occur when there's an errror.

What if there's no error, what if it just HANGS on boot?

How the hell are you supposed to diagnose a hang if you can't see what's running?

--

Re:textulike

[evilandi]

Brody: what does some flying windows or clouds tell you exactly?? well... if they freeze you know your computer has crashed.

No, it tells you that Windows installs every frickin protocol under the sun when you pop in a new network card, and it just sitting there waiting for a nonexistent DCHP server.

--

Re:Not *all* messages

[Just Some Guy]

Linux isn't saying anything at all about a graphical boot, and that graphical boot system is still hugely experimental anyway.

Actually, he is. One of the big quandries about the graphical boot project was how to display all of those various informational messages. Some drivers' licenses (particularly non-Free ones, for example, for certain graphics cards) require that the message be displayed.

Imagine that you're the one writing the graphical boot. How do you decide which messages to display and which to skip? If Linus had his way, and no unimportant messages are printed, then the task would be much easier.

how many people here actually *read* Linus's msg?

[Lally Singh]

Linus was just saying that *NON-INFORMATIONAL* messages are just a waste, and that they shouldn't be done. Information related to DEBUGGING is logged. Error messages are still printed. Relax and get over it.

--

So?

[laertes]

I use OS X, which is a nice happy UNIX (BSD) without pages of boot information. You know what? I survived! Of course, OS X doesn't allow the tinkering that Linux does, but it goes to illutrate a point; people don't need, or even notice, their kernel boot messages.

I personally have done a little kernel programming, and that is the only time I paid any attention to these messages. Every other boot ends up with me just looking at the screen with glazed over eyes, not a single message registering. When things go wrong, those messages are invaluable, but if the kernel can't get it's debugging messsages on the screen during a catastrophe, there's not much chance that dmesg would help anyway.

I've always thought those messages looked pretty interestingly verbose... The more I think about that sentence, the more my head hurts.

Re:Bad move Linus

[Ben Hutchings]

There are two problems here - first, fluff like copyright information is being printk'd at too high an importance level, and second, the minimum level for display defaults to KERN_DEBUG (lowest level). The changes being proposed are that fluff should be printk'd at a lower importance level and the minimum level for display should be higher by default. That would not stop you from specifying a lower minimum level to dmesg (which currently also defaults to KERN_DEBUG) and seeing messages which were suppressed from display at boot time.

saftey

[MenTaLguY]

Dynamic allocation isn't necessarily safe in all the places printk() can be (is) used.

There are some points in the kernel at which you just cannot touch the parts of the VM subsystem that would be necessary to dynamically allocate memory. It's not a design flaw; it's just a constraint of the way locking and reentrancy have to be handled.

With a lot of work, you could probably create a dynamically allocating printk(), but you'd have to introduce a tremendous amount of additional locking just to support printk(), which would kill performance, and would mean that printk() couldn't be used in many places that it is now (certain interrupt handlers where the locking overhead would be too much).

Also, an obvious one ... what happens if the printk() happens to be reporting something due to VM exhaustion? ....from within the VM subsystem?

The static printk() buffer is a necessary design decision.

Typical Slashdot

[morven2]

Can't let a small bit of reading get in the way of being a loser, can we?

Sigh. Don't know why I bother anymore. Oh, I do know why I bother -- I'm bored. Yawn.

A very good move.

[morven2]

As a longtime UNIX admin who's used many UNIX systems, I think this is a good thing. I don't mind text printing on startup, but Linux prints so much ABSOLUTE USELESS CRAP to screen.

90% of them are ego-boosting messages by the authors of each chunk of the kernel. These are in particular what Linus seems to have an issue with.

I don't think boot messages should be completely gotten rid of, but they should be put on a rigorous diet. Let's look at the way other Unices do it. FreeBSD, for example, prints stuff, but it's terse and professional looking. Same with Solaris.

On Linux, the messages scroll so damn fast, especially on a speedy modern system, that you can't even read them. That's bad, folks. I can't even tell if the kernel is printing any error messages, because the credits messages scroll it so quick.

Yes, I don't mind a kernel component telling me it's there and the hardware's functional, but don't be so verbose about it!

Re:A very good move.

[esper]

No, MS isn't producing "a CLI" (which I assumed you parsed as meaning "command-line interface") for BSD/XP. They are working on an implementation of CLI, their new Common Language Infrastructure. It's something completely different.

Re:A very good move.

[mrm677]

Don't count *BSD out yet. Microsoft just announced that they are working on a CLI and a C# compiler for FreeBSD and Windows XP.

Single screenful from boot

[Ed Avis]

On some DOS machines you will get the memory count displayed by the BIOS, then only a few more lines of text before the C:\> prompt. That means you can start using the machine while still on the first screenful of text since it was switched on. I think that's cute.

On Linux you can usually Shift-PageUp to scroll all the way back to the beginning of time, but it's not the same.

BTW - drivers that are likely to crash must print something, so you know what crashed. Although in a production system you wouldn't expect anything to just hang the machine at boot time, so the suggestion of a special 'verbose mode' for kernel troubleshooting is a good one.

Re:Read Linus' post.

[rew]

You don't lose useful information, only useless information. "success" type status messages will be removed

I'm sorry, but it's kind of useful to have the system print an inventory as it finds it. So for example, the fact that it finds a harddisk is something that is highly useful. The fact that it found 512Mb memory is useful.

Now, I agree with Linus that I don't need to know that every kernel since 4 years ago runs with "net-3".

My diskless workstations sometimes don't boot. The "found an RTL8139" message is then ESSENTIAL to determine wether or not I forgot to include the network card again.

The "found hda" message is useful when you're debugging a non-booting Linux system. Stuff like that. But again: Linus is right when he says that plain version information is not very useful.

Sometimes the messages can be condensed:

PIIX4: IDE controller on PCI bus 00 dev 39
PIIX4: chipset revision 1
PIIX4: not 100% native mode: will probe irqs later
ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hda:DMA, hdb:pio
ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:pio, hdd:DMA
hda: QUANTUM FIREBALLlct10 15, ATA DISK drive
hdd: CD-ROM 40X/AKU, ATAPI CD/DVD-ROM drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide1 at 0x170-0x177,0x376 on irq 15
hda: 29336832 sectors (15020 MB) w/418KiB Cache, CHS=1826/255/63, UDMA(33)
Partition check:
hda: hda1 hda2 hda3

How about:
PIIX4 rev 1 at 00/39(14/15): DMA/pio/pio/DMA
hda [hda1 hda2 hda3]: QUANTUM FIREBALLlct10 15: 15020 MB, w/418KiB Cache UDMA(33)

just two lines with all the relevant info.

Roger.

You are not French by any chance

[Archfeld]

are you ? It would explain so many things. What I find most annoying is the folks who can't read past a typo. If we are talking about a serious grammar error that alters the context I can see some uproar, but I'll bet there are more grammar gestapo posts that hot grits posts. You take your choice at which is a bigger TROLL....

yes but that is a test kernel

[Archfeld]

and the messages are more likely to be of use. I like the idea of using a -v parm to get any info beyond a silent load. These things should load quietly and quickly unless the admin wants further info.

Re:yes but that is a test kernel

[orangesquid]

Hmm.. well, it's just that I learned so much more about my computer and about the Linux kernel by watching it boot up once than I ever learned in years of watching Windows boot.

I don't think not displaying "ah-scsi.c 0.3" is going to speed up boot time drastically :)
I used to always hate that about Windows, just having to sit there for six minutes while the hard drive light churned, not knowing what was going on, and always being worried if the light stopped for more than a second.. was it checking for new hardware? expanding a large data table in memory? did it crash?

What would be good is, instead of scrolling messages, have a status bar in the middle of the screen below a beautiful artistic rendition of Tux, and below the status bar messages saying what driver is currently in control of the CPU could pop up, so if the machine died, you could jot down what failed, without having to see your screen fill up with junk.

I know all my whining is moot if there's an option to augment verbosity, but it's a disappointment to see the Linux kernel lose its sense of "underground-ness."

[And to whomever moderated my above post: For future reference, "Redundant" is intended for those times when someone posts the same message three times in a row. If nobody ever repeated anyone else on a /. story, the number of comments would never rise above 200. :)]

Re:yes but that is a test kernel

[markbthomas]

What you want is the Linux Progress Patch, my friend.

the above has a point

[Archfeld]

I'd like to see NEW services and devices etc, but after they are running nicely I'd just as soon they fade in the background and not take up screen time unless an ERROR is generated. As a development system builder I never know what specs the developers will want.

"people want pretty boot-ups"?

[Xunker]

IIRC, doesn't Mandrake-Linux already have an answerf or this request? The Madrake 8 I have on my laptop at home had a 'pretty' boot screen that tells you what 'section' it's on, but other than that it's just a picture/corporate propoganda display.

Why do verbose boot messages matter when you don't even see them?

Re:"people want pretty boot-ups"?

[Wildfire Darkstar]

Why do verbose boot messages matter when you don't even see them?

'Cause if you're trying to go through the startup logs to figure out what went wrong, it gets to be a little ridiculous.

Linus isn't arguing that a nifty graphical startup screen is the answer; his reasoning doesn't even seem particularly aimed at the "casual" user, who may be scared of verbosity, but designed to streamline things for the real power user.

Re:Message classification

[Kitanin]

We need the messages printed at boot time to be classified.

I'm sorry, Citizen, you are not authorized to view the modem status.

I could tell you if your memory is bad, but then I'd have to kill you. Trust the Computer. The Computer is your friend.

...

Erm... Isn't this what we're trying to get away from by moving to Linux? :-)

Utility Wanted

[RichMan]

Ok, so does anyone have a utility to turn the boot log file into a real pretty image to load as a background for graphical logins?
Color highlights and folded into 4 vertical strips would be nice.

Re:I did and I still don't like it

[BrookHarty]

Unless you need to know what device is using IRQ 11, both your NIC and Soundcard.

I think Version numbers, and other informational data should be kept in. Too much information is better than no information.

Re:Seems to me....

[esper]

What about things which are built into the kernel? No modprobe there, but they could still break. The code would have to check whether it's part of the kernel or a module and either print or suppress messages accordingly, plus an additional mechanism may be needed to pass messages from a module to modprobe.

Much simpler for the code to just do the printing itself. (Simpler is often, but not always, better. I'm not sure whether it is in this case or not.)

poor brittle engineering?

[jonbrewer]

By your arguement, we should be calling Digital Unix, Irix, Solaris, and a host of other rock-solid variants of Unix "poor and brittle," because they don't "run" on as many architectures.

And it's a funny thing, but I don't seem to remember any informative messages from Linux telling me why I couldn't bind an IP to my 3com (3c589d) pcmcia ethernet card. The solution to that problem was found not through "transparency and diagnosibility" but through a win2k cd.

Re:poor brittle engineering?

[jimmyphysics]

you missed his point... the poor, brittle engineering is evidenced by the macos' boot sequence and lack of useful information.
sigwinch's point was that apple could get away from this because they only had to support a very small number of configurations.
i can't speak for digital unix or irix, but solaris machines are plenty verbose through the boot sequence, and even more so when something breaks. the fact that these unices only run on a small number of machines has nothing to do with their quality of engineering.

Re:poor brittle engineering?

[jfunk]

Funny, it worked fine when I used it.

In fact, I often demonstrate how cool Linux PCMCIA support is by grabbing a random card (often one from the demonstratee), plugging it into my laptop, plugging in the network cable, and browsing to sites.

I even did it with a WaveLAN card once. It was the first time I plugged a 802.11 card into my laptop ever.

No messages, no popups, no "insert a stupid disk because I'm going to reboot the machine 3 times."

Interestingly enough, I have a tiny Windows partition on my laptop, but I never use it. Windows will lock up hard on boot if my NIC (3c575CT) is inserted. No biggie, Windows is useless to me and my solution was found through a SuSE DVD.

get linux_logo

[jscott]

linux_logo also displays BogoMips!

Re:get linux_logo

[Delrin]

they're so cute... :)

I did and I still don't like it

[Sangui5]

Those non-informational messages do contain information -- in their negative space. Sometimes things fail silently. There are no error messages. If you are expecting a success message, however, and don't recieve one, then the lack of a message indicates failure

On the other hand, some of the messages are useless. When my sound card driver loads, it outputs three lines:
es1371: found chip, vendor id 0x1274 device id 0x1371 revision 0x08
es1371: found es1371 rev 8 at io 0x1080 irq 11
es1371: features: joystick 0xx

That much information is overkill (and unneeded). Go ahead and throw out the results of profiling code. Go ahead and ditch the ReiserFS built in ad. Go ahead and throw out the copyright notices. But please keep the sucess messages - they are just as important as failure ones.

Re:I did and I still don't like it

[jfunk]

If you are expecting a success message, however, and don't recieve one, then the lack of a message indicates failure

Traditionally, in UNIX environments, the lack of a message indicates success. If it worked, why bother me about it?

This is important because I've seen *lot's* of people panic when they get success messages.

Pretend you know nothing of IRQs, IO ports, and hex numbers. If you saw a screen full of them you wouldn't know if it was good or bad. An important part of useability is not confusing the user, because they get very worried when they see something confusing.

Re:OT: Crapflooding

[clifton_]

What? Bear is driving car? How can that be?!

Really off topic, but did you signature came from Clerks the animated series? LOL, that is one of the funniest lines on there. :)

More pointless drivel.

[Pope Slackman]

Thank you for using your +1 bonus on that drivel. I appreciate you wasting my time; after all, I browse at +2 so I can see stupid, pointless comments like yours.

Heh...I love having +1 just for that reason.

Hah-hah!

C-X C-S

Re:OT: Crapflooding

[ethereal]

They are getting gradually whacked - see here. The big question is how he's avoided the 2-minute repost wait. Maybe by posting them all absolutely simultaneously?

Caution: contents may be quarrelsome and meticulous!

Re:No wonder you're not working with linux

[ArsonSmith]

uhh, thats an easy one:

/boot/config-

why would you even need to ask?

Re:A better solution

[warlock]

Not particularly useful when your console is a real tty, not to mention that modern boxes are so fast that you wouldn't see much like that anyway, so there's no point to print it in the first place.

Re:A quote from The Good Book

[sharkey]

You ought to pass that book on to the MS Access developers. Does this sound familiar?
"You are about to modify 0 records. This action cannot be undone."

The Outlook people could use some help, too:
"This graphic does not do anything. For Help on an option, click tho question mark [?], and then click the option."

--

Re:Message classification

[Imperator]

Why don't we have different levels of information for each category? Then I could get info-level messages about failures, and only warning-level messages about credits. And also I'd like to customize these messages for each category: for example, I only want warning-level messages about failures in the filesystem category. And perhaps I'd like only critical-level messages about successes in the boot-message-printing category, to prevent me from getting a notification about a successful notification (thus crashing at boot because some poor driver wanted to tell me that its programmers have big egos). Or alternatively the kernel would be smart enough to check the boot-message-printing options (configured through LILO perhaps? there's lots of unused space in my MBR) to make sure that it won't get stuck in a successful-message loop. Maybe for ease of maintenance we could get ESR to write this in Python, and then compile a small Python interpreter into the portion of the kernel code that gets loaded first...

Yes, this all sounds very good, Mr. Gates. True innovation. We'll implement it at once, sir.

--

Can we have a verbose mode?

[Adam J. Richter]

Out the outset, let me say that I agree that kernel advertising/credits messages hide useful information and make the system unnecessarily confusing to less experienced users. I am glad to see Linus taking aim at them, as it really takes someone very respected in the development world to attack something that has so much ego attached to it.

That said, I would like to still have a verbose mode that includes these messages so that someone who knows what they are doing can verify that a given facility is being initialized at boot. The simplest way to achive this would be to change DEFAULT_MESSAGE_LOGLEVEL in linux/kernel/printk.c, from 7 (KERN_DEBUG) to 6 (KERN_INFO), thereby filtering out KERN_INFO messages. Maintainers who feel that some of these messages really should be printed by default could submit patches to change them to the previously ill-defined KERN_NOTICE (5) level and try to convince Linus to apply them.

It is also trivial for individual Linux distributions and sysadmins to modify this policy by booting with the "debug" argument (sets console_loglevel to 10) or to modify this in a boot script by writing to /proc/sys/kernel/printk as documented in linux/Documentation/sysctl/kernel.txt. A "loglevel=n" kernel boot argument would also be a helpful feature for the future and would be trivial to add.

Re:OT: Crapflooding

[Pope]

You make me believe whole-heartedly in censorship.

Re:OT: Crapflooding

[Pope]

Yep, That entire end sequence made me fall off the couch!

Re:It's all irrelevant.

[saihung]

>('cept to add new hardware of course.)

Bah. Like a marksman who fires his gun between hearbeats, I just listen for the 60-cycle hum and slam my PCI cards in during the trough between signal peaks. Unfortunately my teeth are starting to blacken from gnawing on live IDE ribbon cables, so I may have to reconsider my hardware techniques.

Re:no version numbers?

[ywwg]

[owen@localhost owen]$ cat /proc/scsi/driver_name/0
cat: /proc/scsi/driver_name/0: No such file or directory

that's why.

Re:no version numbers?

[ywwg]

I guess I get the idiot literalist award for today. thanks :) Also it's tough to load the scsi driver as a module when all my disks are scsi -- circular logic leads to kernel panics

no version numbers?

[ywwg]

hey now, I needed to downgrade from the "new" delete-all-your-data adaptec scsi driver (6.x.x) to the "old" doesn't-delete-all-your-data driver (5.x.x), and I knew I had done it properly when I saw the correct version number. leave that in!

Need multiple boot message modes.

[meldroc]

This way, there would be several modes, selected by kernel parameters in the lilo.conf file or LILO prompt.

1. Splash-screen mode. Puts a pretty picture on the screen, stays there until the system gives you a login prompt or (x|k|g)dm login screen. The user should be able to press escape to ditch the splash screen & go to concise mode if problems arise.
2. Concise mode: Displays minimal information on bootup - kernel version, distribution name, machine name. Nothing else is printed unless something is wrong.
3. Verbose mode: Shows everything you never wanted to know about Linux as it boots - handy for kernel debugging or fixing configuration problems.

A quote from The Good Book

[Brento]

"Unnecessary messages are pure evil."

- Everett N. McKay, page 367
Developing User Interfaces for Microsoft Windows
(c) Copyright 1999

Re:A quote from The Good Book

[ttfkam]

How about the error in IE 5.x that you get when you try to programmatically add an option to an empty dropdown box:

Internal Error

A modal dialog box that, when you hit OK, the browser closes. Thank god they didn't put any unecessary messages in there.

Re:A quote from The Good Book

[onosendai]

What about my absolute favourites ...

Unknown Hardware Detected
Finding Device Driver

or

Keyboard Not Found
Hit F1 to Continue

Re:A quote from The Good Book

[Lozzer]

And if the people who wrote SQL*Worksheet for Oracle are listening. I could live without the message box after you exit it saying (from memory) "SQL*Worksheet Exited" with an OK button.

red hat?

[delmoi]

Red hat is doing fine...

Why not add a little statement like...

[gmhowell]

CONFIG_SUPPRESS_VERBOSE_INIT

????

Then RH and the rest can choose that (and lpp) in their default kernels. The rest of us can let 'er rip.

I always wanted more messages. That way when rebooting, I looked extra '1ee7. The only thing more frightening to the employees is when they see a kernel compile running. Or a BSOD (a true NT type one with all of that debugger information or whatever it is. Not the wimpy Win9x one)

Maybe I should read it first

[gmhowell]

Just read it. Maybe others should:

So let's simply disallow versions, author information, and "good status"
messages, ok? For stuff that is useful for debugging (but that the driver
doesn't _know_ is needed), use KERN_DEBUG, so that it doesn't actually end
up printed on the screen normally.

So, he's just talking about copyright notices and that sort of garbage. He's actually keeping the good stuff in. So, this makes sense. Must be a slow newsday, what with no Micro-Soft FUD to report.

Re:Maybe I should read it first

[HongPong]

Must be a slow newsday, what with no Micro-Soft FUD to report.

*PING*... the mystical HongPong has granted your wish!

--

Re:No wonder you're not working with linux

[gmhowell]

Relying on dmesg to configure a kernel (like the person you replied to) is kinda like relying on the author/editor blurb at the top of a Slashdot story to give an accurate impression of the news (like the person you replied to).

You guys want to actually try reading the message?

[Minstrel78]

If you bother to read Linus's message, you'll note that this will not eliminate messages scrolling by on bootup. This will simply eliminate the display of suplimentary information that isn't crucial to the booting process.

For instance, you will no longer see:

Random Device Driver: Initialized
Random Device Driver version 1.23 (c)1999 John Doe

You will simply see:

Random Device Driver: Initialized

The other stuff is available elsewhere, and just adds to the clutter. Simple == beautiful.

ReiserFS

[Pengo]

Thats one thing that sorta anoyed me about ReiserFS kernel module. I don't use SuSE and hate mp3.com so I simply went in and changed the output message to something amusing.

Great thing about open source, if you don't like the messages.. change em! :)




--------------------
Would you like a Python based alternative to PHP/ASP/JSP?

Re:ReiserFS

[Juln]

Previously, resiserfs told you 'Journaling sponsored by MP3.com' and "[something] about SUSE'. More recent version don't do this, apparently.
Thanks for the shift page up tip. i haven't tried that.

ABM Treaty

[Kevin S. Van Horn]

So, in other words, Linus is proposing an ABM (Annoying Boot Message) treaty?

Re:Not *all* messages

[Shotgun]

Linus specifically said that the important kind of messages are the ones that are displayed when something isn't working properly,

You see, the problem is that software often doesn't KNOW when something isn't working properly. The software cannot know unless the developer either:
1)predicts every possible failure case or,
2)checks the inputs at every function call (which will slow your 1000GHz Fitztanium to 8086 speeds)

Version numbers are also important. If you show me your /var/log/messages and I happen to notice that you're running the 2yr old ReiserFS code with the security hole, I'll tell you and you will be happy. But because there isn't a problem, ReiserFS doesn't ever say anything and you get cracked.

Bootup messages are a very important standard that has become an invaluable troubleshooting tool. Even know that something DOES work is good information when troubleshooting. Linus should leave it alone. If I don't wanna see "ReiserFS is brought to you by MP3.com", I'll either modify it (I do have the source), or I'll use IBM's JFS.

Imagine this:

[schon]

Even "useless" information isn't really useless..

Suppose you had a server in another office (maybe in another town) and one of the "helpful" office people decides that he doesn't like the brand of network card installed, so he replaces it with one of his own..

You come in on Monday morning, and everybody is screaming at you because the server will no longer talk to the network..

It's nice to be able to dial in (you do have a modem in those remote machines, right?) do a dmesg and look for notification that the driver was loaded properly..

Now, I know this isn't a (hugely) likey possibility, but there has been more than one occasion when I've needed to know what was in a remote box, and checking dmesg remotely was MUCH faster than physically going to the box.

It's just an example of a good reason to keep the information available somewhere (/proc isn't very good for this, because it tells you what's happening now, but it won't tell you what happened while the machine was booting.)

Re:Imagine this:

[Eil]

It's nice that someone else seems to agree... I don't like seeing the corporate stuff, (mp3.com? sheesh...) but I don't really agree that it's all that annoying to notice a coder's name after his particular driver has just loaded. He wrote the code that makes my computer work and did it for free. Is it really that much to ask that his name appear next to his driver?

I'm afraid that this is one area where I don't agree with Linus. I *like* to know that my computer is doing something sucessfully. I *like* seeing that my serial drivers, ethernet, ppp, scsi, etc have all been detected by the kernel.

Lets not forget that all the stuff that the kernel outputs normally gets stored in logs. You'd be surprised what problems can be solved by noticing a mundane change in kernel output. Also, much of my important configuration stuff stuff comes from the output of dmesg. When I have a brain fart and can't remember the SCSI ID of my CD burner for cdrecord, I simply take a quick glance at dmesg. Ditto for pretty much every other device on my system. dmesg sure beats the hell out of spending 10 minutes searching /proc (which may or may not actually have the information).

Please no one take this as flamebait, but the way Linus worded his ultimatum was that he didn't want *any* kernel output except errors. One thing I absolutely hate is operating systems which offer little to no information about what they are doing while booting. This includes Windows and Linux distributions with Aurora installed by default (Mandrake).

I like Linux because it lets me see more of what my system is doing which gives me the opportunity to act when I notice something that I don't like.

PLEASE, at the very least, Linus, if you're going to do this, give your users some option of verbosity. Don't just tell the kernel hackers to get rid of all non-error messages.

Re:Imagine this:

[The Madpostal Worker]

lspci ? theres a reason for utilities like it.

/*
*Not a Sermon, Just a Thought
*/

Re:Imagine this:

[fors]

That's funny. I've got Aurora and it lets me know if something didn't start right. Doesn't give all of the details but I know where to look for the problem.

Happy medium

[eric2hill]

For all the Windows flames that get thrown around here, I think MS hit the nail on the head with the Windows 9x series. (Well, boot screens that is...) The Windows 9x series had the best of both worlds: A pretty picture (animated even!) that was shown while the system was booting, but if you wanted to see DOS messages, just hit escape and you're greeted with the standard DOS bootup.

I think Linux should do the same thing - leave the bootup messages, debug messages, and whatever else there, but cover it up with a penguin sitting in a speedboat. For 95% of the time that I don't care what happens as long as the thing boots, I'll see my happy Tux, but for the other 5% of the time, I can just hit escape to see all the messages I've grown to know and lov^H^H^Hhate.

Windows 2000 doesn't offer this functionality. Yes, yes, I know about Windows 2000's /CRASHDEBUG option. It's not particularly pretty, but serves kind of the same purpose. It's just not available with an escape key. You can't set the option (easily) if you can't boot the system, so what's the point?

"Of course that's just my opinion, I could be wrong." - Dennis Miller

Re:Happy medium

[rabtech]

Add /SOS to the end of boot.ini and you'll get a listing of kernel drivers loaded during startup, as well as the autochk status stuff.

But neither Win9x or Win2K have the ability to print the same type of info that goes into the boot log file to the screen. I've wished there was a way to do so, but only for my own home machine because I like to tinker. For most users, it is a moot point.
-- russ

Re:how many people here actually *read* Linus's ms

[zmooc]

What actually would even be more interesting, is if there'd be a possibility to set the loglevel during compile time (so you don't get very large kernels) or on the commandline (for maximum flexibility) so one can choose between levels like DEBUG, INFO, WARNING and ERROR or something like that. Then everybody would be happy.

Re:It's all irrelevant.

[David Roundy]

Except, I hope, for those of us in California, who should be turning our home computers off daily to save power... (unless of course running a server at home).

Re:WOW! Linux is perfect...

[TheInternet]

When Linus starts thinking about changing boot messages, that must mean the rest of the kernel is perfect 8)

No, he just spawned a new thread.

- Scott
--
Scott Stevenson
WildTofu

Re:Mac OS X & Informative Boot Screens

[TheInternet]

I would just really really like to avoid being forced into a non-verbose display. By default your bootup sequence should have full verbosity turned on

Perhaps yours should. :)

Assuming you want your OS to become mainstream, you better start letting go of these things that make it decidely ugly and confusing. By default, it should display general high-level information ("starting networking") and provide errors only. Most people don't need to know the filesystem was mounted as read/write.

People aren't going to change for Unix. Unix will have to change for people (as Mac OS X demonstrates).

- Scott
--
Scott Stevenson
WildTofu

Re:So let me get this straight

[ttfkam]

I'd agree with you if MS hadn't put in a handler for the escape key since Win95. Want all those annoying messages? Hit escape on boot. Don't want splash screen to ever appear? Delete the splash screen image.

Don't get your panties in a bunch over nothing. Even NT/2K has a way of watching the boot process. I'd just as soon that distributions shipped with the Linux Progress Page patch. If "real geeks" want to see the messages, since they are recompiling the kernel anyway (right? right!?!), they can just choose not to have the flashy screen.

Isn't choice wonderful?

The new user doesn't understand the messages going by. The new user doesn't care about the messages going by. The new user wants to check their email, browse the web, and write their reports. I think it's quite appropriate that the default should be sans informational messages. The technical user is more adept at changing the option, so leave it to the more adept user to change the option.

If you want a really quiet kernel...

[Rufty]

Then patch /usr/src/linux/init/main.c with:

if (!strcmp(line,"silent")) {
console_loglevel = 1;
continue;
}

Just below the "quiet" option...

Remember to add the "silent" option in /etc/lilo.conf All messages are still accessible by dmesg and you can boot with "debug" if you want to.

Silencing about 50 bazillion init scripts is left as an excercise for the reader ;->

Re:No wonder you're not working with linux

[TRyanC]

Depends on the kernel you're running. You'll need one of these packages:

(RedHat 6.x) kernel-headers-2.2.19-6.2.1
(RedHat 7.0) kernel-source-2.2.19-7.0.1
(RedHat 7.x) kernel-source-2.4.3-12

Look in /usr/src/linux/configs/:

[tryanc@dialup ~]$ rpm -q kernel-headers
kernel-headers-2.2.19-6.2.1
[tryanc@dialup ~]$ ls /usr/src/linux/configs/
kernel-2.2.19-i386-BOOT.config
kernel-2.2.19-i586.config
kernel-2.2.19-i386-smp.config
kernel-2.2.19-i686-enterprise.config
kernel-2.2.19-i386.config
kernel-2.2.19-i686-smp.config
kernel-2.2.19-i586-smp.config
kernel-2.2.19-i686.config
[tryanc@dialup ~]$

Now, on topic: I like the boot messages. It gives some history and culture regarding Linux. If someone wants to make them optional, I don't have a problem with it, but for all my servers I plan to keep them enabled, so I can spot problems in the boot as they happen (drive not being detected, etc.). The only time we reboot the Linux machines at my company (about 20 desktops, and about 25 servers) is when there's a problem, and the messages help in debugging (not just what doesn't work, but also what does).

-Ryan

Re:No wonder you're not working with linux

[TRyanC]

Have you ever built a Linux kernel? There is exactly one, quite non-mythical, ".config" file involved, located in /usr/src/linux/.config

kernel is one thing, rest of bootup another

[Juln]

I don't think they would take away the SYSV init messages, or the debian startup messges (whatever thats called).
Those are totally separate from the kernel messages that are spewed.

Re:Message classification

[p3d0]

LOL. I didn't know you were joking at first. If only I had some mod points.
--

Version numbers ought to stay

[p3d0]

Far be it for me to disagree with Linus, but version numbers don't take up much space, and they often make debugging instantaneous. ("Oh, you only have version 4.5.6. Obviously that's not going to work.")
--

Message classification

[p3d0]

We need the messages printed at boot time to be classified. You can have success messages, failure messages, credits, etc. Then you could configure the system to display the kinds of messages you want to see.

I don't know much about the innards of the kernel, but I suspect something like this already exists. Could it be used for boot messages? Perhaps it could be extended?
--

Ob: Neal Stephenson

[drivers]

In "In the Beginning was the Command Line" Neal Stephenson pointed out all the messages that get printed out when you boot linux.

Cryptic messages began to scroll up the screen. If you had booted a commercial OS, you would, at this point, be seeing a "Welcome to MacOS" cartoon, or a screen filled with clouds in a blue sky, and a Windows logo. But under Linux you get a long telegram printed in stark white letters on a black screen. There is no "welcome!" message. Most of the telegram has the semi-inscrutable menace of graffiti tags.

Then he cuts and pastes the entire scroll of his bootup sequence. Now his book is going to be outdated. :)

kernel messages helped me get a vacation

[mister_jpeg]

I shite you not: Staying late one night at work with my boss, I rebooted, watched the messages and began calling some out before they printed.

Boss knew I was about to lose it and gave me the rest of the week off. (yes, this was on a friday night, but time off is time off:)

dash dash space newline
jpg

Re:textulike

[Pedersen]

Either that, or a simple script which will open up as many connections as possible,send the data, and then each of those connections does a near simultaneous hit of the 'submit' button (or rather, the equivalent of doing so). Simple trick, doesn't require a lot of bandwidth, and would probably get past the race condition that's been found and is being more than slightly abused.

But success is important information

[Isomer]

Linus is quoted in the article stating that he wants to get rid of all the successful loading messages as well. I think this is a very unwise idea. Recently when debugging a hang during boot, I was easily able to track it down by looking at what the last thing was that successfully initialised, and what the next boot message *should* have been. That message didn't appear, so I was able to deduce that that driver was probing something that was hanging my machine.

Sure, get rid of all the fluff that the drivers print out, but I believe that a lot of this information is important, even if it is verbose.

Having too much information is far better than not
having any. You can always throw away information thats not relevant. You can't just "make up" information.

Perhaps you should read it more carefully

[Rix]

Linus wasn't suggesting removing "it worked" messages, just shunting them to the log files.

Perhaps a happy middle ground would be to have a verbosity flag in the config process.

Wrong

[Rix]

Linus also wants "good status" messages taken out.

ie) If it's not broken don't tell me about it.

Boot screen?

[Old Wolf]

I don't give a fuck what the boot screen says.

Now, since this story says 124 of 357 posts at threshhold 1, I'm going to see what I'm missing out on...

Re:Mac OS X & Informative Boot Screens

[mr100percent]

No, by deafult my grandmother's OS should not be verbose, because I set hers up right, she should only see a progress bar by default, with a verbose option. She doesn't do anything hardware related or hacking. As long as Word and AOL runs, the avergae user is happy.

Re:Please oh PLEASE!!???

[glitch!]

I read the article.

And I think removing versions and "good status" messages is just plain stupid. When I boot up, I
WANT to know what is running, and that everything is "just swell". What is the problem,
anyway? BOO HOO! BOO HOO! "I cannot deal with all those confusing boot messages. BOO HOO!"

And how many times do you reboot, anyway? Once a week? Twice a year? Every day? Whatever the
interval, a few pages of messages aren't going to hurt anyone. What if you have a problem and now
NEED the messages from last week? On many systems, that will still be in the message logs.

There are a lot of ways to get information, but the message logs are often the most convenient.
I don't see why someone would be so upset about a few K of storage or a few extra kprintf calls.

Re:Please oh PLEASE!!???

[glitch!]

The boot messages are annoying and obscure actual errors.

You state that as if it were an objective fact. Actually it is your opinion. My opinion is otherwise.

The discussion is about NOT DISPLAYING THE USELESS ONES during booting.

Useless is also a matter of opinion. And the messages do not add any appreciable delay. I have

seen kernel messages go by at hundreds per second on occasion. Mere dozens of "useless" messages won't make any difference in boot time.

I read the article. I was paying attention.

No, I think we are confusing issues.

[clump]

My analysis isn't that Linus is killing dmesg, but rather he is responding to advertisements and junk clogging up needed boot and debugging messages. I agree with this. Have you seen what an AOpen motherboard boot screen looks like? On some models there is a right frame with all sorts of advertisements for all sorts of products. That is insane.

Linus isn't killing boot information but rather not placing everything from URLs to copyright information in the same place where you find out if your networking interfaces are functioning properly.

Its a good move.

[clump]

Personally I don't use Linux anymore so I don't really care, but I think it's a dumb move on Linus' part and here's why. Suppose you've just been hired as a sys admin in a company and have to recompile kernels across a couple dozen perhaps hundreds of Linux based servers... Suppose that company was crappy via way of having things documented.

I use a variety of operating systems including Linux and I disagree. In Linus' message he stated "So let's simply disallow versions, author information, and "good status" messages, ok?". That doesn't mean you can't find out what hardware you were using or what happened at boot. Keeping dmesg free of advertisements and data that could be confusing is a good idea, and elimiates the redundency of having things like boot.log, messages, and syslog all saying exactly the same thing.

Not *all* messages

[PurpleBob]

Some people seem to be confused as to which messages will be removed. The messages that Linus was specifically lashing out against were the kind that tell you the version number of every driver the kernel is loading - for example, from my dmesg, "Linux NET4.0 for Linux 2.4 - Based upon Swansea University Computer Society NET3.039". I really don't care. If something goes wrong with NET4.0, *then* I'd like to hear about it. If it's Swansea University's fault, then maybe I'd want to hear about them too. :P Another example is the famed "ReiserFS is brought to you by MP3.com" message.

Linus specifically said that the important kind of messages are the ones that are displayed when something isn't working properly, so no more whining that you think those messages are going to disappear, okay?

Moreso, some people seem to be under the delusion that the linked Slashdot article with the "pretty bootup" is related to what Linus said. No. Linux isn't saying anything at all about a graphical boot, and that graphical boot system is still hugely experimental anyway.
--

Re:Not *all* messages

[zoftie]

All the information is important, especially when
you are tracking down features that were inserted
into kerenel, by suse people and break my
network card drivers. All I knew my bootup was
failing, but I did not know where. Proximity
of failures maybe keypoint in determening why
current kernel does not work. I like to see
all subsystems reporting to me, at boot. For
ones who want pretty screens you can use Aurora,
it works well without any modifications, but
adding framebuffer console to kernel.

I agree "Reiserfs brought to you by MP3.COM"
should be gone. I could not believe my eyes,
whe 2.4.0 loaded.

Re:Not *all* messages

[mccalli]

>If something goes wrong with NET4.0, *then* I'd like to hear about it.
What if the bug you're dealing with is a failure to print information when something goes wrong? Worse still, what if it doesn't know something is about to go wrong?

Take this example:

• Foo 3.6 is compatible with Bar 2.8
• Bar 2.8 loads after Foo 3.6
• Bar is upgraded to 2.9, which breaks Foo 3.6

So Foo 3.6 is unable to say that something is about to go wrong, because it doesn't know. Bar 2.9 may still function as it was never aware of Foo 3.6 anyway. Result? No error message, but a broken system.

Fixing this today, I'd search the newsgroups, looking for problems reported with a similar driver setup to mine. I'd know which drivers I had because dmesg would tell me. Tomorrow under the 'no version messages' policy? I wouldn't necessarily know which drivers I had installed, and would have less chance of resolving the problem.

An aside: The people least likely to want a verbose boot are also those most likely to not know which drivers their distro gave them. Without a more technical person being able to ask "Show us the result of dmesg" on the newsgroups it will be extremely hard for them to get out of this situation.

My solution? I support the multiple boot mode option that others are discussing. Make the standard boot pretty - I'd use that mode most of the time. However, I think that a more verbose mode should always be allowed for when problems arise.

Ian

Bad move Linus

[joq]

Personally I don't use Linux anymore so I don't really care, but I think it's a dumb move on Linus' part and here's why. Suppose you've just been hired as a sys admin in a company and have to recompile kernels across a couple dozen perhaps hundreds of Linux based servers... Suppose that company was crappy via way of having things documented.

It's so easy to just do a dmesg and see exactly what's being used in order to recompile the kernel, as well as determine should something be replaced, or removed to make things better for the network. Anyways it's his dictatorship... err... OS, however it'll just makes things a bit more difficult for some

Re:Bad move Linus

[Andrewkov]

You could always use lsmod for a list of loaded modules, and also, if the machine in question has a non-standard kernel compile, you could run make config to see what was originally done.

Just out of curiosity, what OS are you running since you gave up on Linux? (Please don't tell me Win2K!)

---

Some users don't want pretty boot pics

[quartz]

I hope this will go towards clearing the bootup logs so that they only show useful debugging info, not towards establishing some sort of "pretty boot" as default. At the very least I hope it will leave me with a choice. If I wanted my computer to show pretty pictures when it boots, I would have bought a Macintosh or something.

Maybe it'll look a little more professional

[Zigg]

I've always felt since day 1 of using Linux that about half of the kernel output was ego-stroking, and very little useful information. It made the whole thing feel less like a unified project and more like a haphazard collection of little bits and pieces.

I hope the opportunity will also be taken to give everything a more unified look, a la commercial UNIX and *BSD kernels; it will certainly make finding what you're looking for in the dmesg much easier rather than having to sift through rabble.

Re:Annoying boot messages

[deKernel]

Dude, you just made my day by making me laugh real hard.

Thanks. I appreaciate the humor.

you guys are missing something important....

[23]

apart from talking common sense as usual, our dear Linus doesn't do justice to the grand master Tolkien. so sad...:

... and if somebody wants to change the name to "Frodo Rabbit ", I wouldn't holler loudly...

I guess Linus should grab his old copy of The Hobbit or LOTR and read in it a little again. It's always worth it! Would take his mind off nasty kernel-flamewars...
;-)
cheers,
Roland

Re:So in shocking news...

[Phork]

os7, and i think 6(not sure though) did the exact same thing with icons.

Don't get rid of "OK" messages!

[Webmoth]

I agree that the messages that don't add to debugging shouldn't be there in the first place, but I really don't want to see the "OK" messages disappear.

If I get neither an OK or a FAILED message, then I know that the thingy isn't even attempting to load.

Re:Don't get rid of "OK" messages!

[RobNich]

Look here, Redhat user, those are not kernel messages.

"the thingy" BAH!

Re:I hope it will be optional

[orangesquid]

All too often I'll boot an experimental kernel only to have it die on bootup -- no error message, no warning, but the last message will be, for example, "SCSI: ncr53c8xx version 0.3" -> then I know about where it died, and it gives me a good idea of what to take out of the kernel to try to get it to boot. And you can learn a *ton* of stuff about exactly what is being initialized in what order... Linus said something like "not to notify the user of every single driver loading" -- but that's *exactly* what I want! It's _my_ damn processor, and I better know what it's doing. Grr.

Re:I hope it will be optional

[orangesquid]

I find that if I disable everything experimental except for my system's drivers -- the machine doesn't boot. If I had a choice over what hardware I was using, I could pick something with a more developed driver, but until the day I'm rich enough to buy reliable and tested components... Believe me, I spend plenty of time making sure I'm not loading extraneous drivers nor anything I haven't tested before (except for new hardware or "improved" drivers which it seems more often than not need to be patched to work); however, I consistently run into problems with the firewall drivers... They don't compile as modules, which means I have to recompile, install, reboot, test, recompile, install, reboot, test, ...

They seem to have fixed a lot of the firewall problems in 2.4, but some 2.2.x kernels were headaches under which to get running ipchains properly.

And, by the way, I don't need to call in a contractor. I'm very careful about what I do and a decent observer, and I've never found a problem I couldn't fix.

I wouldn't have money to pay a contractor anyway :)

Yes, I always RTFM.

(I guess my hardware's drivers really *are* that experimental.)

Re:So let me get this straight

[fred911]

Don't get me wrong.. but what's so hard about hitting the ESC key?

(me thinks you have to do it twice to see all the drivers load)

Re:I hate to sound like an old bore, but...

[kimihia]

is to suggest that maybe some sort of lilo option for verbosity my be the solution here

I believe Mandrake 8.0 has just that.

Hmmm, sort of changes the meaning of 'boot image' doesn't it? :-)

Re:No wonder you're not working with linux

[LordNimon]

Config file - you mean .config? Tell me, when I do a stock installation of Red Hat Linux, where is the .config file that represents my system?
--
Lord Nimon

Re:No wonder you're not working with linux

[LordNimon]

Let me reword my question. If I install Linux (whichever distribution) and then go to /usr/src/linux and type in "make", will the resulting vmlinux (or vmlinuz) be identical to the kernel that I'm running right now? I don't think so. It will be different. My network adapter probably won't be recognized because it's not the "default".
--
Lord Nimon

Re:Please oh PLEASE!!???

[RobNich]

We are not talking about servers, we are talking about DESKTOPS. They are turned off and on frequently. The boot messages are annoying and obscure actual errors.

An noone said to REMOVE THE MESSAGES. The discussion is about NOT DISPLAYING THE USELESS ONES during booting.

If you read the article, you weren't paying attention.

"good status" messages

[Greyfox]

If I read that right, you can't get rid of those. I use "good status" messages to easily determine that my hardware is detected. I particuarly use the Ethernet and Sound driver messages on a regular basis.

Sure I could go digging around in proc and eventually turn that crap up, but it's usually much easier to grep dmesg or watch the kernel bootup. You might be able to squeeze everything that's needed into one line, but you still need some message during bootup.

Re:noooooooooooooooooo!

[T-Punkt]

Exactly!

Some time ago I've written a small perl script that can parse the NetBSD dmesg (of all architectures! - well, I tried it on 10) and create a device tree from it. Try that with linux!

BTW: Here's an example (actually this is the shortest dmesg of all my machines):

NetBSD 1.5W (XXX) #36: Sat Jun 9 22:48:54 MEST 2001
XXX@XXX:/usr/src/sys/arch/vax/compile/XXX

VAXstation 3100/m{30,40}
cpu: KA41/42
cpu: Enabling primary cache, secondary cache
total memory = 8076 KB
avail memory = 5732 KB
using 126 buffers containing 504 KB of memory
mainbus0 (root)
vsbus0 at mainbus0
vsbus0: interrupt mask 8
le0 at vsbus0 csr 0x200e0000 vec 120 ipl 14 maskbit 5 buf 0x254000-0x263fff
le0: address 08:00:2b:16:d8:1c
le0: 32 receive buffers, 8 transmit buffers
dz0 at vsbus0 csr 0x200a0000 vec 304 ipl 14 maskbit 6
dz0: 4 lines
boot device: le0
root on le0
nfs_boot: trying DHCP/BOOTP
nfs_boot: DHCP next-server: 10.0.0.1
nfs_boot: my_name=XXX
nfs_boot: my_domain=XXX
nfs_boot: my_addr=10.0.0.10
nfs_boot: my_mask=255.255.255.0
nfs_boot: gateway=10.0.0.1
root on XXX:/XXX/XXX
root file system type: nfs

parsed you get a tree like this:
+root
|VAXstation 3100/m{30,40}
+cpu
|KA41/42
|Enabling primary cache, secondary cache
+ram
|8076 KB
+mainbus0
+vsbus0
|interrupt mask 8
+le0 csr 0x200e0000 vec 120 ipl 14 maskbit 5 buf 0x254000-0x263fff
|address 08:00:2b:16:d8:1c
|32 receive buffers, 8 transmit buffers
+dz0 csr 0x200a0000 vec 304 ipl 14 maskbit 6
|4 lines

Hold music

[Twitch42]

Useful or not, I want to see them. Heck, I like to watch disks defrag. Window boots are torture; the visual equivalent of hold music. You mean I'm supposed to sit quietly for a full 60 seconds with nothing to do? Besides, they're impressive. Especially if you throw in a few lines like echo 'Plutonium containment field initialization failure' in your boot scripts. The more the better, I say.

So?

[jarran]

Jeez! Why bother? What difference does it make? How often to you actually boot your computer anyway?

I mean, I start up my computer once a day. I know, I know, it's a lot. And yet somehow I manage to cope with the daily trauma of my computer displaying messages as it boots. In fact, some days, I barely even notice them!

Does anyone else get this, or am I unusually Zen?

No wonder you're not working with linux

[BierGuzzl]

Relying on dmesg output to decide what to configure a kernel with is error prone and SLOW. any decent admin would have the config file handy.

Re:No wonder you're not working with linux

[denshi]

No, config file means:

/etc/rc.d
modinfo
/etc/modules.conf
whatever kind of package management tools you run on your system.

Logging info you would look for in dmesg:

/var/log/syslog
/var/log/messages

Stop taking shit out of context. You know the poster didn't mean one mythical '.config' file.

Re:No wonder you're not working with linux

[wierdo]

Config file - you mean .config? Tell me, when I do a stock installation of Red Hat Linux, where is the .config file that represents my system?

Perhaps in the kernel-source RPM?

I don't know about RedHat, but on Mandrake if you make menuconfig in /usr/src/linux, on a virgin install, it has the options the default kernel was compiled with selected.

-Nathan

Care about freedom?

Re:No wonder you're not working with linux

[wierdo]

If I install Linux (whichever distribution) and then go to /usr/src/linux and type in "make", will the resulting vmlinux (or vmlinuz) be identical to the kernel that I'm running right now?

In Red Hat and Mandrake, (and most others, probably), the drivers for almost everything (sans IDE disks), are built as modules, and a suitable /etc/modules.conf is generated at install time. Then on each boot Kudzu checks for new hardware and adds any necessary gunk to modules.conf to get it to load when you try to access the device.

So if you do make menuconfig (exit, yes save this), make dep;make clean;make bzImage;make modules, the generated files should indeed match the default Red Hat or Mandrake setup. As soon as you start compiling drivers into the kernel directly, you cease running the distribution's kernel.

-Nathan

Care about freedom?

Read Linus' post.

[BierGuzzl]

You don't lose useful information, only useless information. "success" type status messages will be removed as well as other messages containing things like people's names,etc, leaving behind only stuff like "hey you there, something's broken" thereby making dmesg 10 times more useful for debugging.

Naturally, this means that a dmesg output doesn't give you any idea of what a system is running, but we have other better ways of displaying that through for example the proc interface.

Re:Read Linus' post.

[stilwebm]

Success is in fact useful, not only for the reason you mention. I find it quite usefull, especially when installing new kernels on fringe hardware (say a new G4 PPC). It is good to see that the usb driver loaded and detected a OHCI hub with xyz device connected. And when the new kernel doesn't work at boot, you don't always get usefull error messages. But with success messages, you can tell what IS working, and usually figure out what caused the freeze. I wouldn't mind having a progress bar or status indicator like others mentioned, as long as I can still get the same basic information from my boot logs.

Re:Read Linus' post.

[kbeast]

aren't success logs helpful as well? What if your new to the scene and trying to learn...

.kb

I'm going to hell

[Pedrito]

I know someone is going to put a hit out on me for saying this, but why not do what Win95/98 do. Put up a picture of something and if the user wants to see diagnostic messages, they just hit a key...

Re:A better solution

[goldfndr]

I think you missed the word "hang".

Stephenson might disapprove

[crucini]

Neil Stephenson wrote this:

Cryptic messages began to scroll up the screen. If you had booted a commercial OS, you would, at this point, be seeing a "Welcome to MacOS" cartoon, or a screen filled with clouds in a blue sky, and a Windows logo. But under Linux you get a long telegram printed in stark white letters on a black screen. There is no "welcome!" message. Most of the telegram has the semi-inscrutable menace of graffiti tags.

In In the Beginning was the Command Line. A little transparency is lost. The Eloi/Morlock split of which Stephenson wrote is reinforced - you won't know how to view the kernel messages unless you're already a member of the technical tribe.
However I'm still not clear whether Linus wants to squelch all 'happy messages' or only author attributions. If it's the latter, I don't care.

No more boot messages?

[Delrin]

But how am I going to know how many bogomips my processor is? ;-)

Re:No more boot messages?

[cicadia]

Try cat /proc/cpuinfo

No penguins... but you do get to see the stepping number of your CPU!

Re:You guys want to actually try reading the messa

[shayne321]

I dunno, I certainly understand pulling out all of the "I'd like to give a shoutout to my peeps back in Brooklyn -- (c) G'love" type of messages, but I think versions are useful. There have been times when tracking down a problem with a NIC or SCSI card I need to check the version of the driver in the kernel. It's 1000 times easier to dmesg then look for the driver version than it is to go walking through the kernel source tree and lessing through the .h and .c files looking for version info.

Of course, if rather than printing to the screen if the drivers would drop a file to (say) /proc/drivers/((device)) so that you could ls -l /proc/drivers to see what's compiled in your kernel, or cat /proc/drivers/((device)) for all of the startup info, version info, shoutouts, etc - that would ROCK.

Shayne

Useful for configuration

[Krellan]

I disagree with this. I have used the Linux boot process to identify an unknown system that a friend wanted Windows installed on, so that I'd know which Windows drivers to look for! I find one of the best ways to take hardware inventory of an unknown computer is to attempt to boot Linux on it. Even if the boot is unsuccessful, the console messages tell a lot about the hardware and what's in there!

I sincerely hope the drivers aren't edited down to prevent this logging. All of the text should be available to read with dmesg. What I hope is that a compromise can be reached: default to a pretty-boot process, but have a way for the knowledgeable user to still see the console (perhaps by pressing the ESC key).

Or, use the priority system of printk to prioritize each message, and have an optional boot parameter that would control how many messages get printed. It could be something like bootverbosity=8, given on the boot command line. This already exists, look at the argument to klogd -c!

Doing a kernel-wide purge of all informational messages is just the Wrong Thing, as these messages will then be lost and there will be no way to print them when they are needed to solve problems!

Super eurobeat from Avex and Konami unite in your DANCE!

cool your overreactive jets folks

[RestiffBard]

um the way i read this is that linus doesn't want our dmesg full of crap about the maintainer's email address and other gobbledy gook that just doesn't matter. all i want to see on boot up is
keyboard.......ok
mouse..........ok
modem..........hosed---->read such and such to fix problem.
monitor........doing better than the modem apparently

Could this be made optional

[heikkile]

There seems to be some agreement that personal names, copyrights, etc. are harmful. The opinions here seem to be divided on the usefulness of successfull messages "Driver Foo loaded OK and detected 2 Foo cards at oxF00 and oxBAR". Those messages annoy some, and are very useful for some. So, why not add a boot parameter, a key, or something, to make them visible when needed. When not needed, they could be kept away from confusing the poor users...

Re:So let me get this straight

[KingAdrock]

Well.. It isn't as detailed as it is in linux, but if you hit Esc as the windows start up screen is displayed you are kicked out to a screen where it shows certain drivers (CD..) being loaded. I think if that Linux ever wants to become mainstream (insert grandpartent example here) they need to hide the stuff that would scare the average computer user. This is probably a step in the right direction as long as the info can still be seen in logs, or via escaping the startup screen.

Well Then...

[KingAdrock]

The linux community is always complaining that they have the better OS and they are poised to claim their throne by toppling Microsoft in the OS market.

I'm not sure I believe this at all, but if Linux hopes to get any sort of userbase they are going to need to make things much simpler.

I'm not saying that this is a earth shattering development, but I am saying that it is a step in the right direction!

What's wrong with both ways?

[flsquirrel]

As a sys admin, I have to agree with the greater sentiment posted so far that I want my boot messages. I don't care about seeing a pretty graphic of Tux dancing while my computer boots. But for a long time, the debate has been raging over what it's gonna take to get Linux on the desktop. It's about time we realize that obscure text messages scare the common user. I'm sure the logs will still be there, but why scare the bejeezus out of a guy if you don't have too. This is only one more step in making linux seem a Little less elite to the geek community.

Perhaps what should be done(and may actually be the case, I really don't know) is that this could be set with a boot parameter. And the distro when you install it could ask you which you would prefer and adjust lilo accordingly. If a normal user with a graphic boot was having trouble, the tech guy could always pass the boot parameter for verbose messaging to disable the graphic boot.

There goes my claim to fame

[Sits]

Based upon Swansea University Computer Society NET3.039

No fair! I've been trying to pick up kudos by pointing out that the society I'm in is mentioned in the startup sequence of a world famous OS (even if the credit should have really gone to Alan Cox). I just hope the department gets round to changing the prospectuses in time...

Funny you should say that...

[Sits]

Although I'm the current president of the Swansea University Computer Society I doubt I could fix the Linux TCP/IP stack.

Still, as I pointed out in another post, I doubt the University itself nor its (excellently rated) Computer department had that much to do with it. Alan Cox really should have got the credit and now Linus has seen to it that I can't even grab false fame by association. Oh well.

Put two notices together....

[p0et]

So, Just after Microsoft verdict being vacated, Linus anounced that he didn't wanted to see any annoying boot messages... : ) Hum...

Is something fishy going on... : )

Messages like..

[Dr Caleb]

"Recieved Signal 11"?

Oh, wait, that was an annoying message on /.

You missed the point

[Adam Jenkins]

Does Windows or Solaris sit there when you boot up saying "yes! I found the keyboard!". "YES! I'm using an Intel Pentium CPU" "Yes, you've got a hard drive". How bloody useless.

The BIOS boot screen tells you things like what CPU, how much memory you're running. If you really want to know what irq your soundcard is using, then you can do it after boot. That's what I just don't understand. Sure, great you can see what your machine is running, and your hardware settings. But.. No News is Good News. Ie, if when I boot up, the soundcard isn't found.. THEN tell me something is wrong. Otherwise I'll assume it's found and would really rather not have my hand held at every step of the way, thanks all the same. Is Linux really so flaky that it's so necessary to say that things ARE working rather than aren't? I don't have a problem with the messages being ugly/looking technical etc. I do have a problem with their relevance.

To each their own; I guess I don't have the correct Linux mindset and should stick with real Unix.
--
Never try to teach a pig to sing. It wastes your time and annoys the pig.

Re:No more Bogomips or Tux then.

[Robert Borkowski]

Why should Linux be dumbed down for the 'my foot pedal isn't working'/'The internet is broken' crowd? I'll have my boot messages, thank you.

My Computer

[Animats]

I hate to say it, but Microsoft's "My Computer" access to the hardware tree is much closer to the right way to look at the hardware situation.

In fact, there's much to be said for presenting messages as openable trees, rather than text streams. I came across this in an application from Eastern Europe written by someone who was very good, but did everything their own way. The message file was a tree, and you could open up messages and see submessages. This is way ahead of using "grep" on text files.

Misconceptions!

[dorward]

Judging from the comments it appears a lot of people are inturpreting this as "Linus thinks we should have a pretty graphical start up with displays no information at all", this isn't so.

If you read the email linked to by the artical it is suggesting the zapping of information that does not need to be displayed at startup such as what version of an obsure driver you are using.

Linus made no mention of the pretty bootups, they are a seperate an only indirectly related item.

I say we keep 'em ....

[Enonu]

Otherwise I won't get to gloat that my BOGO MIPS is higher than yours.

Re:Um, isn't LINUS'S NAME IN THE MESSAGES?

[krogoth]

Did you actually read the post? (did I even need to ask that question?) Linus says that his name, in the one place it shows up in this log, was placed there by someone else, and it could be changed to any name for all he cared.
---

Re:But I always wanted to see

[krogoth]

I think you missed the point - Linus is saying the boot process shouldn't look like a movie with the credits at the start. He is telling developers that the users don't need to be told the version and authors of every module each time they boot. I agree.
---

Please read the real story before replying...

[krogoth]

I think the link to the other slashdot story, the one everyone follows first, is confusing everyone. This is not about removing useful information from the boot process!! Linus was just saying that we don't need to be told the version and authors of every module and driver every time we start our computers. AFAIK, he is for useful messages at boot time, he just doesn't want the boot process to look like a movie with the credits at the start.

And for all you people who want this to be optional - if it is done (and I hope it is), it will be fully optional. If you're one of the 0.0000000001% of people who actually need to know this, you can always find the version information and credits for any driver. That's all Linus wants to remove from the boot messages, and I hope he gains support on this, because I want to use my computer, not read the full credits for the kernel 5 times a day.

note: I do not currently use linux, but I plan to install the Mandrake Powerpack I recently bought as soon as I get my new computer running. This may cause a few innacuracies in my post, but I belive that I know enough (and I do have some linux experience) to correct the mistaken ideas too many people seem to have gotten from this story
---

Re:Please read the real story before replying...

[krogoth]

Because I may not want to leave it on all the time...
---

Re:I hope it will be optional

[krogoth]

This isn't about the same thing as the slashdot story linked from it. This is about not telling you you're using HardwareDriver 1.34.135 by Abjkl Magdinadf. That is exactly the kind of thing we don't need - if I want to know the name of the person who wrote my drivers, i'll find the information myself. I don't need to be told every time I start my computer. This is optional - the messages like this should be removed from the boot process, and if you want to know you can just get the version information.
---

Re:So in shocking news...

[DrgnDancer]

This is also only refering to the the kernel startup. That goes by so fast that you couldn't read it if you wanted to (at least it does on all my machines). If your distro alerts you of daemon startup statues, as I know Red Hat and Suse do, that won't be affected. Since 99% of the time, that is the data you really want/need at startup, (and even if you needed the kernel startup data, you'd have to search the logs, 'cause you can't read it at start up) I think reducing the Kernel dialog to the screen is a good thing.

Re:I hope it will be optional

[nomadic]

if I want to know the name of the person who wrote my drivers

Sounds like Linus wants only his name on the system...
--

Re:I hope it will be optional

[nomadic]

To me this does NOT sound like an ego-maniac wanting his name and only his name all over the system.
I know, I just like provoking people who subscribe to that idiotic cult of personality. Actually I always assumed he called it linux to rhyme with minix.
--

Please oh PLEASE!!???

[silicon_synapse]

Please read the article folks. It doesn't take long. The only messages that will not be displayed during boot are those that don't deal with how the system is running, etc. That means no version numbers, no author names, etc, etc.

...So let's simply disallow versions, author information, and "good status" messages, ok? For stuff that is useful for debugging (but that the driver doesn't _know_ is needed), use KERN_DEBUG, so that it doesn't actually end up printed on the screen normally.


--

Re:I hope it will be optional

[Ravenseye]

I just junked the graphical boot screen on NetMax because I don't like the idea of things grinding away under a pretty (or ugly) graphic. Even if I don't like the text, at least I can see what the system is doing....or what it's stuck on. If we're going to go graphical, then we need to be able to hit a key, change a config or use a LILO option to boot textually. But I think what Linus is saying is that we should be seeing boot messages that mean something, not ads for someone or chest-pounding geek speak designed to echo how much tech talk we can slap together. If you want to tell me something, tell me. Don't show me a picture nor read me the spec sheet.

Actually, that'd be quite _unlike_ windows...

[drycht]

I don't remember Windows having anything over a crummy VESA 8-bit display for boot-up.
I'm all for graphical-stunningness in the bootup. I can handle dancing penguins as long as there's a little corner of the screen where there are boot messages scrolling up in a obfuscationally pleasant manner. That or if (infinitely niftier) there was a graphical representation of the boot process displayed [e.g. an fsck where little sectors were displayed on a disk and little red portions appeared where problems were found].
Anyhow, all I'm saying is that, though there is plenty of evidence to the contrary, shifting into a more graphical computer interface is not necissarily a bad thing.

-- nath

Re:I think we're taking this wrong

[kableh]

Isn't one of the driving factors behind contributing to Open Source software getting mad propz? Maybe a better idea would be a pretty bitmap (ala lilo on RH 7.x) and you can hit esc to see the full output?

it sounds good, but...

[cyberconte]

even removing the "useless" stuff like "driver (female) loaded successfully" can be helpful...i mean, what if the next message is "driver (male) failed: seat full (IRQ 11)"? without the previous message, all we know is *something* is using IRQ 11. even if the first driver didn't specify what IRQ it was using (as in my example) we know its one of the drivers before driver(male) and not after. When you load lots and lots of drivers, this can be a big help! Insert witty comment here.

Re:So in shocking news...

[QuaZar666]

according to the ads on /. it probally can make coffee. (the ads speak of OSX being the best java platform)

Qua

Annoying boot messages

[rizzo242]

Besides, who wants to see that stupid self-promoting "Kernel panic" message, anyway...all it does it hold up the boot process. Sometimes indefinitely!


"Sweet creeping zombie Jesus!"

Re:I hope it will be optional

[Eric Gibson]

That's the whole point, all of the information in here is useless because more precise/the same information can be found elsewhere. Things like errors will obviously be left in. What we have now is a just a rolling of credits for linux developers, basically. Other OS's don't do this, even commercial UNIX, because if you don't know how to find this information the right way, you shouldn't be using it!

I know, I know people will go on about that last statement, "We need wider acceptance. Linux needs to be easier to use...", blah! I've been doing private consulting for linux based networks, and let me tell you I've seen some VERY FUCKED NETWORKS that were caused by linux distro's that were too easy to use! This is the whole reason that the tech economy is busting, you have all this easy to use, "automated" software. So nobody spends anytime on design and understanding what really goes on under the hood and in result make crappy setups. So in conclusion, keep the bullshit out of my bootup messages, make me learn where to find it the right way, and make everything harder. Because in the end it's going to make my life, as somebody that actually takes the time designing, writing requirements and documentation, a hell of a lot easier!

Re:I hope it will be optional

[Eric Gibson]

All too often I'll boot an experimental kernel only to have it die on bootup -- no error message, no warning, but the last message will be, for example, "SCSI: ncr53c8xx version 0.3" -> then I know about where it died, and it gives me a good idea of what to take out of the kernel to try to get it to boot.

When I build a new kernel, I usually spend about 20 minutes going thru every single option. Making sure my modules.conf is correct and all my settings are right. Then... I got thru them all again to make sure I didn't miss something. I rarely if ever have a problem that leaves my system unbootable, most of the time it works the first time. If you use whether your system boots or not as an indicator of whether your kernel is built correctly you deserve what you get. You sound exactly like one of admins I replace as a contractor, so rock on buddy, thanks for the cash.

Re:OT: Crapflooding

[2Bits]

Ok, Taco, where's that filter of 2-minute waiting for each post to prevent this kind of crap flooding, when it is really needed?

Re:OT: Umm shouldn't the lameness filter ...

[neo-phyter]

mmm hmm. and you -1. Just plain weird.

A better solution

[Viadd]

A better solution for pretty boots would be to print "Installing whizzomatic driver v 3.14" when the driver starts to install, then after the installation is complete, printing "\r_________\r" to erase the line and go back to the start.

This way, the message is only on the screen for the duration of the time the installation is running. If it hangs, you have a mea culpa sitting there. If it doesn't hang, but the next thing does, the whizzomatic is absolved.

Good == Good Bad == Failed

[DA_MAN_DA_MYTH]

When I first started using linux all I saw during the boot process was this failed failed failed... But seeing OK, GOOD OR PASSED is like a daily affirmation, 'I am learning this stuff', or 'at least I haven't fscked up my machine that badly yet.'

Does this also mean that they are killing it when you close down your machine too? It's cool to know what is taking so long to shut down your machine, unlike Windoze's descriptive message: 'Windows is now shutting down.'

The worst vice is advice...

need...Re:Need multiple boot message modes.

[leuk_he]

1.Splash-screen mode.
windows 95 mode. show nothing, log it too boottxt.log if pressed F5 on startup.
2. Concise mode:
Dos 6.0 mode. show starting up..... (going init level 6)
3. Verbose mode
Geek mode. (like it always was) maybe it should also be possible to confirm every driver that is loaded with Y/N

However if this makes it in linux 2.6(?) i will be missing the slashdot "linux now runs on Risc-Q" see the Boot log here on the /. home page.

No more Bogomips or Tux then.

[Mtgman]

When I boot Linux I see a little pic of Tux and a line saying something like "ABCXYZ Bogomips"

According to Linus's definition, this information is useless. It doesn't help me debug anything and isn't necessary for the system. I think it's a cool thing personally, but I can undestand concern that this bootlog space would be used for spamming the Linux using community. And whenI say Linux using community I'm not talking about sysadmins who tweak out their systems to the utmost. You're not users guys, you're not the audience Linux needs to reach.

Steven

I hope we get a boot option for it

[Urd]

I like the idea of seeing less. The less junk the more usefull the bootup messages are.

For use in a general user environment with stable hardware and a chosen kernel it doesn't make sense to see the same thing over and over again. At most one would count on using this when installing a new kernel or some new hardware.

Going further on the idea of levels of verbosity however, we really could use a boot option to help with possible diagnostic boots.

And let not forget for the real hardcore computer kids that a nice look adds a great deal to the end-user experience.

Me.

Re:So let me get this straight

[jawtheshark]

Well I like the boot messages....At least you see what your computer is working on. A lot of modern computers show a bitmap while doing POST (Dell screen, Toshiba screen, Tulip screen, whatever...). I disable it when possible because I like to see my puter check it's memory, giving a report of IRQ usage, PCI bus stuff, SCSI finding the drives, etc...
I just like to be informed...That's all...

More is better than less

[mirwor]

I think Linus is wrong, we should have a lot more messages for the individual drivers cause when more is printed on the screen while booting, the text ist scrolling by faster, and the faster the text is scrolling by, the faster linux seems to be. Just show all that windows lusers how fast Linux is with our ultra-fast Linux bootup ;)

Re:So let me get this straight

[Denial of Service]

Hey, who wouldn't want a hot burnette in a skimpy bikini...

Carol Burnette? Man, I am with you there.

---

I hope it will be optional

[iseletsk]

I will miss thouse messages, without them linux will loose some of its charm for me ;-)

Re:I hope it will be optional

[Evil Grinn]

I wish there was a way to get some of the same information to show up when I boot Windows 2000

Hit F8 and choose "enable boot logging".

Re:I hope it will be optional

[triticale]

It got passed my spell checker. It musth be correct sew what are you complaining about?

Re:I hope it will be optional

[einhverfr]

I hope so too. Imagine-- you write a startup script for, say, Tomcat JSP server and suddenly your system will no longer boot (OK, technically this is part of the init process and not the kernel boot but the logic applies there too).

These messages are usefull and I hope will either be logged or better yet be optionally echoed to the screen instead of prty pics. Wven WinDoS has an option for bootlogging.

A more real example-- troubleshooting an ethernet card you just installed. The boot messages can tell you quite a bit about its initializaton on the kernel level before your network script is run. This can tell you whether you need to recompile your kernel.

Re:I hope it will be optional

[FuegoFuerte]

Sounds like Linus wants only his name on the system...

Did you happen to follow the first link in the blurb? It goes to a mailing list archive, in which someone points out the Linus's name is printed next to some driver. Linus replies: "It wasn't addded by me, though, and if somebody wants to change the name to "Frodo Rabbit", I wouldn't holler loudly." To me this does NOT sound like an ego-maniac wanting his name and only his name all over the system.

Re:I hope it will be optional

[lvv]

Spelling police again. I remember black driver in NYC was boasting that he knows English better than any other cabbie in Manhattan. And his language was pretty far from English. Why people in US are so proud that they know only one language? And why they are so intolerant to any one who doesn't know theirs language perfectly?

English is not my first language too. And not second either. Isn't it obvious by name of mis-spe ller, that he is not Anglo-Saxon origin? I can bet that he didn't have English in his first grade.

Re:I hope it will be optional

[SilentChris]

Same here. After installing Linux on a friend's computer, she was in awe of all the textual information on boot. Of course, she wanted a graphical login menu to show up in a few minutes (as most of us -- myself included -- do). I kind of like all the professional sounding terms that come up while you boot (although seeing a driver in 0.09 stage is a little frightening).

Hopefully it'll be optional, although I really doubt it wouldn't. That kind of thing needs to be at least a little understood (I wish there was a way to get some of the same information to show up when I boot Windows 2000. Good [expensive] OS, lousy under the hood documentation.)

Re:I hope it will be optional

[zangdesign]

If you want Ma and Pa Walmart to accept Linux (which presumably is the case) lose the messages. I suspect most people really don't care if there's a monkey with a hammer and chisel inside the iron as long as it works and isn't confusing. HOWEVER, that being said, superusers need to be able to view EVERYTHING just to make sure all is copacetic. Little Bobby UTGamer doesn't.

Re:I hope it will be optional

[bigboard]

It's not 'their' language. And they officially spell loads of words incorrectly. ;) But to be honest, who cares about the spelling if you can understand the meaning? Life's to short to worry about such things.

Re:I hope it will be optional

[stark_fist_05]

maybe it will be a patch

Re:I hope it will be optional

[nixxy]

I too hope they will be optional. I enjoy seeing what my system actually does when it restarts (on the rare occasion that it does) not just selected errors
------------

anyone bother to *read* the email?

[EllF]

no, linus doesn't want to replace the boot information with clouds, penguins, or anything else. FUD, FUD, FUD.

what he said was: "'Informational' messages aren't informational, they're just annoying, and they hide the _real_ stuff."

The "real" stuff is still there. What should be cut (and right on, Linus!) are "version strings, author information, and "good status" messages".

bleh. go read it.

Re:It's all irrelevant.

[Eunuchswear]

Everyone knows we never have reboot our Linux boxen anyway. ('cept to add new hardware of course.)

You reboot to add new hardware? How primitive, you've still not caught up with 1960's technology.

Linux - 40 years out of date and proud of it. :-)

OK/FAILED/Pass is NOT what LInus is talking about.

[aussersterne]

Some readers are operating under lack of information.

If you see "OK...OK...OK...FAILED...OK" when you boot up or "Pass...Pass...Pass...FAILED..." when you boot up, then you're thinking of the wrong messages. To see the messages Linus is talking about (the endless copyrights and chit-chat of the kernel) type 'dmesg' as root after a fresh boot.

The status messages that give you an OK or a FAILED on a system-by-system basis are (in the most broad sense) init messages rather than kernel messages and are added by your distribution. These will not go away under the "new deal" Linus is proposing...

Re:So let me get this straight

[Carpathius]

I didn't get that _at_all_ from what Linux said. Given that he first said that there weren't any extranious messages, I think that what he meant was that only those items of a completely non-useful nature would be removed.

What does that mean? I think it means that stuff like:

Jun 6 13:10:04 localhost kernel: hdc: ATAPI 1X CD-ROM drive, 32kB Cache
eth0: PCnet/PCI II 79C970A at 0x1000, 00 50 56 8e 6e 4d assigned IRQ 9.

will stay around. But stuff like:

pcnet32.c:v1.25kf 26.9.1999 tsbogend@alpha.franken.de

which (arguably) provides little information will go away.

I really don't think there's any intent to take away boot messages.

Sean.

DevFS

[Cardhore]

Rock Linux uses the Device Filesystem by default.

Re:What about credit and peer reckognition?

[megalomang]

This is a good point. For some, this may be the reason why they contribute in the first place. It surely isn't for the money...

Hope we don't lose our verbose logs with this

[Arethan]

Sure, pretty boots are nice and all, but I can't imagine what I'd do without my boot logs. It's kind of hard to catch those first few kernel messages as it is, and I'm usually rummaging through my log to see why the hell something is no longer working like it was on the last boot.

jihad?

[heataz]

did someone say jihad?

Seems to me....

[Xibby]

That what's being said is that the modules shouldn't be spitting out this information as part of their code, as modprobe could handle that function. It could then be turned on and off as needed. Makes sense to me, why does every kernel module need to have code in it to print out information when the software loading the modules could do it, it would then only be written once, could be configurable. (verbose level = X, or modprobe -v, modprobe -vv, modprobe -vvv, etc.)

If it fails, print out a message "Module X failed to load. Noncritical error, continuing to load linux. Fix me if you dare."

Less scarier...

[C0vardeAn0nim0]

for newbie users, for sure.

But itll be a Good Thing(tm) to be able to suply a kernel parameter via LILO prompt or loadlin to have the debug messages back, to allow an experienced guru to diagnose the stuff if something goes wrong, specially when the kernel panicks on you during boot.

--

Same info in Win2k

[Moghedien]

Easy. Add the /SOS-switch to the win2k-entry in boot.ini. It then displays the files it loads.

This means.....

[Quazion]

that i dont keep pressing shift-pageup the whole time to read the bogomips anymore.....

Owh you can do dmesg after a boot ? why would i wanna do that, i wanna spill that useless boot time on something usefull....that is if i reboot...damn why did i buy that NVIDIA card =(

This is a very very bad move, people!

[nidarus]

You see, eliminating the ugly and distracting unnecessary kernel messages will make Linux look somewhat more elegant, prettier and friendlier to the user. This is, of course, something that shouldn't be allowed to happen.

Why? When you can handle a system that prints hundreds of lines of meaningless and wonderfully cryptic information when it boots, you look cool. You are a revered computer guru. If you can handle a system that prints lines like "Loading modem...done" (or say, a pretty loading screen with an animated and uninformative stripe at the bottom just to show that the system hasn't crashed, but I'm digressing) when it loads, you're just someone who can turn on a computer. Big deal. Even I (says the PHB) can do that.

Don't get me wrong here, it's not just an ego thing. It has a direct effect on the average Linux hacker's/admin's salary. If you can handle a system that cryptic, you are indispensable. If you're indispensable, you get paid more.

In fact, I suggest that the kernel should actually print more messages, and make them much more cryptic and weird. Why not use more braces, asterisks, hashes, ampersands and "at" signs (ooh, those "at" signs)? Why not add blinking (and totally meaningless) messages in several different colors (I mean, several different colors within the same message)? Why not add some text-animations depicting obscure internal operations of the kernel?

I'm telling you, if these changes will be inserted in the kernel (and in the distributions - look at the atrocity that is Mandrake 8!), making a living by working with Linux will really start to pay.

Re:So let me get this straight

[The Vulture]

Actually, Windows doesn't show the output of ANY Windows drivers that are being loaded - they're just loaded (or fail) without any sort of an indication. The text that you're seeing on the screen is that from a DOS driver (you really shouldn't be loading ASPICD.SYS or MSCDEX.EXE with Windows 95 and higher, that can actually cause more problems then it solves - mind you, if your CD-ROM isn't supported through Windows native drivers, then by all means, go for it). Myself, I like the idea that Windows NT has, where any problems are logged in the Event Viewer (dmesg in Linux), and a message appears when you login (or maybe at bootup) when something went wrong, and it tells you to check the Event Log. -- Joe

Re:So let me get this straight

[kbeast]

its much prettier to see scrolling text (especially in a darker green color on black) then do see something flying...

.kb

Mac OS X & Informative Boot Screens

[owenc]

When Mac OS X boots up you see a box come up somewhat early in the boot process with a progress bar also labels what is being started as boot progresses ("Welcome To Macintosh", "Tuning System","Starting Network Services","Starting Apache","Starting Sendmail" etc). The progress bar itself looks really nice too, as it has that cool aqua water animation pulsing through it as it moves (cool!)... This to me seems a better mix between "eye candy" and functionality than simply throwing a full screen windows logo or a penguin on the screen.

Of course, you can boot up in "verbose mode" (scrolling text) by holding down command-v or something.

Re:Mac OS X & Informative Boot Screens

[owenc]

Yea, but this isn't Mac OS 8 :)

Besides, if you were the kind of person who'd want to see that kind of system text during the boot process, you'd be smart enough to activate it.

It's all irrelevant.

[Fizzlewhiff]

Everyone knows we never have reboot our Linux boxen anyway. ('cept to add new hardware of course.)

Re:It's all irrelevant.

[discogravy]

Everyone knows we never have reboot our Linux boxen anyway. ('cept to add new hardware of course.)

you have to reboot to detect new hardware? this is why windows is better than linux, it's Plug 'n' Play bug^H^H^Hfeature means i never have to reboot...the machine usually just crashes or GPFs for me -- i don't have to do anything!

-d.

--
Slashdot: When News Breaks, We Give You The Pieces

3 Options...

[nexex]

I like the idea of a graphical boot screen. Some like the standard text, as it is now. I like what Mandrake 8.0 has, plain old text, or you can have a static graphic similiar to Windows 95 etc, or you can also have a graphic one with icons across the screen that light up when the coresponding program/service is starting, which I think is the best of both worlds. Perhaps there should be a combination of these standardized?

I hate to sound like an old bore, but...

[RexxFiend]

OS/2 could do all this years ago. You had default driver messages printer to the screen at boot time more or less at the delelopers whim but if you needed to you could press alt-F2 (I know its a little obscure) at boot time and it would display the file name of each driver as it loaded so you could quickly work out which driver file was hosing the system and take steps to remedy it.
The point of all this rambling is to suggest that maybe some sort of lilo option for verbosity my be the solution here - the default is a picture of a nice fluffy penguin with a message saying "Loading" (or something) but you can type linux verbose (or something) into lilo to get all the nice driver messages if you need them.
Personally I like the driver messages - it reminds me of OS/2 :-)

A crash reduces
Your expensive computer

Re:Ummmm...

[KupekKupoppo]

Well, it's a good thing that Linus is calling for a standard way to do it then. If people know "it's expected to be silent on success, bitch loudly on failure," then the drivers will be written that way.

<HYPERBOLE>It's not like 99% of Linux isn't rewritten every day.</HYPERBOLE>

-k.

Who cares?

[glrotate]

If somebody who contributed usefull code into to kernel wants to put in a vantity message who cares? If you don't like it look away.

Messages for Device 'Drivers' are essential

[Myrv]

Ok, knowing what version of filesytem you have isn't really required but from my experience seeing what device driver is being loaded is essential.

For example, when I have multiple NICs installed in my box, being able to see what driver is being loaded helps significantly in debugging (particularly if the NICs are different). Knowing that eth1 or eth0 didn't come up isn't particularly useful when for new boxes I don't even know which card is eth0. Watching

eth0: 3Com 3c905B Cyclone 100baseTx at 0x6100, 00:50:da:1e:7e:51, IRQ 12
8K byte-wide RAM 5:3 Rx:Tx split, autoselect/Autonegotiate interface.
MII transceiver found at address 24, status 786d.
MII transceiver found at address 0, status 786d.
Enabling bus-master transmits and whole-frame receives.
eth1: RealTek RTL8139 Fast Ethernet at 0x6200, IRQ 9, 00:4e:4f:03:9e:c0

come up makes my day when booting a new router.

Of course these messages may not fall under the banned catagory.

Re:textulike

[fatgraham]

what does some flying windows or clouds tell you exactly??
its like a driving test (uk at least)

"prepare to stop"

WOW! Linux is perfect...

[bzzzt]

When Linus starts thinking about changing boot messages, that must mean the rest of the kernel is perfect 8)

A good move, if it can be undone at boottime

[TheAcousticMotrbiker]

Yes, 90% is absolute crap,
but the 10% that isn't is there to help me.

I like version numbers in my bootup, I've more than once solved a bootup problem simply by seeing the version numbers floating by on the screen.

As for Solaris, I've been forced to do a set -x in too many /etc/rc2.d/SXXxxxxx scripts too many times, just to be able to figure out what exactly was hanging my boot process ..

Ahhh, I see, it's fsck-ing my 420G harddisk, that's why it's sitting there idle for over an hour !!

As someone else has suggested, make kernel /modules verbosity either a kernel config option (those of us who need this info are usually the ones who built their own kernels anyway) or make it a lilo settable (append="bootup=verbose") option.

The latter would be great if you have to diagnose a box for which you haven't built a kernel .

Re:textulike

[roguerez]

I suspect he automated the process of posting from different ip's.

Re:textulike

[Marcus Brody]

Man I wish i hadnt posted now.....

I could have used those moderator points (i really do have some ;-) against that C*** SpanishExtermination (scroll down and you will understand) . He just posted that message 37! times (and yes, i am sad). By the 2-minute repost rule that makes, er..... 74 minutes. The story has been up.... what? 15 minutes.

any ideas how he does that anyone?

textulike

[Marcus Brody]

contacting host slashdot.org..... connected
slashdot.org contacted.... waiting for reply
downloading slashdot.org 10.5k/s
document:done
welcome marcus brody
congratulations! you have moderator satatus...
please go forth and moderate

words speak a thousand pictures. so much so, we often take the information they contain for granted.

what does some flying windows or clouds tell you exactly??
well... if they freeze you know your computer has crashed.

Re:So let me get this straight

[Marcus Brody]

Linus is saying that MS was right in hiding bootup information from the user and masking it with a pretty picture?

Linus: "Let's make it policy that we _never_ print out annoying messages that have no useful purpose for debugging or running the system, ok?"

Dont worry - Linus isnt going all Mac on us! As ever he is making more sense than your average hacker. I know what OS I'm running - and which kernel version, and even what modules im using. but yes some of this information is occasionly useful, and these are the bits that should be left. Maybe then they might not scroll off the screen so fast that we never read them anyhow.

I think we're taking this wrong

[Chakat]

Linus isn't saying that we should go the way of Windows and have a 1024x768x24bit bootup splash screen with OpenGL penguins frolicking in the background. All he's saying is that he thinks that the module loading messages should be shortened to such things as: Loading FooDriver.

The more I think about it, the more I feel this is a win here; we get a bootlog that's easily eyeball parsed without all the cruft of a driver maker sending out mad props to all the sponsors.

Though what I'd really be interested in seeing one of the big vendors at least experimenting with the DevFS system. It looks like it could clear up confusion and make things drivers/programs easier to write.

They should stay

[flippety_gibbet]

How on earth will people know how clever I am if my PC doesn't spew pages of abstruse messages when I boot?

Verbose Option

[nate1138]

Hopefully Linux will retain the verbose messages as an option at least, I for one kinda like seeing everything that's going on.

So in shocking news...

[Tyler Eaves]

Linux becomes more like Windows and MacOS?

Re:So in shocking news...

[srvivn21]

As a very talented webcomic says on his site: The First one is always Free.

Welcome to the first day of the end of your free time. Yeah, I know. It's off topic. But I haven't seen a proper welcome wagon...

Re:So in shocking news...

[jeffy124]

You really should only be concerned with Linux starting to resemble older versions of Windows. MacOS (any recent version) is able to show how far along the boot-up process is. OS 8 and 9 use those little icons along the bottom of the screen, OS-X uses a progress bar and a message saying what's currently being loaded. Win2K also uses a progress bar, but no messages.

I don't think the messages in Linux are gonna disappear completely, just not as many in number.

Re:So in shocking news...

[jeffy124]

very kool!!! learn something new everyday......

Re:So in shocking news...

[WoofLu]

Small precision on OSX:
it has a verbose option ("Apple key+ v" at system boot) wich looks like good ol' *nix boot.

And it supports color !

I bet it can make coffee too =)

--
Arnaud Willem
1st post of my life on /. =)

if you read the mail mesg ...

[yukonbob]

... you'll see that Linus is suggesting a change in policy ... ie: printk() isn't being disabled.

Let's make it policy that we _never_ print out annoying messages that have no useful purpose for debugging or running the system, ok?

-bch

I don't see the problem.

[Flying Headless Goku]

Flying Headless Goku - version 0.8 (the other 0.2 would be the head)

thinking of idea

considering approaches

initialization complete, beginning typing

I dont' se e the protlem with verbose messages, you can just ignroe tem if you're not interested in what they say.

spellcheck:
dont': don't dent donut
don't
se: sea see is se.cx
see
e: a

ignroe: ignore
ignore
tem: Tim term team them
them

finished processing!

I don't see the problem with verbose messages, you can just ignore them if you're not interested in what they say.
--

I like the current.

[hackie]

Please don't change those, or atleast make it optional. It always gives me a warm "I'm home"-feeling. And those messages seem 31337 to some Windoze users.

Re:BUT I LIKE REPETITIVE MESSAGES

[haruharaharu]

Karma to burn, right?

Personally...

[cuyler]

I don't mind the non-pretty cool logo bootup. The idea that I have with my computer systems is that I boot them one in a while. On my BSD box I've only seen the boot messages twice now and I've had that system for almost 4 months now.

The only real time that I'm booting up the systems are when something has gone wrong (power loss in house for instance) and in that case I'd like to see all the boot up messages. I don't really need to see "Network , brought to you by your friends at Coca-Cola" but it's better than seeing a flashy logo and wondering what exactly my computer is doing.

-Cuyler

PATHETIC

[thesk8ingtoad]

Why is it that people always complain about how they don't like to see linux go in this direction or that but do nothing to stop it? The entire point of Open-Source is the ability to take it and do what you want with it. If you don't like the way it's going, change it.

People act as if they are at the mercy of an all-powerful entity forbidding them to change the system. Basically, what I'm trying to say is quit bitching and remember what open-source is all about: Freedom. People who do no more than bitch about the way linux-or any other O.S. project- are throwing away this freedom and are no better than the alcoholic on welfare bitching about the government.

They aren't that annoying

[s4ltyd0g]

Well at least it doesn't go TADAA! evry time it performs some basic task.

What about credit and peer reckognition?

[s4ltyd0g]

IMHO it is perfectly acceptable to see who contributed to what on startup. How often do people have to reboot their systems anyway?

what a stupid idea

[m08593]

I have used systems that have eliminated these kinds of boot messages, and they are a pain to work with.

It may be obvious to Linus which version of the driver a particular kernel version uses, it isn't obvious to other people. In fact, in the presence of driver patches, there is just no way to tell. The bit of vanity, having people's names print out who put in a lot of effort into writing a driver, seems justifiable to me as well: these people don't get money, the least they can get is a little publicity. Besides, some licenses require copyright information to be printed.

The Linux kernel has lots of serious problems in terms of configurability, usability, and lack of support for crucial standards. The bootup messages are completely harmless, take almost no resources, and are useful to many people. Why fix something that isn't broken? I'm disappointed that such a stupid idea would even get seriously considered.

Actually...

[return 42]

I think Linus is just talking about messages that don't carry any useful information, for example, "module foobar (c) 1998 Yoyodyne Inc."

I suppose that kind of message has some limited usefulness if your boot hangs midway - you can pin down more exactly how far it got before it went kaka. But in general I think we can do without them.

So let me get this straight

[fa098h23fra]

Linus is saying that MS was right in hiding bootup information from the user and masking it with a pretty picture? Why! Windows obviously does this and often times fails to load drivers, or loads useless ones, etc, and the user has no clue why their machine takes 10 minutes to load. Isn't the point of linux to feel like a real geek who doesn't need technical details hidden from them? Maybe not, but I like my verbose kernel boot messages.

Linux Forum

[sonnyjz]

Come Discuss this topic @ http://forums.wontstop.com

Re:I hope it will be optional (Hear ye, hear ye)

[LoneSocialist]

I am fairly certain that this new setup will be optional. The boot messages are often -very- necessary to discover problems that plague a system, and I'm sure Mr. Torvalds is aware of that. I just hope that it will be a straightforward procedure to re-enabling the messages again, or else some people will be in for problems... The Lone Socialist