Re: Kernel panic by write to console while switching consoles.

Re: Kernel panic by write to console while switching consoles.

[Carlos Manuel Duclos Vergara]

Hi,

[...]

the console is enabled very early in the boot process in order to print some 
info about the boot process itself. Read start_kernel() in 
$(KERNEL_SOURCE)/init/main.c
for more info

bye


-- 
Carlos Manuel Duclos Vergara
carlos@embedded.cl


-------------------------------------------------------
This SF.Net email is sponsored by OSTG. Have you noticed the changes on
Linux.com, ITManagersJournal and NewsForge in the past few weeks? Now,
one more big change to announce. We are now OSTG- Open Source Technology
Group. Come see the changes on the new OSTG site. www.ostg.com

Kernel panic by write to console while switching consoles.

[Jaap van den Handel]

Hi all, 

The framebuffer driver for our vga device is working.
But when we tried to boot with the console directing to it, it crashed.
We are running linux 2.4.20 on an embedded MIPS @ 175MHz with a terminal 
attached to a serial port. When we specify "console=tty0" the console is 
switched from the serial console to the framebuffer console when our 
framebuffer driver is loaded.

we found that while switching the console, it is not disabled. The console is 
switched and before it is fully initialzed, something is already printing text 
to it.

when we disabled the console during a switch by setting 
"printable = 0;" at the beginning of "take_over_console.c" in drivers/char.
at the end of "take_over_console" we set it back to 1.

after this we were able to boot and see the pinguin logo with a cursor.
since it is not possible to attach a keyboard we tried 
"echo some text > /dev/tty0" and "some text" does appear at the framebuffer 
console :-).

Did we miss a setting to disable writes to the console while initializing our 
driver?
Is it a bug which only comes forward in combination with a serial console or 
relative slow cpu's?

Greetings,
Jaap van den Handel




Confidentiality Note:
This e-mail and any attachments are confidential and the information herein
is property of Quest Innovations B.V.. If you are not the intended recipient(s) of this e-mail,you are hereby notified that any dissemination, distribution, 
copying, use of or reliance upon the information contained within this e-mail 
or attachments is unauthorized and strictly prohibited. If you have received
this e-mail in error, please notify us immediately by returning it to the
sender and delete this copy from your system.

Quest Innovations B.V.


-------------------------------------------------------
This SF.Net email is sponsored by OSTG. Have you noticed the changes on
Linux.com, ITManagersJournal and NewsForge in the past few weeks? Now,
one more big change to announce. We are now OSTG- Open Source Technology
Group. Come see the changes on the new OSTG site. www.ostg.com

Re: Kernel panic by write to console while switching consoles.

[Antonino A. Daplas]

On Thursday 29 July 2004 23:57, Jaap van den Handel wrote:
>
> Did we miss a setting to disable writes to the console while initializing
> our driver?
> Is it a bug which only comes forward in combination with a serial console
> or relative slow cpu's?

This is a problem in 2.4, hopefully fixed by BenH in 2.6 by acquiring the
console semaphore at strategic places.

Try placing acquire_console_sem(), release_console_sem() around the loop
in take_over_console() to disable early printing during console initialization.

Tony 




-------------------------------------------------------
This SF.Net email is sponsored by OSTG. Have you noticed the changes on
Linux.com, ITManagersJournal and NewsForge in the past few weeks? Now,
one more big change to announce. We are now OSTG- Open Source Technology
Group. Come see the changes on the new OSTG site. www.ostg.com

Re: Kernel panic by write to console while switching consoles.

[Jaap van den Handel]

> the console is enabled very early in the boot process in order to print some 
> info about the boot process itself. Read start_kernel() in 
> $(KERNEL_SOURCE)/init/main.c 
> for more info 

The console can be switched after the boot process has finished right? 
When loading a driver module for example. Then it would also crash when sombody 
prints to the screen. Only when booting this change is quite big. 

> This is a problem in 2.4, hopefully fixed by BenH in 2.6 by acquiring the 
> console semaphore at strategic places. 
> 
> Try placing acquire_console_sem(), release_console_sem() around the loop 
> in take_over_console() to disable early printing during console 
> initialization. 
> 

Thanks for the reply, glad to know somebody is working on it. 

Greetings 
Jaap van den Handel

Confidentiality Note:
This e-mail and any attachments are confidential and the information herein
is property of Quest Innovations B.V.. If you are not the intended recipient(s) of this e-mail,you are hereby notified that any dissemination, distribution, 
copying, use of or reliance upon the information contained within this e-mail 
or attachments is unauthorized and strictly prohibited. If you have received
this e-mail in error, please notify us immediately by returning it to the
sender and delete this copy from your system.

Quest Innovations B.V.


-------------------------------------------------------
This SF.Net email is sponsored by OSTG. Have you noticed the changes on
Linux.com, ITManagersJournal and NewsForge in the past few weeks? Now,
one more big change to announce. We are now OSTG- Open Source Technology
Group. Come see the changes on the new OSTG site. www.ostg.com