Hello World,
this eMail shall be a means of bringing up again a topic I believe has
already been discussed extensively. Wait! Don't delete, read further
please!
[ Note: please cc: me in any replies as <[email protected]>,
since (a) I'm at work and (b) not subscribed to the list. Thanks. ]
It is my very understanding one can not have, conveniently it should be,
a simple *bootscreen* under Linux. With that I mean a picture of at
least 256 (indexed) colours at a size of 640x480 pixels. Doesn't have
to be a higher resolution. And yes, I'm taking the standpoint that every
computer nowadays [where this shall be possible] *can* do that resolution.
Framebuffer, I hear people shouting? Well. During the last *two days*,
which includes one full night, I've been trying to get my v2.4.20 kernel
to display such a bootscreen. All I get is segfaults. I've tried what I
believe to be every tool out there: pnmtologo, fblogo, boot_logo, the
GIMP plugin. You name them. None of which wouldn't have required any
hacking to work with 2.4.20, by the way...
And maybe it's right, maybe I demand too much from the (VESA) framebuffer.
Maybe my picture is also too complex, but I've tried simple ones as well.
And anyway: I don't *want* any simple picture, I want as complex a picture
as it gets. In 640x480. At 256 indexed colours.
So although I'm just learning C and can't code it myself, here's an idea:
If Syslinux can display this kind of images, and if LILO can, so why would
Linux be unable to display it? VESA was the term, if I right remember?
If this request is too much of an effort to implement, then couldn't there
be a kernel configuation option that simply tells Linux to leave the screen
as it is, until some user space software (X) changes it? (In conjunction
with console=/dev/null or something). I just want my picture remain there.
I realize these ideas may sound kind of alien to you, but they make sense.
Windows, MacOS all have bootscreens. There really is no way why Linux
shouldn't.
In that veine, another thing I've been puzzled with... can you somehow
disable
virtual consoles (Alt-Fx) completely while still maintaining an interface
for
X to come up on?
Thanks for reading through until here. Thanks for any considerations in
advance.
Your truly, Raphael
> It is my very understanding one can not have, conveniently it should be,
> a simple *bootscreen* under Linux.
Linux machines shouldn't ever need to be rebooted, so you'd only ever
see it once, (on each machine) :-).
John.
Raphael Schmid, Tue, Jan 28, 2003 11:43:28 +0100:
> > I didn't mean use it with 2.5, i meant to port them to 2.4 ;)
> Ouch. Now that would be one hell of an exercise for me.
> I'm only learning how single linked lists work these days.
we all trust in you
> > why not write a prog which does all the setup
> > and execs xinit at the end?
> I have that. It's in effect an /sbin/init replacement.
Oh.
> But this is also a question of attitude to me. Even if
> only a fake, I want Linux to be a *graphical* OS. *g*
put svgalib in kernel and export it's interface?
> > You can redirect the output off the default
> > console. In this case you will see nothing.
> > Probably.
> I've just been asking this in another answer to LKML.
> Would mean the bootloader is the one reinitializing
> the screen, right?
not exactly. What you mean "reinitializing"?
It sets some video-mode and restores it afterwards.
Hi Raphael :)
> In that veine, another thing I've been puzzled with... can you
> somehow disable virtual consoles (Alt-Fx) completely while still
> maintaining an interface for X to come up on?
Disable the gettys in your inittab and boot in runlevel 3? (don't
remember the runlevel for the xdm). You need at least one, anyway...
Ra?l
Hello there,
On Tue, 28 Jan 2003, Raphael Schmid wrote:
> It is my very understanding one can not have, conveniently it should be,
> a simple *bootscreen* under Linux. I don't *want* any simple picture, I
> want as complex a picture as it gets.
>
> I realize these ideas may sound kind of alien to you, but they make sense.
> Windows, MacOS all have bootscreens. There really is no way why Linux
> shouldn't.
There is a very simple reason why Linux shouldn't have a "bootscreen" -
its a lame idea. We have copied enough of the bad "features" of Windows et
al into Linux already, IMHO.
FWIW, I usually remove the graphics from LILO/GRUB config as installed by
default these days.
Most of the machines I maintain are very seldom rebooted, but if someone
was to do a reboot, I would want them to be able to observe any errors or
other abnormal output from the boot-up process. A "bootscreen" makes it
more likely that such an error message would be more likely to go
unnoticed - and, if they became commonplace, may eventually result in
developers making the on-boot output less verbose/informative/etc, on the
basis that it isn't likely to be seen in the first place!
The stuff you see on your screen when your Linux installation boots up
weren't put there for no reason. Please lets leave them there!
Robert Morris
08707 458710
http://www.r-morris.co.uk/
On Tue, 28 Jan 2003, Alex Riesen wrote:
> > But this is also a question of attitude to me. Even if
> > only a fake, I want Linux to be a *graphical* OS. *g*
>
> put svgalib in kernel and export it's interface?
Fbcon, Kernel Graphics Interface (http://kgi-wip.sourceforge.net, no we
don't use ioctls or kernel space for accelleration anymore). Enough choice
for your "graphical OS" ;-)
Jos
> > It is my very understanding one can not have, conveniently it should be,
> > a simple *bootscreen* under Linux. I don't *want* any simple picture, I
> > want as complex a picture as it gets.
> >
> > I realize these ideas may sound kind of alien to you, but they make sense.
> > Windows, MacOS all have bootscreens. There really is no way why Linux
> > shouldn't.
>
> There is a very simple reason why Linux shouldn't have a "bootscreen" -
> its a lame idea.
For a desktop or server machine, yes.
> Most of the machines I maintain are very seldom rebooted, but if someone
> was to do a reboot, I would want them to be able to observe any errors or
> other abnormal output from the boot-up process.
Agreed, for standard desktops and servers.
> A "bootscreen" makes it more likely that such an error message would
> be more likely to go unnoticed - and, if they became commonplace,
> may eventually result in developers making the on-boot output less
> verbose/informative/etc, on the basis that it isn't likely to be
> seen in the first place!
There is no reason why the boot data can't go to a secondary display,
or a serial terminal, or a printer, or a speaker as a bleep code, etc.
> The stuff you see on your screen when your Linux installation boots up
> weren't put there for no reason. Please lets leave them there!
There are applications where it is not appropriate to have it, though,
what if you were using Linux in an embedded device such as a set top
box?
It's perfectly possible that somebody might want to make a television
set top box out of a standard x86 motherboard and VGA card, and not
have anything displayed until X starts, because the television would
not accept the standard VGA scanrate, but X can easily re-program that
to around 15 Khz.
In this case, boot data could be sent to a serial port, and the
graphics card initialised by the boot loader to display a "Please
wait, set top box booting up" screen, using a scan rate, which would
be acceptable to the television. In this case, we do not want the
kernel to change the video card setup at all.
John
Previously John Bradford wrote:
> There are applications where it is not appropriate to have it, though,
> what if you were using Linux in an embedded device such as a set top
> box?
Kiosks and things like ATMs are another place where you do not want
a bootscreen. You do not want to possibly confuse customers with
stuff that they can't understand but show a nice friendly message saying
'the system is currently unavailable'.
Wichert.
--
Wichert Akkerman <[email protected]> http://www.wiggy.net/
A random hacker
Hello there,
On Tue, 28 Jan 2003, John Bradford wrote:
> > There is a very simple reason why Linux shouldn't have a "bootscreen" -
> > its a lame idea.
>
> For a desktop or server machine, yes.
>
> There are applications where it is not appropriate to have it, though,
> what if you were using Linux in an embedded device such as a set top
> box?
I agree that it may be less inappropriate for certain specialised
applications, such as the one you suggested, but Raphael made specific
reference to Windows and Mac OS, which implies desktop use.
I am totally fed up with the quest to make Linux into as close to a copy
of Windows as possible.
> It's perfectly possible that somebody might want to make a television
> set top box out of a standard x86 motherboard and VGA card, and not
> have anything displayed until X starts, because the television would
> not accept the standard VGA scanrate, but X can easily re-program that
> to around 15 Khz.
OK, but in this case you would have problems with BIOS output etc. If you
left Linux alone, but fixed the BIOS to output at the required
frequencies, it would work - and using the quiet option, together with
appropriate output from the init scripts (which would presuambly be
heavily customised, in such an application) would yield a similar result.
And I question the approach of automatically deciding to hide startup
output from the user, in any case. I can imagine a set-top box user on the
phone to the support department saying "it gets to the Starting - Please
Wait screen, then just hangs", which would then require an engineer visit,
as opposed to, for example, "it says Obtaining IP Address... then hangs"
which would give the support techie the opportunity to tell the user to
check the ethernet cable is plugged in correctly, etc, before resorting to
sending out an engineer.
Robert Morris
08707 458710
http://www.r-morris.co.uk/
> > It's perfectly possible that somebody might want to make a television
> > set top box out of a standard x86 motherboard and VGA card, and not
> > have anything displayed until X starts, because the television would
> > not accept the standard VGA scanrate, but X can easily re-program that
> > to around 15 Khz.
>
> OK, but in this case you would have problems with BIOS output etc.
Yeah, I was thinking along the lines of having a simple colour bar
generator for when the VGA output wasn't in the valid frequency range,
but your idea is better as long as the custom BIOS was easily
do-able.
> And I question the approach of automatically deciding to hide startup
> output from the user, in any case. I can imagine a set-top box user on the
> phone to the support department saying "it gets to the Starting - Please
> Wait screen, then just hangs", which would then require an engineer visit,
> as opposed to, for example, "it says Obtaining IP Address... then hangs"
> which would give the support techie the opportunity to tell the user to
> check the ethernet cable is plugged in correctly, etc, before resorting to
> sending out an engineer.
I agree, but how many set top boxes are designed like that? I would
prefer verbose output, but it's generally hidden from end users :-(.
John
Hello there,
> Kiosks and things like ATMs are another place where you do not want
> a bootscreen. You do not want to possibly confuse customers with
> stuff that they can't understand but show a nice friendly message saying
> 'the system is currently unavailable'.
I once saw an ATM with a Windows NT "blue screen" error message :-)
FWIW, I wouldn't have responded to Raphael's post in the way I did, if he
said something along the lines of "I'm using Linux in a TV set-top
box|ATM|kiosk application, and I want to hide the standard boot-up output
with a nice graphic".
Robert Morris
08707 458710
http://www.r-morris.co.uk/
* Raphael Schmid <[email protected]> [030128 10:01]:
> It is my very understanding one can not have, conveniently it should be,
> a simple *bootscreen* under Linux. With that I mean a picture of at
> least 256 (indexed) colours at a size of 640x480 pixels. Doesn't have
> to be a higher resolution. And yes, I'm taking the standpoint that every
> computer nowadays [where this shall be possible] *can* do that resolution.
It's not too much to even state that almost any computer working with
Linux 2.4+ can do 800x600 or 1024x768. Anything below that can be
considered a special case, regarding the numbers out there. But that
does not influence the possibility of using a bootsplash graphics.
On a system you can't use it properly, you probably also would not
want it (i.e. use normal text mode boot instead of a framebuffer
driver)
> Framebuffer, I hear people shouting? Well. During the last *two days*,
> which includes one full night, I've been trying to get my v2.4.20 kernel
> to display such a bootscreen. All I get is segfaults. I've tried what I
> believe to be every tool out there: pnmtologo, fblogo, boot_logo, the
> GIMP plugin. You name them. None of which wouldn't have required any
> hacking to work with 2.4.20, by the way...
Have a look at ftp.suse.com/pub/people/stepan/bootsplash/ - There you
find kernel patches, user space utilities and such to display a
bootsplash screen. You can either choose to have a picture put "behind"
your text, or have a picture _instead_ of text. (triggerable with a
boot parameter so anybody is happy). And yes, it _does_ look cool to see
your kernel messages scrolling up on a background of a slightly faded
out penguin, looking like a water sign. ;-)
> And maybe it's right, maybe I demand too much from the (VESA) framebuffer.
> Maybe my picture is also too complex, but I've tried simple ones as well.
> And anyway: I don't *want* any simple picture, I want as complex a picture
> as it gets. In 640x480. At 256 indexed colours.
My patch above includes a small and efficient jpeg decoder (8k), which
allows you to read any jpg picture from an initrd.
> I realize these ideas may sound kind of alien to you, but they make sense.
> Windows, MacOS all have bootscreens. There really is no way why Linux
> shouldn't.
It's not alien, and it does make sense. I, speaking for myself, know the
kernel boot messages by heart and I don't expect them to change with the
2957596. bootup of my linux box. ;)
Any comments?
Stefan
--
The use of COBOL cripples the mind; its teaching should, therefore, be
regarded as a criminal offense. -- E. W. Dijkstra
On Tue, 2003-01-28 at 11:48, Wichert Akkerman wrote:
> Previously John Bradford wrote:
> > There are applications where it is not appropriate to have it, though,
> > what if you were using Linux in an embedded device such as a set top
> > box?
>
> Kiosks and things like ATMs are another place where you do not want
> a bootscreen. You do not want to possibly confuse customers with
> stuff that they can't understand but show a nice friendly message saying
> 'the system is currently unavailable'.
The real question is whether you want to do this in the kernel or simply at
the moment the kernel flips to user space. An init can easily open vt2,
draw a pretty boot screen with something like nanogui or bogl and then
continue to spew the text to vt1 so anyone can see the text messages if
need be.
Previously Alan Cox wrote:
> The real question is whether you want to do this in the kernel or simply at
> the moment the kernel flips to user space. An init can easily open vt2,
> draw a pretty boot screen with something like nanogui or bogl and then
> continue to spew the text to vt1 so anyone can see the text messages if
> need be.
It takes a while before the kernel starts init though, especially if you
have things like SCSI controllers to initialise. If you do not use fb
you can have your bootloader setup a pretty bootscreen, but if you need
fb I don't see how you can prevent the textscreen with kernel messages.
Wichert.
--
Wichert Akkerman <[email protected]> http://www.wiggy.net/
A random hacker
On Tue, 28 Jan 2003, Alan Cox wrote:
> On Tue, 2003-01-28 at 11:48, Wichert Akkerman wrote:
> > Kiosks and things like ATMs are another place where you do not want
> > a bootscreen. You do not want to possibly confuse customers with
> > stuff that they can't understand but show a nice friendly message saying
> > 'the system is currently unavailable'.
>
> The real question is whether you want to do this in the kernel or simply at
> the moment the kernel flips to user space. An init can easily open vt2,
> draw a pretty boot screen with something like nanogui or bogl and then
> continue to spew the text to vt1 so anyone can see the text messages if
> need be.
>
I agree with you that it's not the kernels main task to draw logos, though
it might be a little late to handle it in init. 2.4 kernels take quite
some time before entering init. True, 2.5 kernels are a lot faster
already, but, take an embedded device: all drivers might be built into the
kernel, and when the kernel is entering init, most stuff is done already.
Besides: There is no need for a user to see that the kernel detected the
CPU again. Maybe the kernel messages should automagically show up in case
of errors, and flip away the logo. Of course this implies the logo is
handled completely by the kernel.
I think you are mostly done by tweaking some fbcon code, it has
implemented some logo code already.
Jos
> > Most of the machines I maintain are very seldom rebooted, but if someone
> > was to do a reboot, I would want them to be able to observe any errors
or
> > other abnormal output from the boot-up process.
>
> Agreed, for standard desktops and servers.
Well, I really don't know about you, but I for one reboot my desktop every
morning. Maybe this is a German attitude, but I generally consider it a
waste
of resources to have my workstation run during the night. For downloads and
the like, I got a headless server which does good power management in the
closet room. Besides, again: everyone [who is not a hacker] likes eyecandy.
I wouldn't normally say that if it wasn't about this discussion.
> There is no reason why the boot data can't go to a secondary display,
> a serial terminal, or a printer, or a speaker as a bleep code, etc.
Absolutely!
> In this case, boot data could be sent to a serial port, and the
> graphics card initialised by the boot loader to display a "Please
> wait, set top box booting up" screen, using a scan rate, which would
> be acceptable to the television. In this case, we do not want the
> kernel to change the video card setup at all.
Yes! This is exactly what I want!
* Robert Morris <[email protected]> [030128 12:20]:
> There is a very simple reason why Linux shouldn't have a "bootscreen" -
> its a lame idea. We have copied enough of the bad "features" of Windows et
> al into Linux already, IMHO.
Then why not make something better instead of denying it completely.
"It's lame" is not a reason. Seeing the same "Oh, I got my name and
copyright visible in this and that driver" every time I boot a system
is about as lame, and the boot messages are far more a try of a hall of
fame than pure system status in a standardized way. Don't get me wrong,
I don't think it's completely wrong, or should be changed generally.
Rather, Linux has been the operating system of free choice, as long as
I can think of it. And this free choice might also mean that people want
some graphics surrounding their boot messages or replace them.
> FWIW, I usually remove the graphics from LILO/GRUB config as installed by
> default these days.
Which is your perfect right, if you like it that way. But there are
people who think different. Don't call them lame.
> Most of the machines I maintain are very seldom rebooted, but if someone
> was to do a reboot, I would want them to be able to observe any errors or
> other abnormal output from the boot-up process.
A graphical boot does not necessarily hide all kernel messages. Look
at my bootsplash patches how to do it in a technical sane way.
ftp.suse.com/pub/people/stepan/bootsplash/. If you have an embedded
device with a fixed system and non-variable hardware you just don't need
the information from the kernel messages.
> A "bootscreen" makes it
> more likely that such an error message would be more likely to go
> unnoticed - and, if they became commonplace, may eventually result in
> developers making the on-boot output less verbose/informative/etc, on the
> basis that it isn't likely to be seen in the first place!
Make it configurable, so that everybody can turn it on and off, and
don't turn it on by default. This is done with the fbcon stuff anyways,
which is mandatory for a splash screen. Where's the problem in just not
switching such a function on?
> The stuff you see on your screen when your Linux installation boots up
> weren't put there for no reason. Please lets leave them there!
There's always a reason for someone. This does not mean that it's my
reason, too. Microsoft have their reasons as well with their backdoor
politics...
Stefan
--
The use of COBOL cripples the mind; its teaching should, therefore, be
regarded as a criminal offense. -- E. W. Dijkstra
Hello there,
On Tue, 28 Jan 2003, Stefan Reinauer wrote:
> Seeing the same "Oh, I got my name and
> copyright visible in this and that driver" every time I boot a system
> is about as lame
Agreed
> people who think different. Don't call them lame.
That's *not* what I said!
> Make it configurable, so that everybody can turn it on and off, and
> don't turn it on by default. This is done with the fbcon stuff anyways,
> which is mandatory for a splash screen. Where's the problem in just not
> switching such a function on?
The distribution vendors will turn it on by default - as already happened
with graphical bootloader screens, for example - and then the majority of
users will not turn it off. Then it will become the norm...
Most Windows users notionally have the choice to download another web
browser such as Mozilla. But how many actually do, when Internet Explorer
is installed already? The consequence of this is that, de facto, IE
becomes the predominant browser, then web developers disregard support of
other browsers, and then users of Mozilla are stuck.
Your point that everyone has a choice is correct in theory, but does not
take into account the very great power of influence that software
distributors (be they Microsoft, Red Hat, or Suse) have. Yes, Linux should
be a platform where people have a choice - but lets make the default
option a sensible one, and not simply copy as much of the Windows/Mac OS
environment as possible to try to gain favour with users of those
platforms, at the expense of our own user community.
The day when a default Red Hat install covers up all the startup output
with a pretty graphic will be a very sad one indeed for me.
Robert Morris
08707 458710
http://www.r-morris.co.uk/
On Tue, 2003-01-28 at 13:09, Wichert Akkerman wrote:
> It takes a while before the kernel starts init though, especially if you
> have things like SCSI controllers to initialise. If you do not use fb
> you can have your bootloader setup a pretty bootscreen, but if you need
> fb I don't see how you can prevent the textscreen with kernel messages.
I'd not really pondered people who compile many drivers into their kernel
instead of into the initrd. I guess a few people still do that.
As to the messages, if you are in graphical mode then you are right, the
messages could be supressed and sent via user space but not directly. I
guess the other way then is to use a custom font, set the configuration to
not leave gaps between characters and draw a boot graphical view in that
font ?
> > There is a very simple reason why Linux shouldn't have a "bootscreen" -
> > its a lame idea.
>
> Then why not make something better instead of denying it completely.
Surely the most sensible lines to think along are:
* Make boot times as short as possible
* Support, and encourage the use of more efficient CPU designs, so
that it becomes sensible to leave machines on all the time.
Then, the issue of a bootscreen becomes a non-issue.
John.
On Tuesday 28 January 2003 14:32, Stefan Reinauer wrote:
> * Robert Morris <[email protected]> [030128 12:20]:
> > There is a very simple reason why Linux shouldn't have a "bootscreen" -
> > its a lame idea. We have copied enough of the bad "features" of Windows
> > et al into Linux already, IMHO.
I'm working for a company doing VoD and IPTV applications, and you surely
don't want some verbose kernel output upon booting set-top-boxes. At least -
the customer doesn't want it, meaning you shouldn't have it. Then it's better
to have some LED flashing in case of error.
roy
--
Roy Sigurd Karlsbakk, Datavaktmester
ProntoTV AS - http://www.pronto.tv/
Tel: +47 9801 3356
Computers are like air conditioners.
They stop working when you open Windows.
> > It takes a while before the kernel starts init though, especially if you
> > have things like SCSI controllers to initialise. If you do not use fb
> > you can have your bootloader setup a pretty bootscreen, but if you need
> > fb I don't see how you can prevent the textscreen with kernel messages.
>
> I'd not really pondered people who compile many drivers into their kernel
> instead of into the initrd. I guess a few people still do that.
I don't usually compile support for modules in at all, and even on
486s the boot time isn't *that' long. 30 Seconds or so on a 4 Mb RAM,
486 SX 20 for a complete boot, only about half than until init starts.
I personally don't think a few seconds of text mode is a major
problem.
What I'm wondering is how long it will be before text mode is no
longer implemented in VGA cards at all. Once the BIOS stops using it,
who needs it anymore?
John
> And I question the approach of automatically deciding to hide startup
> output from the user, in any case. I can imagine a set-top box user on the
> phone to the support department saying "it gets to the Starting - Please
> Wait screen, then just hangs", which would then require an engineer visit,
> as opposed to, for example, "it says Obtaining IP Address... then hangs"
> which would give the support techie the opportunity to tell the user to
> check the ethernet cable is plugged in correctly, etc, before resorting to
> sending out an engineer.
You'll never send an engineer out to replace a set-top-box. You'll just wait
for the customer to return the box and send out a new one. Software doesn't
fail on those - it's some 99.9% hardware failure. If you get a hang or panic
or something, the boxes usually have watchdogs to take care of that (and then
reboot automatically). The average computer-frightened user getting an STB
from the VoD/IPTV company (or her ISP) don't want to see any kernel
gibberish. They just want a nice splash screen telling them "everything's
gonna be alright in 45 seconds" or something. Trouble shooting is done in the
lab after the box is returned
roy
--
Roy Sigurd Karlsbakk, Datavaktmester
ProntoTV AS - http://www.pronto.tv/
Tel: +47 9801 3356
Computers are like air conditioners.
They stop working when you open Windows.
On Tue, 2003-01-28 at 09:06, John Bradford wrote:
> > > It takes a while before the kernel starts init though, especially if you
> > > have things like SCSI controllers to initialise. If you do not use fb
> > > you can have your bootloader setup a pretty bootscreen, but if you need
> > > fb I don't see how you can prevent the textscreen with kernel messages.
> >
> > I'd not really pondered people who compile many drivers into their kernel
> > instead of into the initrd. I guess a few people still do that.
>
> I don't usually compile support for modules in at all.
Me neither. Guess we're a part of the "few"
Previously Alan Cox wrote:
> I'd not really pondered people who compile many drivers into their kernel
> instead of into the initrd. I guess a few people still do that.
Agreed - what you probably want to do is have a minimal kernel that
boots an initrd which loads modules for the rest. If the kernel is
small enough you don't care for its boot messages: if it fails the
hardware is screwed and your box has to be repaired (esp. if you are
dealing with embedded/special purpose systems where the people using
the box can't even see the hardware).
Then have the bios/bootloader setup your pretty bootscreen and reset
it in the initrd when you load a fb driver.
Wichert.
--
Wichert Akkerman <[email protected]> http://www.wiggy.net/
A random hacker
> I do have a solution for that. Just make the image 640x440 instead
640x480,
> and have the initscripts output on one of the lower lines only, always
over-
> writing the previous message. That way, the support engineer would know
> what's going wrong and you'd still have a cute picture.
There's a good way to encode startup sequence info into the screen...it
doesn't require text.
WinXP outputs an image, starts it dim, fades it up to bright, starts a
sliding indicator, moves the slider back and forth at various speeds, then
starts the gui (?) and it goes more various steps.
I imagine someone with the right documentation could say exactly what's
going on at each step.
Hello there,
On 28 Jan 2003, Richard B. Tilley (Brad) wrote:
> On Tue, 2003-01-28 at 09:06, John Bradford wrote:
> > I don't usually compile support for modules in at all.
>
> Me neither. Guess we're a part of the "few"
<fx: waves hand from the back of the room> :-)
I don't normally use modules on production machines, unless its one of a
few specific (broken, IMHO) things which can *only* be built as modules
(eg PPP compressors).
Robert Morris
08707 458710
http://www.r-morris.co.uk/
Le mar 28/01/2003 ? 14:13, Raphael Schmid a ?crit :
> > > Most of the machines I maintain are very seldom rebooted, but if someone
>
> > > was to do a reboot, I would want them to be able to observe any errors
> or
> > > other abnormal output from the boot-up process.
> >
> > Agreed, for standard desktops and servers.
> Well, I really don't know about you, but I for one reboot my desktop every
> morning. Maybe this is a German attitude, but I generally consider it a
> waste
> of resources to have my workstation run during the night. For downloads and
> the like, I got a headless server which does good power management in the
> closet room. Besides, again: everyone [who is not a hacker] likes eyecandy.
> I wouldn't normally say that if it wasn't about this discussion.
Yeah, I'd really like a stable working swsusp (on a working kernel) to
shortcut the fscking boot. Go pawel !
Xav
Previously John Bradford wrote:
> Surely the most sensible lines to think along are:
>
> * Make boot times as short as possible
So with a short boot time instead of seeing text messages for a while
you'll get some flickering on the screen - I don't call that an
improvement.
> * Support, and encourage the use of more efficient CPU designs, so
> that it becomes sensible to leave machines on all the time.
Unfortunately in the real world we are dealing with existing cheap
hardware.
Wichert.
--
Wichert Akkerman <[email protected]> http://www.wiggy.net/
A random hacker
Previously Jos Hulzink wrote:
> Why are you guys walking around the issue ? There are people demanding for
> a feature, and they are marked as complete idiots.
Did you completely miss my point?
> Oh, and using modules is a (minor) security issue. I have all my drivers
> compiled in the kernel. I like it and it is secure.
Not using modules does not add any security at all, that is a myth that
refuses to die.
Wichert.
--
Wichert Akkerman <[email protected]> http://www.wiggy.net/
A random hacker
On Tue, 2003-01-28 at 12:10, Stefan Reinauer wrote:
> It's not too much to even state that almost any computer working with
> Linux 2.4+ can do 800x600 or 1024x768. Anything below that can be
> considered a special case, regarding the numbers out there. But that
> does not influence the possibility of using a bootsplash graphics.
> On a system you can't use it properly, you probably also would not
> want it (i.e. use normal text mode boot instead of a framebuffer
> driver)
Lots of systems cannot do 800x600 or 1024x768. Some of them cannot
do 640x480 very well but 640x480 is safe except for weird kit because
of the VGA mode support.
> I imagine someone with the right documentation could say exactly what's
> going on at each step.
Hmmm, I think the traditional text diagnostic messages are best kept
as they are, otherwise we'll end up with bug reports like this:
Date: Jan 28 14:39:29 2006
Subject: Kernel 3.6.2 boot failiure
To: [email protected]
Hi,
I just upgraded from 3.6.1, which booted fine, to 3.6.2, which stops
after Tux has waved twice, and winked his left eye.
John
On Tue, 28 Jan 2003, Wichert Akkerman wrote:
> Agreed - what you probably want to do is have a minimal kernel that
> boots an initrd which loads modules for the rest. If the kernel is
> small enough you don't care for its boot messages: if it fails the
> hardware is screwed and your box has to be repaired (esp. if you are
> dealing with embedded/special purpose systems where the people using
> the box can't even see the hardware).
Why are you guys walking around the issue ? There are people demanding for
a feature, and they are marked as complete idiots. Linux is no kernel for
your computer only, it is a general purpose kernel, used in PDAs, set top
boxes, cars, mp3 players, servers, oh, and desktop computers. The nice
thing of features of the Linux kernel is that you can disable them if you
don't like them. I think the fbcon system is shit, so I disable it,
instead of telling everyone who uses it they are morons.
Oh, and using modules is a (minor) security issue. I have all my drivers
compiled in the kernel. I like it and it is secure.
Jos
On Tue, Jan 28, 2003 at 12:55:29PM +0000, Alan Cox wrote:
> The real question is whether you want to do this in the kernel or simply at
> the moment the kernel flips to user space. An init can easily open vt2,
> draw a pretty boot screen with something like nanogui or bogl and then
> continue to spew the text to vt1 so anyone can see the text messages if
> need be.
An ATM here in delft showed
OS/2 booting
and all the BIOS stuff when I once caught it in the act of booting...
Rogier.
--
** [email protected] ** http://www.BitWizard.nl/ ** +31-15-2600998 **
*-- BitWizard writes Linux device drivers for any device you may have! --*
* The Worlds Ecosystem is a stable system. Stable systems may experience *
* excursions from the stable situation. We are currently in such an *
* excursion: The stable situation does not include humans. ***************
Previously John Bradford wrote:
> Well, if the machine boots up OK, you can always review the messages
> with dmesg. If it doesn't, the messages are there to review.
I think we're talking about different things here. Personally I'm more
interested in having this work in environments where you can not look
at dmesg or anything else: appliances that do a single job and have to
do it well. Think ATMs, point of sale machines, settop boxes, etc.
If they break you don't go in and look at dmesg, you replace them.
Wichert.
--
Wichert Akkerman <[email protected]> http://www.wiggy.net/
A random hacker
* Robert Morris <[email protected]> [030128 14:46]:
> The distribution vendors will turn it on by default - as already happened
> with graphical bootloader screens, for example - and then the majority of
> users will not turn it off. Then it will become the norm...
WRT distribution vendors, UnitedLinux, SuSE and Mandrake already come
with a bootsplash screen per default. But this is a seperate issue.
> Most Windows users notionally have the choice to download another web
> browser such as Mozilla. But how many actually do, when Internet Explorer
> is installed already? The consequence of this is that, de facto, IE
> becomes the predominant browser, then web developers disregard support of
> other browsers, and then users of Mozilla are stuck.
I think you are exaggerating with the consequences of having a
bootsplash as an _option_.
The problem with IE is that you cannot remove it from a system, so you
have to have 2 browsers installed. With the above mentioned Linux
distributions you can, with no additional downloads from the internet,
change one config variable in /etc/sysconfig/.. and you will not get
a bootsplash anymore, plus no memory is wasted for the pictures as well.
> Your point that everyone has a choice is correct in theory, but does not
> take into account the very great power of influence that software
> distributors (be they Microsoft, Red Hat, or Suse) have.
If one of these companies wants a graphical bootsplash, they can, due to
the freedom of choice, do so. Whether or not this code ends up in the
kernel is a completely different thing and decided by the maintainers of
the several subsystems, not by any software vendor. Don't search an
enemy where there is none.
> Yes, Linux should
> be a platform where people have a choice - but lets make the default
> option a sensible one, and not simply copy as much of the Windows/Mac OS
> environment as possible to try to gain favour with users of those
> platforms, at the expense of our own user community.
I still don't get the connection between "bootsplash is bad" and
"windows has a bootsplash".
Windows has a gui and modal dialogs, by now even a virtual filesystem
layer. Should we get rid of that, because Microsoft uses something
similar? No, for sure not. "Ours" is better, because everybody can
choose to use it or not, and to change it to own needs and wishes.
> The day when a default Red Hat install covers up all the startup output
> with a pretty graphic will be a very sad one indeed for me.
Hey, maybe I can help you out with a little hint... If they ever do,
"vga=normal" will get you back to the good old times and 10 key strokes
make your day.
Stefan
--
The use of COBOL cripples the mind; its teaching should, therefore, be
regarded as a criminal offense. -- E. W. Dijkstra
> > Surely the most sensible lines to think along are:
> >
> > * Make boot times as short as possible
>
> So with a short boot time instead of seeing text messages for a while
> you'll get some flickering on the screen - I don't call that an
> improvement.
Well, if the machine boots up OK, you can always review the messages
with dmesg. If it doesn't, the messages are there to review.
> > * Support, and encourage the use of more efficient CPU designs, so
> > that it becomes sensible to leave machines on all the time.
>
> Unfortunately in the real world we are dealing with existing cheap
> hardware.
Well, my current IA-32 box is hopefully going to be my last :-).
John.
Am Dienstag, 28. Januar 2003 15:32 schrieb Wichert Akkerman:
> Previously Alan Cox wrote:
> > I'd not really pondered people who compile many drivers into their kernel
> > instead of into the initrd. I guess a few people still do that.
>
> Agreed - what you probably want to do is have a minimal kernel that
> boots an initrd which loads modules for the rest. If the kernel is
Why? If you do an embedded system you go for minimal memory.
You'd probably compile a kernel without module support. You know
which hardware is to be supported, so you gain nothing at all
by using modules.
Frankly, even for a normal system, why compile an initrd for drivers
you need during everyday operation? If you compile yourself at all, why
pay the price in memory and TLB misses?
Regards
Oliver
On Tue, 2003-01-28 at 09:47, Wichert Akkerman wrote:
> Not using modules does not add any security at all, that is a myth that
> refuses to die.
Could you explain this in more detail? It seems to me that if the kernel
does not support loadable modules that it would be inherently more
secure because it could not be dynamically modified with a module. How
is my understanding of this wrong?
Previously Richard B. Tilley (Brad) wrote:
> Could you explain this in more detail? It seems to me that if the kernel
> does not support loadable modules that it would be inherently more
> secure because it could not be dynamically modified with a module. How
> is my understanding of this wrong?
You can fiddle with kernel memory by hand and insert code and modules.
There are a couple of tools available to do that for you, google can
probably find them for you.
Wichert.
--
Wichert Akkerman <[email protected]> http://www.wiggy.net/
A random hacker
OK, you win.
Were can I find these simple tools that modify a kernel that does not
support loadable modules as easily as insmod modifies a kernel that does
support them?
On Tue, 2003-01-28 at 10:14, Wichert Akkerman wrote:
> Previously Richard B. Tilley (Brad) wrote:
> > Then, not supporting loadable modules *is* more secure. By not
> > supporting them, you are decreasing the ease in which the kernel can be
> > modified. There are fewer people who can "fiddle with memory by hand"
> > than there are that can insert a loadable module... a lot fewer, don't
> > you agree?
>
> No, there are simple tools to do that which are just as easy to use as
> insmod.
>
> Wichert.
>
> --
> Wichert Akkerman <[email protected]> http://www.wiggy.net/
> A random hacker
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
On Tue, 2003-01-28 at 14:45, Jos Hulzink wrote:
> Oh, and using modules is a (minor) security issue. I have all my drivers
> compiled in the kernel. I like it and it is secure.
Myth I am afraid. Code for loading modules into a kernel with no module support
by poking in /dev/*mem exists and is published.
* Alan Cox <[email protected]> [030128 15:50]:
> On Tue, 2003-01-28 at 12:10, Stefan Reinauer wrote:
> > It's not too much to even state that almost any computer working with
> > Linux 2.4+ can do 800x600 or 1024x768. Anything below that can be
> > considered a special case, regarding the numbers out there. But that
> > does not influence the possibility of using a bootsplash graphics.
> > On a system you can't use it properly, you probably also would not
> > want it (i.e. use normal text mode boot instead of a framebuffer
> > driver)
>
> Lots of systems cannot do 800x600 or 1024x768. Some of them cannot
> do 640x480 very well but 640x480 is safe except for weird kit because
> of the VGA mode support.
Safe or not safe depends highly on the kind of hardware we are talking
about. To clarify: 1024x768 is not a problem on most new PCs bought
today. For a PDA or embedded system you maybe only have 320x244.
My point was rather, if the graphics hardware cannot display a graphical
bootup without drawbacks, it does not hurt to disable it and go back to
text.
Stefan
--
The use of COBOL cripples the mind; its teaching should, therefore, be
regarded as a criminal offense. -- E. W. Dijkstra
> Oh, and using modules is a (minor) security issue. I have all my drivers
> compiled in the kernel. I like it and it is secure.
You can patch code in to a running kernel without module support.
John
> > It's not too much to even state that almost any computer working with
> > Linux 2.4+ can do 800x600 or 1024x768. Anything below that can be
> > considered a special case, regarding the numbers out there. But that
> > does not influence the possibility of using a bootsplash graphics.
> > On a system you can't use it properly, you probably also would not
> > want it (i.e. use normal text mode boot instead of a framebuffer
> > driver)
>
> Lots of systems cannot do 800x600 or 1024x768. Some of them cannot
> do 640x480 very well but 640x480 is safe except for weird kit because
> of the VGA mode support.
Infact, better than using text mode with a custom font to draw a logo,
because some LCD screens display 720x400 as 640x400 by missing out one
pixel of each character.
John
Previously Richard B. Tilley (Brad) wrote:
> Then, not supporting loadable modules *is* more secure. By not
> supporting them, you are decreasing the ease in which the kernel can be
> modified. There are fewer people who can "fiddle with memory by hand"
> than there are that can insert a loadable module... a lot fewer, don't
> you agree?
No, there are simple tools to do that which are just as easy to use as
insmod.
Wichert.
--
Wichert Akkerman <[email protected]> http://www.wiggy.net/
A random hacker
Then, not supporting loadable modules *is* more secure. By not
supporting them, you are decreasing the ease in which the kernel can be
modified. There are fewer people who can "fiddle with memory by hand"
than there are that can insert a loadable module... a lot fewer, don't
you agree?
On Tue, 2003-01-28 at 09:58, Wichert Akkerman wrote:
> Previously Richard B. Tilley (Brad) wrote:
> > Could you explain this in more detail? It seems to me that if the kernel
> > does not support loadable modules that it would be inherently more
> > secure because it could not be dynamically modified with a module. How
> > is my understanding of this wrong?
>
> You can fiddle with kernel memory by hand and insert code and modules.
> There are a couple of tools available to do that for you, google can
> probably find them for you.
>
> Wichert.
>
> --
> Wichert Akkerman <[email protected]> http://www.wiggy.net/
> A random hacker
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
Previously Richard B. Tilley (Brad) wrote:
> Were can I find these simple tools that modify a kernel that does not
> support loadable modules as easily as insmod modifies a kernel that does
> support them?
Ask google. Phrack 58 for example seems to have a decent description and
example code.
Wichert.
--
Wichert Akkerman <[email protected]> http://www.wiggy.net/
A random hacker
On Tue, 28 Jan 2003 09:34:46 EST, jeff millar said:
> I imagine someone with the right documentation could say exactly what's
> going on at each step.
That's the problem with that right there - "with the right documentation".
The current boot system, you can *SEE* "oh, it wedged trying to bring up
the sound card". No "Well, if the slider is all the way to the left and
the screen is half-dim, then it's the sound card, else if it's all the
way to the left but the screen is bright, it's the joystick port, unless
you've also installed...."
On Tue, 28 Jan 2003 14:32:52 +0100, Stefan Reinauer said:
> Then why not make something better instead of denying it completely.
> "It's lame" is not a reason. Seeing the same "Oh, I got my name and
> copyright visible in this and that driver" every time I boot a system
> is about as lame, and the boot messages are far more a try of a hall of
> fame than pure system status in a standardized way.
/usr/src/linux-2.5.59/COPYING:
If the program is interactive, make it output a short notice like this
when it starts in an interactive mode:
Gnomovision version 69, Copyright (C) year name of author
Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
This is free software, and you are welcome to redistribute it
under certain conditions; type `show c' for details.
I'll let the lawyers argue about this one.. ;)
(btw - thanks for bootsplash, looks pretty cool.. ;)
> I realize these ideas may sound kind of alien to you, but they make sense.
> Windows, MacOS all have bootscreens. There really is no way why Linux
> shouldn't.
I think it's a better plan to justify new features with an explantion
of why we should have something, rather than than saying there's no
reason we shouldn't.
M.
On Tue, Jan 28, 2003 at 02:50:42PM +0000, Alan Cox wrote:
> Lots of systems cannot do 800x600 or 1024x768. Some of them cannot
> do 640x480 very well but 640x480 is safe except for weird kit because
> of the VGA mode support.
Hey, we have aalib for that :)
Erik
--
J.A.K. (Erik) Mouw
Email: [email protected] [email protected]
WWW: http://www-ict.its.tudelft.nl/~erik/
On Tue, Jan 28, 2003 at 10:01:37AM +0100, Raphael Schmid wrote:
> Hello World,
>
> this eMail shall be a means of bringing up again a topic I believe has
> already been discussed extensively. Wait! Don't delete, read further
> please!
>
> [ Note: please cc: me in any replies as <[email protected]>,
*plonk*
The linux progress patch could be what you want. I tried porting this to
2.4.18 and was successful in just a couple of hours. I have tested the
thing on many systems. It worked well. If any one is interested maybe i
can mail you the patch for 2.4.18-4.
Prasad.
On Tue, 28 Jan 2003, Raphael Schmid wrote:
> Hello World,
>
> this eMail shall be a means of bringing up again a topic I believe has
> already been discussed extensively. Wait! Don't delete, read further
> please!
>
> [ Note: please cc: me in any replies as <[email protected]>,
> since (a) I'm at work and (b) not subscribed to the list. Thanks. ]
>
> It is my very understanding one can not have, conveniently it should be,
> a simple *bootscreen* under Linux. With that I mean a picture of at
> least 256 (indexed) colours at a size of 640x480 pixels. Doesn't have
> to be a higher resolution. And yes, I'm taking the standpoint that every
> computer nowadays [where this shall be possible] *can* do that resolution.
>
> Framebuffer, I hear people shouting? Well. During the last *two days*,
> which includes one full night, I've been trying to get my v2.4.20 kernel
> to display such a bootscreen. All I get is segfaults. I've tried what I
> believe to be every tool out there: pnmtologo, fblogo, boot_logo, the
> GIMP plugin. You name them. None of which wouldn't have required any
> hacking to work with 2.4.20, by the way...
>
> And maybe it's right, maybe I demand too much from the (VESA) framebuffer.
> Maybe my picture is also too complex, but I've tried simple ones as well.
> And anyway: I don't *want* any simple picture, I want as complex a picture
> as it gets. In 640x480. At 256 indexed colours.
>
> So although I'm just learning C and can't code it myself, here's an idea:
>
> If Syslinux can display this kind of images, and if LILO can, so why would
> Linux be unable to display it? VESA was the term, if I right remember?
> If this request is too much of an effort to implement, then couldn't there
> be a kernel configuation option that simply tells Linux to leave the screen
> as it is, until some user space software (X) changes it? (In conjunction
> with console=/dev/null or something). I just want my picture remain there.
>
> I realize these ideas may sound kind of alien to you, but they make sense.
> Windows, MacOS all have bootscreens. There really is no way why Linux
> shouldn't.
>
> In that veine, another thing I've been puzzled with... can you somehow
> disable
> virtual consoles (Alt-Fx) completely while still maintaining an interface
> for
> X to come up on?
>
> Thanks for reading through until here. Thanks for any considerations in
> advance.
>
> Your truly, Raphael
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
--
Failure is not an option
Followup to: <[email protected]>
By author: Roy Sigurd Karlsbakk <[email protected]>
In newsgroup: linux.dev.kernel
>
> You'll never send an engineer out to replace a set-top-box. You'll just wait
> for the customer to return the box and send out a new one. Software doesn't
> fail on those - it's some 99.9% hardware failure. If you get a hang or panic
> or something, the boxes usually have watchdogs to take care of that (and then
> reboot automatically). The average computer-frightened user getting an STB
> from the VoD/IPTV company (or her ISP) don't want to see any kernel
> gibberish. They just want a nice splash screen telling them "everything's
> gonna be alright in 45 seconds" or something. Trouble shooting is
> done in the lab after the box is returned
>
Right, which is why those boxes when running Linux invariably direct
their kernel messages to a debugging port of some sort (usually a
serial port.)
-hpa
--
<[email protected]> at work, <[email protected]> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt <[email protected]>
On Tue, 28 Jan 2003 18:41:39 +0100
Erik Mouw <[email protected]> wrote:
> On Tue, Jan 28, 2003 at 02:50:42PM +0000, Alan Cox wrote:
> > Lots of systems cannot do 800x600 or 1024x768. Some of them cannot
> > do 640x480 very well but 640x480 is safe except for weird kit because
> > of the VGA mode support.
>
> Hey, we have aalib for that :)
yeah, why to do it inside the kernel?
Just run a userspace logo for the first thing in the
system in the init screens. I don't see a real reason why
that thing should be put in kernel. Where would you put the
800x600 image (since you have nothing mounted)?
Just run the first task with something that puts
a fb logo; and send nothing to the screen until you run
xdm. That would be nice for the users that doesn't
want to see those horrible "debug" things.
If i remember correctly, xp doesn't shows the logo
since the start neither. It does a bit of job before.
A linux kernel doesn't take too much time to boot
(the ide detection is the slower part i remember)
And the kernel messages always were, always will be,
useful. To get a clean screen perhaps we could have
something like a boot parm called silentdmesg, and then
do the previous thing.
>
>
> Erik
>
> --
> J.A.K. (Erik) Mouw
> Email: [email protected] [email protected]
> WWW: http://www-ict.its.tudelft.nl/~erik/
>
In article <[email protected]> you wrote:
> If i remember correctly, xp doesn't shows the logo
> since the start neither. It does a bit of job before.
w2k and xp show ascii progress bars and menus before they show vga logos.
And speaking of grafical boot screens, they suck but look cute :)
Greetings
Bernd
--
eckes privat - http://www.eckes.org/
Project Freefire - http://www.freefire.org/
though i have recently subscribed to this list (OMG, 30 mails/hour!), I've
read the whole thread about this bootscreen thing... I see absolutely _no_
reason why it should not be included in the kernel configuration, since there
are patches available already to make it work, and there are already soo many
useless options that adding another won't make a difference...
Many people will choose to stay away from it, including me, but those users
coming from the windows world are scared out of their wits when they see the
kernel booting, and I've seen it myself...
I don't think adding the option to the kernel configuration would do any
harm...exept that the kernel source may get enlarged by (200 kb?), and the
kernel source gets enlarged every day anyway... 2.5 is HUGE compared to
2.2...
On Wednesday 29 January 2003 01:41, Arador wrote:
> yeah, why to do it inside the kernel?
>
> Just run a userspace logo for the first thing in the
> system in the init screens. I don't see a real reason why
> that thing should be put in kernel. Where would you put the
> 800x600 image (since you have nothing mounted)?
>
>
> Just run the first task with something that puts
> a fb logo; and send nothing to the screen until you run
> xdm. That would be nice for the users that doesn't
> want to see those horrible "debug" things.
>
> If i remember correctly, xp doesn't shows the logo
> since the start neither. It does a bit of job before.
>
> A linux kernel doesn't take too much time to boot
> (the ide detection is the slower part i remember)
>
> And the kernel messages always were, always will be,
> useful. To get a clean screen perhaps we could have
> something like a boot parm called silentdmesg, and then
> do the previous thing.
does a slient screen work when you have framebuffer enabled?
> though i have recently subscribed to this list (OMG, 30 mails/hour!), I've
> read the whole thread about this bootscreen thing... I see absolutely _no_
> reason why it should not be included in the kernel configuration, since there
> are patches available already to make it work, and there are already soo many
> useless options that adding another won't make a difference...
Many old options are also removed, though.
> I don't think adding the option to the kernel configuration would do any
> harm...exept that the kernel source may get enlarged by (200 kb?), and the
> kernel source gets enlarged every day anyway... 2.5 is HUGE compared to
> 2.2...
That's partly because it's a development tree. There are big efforts
to trim it down as much as possible.
I don't see any real advantage to putting in to the mainstream kernel
something which can be achieved easily with a custom bootloader, and
kernel options.
John.
On Tue, 28 Jan 2003, Roy Sigurd Karlsbakk wrote:
> On Tuesday 28 January 2003 14:32, Stefan Reinauer wrote:
> > * Robert Morris <[email protected]> [030128 12:20]:
> > > There is a very simple reason why Linux shouldn't have a "bootscreen" -
> > > its a lame idea. We have copied enough of the bad "features" of Windows
> > > et al into Linux already, IMHO.
>
> I'm working for a company doing VoD and IPTV applications, and you surely
> don't want some verbose kernel output upon booting set-top-boxes. At least -
> the customer doesn't want it, meaning you shouldn't have it. Then it's better
> to have some LED flashing in case of error.
Then don't output kernel messages to your default display device. No
Kernel Hacking Required (tm)
Zwane
--
function.linuxpower.ca
G'morning everyone,
> The linux progress patch could be what you want. I tried porting this to
> 2.4.18 and was successful in just a couple of hours. I have tested the
> thing on many systems. It worked well. If any one is interested maybe i
> can mail you the patch for 2.4.18-4.
To the farthest of my knowledge, the LPP is just an add-on to the fbdev's
(fake, ahem) capability of showing a picture at startup. But if you say it
worked for you, I'd be eager to trying out your patch. Only question: what
are you referring to with "-4"? Release 4 of your distro's 2.4.18 kernel?
2.4.18-pre4?
- Raphael
> yeah, why to do it inside the kernel?
>
> Just run a userspace logo for the first thing in the
> system in the init screens. I don't see a real reason why
> that thing should be put in kernel. Where would you put the
> 800x600 image (since you have nothing mounted)?
Dunno about you, but for me the kernel itself takes some time
to come up and handle things over to init. Not covering that
time would just feel... unfinished.
> If i remember correctly, xp doesn't shows the logo
> since the start neither. It does a bit of job before.
I'd rather not compare ourselves with XP all too closely.
> A linux kernel doesn't take too much time to boot
> (the ide detection is the slower part i remember)
You have SCSI disks only, I presume?
> And the kernel messages always were, always will be,
> useful. To get a clean screen perhaps we could have
> something like a boot parm called silentdmesg, and then
> do the previous thing.
Again, we've already encountered the diversity of (emotional)
opinions on this matter. I believe the final compromise "every-
one what s(he) needs" will do us just a fine service.
Not meaning to be offensive.
Roy Sigurd Karlsbakk <[email protected]> said:
[...]
> You'll never send an engineer out to replace a set-top-box. You'll just
> wait for the customer to return the box and send out a new one. Software
> doesn't fail on those - it's some 99.9% hardware failure. If you get a
> hang or panic or something, the boxes usually have watchdogs to take care
> of that (and then reboot automatically). The average computer-frightened
> user getting an STB from the VoD/IPTV company (or her ISP) don't want to
> see any kernel gibberish. They just want a nice splash screen telling
> them "everything's gonna be alright in 45 seconds" or something. Trouble
> shooting is done in the lab after the box is returned
Right. And this is a very specialized application, where the vendor will
install a heavily hacked system anyway, as they won't have standard VGA or
keyboard or...
It has no relevance whatsoever to the standard, for-your-typical-box
kernel.
--
Dr. Horst H. von Brand User #22616 counter.li.org
Departamento de Informatica Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria +56 32 654239
Casilla 110-V, Valparaiso, Chile Fax: +56 32 797513
Le mer 29/01/2003 ? 10:13, Horst von Brand a ?crit:
> Roy Sigurd Karlsbakk <[email protected]> said:
>
> [...]
>
> > You'll never send an engineer out to replace a set-top-box. You'll just
> > wait for the customer to return the box and send out a new one. Software
> > doesn't fail on those - it's some 99.9% hardware failure. If you get a
> > hang or panic or something, the boxes usually have watchdogs to take care
> > of that (and then reboot automatically). The average computer-frightened
> > user getting an STB from the VoD/IPTV company (or her ISP) don't want to
> > see any kernel gibberish. They just want a nice splash screen telling
> > them "everything's gonna be alright in 45 seconds" or something. Trouble
> > shooting is done in the lab after the box is returned
>
> Right. And this is a very specialized application, where the vendor will
> install a heavily hacked system anyway, as they won't have standard VGA or
> keyboard or...
What we did with our appliances is make them display the boot messages
(i.e. the console) on a hidden serial port (you had to solder pins on
the "motherboard" to get to it). That way the screen (which is only an
LCD) is free for whatever boot message the bootloader wants to print.
Xav
Jos Hulzink <[email protected]> said:
[...]
> Oh, and using modules is a (minor) security issue. I have all my drivers
> compiled in the kernel. I like it and it is secure.
There are ways of installing rogue "modules" even without module support.
Google and ye shall find... [One of the online cracker mags had an
extensive article on module/kernel based rootkits some 6-10 months
ago]. And if "someone" gets root, they can leave behind a doctored kernel
any time they want. You'll never notice...
--
Dr. Horst H. von Brand User #22616 counter.li.org
Departamento de Informatica Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria +56 32 654239
Casilla 110-V, Valparaiso, Chile Fax: +56 32 797513
"Richard B. Tilley " "(Brad)" <[email protected]> said:
> Then, not supporting loadable modules *is* more secure. By not
> supporting them, you are decreasing the ease in which the kernel can be
> modified. There are fewer people who can "fiddle with memory by hand"
> than there are that can insert a loadable module... a lot fewer, don't
> you agree?
There are published programs that do the fiddling automatically, no
expertise needed. Just as Aunt Tillie is able to use Linux without the
slightest clue about what goes on inside...
--
Dr. Horst H. von Brand User #22616 counter.li.org
Departamento de Informatica Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria +56 32 654239
Casilla 110-V, Valparaiso, Chile Fax: +56 32 797513
> Hmmm, I think the traditional text diagnostic messages are best kept
> as they are, otherwise we'll end up with bug reports like this:
>
> Date: Jan 28 14:39:29 2006
> Subject: Kernel 3.6.2 boot failiure
> To: [email protected]
>
> Hi,
>
> I just upgraded from 3.6.1, which booted fine, to 3.6.2, which stops
> after Tux has waved twice, and winked his left eye.
The point is that Linux should allow for a user-friendly image (yes! possibly
with Tux winking with the eyes or something - in a Mac sorta way). This will
allow for higher user-friendlyness, but should be turned off by default. That
way, SuSE, RedHat and the rest can turn it on if they want to do support
without the verbose messaging. Perhaps do it like 'if splash screen's active,
one can disable it by holding SHIFT or something pressed down'.
I don't know about you, but most non-technical people DO NOT LIKE verbose
messages they can't understand. My father was scared by the linux bootup when
I installed Linux on their PC. I beleive most non-techies like the 'windows
is now hopefully starting' screen. It shields the info they don't want.
So please - don't scare the 'normal' people with (in their eyes) verbose crap.
roy
--
Roy Sigurd Karlsbakk, Datavaktmester
ProntoTV AS - http://www.pronto.tv/
Tel: +47 9801 3356
Computers are like air conditioners.
They stop working when you open Windows.
On Wednesday 29 January 2003 09:06, Zwane Mwaikambo wrote:
> > I'm working for a company doing VoD and IPTV applications, and you surely
> > don't want some verbose kernel output upon booting set-top-boxes. At
> > least - the customer doesn't want it, meaning you shouldn't have it. Then
> > it's better to have some LED flashing in case of error.
>
> Then don't output kernel messages to your default display device. No
> Kernel Hacking Required (tm)
Still - it'd be cute to have a 'Pronto Television blah starting' splash.
Why not include it? It won't break anything!
roy
--
Roy Sigurd Karlsbakk, Datavaktmester
ProntoTV AS - http://www.pronto.tv/
Tel: +47 9801 3356
Computers are like air conditioners.
They stop working when you open Windows.
> Right. And this is a very specialized application, where the vendor will
> install a heavily hacked system anyway, as they won't have standard VGA or
> keyboard or...
It's a standard system. The keyboard and remote control are just hacked into
the PS/2 input. It _does_ have standard VGA etc. This is the normal way of
doing set-top-boxes today.
> It has no relevance whatsoever to the standard, for-your-typical-box
yes, it does.
--
Roy Sigurd Karlsbakk, Datavaktmester
ProntoTV AS - http://www.pronto.tv/
Tel: +47 9801 3356
Computers are like air conditioners.
They stop working when you open Windows.
On Wed, 2003-01-29 at 08:09, Roy Sigurd Karlsbakk wrote:
> > Hmmm, I think the traditional text diagnostic messages are best kept
> > as they are, otherwise we'll end up with bug reports like this:
> >
> > Date: Jan 28 14:39:29 2006
> > Subject: Kernel 3.6.2 boot failiure
> > To: [email protected]
> >
> > Hi,
> >
> > I just upgraded from 3.6.1, which booted fine, to 3.6.2, which stops
> > after Tux has waved twice, and winked his left eye.
>
> The point is that Linux should allow for a user-friendly image (yes! possibly
> with Tux winking with the eyes or something - in a Mac sorta way). This will
> allow for higher user-friendlyness, but should be turned off by default. That
> way, SuSE, RedHat and the rest can turn it on if they want to do support
> without the verbose messaging. Perhaps do it like 'if splash screen's active,
> one can disable it by holding SHIFT or something pressed down'.
>
> I don't know about you, but most non-technical people DO NOT LIKE verbose
> messages they can't understand. My father was scared by the linux bootup when
> I installed Linux on their PC.
How do positive terms such as OK or YES scare people?
Hi there,
On Tue, Jan 28, 2003 at 09:48:55AM -0800, Martin J. Bligh wrote:
[Linux and Bootscreens]
> I think it's a better plan to justify new features with an explantion
> of why we should have something, rather than than saying there's no
> reason we shouldn't.
Ok, I would say there are several reasons:
- People like themes and this makes the theme madness more
complete
- Some people get nervous, if they see text (esp. slow
readers for obvious reasons)
- Other people consider graphics archaic and "uncool"
The last 2 apply to kids as well.
So there are usability concerns and the boot might be the right
place to implement that kind bootscreen retainment.
Showing the dmesg log buffer on panic or BUG would be a nice
thing, to retain usability in that case as well.
Regards
Ingo Oeser
--
Science is what we can tell a computer. Art is everything else. --- D.E.Knuth
> How do positive terms such as OK or YES scare people?
The problem is not necessarily the init scripts. Each
distributor can customise these quite easily. The problem
is (how I see it) divided into two parts:
I. The kernel messages. (And yes, I do know
about both, "quiet" and "console=/foo/bar)
II. Both, using "quiet" or "console=/foo/bar",
and customising the init scripts only leaves
you with a blank, black screen. Now, you might
argue how "classic" or "beatieful" black is.
But it should be obvious this is an absolute
matter of taste. A nice picture is far more
appealing, and pleasing, and whatnot.
- Raphael
> The problem is not necessarily the init scripts. Each
> distributor can customise these quite easily. The problem
> is (how I see it) divided into two parts:
>
> I. The kernel messages. (And yes, I do know
> about both, "quiet" and "console=/foo/bar)
>
> II. Both, using "quiet" or "console=/foo/bar",
> and customising the init scripts only leaves
> you with a blank, black screen. Now, you might
> argue how "classic" or "beatieful" black is.
> But it should be obvious this is an absolute
> matter of taste. A nice picture is far more
> appealing, and pleasing, and whatnot.
I don't think it is appropriate to include graphical boot screen
capabilities in the kernel. There are better ways to achieve what you
want to do:
* Verbose start-up messages on the console
Leave things how they are
* No start up messages on the console
Either use the quiet option, or re-direct the output to serial console
or another virtual terminal.
* Graphical boot screen
Do not have the VGA card configured as a console device at all. Set
the console output to a serial port for debugging, if necessary. Have
the bootloader configure the VGA card, and display the graphic. Boot
in to X, and let X re-configure the VGA card at startup.
Where no console is available for diagnostic messages, the keyboard
LEDs could be lit up in sequence to indicate that the system is
booting.
John.
> How do positive terms such as OK or YES scare people?
Goblodoing the frungled devil [OK]
Escraping various gunfred [OK]
Erasing sdfjebr documents [OK]
.. ad nauseum
now who's scared ?
How About This:
Reading Your Credit Card Number [OK]
Obtaining Your Date Of Birth [OK]
Bringing Up Network Interface [OK]
Transferring Your Personal Info to Finland [OK]
Guess that would scare people.
On Wed, 2003-01-29 at 08:58, Xavier Bestel wrote:
>
> > How do positive terms such as OK or YES scare people?
>
> Goblodoing the frungled devil [OK]
> Escraping various gunfred [OK]
> Erasing sdfjebr documents [OK]
>
> .. ad nauseum
>
> now who's scared ?
>
> - People like themes and this makes the theme madness more
> complete
That a problem? Anyway, when you need to compile something
in order to make a change (or append to something as strange
as an initrd, for that matter), I do not consider this themeing.
> - Some people get nervous, if they see text (esp. slow
> readers for obvious reasons)
Humans like to be in control of things. If something's too fast
for them, or they do not understand it (or even worse: both),
they try to keep away from it. This, I believe, is only superceded
by curiousity. (Which would explain initial interest in things
such as Linux).
> - Other people consider graphics archaic and "uncool"
Seems like here, the need for control is even stronger: When
you understand something, and know it is a mighty tool, you
also wish to control it. On the other hand, you want your
mighty tool to always keep you posted on what it's up to.
(See "The Seven Habits of Seedy Assistants (tm)" for reference).
> The last 2 apply to kids as well.
What do you mean? Gets my attention...
> So there are usability concerns and the boot might be the right
> place to implement that kind bootscreen retainment.
Yup. Not to forget that we like eyecandy.
> Showing the dmesg log buffer on panic or BUG would be a nice
> thing, to retain usability in that case as well.
Maybe. Know how ill-reputed the Windows Bluescreen is? Let's
discuss this a bit...
On Wed, Jan 29, 2003 at 02:58:21PM +0100, Xavier Bestel wrote:
>
> > How do positive terms such as OK or YES scare people?
>
> Goblodoing the frungled devil [OK]
> Escraping various gunfred [OK]
> Erasing sdfjebr documents [OK]
>
> .. ad nauseum
>
> now who's scared ?
>
Seems to me the best place to do the graphical thing is at the distro.
Mandrake has been messing with GUI bootscreens for a few versions, now, and
while I haven't been excited by the results I think it's the right place. My
only reason for not liking Mandrake's screens is that I *like* plain text.
The distros will want to tailor the screens themselves, anyway, so let em
do the code, too.
--
Murray J. Root
On Wed, 2003-01-29 at 13:19, Richard B. Tilley (Brad) wrote:
> How do positive terms such as OK or YES scare people?
Because they contain phrases that are not understood. Sometimes
they also contain phrases which trigger concerned responses
even when that isnt appropriate. I've for example been tweaking
some messages so it is clear that the problem the kernel found
it also fixed.
People find the PC boot intimidating (ask PC support people about
people who call in 'lost in the cosmos' and other such gems). Thats
why the PC boot has often gone graphical.
Alan
> [[email protected]]
>
> Seems to me the best place to do the graphical thing is at the distro.
> Mandrake has been messing with GUI bootscreens for a few versions, now, and
> while I haven't been excited by the results I think it's the right place.
"Don't worry, Mr B., I have a cunning plan to solve the problem."
"Yes Baldrick, let us not forget that you tried to solve the problem
of your mother's low ceiling by cutting off her head."
"But this is a really good one. You become a dashing highwayman, then
you can pay all your bills and, on top of that, everyone'll want to
sleep with you."
Okay, anyone else has found Mandrake resemble Baldrick? Maybe it's not
so ugly-looking but their ideas are clearly of the same philosophy.
--
Tomas Szepe <[email protected]>
On 28 January 2003 23:48, Balram Adlakha wrote:
> though i have recently subscribed to this list (OMG, 30 mails/hour!),
> I've read the whole thread about this bootscreen thing... I see
> absolutely _no_ reason why it should not be included in the kernel
> configuration, since there are patches available already to make it
> work, and there are already soo many useless options that adding
> another won't make a difference... Many people will choose to stay
> away from it, including me, but those users coming from the windows
> world are scared out of their wits when they see the kernel booting,
> and I've seen it myself...
Yeah, dude, let's dumb down our users... don't allow them
to become curious and start learning.
Grown up, mature people are scared when they see letters
and numbers on the screen? On what planet am I?
And if someone *is* scared and totally non-curious, well...
do you want to have such a user? /me not.
--
vda
On Wed, 2003-01-29 at 09:46, Denis Vlasenko wrote:
> Yeah, dude, let's dumb down our users... don't allow them
> to become curious and start learning.
As I recall, the way Windows 95/98/ME operated with the bootscreen (and
this might be wise for Linux as well) is to display a bootscreen, but
have it disappear if someone taps the escape key and returned to the
console where they CAN OPTIONALLY read the messages if their heart
desires. In Windows 2000/XP, this is no longer, sadly, how it works,
but there still was (I believe) a mode in that OS where it displays all
the boot messages (/SOS option or similar).
In Novell Netware 5, which uses X-Windows incidentally (I believe), the
way out of the bootscreen was alt-escape if you wanted to switch over to
the boot console, IIRC.
The answer is that there is nothing they should have to see in the boot
messages that is useful to the end users unless an error occurs. Even
if there is, the boot messages do and should scroll by too quickly to be
meaningful.
It's not called dumbing down users, it's called abstracting the boot
process so that the users don't have to think about it and instead can
think about higher order tasks.
Incidentally, a lot of the init "scripted" startup processes, like
"initializing hotplug system: usb" which stop for several seconds really
should occur in parallel rather than sequentially occuring. However,
the nature of expecting their output to be plopped onto a text display
almost require them to be serialized. If they were instead
parallellized, and set up to output to a system log when
failures/succcesses occured, sort of like the windows event log (or the
existing linux dmesg log) the boot process might occur much more
quickly.
Of course, that last paragraph is admittedly off-topic for the lkml,
since the init scripts are not a kernel issue, they are in userland.
The kernel spawns them, though, so I consider it partially relevant (and
the bootscreen, theoretically would still be displayed as they executed,
so I still consider it relevant to this thread).
-Rob
p.s. I'm hoping this thread is about the possibility of putting some
sort of graphical bootscreen up. I'm tuning in late, so if I'm on the
wrong page, just ignore me. Most people on this list have already
filtered me out (plonked me), so you can too. I won't be offfended.
On Wed, 29 Jan 2003, Roy Sigurd Karlsbakk wrote:
> > Then don't output kernel messages to your default display device. No
> > Kernel Hacking Required (tm)
>
> Still - it'd be cute to have a 'Pronto Television blah starting' splash.
> Why not include it? It won't break anything!
Then display that on your default display device...
Zwane
--
function.linuxpower.ca
On Wed, 2003-01-29 10:31:00 -0500, Rob Wilkens <[email protected]>
wrote in message <[email protected]>:
> On Wed, 2003-01-29 at 09:46, Denis Vlasenko wrote:
> > Yeah, dude, let's dumb down our users... don't allow them
> > to become curious and start learning.
>
> As I recall, the way Windows 95/98/ME operated with the bootscreen (and
> this might be wise for Linux as well) is to display a bootscreen, but
> have it disappear if someone taps the escape key and returned to the
> console where they CAN OPTIONALLY read the messages if their heart
> desires. In Windows 2000/XP, this is no longer, sadly, how it works,
> but there still was (I believe) a mode in that OS where it displays all
> the boot messages (/SOS option or similar).
Well, what do you do (think 2.5.x) when you don't have keyboard drivers
compiled in? I currently do use them as loaded modules, *if* at all I do
need them. Most of the time, my alphas only have two (or three) cables
plugged in: power, network and possibly serial console... If there's a
monitor, it's wired to a monitor switch (so you can see the messages,
but basically, since there's no keyboard support loaded, it's useless),
but it won't be used. Serial console is my way to go...
MfG, JBG
--
Jan-Benedict Glaw [email protected] . +49-172-7608481
"Eine Freie Meinung in einem Freien Kopf | Gegen Zensur
fuer einen Freien Staat voll Freier B?rger" | im Internet!
Shell Script APT-Proxy: http://lug-owl.de/~jbglaw/software/ap2/
Maybe we should replace Tux with a dancing paper clip during bootup.
That will make everybody feel right at home and then they won't mind
when Mozilla or X locks up on them; bug reports will go down to zero,
since users will accept them as special features and this mailing list
will have much less crud messages like this one... ;-)
Balram Adlakha <[email protected]> said:
> though i have recently subscribed to this list (OMG, 30 mails/hour!),
> I've read the whole thread about this bootscreen thing... I see
> absolutely _no_ reason why it should not be included in the kernel
> configuration, since there are patches available already to make it work,
> and there are already soo many useless options that adding another won't
> make a difference...
Patches to clean up the useless options are wellcome... but first check if
they are really useless or just there for some hardware you happen not to
have.
--
Dr. Horst H. von Brand User #22616 counter.li.org
Departamento de Informatica Fono: +56 32 654431
Universidad Tecnica Federico Santa Maria +56 32 654239
Casilla 110-V, Valparaiso, Chile Fax: +56 32 797513
On Wed, 29 Jan 2003, Roy Sigurd Karlsbakk wrote:
> The point is that Linux should allow for a user-friendly image (yes!
> possibly with Tux winking with the eyes or something - in a Mac sorta
> way). This will allow for higher user-friendlyness,
0 2 4 6 8 10
/
/
/
/
/
/
/
TROLL-O-METER
Even though you might not have intended it, the effect on
lkml will be the same ;)
Rik
--
Bravely reimplemented by the knights who say "NIH".
http://www.surriel.com/ http://guru.conectiva.com/
Current spamtrap: <a href=mailto:"[email protected]">[email protected]</a>
> Reading Your Credit Card Number [OK]
> Obtaining Your Date Of Birth [OK]
> Bringing Up Network Interface [OK]
> Transferring Your Personal Info to Finland [OK]
Checking for intelligent life [FAILED]
--
Lab tests show that use of micro$oft causes cancer in lab animals
> The point is that Linux should allow for a user-friendly image (yes!
> possibly with Tux winking with the eyes or something - in a Mac sorta
> way). This will allow for higher user-friendlyness,
Had not caught that message before. Actually I think animations are not
necessary, and especially shall a bootscreen be configurable.
> 0 2 4 6 8 10
>
> /
> /
> /
> /
> /
> /
> /
> TROLL-O-METER
Lol! Kudos for this one Rik. Had me nearly lying on the floor there :-)
On Wed, 29 Jan 2003 09:35:40 +0100
Raphael Schmid <[email protected]> wrote:
> Dunno about you, but for me the kernel itself takes some time
> to come up and handle things over to init. Not covering that
> time would just feel... unfinished.
Just very few seconds in most of the systems, would we have to
bother with that very few time? A black screen isn't so bad.
> I'd rather not compare ourselves with XP all too closely.
Agree :) But if they could put a mpeg video while booting
the kernel, they'd have done it ;)
> You have SCSI disks only, I presume?
Only ide here. That's the only part that seems to "stall" more
than other parts.
Diego Calleja
Rob,
--On 29 January 2003 10:31 -0500 Rob Wilkens <[email protected]> wrote:
[huge snip]
> p.s. I'm hoping this thread is about the possibility of putting some
> sort of graphical bootscreen up. I'm tuning in late, so if I'm on the
> wrong page, just ignore me.
The LKML message I/O scheduling rules suggest that if a fair number of
reads are not processed prior to write activity, then a lot of bandwidth
can be wasted and the medium can eventually become write-only due to
messsage storms. This can be avoided by ensuring writers block to allow
their read activity to progress first - the write should in general be
dependent on the reads in any case. As per Andrew Morton's post on
anticipatory scheduling, a short delay before doing the write to allow
further reads to take place is often a good idea in this context too.
--
Alex Bligh
On Wednesday 29 January 2003 20:16, you wrote:
> Yeah, dude, let's dumb down our users... don't allow them
> to become curious and start learning.
>
> Grown up, mature people are scared when they see letters
> and numbers on the screen? On what planet am I?
>
> And if someone *is* scared and totally non-curious, well...
> do you want to have such a user? /me not.
yes they do get scared!!!
Alright I guess they are not mature...
So 95% of all people are not mature...
So linux is only for the remaining 5%?
On Wednesday 29 January 2003 20:38, Alan Cox wrote:
> On Wed, 2003-01-29 at 13:19, Richard B. Tilley (Brad) wrote:
> > How do positive terms such as OK or YES scare people?
>
> Because they contain phrases that are not understood. Sometimes
> they also contain phrases which trigger concerned responses
> even when that isnt appropriate. I've for example been tweaking
> some messages so it is clear that the problem the kernel found
> it also fixed.
>
> People find the PC boot intimidating (ask PC support people about
> people who call in 'lost in the cosmos' and other such gems). Thats
> why the PC boot has often gone graphical.
>
> Alan
Exactly...
So now its justified that there SHOULD be such a thing, and it SHOULD be
WITHING THE KERNEL SOURCES because so many people will be using it and
we don't want so many patched kernels do we?
So now that there SHOULD be sucha thing, why not create a few implimentations
of it which actually work well and put them into the sources?
People seem to like MacOS X don't they? With a nice bootscreen, and with the
latest version of kde, won't linux be similiar/better than macOS to the
"click click" type users?
> > People find the PC boot intimidating (ask PC support people about
> > people who call in 'lost in the cosmos' and other such gems). Thats
> > why the PC boot has often gone graphical.
>
> Exactly...
> So now its justified that there SHOULD be such a thing,
Possibly.
> and it SHOULD be WITHING THE KERNEL SOURCES
No.
> because so many people will be using it and we don't want so many
> patched kernels do we?
It should be within the bootloader, and the kernel should have an
option not to switch to the console until the login prompt, (I.E. use
two pages of VGA memory, and have the console initialised and being
written to, but not displayed until it either oopses, (in which case
you just toggle the bits in the VGA memory to switch to the other
page, simple enough to do), or as the last task before init is run).
> So now that there SHOULD be sucha thing, why not create a few implimentations
> of it which actually work well and put them into the sources?
For the same reason we got rid of the in kernel boot loader. If it's
there, it'll get used, which is a bad thing.
John
On 29 January 2003 21:52, Balram Adlakha wrote:
> On Wednesday 29 January 2003 20:16, you wrote:
> > Yeah, dude, let's dumb down our users... don't allow them
> > to become curious and start learning.
> >
> > Grown up, mature people are scared when they see letters
> > and numbers on the screen? On what planet am I?
> >
> > And if someone *is* scared and totally non-curious, well...
> > do you want to have such a user? /me not.
>
> yes they do get scared!!!
> Alright I guess they are not mature...
> So 95% of all people are not mature...
>
> So linux is only for the remaining 5%?
This presumes that people don't change. They do,
especially younger ones. Linux gives them a chance.
I am sure you have seen Windows generation geeks:
even those who can potentially learn did not do that
because Windows did not let them do that.
Bootscreen blocked it all.
--
vda 3307
In article <[email protected]> you wrote:
> I am sure you have seen Windows generation geeks:
> even those who can potentially learn did not do that
> because Windows did not let them do that.
Well actually I do not know much people who start their windows with /SOS or
other verbose boot options, even if this is possible with windows.
I agree, that a clean boot screen where only critical information is displayed
(and possibly with some fancy animation) is good for most desktop users. It
helps them more to learn than it will stop them. There is no reason to not
add a "verbose" option.
Greetings
Bernd
--
eckes privat - http://www.eckes.org/
Project Freefire - http://www.freefire.org/
Balram Adlakha wrote:
>
> On Wednesday 29 January 2003 20:16, you wrote:
>
> > Yeah, dude, let's dumb down our users... don't allow them
> > to become curious and start learning.
> >
> > Grown up, mature people are scared when they see letters
> > and numbers on the screen? On what planet am I?
> >
> > And if someone *is* scared and totally non-curious, well...
> > do you want to have such a user? /me not.
>
> yes they do get scared!!!
Instruct tech support to
tell the user to not waste his (or the companys) time staring at
the booting pc. He should do something useful like reading snail mail or
cleaning up his desk. The login window appear soon enough and then
the user forgets about that strange text. The only time you
get a worried user reading the text is when the machine
actually hang and leave the text long enough for him to read.
And then the messages _will_ be useful for support.
_Scared_ by _text_? Sure. The first time. And perhaps the
next time. And then it is just one of lifes little oddities,
"the pc pukes some boring text while starting up, nothing
exciting about that, I can login and start working faster than the
other guys anyway."
> Alright I guess they are not mature...
> So 95% of all people are not mature...
>
> So linux is only for the remaining 5%?
Hey, I turned on the pc, and some lights lit up!
- The power led?
Some of them were RED! I'm scared!...
Helge Hafting
* Denis Vlasenko <[email protected]> [030129 15:46]:
> Yeah, dude, let's dumb down our users... don't allow them
> to become curious and start learning.
How would a graphical bootscreen keep them away from becoming curious?
With my bootsplash patch you can either choose to hide the kernel
messages completely, or have the boot messages scroll down on a
graphical background. No information missing, it just looks a lot
better.
> Grown up, mature people are scared when they see letters
> and numbers on the screen? On what planet am I?
The world is just not just technophile freaks, sorry if this
is new to you ;-)
> And if someone *is* scared and totally non-curious, well...
> do you want to have such a user? /me not.
Freedom for those who are more equal than others.
Regards
Stefan
--
The use of COBOL cripples the mind; its teaching should, therefore, be
regarded as a criminal offense. -- E. W. Dijkstra
* Balram Adlakha <[email protected]> [030129 20:51]:
> Exactly...
> So now its justified that there SHOULD be such a thing, and it SHOULD be
> WITHING THE KERNEL SOURCES because so many people will be using it and
> we don't want so many patched kernels do we?
> So now that there SHOULD be sucha thing, why not create a few implimentations
> of it which actually work well and put them into the sources?
> People seem to like MacOS X don't they? With a nice bootscreen, and with the
> latest version of kde, won't linux be similiar/better than macOS to the
> "click click" type users?
There's one implementation that is used by UnitedLinux, Mandrake, SuSE,
(and Gentoo?), patches are available at
ftp.suse.com/pub/people/stepan/bootsplash/
Stefan
--
The use of COBOL cripples the mind; its teaching should, therefore, be
regarded as a criminal offense. -- E. W. Dijkstra
* John Bradford <[email protected]> [030129 14:48]:
> I don't think it is appropriate to include graphical boot screen
> capabilities in the kernel. There are better ways to achieve what you
> want to do:
>
> * Verbose start-up messages on the console
> Leave things how they are
agreed
> * No start up messages on the console
> Either use the quiet option, or re-direct the output to serial console
> or another virtual terminal.
> * Graphical boot screen
> Do not have the VGA card configured as a console device at all. Set
> the console output to a serial port for debugging, if necessary. Have
> the bootloader configure the VGA card, and display the graphic. Boot
> in to X, and let X re-configure the VGA card at startup.
Some method to switch between verbose and non-verbose mode on the screen
interactively would be useful and it should not loose old messages. Not
have vga configured as console device kind of gives you a hard time
here. My bootsplash patch takes care of this by only "hiding" the text
instead of not redirect it to vga, so the text buffers are valid all the
time and visible as soon as you switch them on
Stefan
--
The use of COBOL cripples the mind; its teaching should, therefore, be
regarded as a criminal offense. -- E. W. Dijkstra
Roy wrote:
[Bootscreen]
> I don't know about you, but most non-technical people DO NOT LIKE verbose
> messages they can't understand. My father was scared by the linux bootup when
> I installed Linux on their PC. I beleive most non-techies like the 'windows
> is now hopefully starting' screen. It shields the info they don't want.
>
> So please - don't scare the 'normal' people with (in their eyes) verbose crap.
Till Windows Me, every Windows user got textmode messages at boot time.
And they survived it.
--
CU Lars
Lars 'Cebewee' Noschinski wrote:
> Till Windows Me, every Windows user got textmode messages at boot time.
> And they survived it.
I don't know what you're running, but I get nothing between "starting
windows 98" and the gui starting up. Win95 was the same way."
Chris
--
Chris Friesen | MailStop: 043/33/F10
Nortel Networks | work: (613) 765-0557
3500 Carling Avenue | fax: (613) 765-2986
Nepean, ON K2H 8E9 Canada | email: [email protected]
On Thu, Jan 30, 2003 at 07:53:03PM +0100, Lars 'Cebewee' Noschinski wrote:
> Till Windows Me, every Windows user got textmode messages at boot time.
> And they survived it.
Most Win98 boxes (at least in my experience) don't show text-mode
message as boot either. Same for *all* of the (now rare) Win95 boxes
I've seen lately.
It depends on whether you have any DOS programs installed in
autoexec.bat or config.sys that output text. AFAICT many old antivirus
programs used to spew stuff at boot, but as people update their
antivirus programs over time, this is becoming less common.
Also, Win95/98 rarely displayed more than a screenful of text, virtually
never displayed more than two screenfuls, and usually displayed only a
few lines. Compare with Linux, spewing a multi-screen waterfall of text
before init even starts. Some people respond by backing away from the
computer in fear. Others say things like "Daaaaaaamn!" when they
suddenly recognize the true speed of their video hardware.
I haven't tried confronting an average person with "quiet" added to the
boot command line arguments yet. That might be sufficient to fix the
problem. (OTOH I haven't read the full thread yet so I don't know if
anyone else has tried this.)
-Barry K. Nathan <[email protected]>
Barry K. Nathan wrote:
>On Thu, Jan 30, 2003 at 07:53:03PM +0100, Lars 'Cebewee' Noschinski wrote:
>
>
>>Till Windows Me, every Windows user got textmode messages at boot time.
>>And they survived it.
>>
>>
>
>Most Win98 boxes (at least in my experience) don't show text-mode
>message as boot either. Same for *all* of the (now rare) Win95 boxes
>I've seen lately.
>
>
FYI:
If you press "ESC" during those SO's bootup, you'll get the text
(actually I think you have to press it twice (one for pre-OS initial
loading and other for until display adapter is reinitialized)).
Also there's an option if I recall in msdos.sys that will disable the
M$W logo at bootup.
--
?lvaro Lopes
---------------------
A .sig is just a .sig
Hi!
> Yeah, I'd really like a stable working swsusp (on a working kernel) to
> shortcut the fscking boot. Go pawel !
Well, with right scripts you can probably boot
faster than do resume (and you can certainly
shutdown faster than suspend). OTOH, if you
turn off ide-scsi in 2.5.59, swsusp should just
work.
--
Pavel
Written on sharp zaurus, because my Velo1 broke. If you have Velo you don't need...
On Thu, 30 January 2003 08:25:22 +0100, Pavel Machek wrote:
>
> Well, with right scripts you can probably boot
> faster than do resume (and you can certainly
> shutdown faster than suspend). OTOH, if you
> turn off ide-scsi in 2.5.59, swsusp should just
> work.
But it takes quite a while to open all those editor windows again, let
alone remembering, where you were when you closed them. This is the
real benefit I see in software suspend.
J?rn
--
Homo Sapiens is a goal, not a description.
-- unknown
Hi!
> > Well, with right scripts you can probably boot
> > faster than do resume (and you can certainly
> > shutdown faster than suspend). OTOH, if you
> > turn off ide-scsi in 2.5.59, swsusp should just
> > work.
>
> But it takes quite a while to open all those editor windows again, let
> alone remembering, where you were when you closed them. This is the
> real benefit I see in software suspend.
Hey, I'm developing swsusp, no need to tell *me* its usefull. Primary
application here will be to be able to do long-running computation
(lingvistic experiments) and *still* sleep at night ;-).
Pavel
--
Casualities in World Trade Center: ~3k dead inside the building,
cryptography in U.S.A. and free speech in Czech Republic.