WelcomeWelcome | FAQFAQ | DownloadsDownloads | WikiWiki

Author Topic: building realtime kernel & remastering for tiny core linux  (Read 15982 times)

Offline webb

  • Jr. Member
  • **
  • Posts: 52
Re: building realtime kernel & remastering for tiny core linux
« Reply #30 on: November 26, 2012, 04:55:12 AM »
interestingly, if i build a 64-bit 3.0.21 kernel with tinycore patches and config-3.0.21-tinycore64, but set the version string to 3.0.21-w64, it does boot. if i compile with the same .config and patches but with version string 3.0.21-tinycore64, then i cannot boot, just as with official tinycore vmlinuz64. ("fixing recursive fault" error)

-webb

Online Rich

  • TinyCore Moderator
  • Hero Member
  • *****
  • Posts: 6888
Re: building realtime kernel & remastering for tiny core linux
« Reply #31 on: November 26, 2012, 10:14:12 AM »
Hi webb
Another forum member  solved his  "fixing recursive fault"  error using the  maxcpus=0  boot code.

Offline webb

  • Jr. Member
  • **
  • Posts: 52
Re: building realtime kernel & remastering for tiny core linux
« Reply #32 on: November 27, 2012, 05:19:03 PM »
yes but i like all of my four cores!

here's a hypothesis about this whole recursive fault business:

Code: [Select]
3.0.21-tinycore64
^^^^^^^^^^^^^^^^^ <-- 17 character version string, doesn't boot

3.0.3-tinycore64
^^^^^^^^^^^^^^^^ <-- 16 character version string, boots

3.0.21-w64
^^^^^^^^^^ <-- 10 character version string, compiled with config-3.0.21-tinycore64 but with version string changed. boots.

perhaps the kernel expects a maximum of 16 bytes in the version string.

-webb

Offline althalus

  • Sr. Member
  • ****
  • Posts: 351
Re: building realtime kernel & remastering for tiny core linux
« Reply #33 on: November 27, 2012, 06:00:51 PM »
yes but i like all of my four cores!

here's a hypothesis about this whole recursive fault business:

Code: [Select]
3.0.21-tinycore64
^^^^^^^^^^^^^^^^^ <-- 17 character version string, doesn't boot

3.0.3-tinycore64
^^^^^^^^^^^^^^^^ <-- 16 character version string, boots

3.0.21-w64
^^^^^^^^^^ <-- 10 character version string, compiled with config-3.0.21-tinycore64 but with version string changed. boots.

perhaps the kernel expects a maximum of 16 bytes in the version string.

-webb
Seems coincidental. The version string doesn't change just by adding maxcpus=0, but doing so lets my system boot.

Idle thought I haven't had a chance to test and/or research yet - Setting maxcpus=0 actually does two things - uses only 1 cpu, AND disables IO APIC. Maybe there's another option that can disable APIC whilst still keeping all of your CPUs, and maybe that will be enough to get around the recursive fault erro

Online Rich

  • TinyCore Moderator
  • Hero Member
  • *****
  • Posts: 6888
Re: building realtime kernel & remastering for tiny core linux
« Reply #34 on: November 27, 2012, 06:11:32 PM »
Hi webb
Quote
yes but i like all of my four cores!
That may be, but one core that boots runs better than four that don't.
Quote
perhaps the kernel expects a maximum of 16 bytes in the version string.
Perhaps. Or maybe the longer string just pushes some data in the kernel into a position that causes a
seg fault.

Online Rich

  • TinyCore Moderator
  • Hero Member
  • *****
  • Posts: 6888
Re: building realtime kernel & remastering for tiny core linux
« Reply #35 on: November 27, 2012, 06:20:36 PM »
Hi althalus
If you go to  http://www.mjmwired.net/kernel/Documentation/kernel-parameters.txt  you can use your browser
to search for instances of  IOAPIC  for other possible options.

Offline webb

  • Jr. Member
  • **
  • Posts: 52
Re: building realtime kernel & remastering for tiny core linux
« Reply #36 on: November 27, 2012, 06:38:49 PM »
althalus wrote:
> Seems coincidental

unlikely. i compiled my own 3.0.21-tinycore64 and 3.0.21-w64 kernels, one after the other, and using the same config-3.0.21-tinycore64 kernel configuration, varying only the version string. the kernel with the 17 character version string won't boot, 16 or fewer will. i'll also try 3.0.21-tinycor64 and 3.0.21-tinycore64b, and also with a newer kernel version (e.g., 3.0.52) to see if this is a kernel bug that has been fixed.

rich wrote:
> maybe the longer string just pushes some data in the kernel into a position that causes a seg fault

sounds plausible.

-webb

Offline bricolodu13

  • WikiUser
  • *
  • Posts: 10
Re: building realtime kernel & remastering for tiny core linux
« Reply #37 on: June 19, 2013, 12:08:17 PM »
Hello,

I'm absolutly newbie but want to learn...

I've setup a basic i386 install for a squeezelite setup.
I want RT Kernel.

when I type "git clone git://gitorious.org/tinyrt/linux.git
I get the following message "-sh: git: not found" !
What am I missing ?

Thank you
Jean

Online Rich

  • TinyCore Moderator
  • Hero Member
  • *****
  • Posts: 6888
Re: building realtime kernel & remastering for tiny core linux
« Reply #38 on: June 19, 2013, 12:34:44 PM »
Hi bricolodu13
You can find the RT kernel here:
http://tinycorelinux.net/4.x/x86/contrib/rt-kernel/
Or you can go to:
https://gitorious.org/tinyrt
and as it says in the first post:
Quote
click linux-->source tree and download master as tar.gz

just run ./build.sh

Offline bricolodu13

  • WikiUser
  • *
  • Posts: 10
Re: building realtime kernel & remastering for tiny core linux
« Reply #39 on: June 21, 2013, 12:30:00 PM »
Thank you Rich,

I have downloaded the  tinyrt-linux-master.tar.gz file and have copied it through ssh (WinSCP) to the TC directory.
I have also made the build.sh and two other .sh files with vi to the same directory.
I have made them usable with chmod a+x
but still, when I do ./build.sh, I got the following message :
./build-kernel.sh: .: line 15: functions.sh: not found
./build-initrd.sh: .: line 15: functions.sh: not found

Sorry to be so stupid !  :'(

Best regards,
Jean


Offline tinypoodle

  • Hero Member
  • *****
  • Posts: 3857
Re: building realtime kernel & remastering for tiny core linux
« Reply #40 on: June 21, 2013, 01:54:46 PM »
Blind guess, perhaps you would use a non compatible filesystem?
"Software gets slower faster than hardware gets faster." Niklaus Wirth - A Plea for Lean Software (1995)