net 5501 and RHEL5.2

[Przemyslaw Maszenda]

Hi

I try to boot and install RedHat Enterprise Linux 5.2 on net5501-70
(with sandisk CF but this doesn't matter, I think), but without success.

With original RHEL5.2 kernel booting from CF (CF was installed on
another PC) or booting over PXE I received:

############################################################################
Linux version 2.6.18-92.1.1.el5
(brewbuilder@...) (gcc v
ersion 4.1.2 20071124 (Red Hat 4.1.2-41)) #1 SMP Thu May 22 09:01:29 EDT
2008
BIOS-provided physical RAM map:

[...]

TCP bic registered
Initializing IPsec netlink socket
NET: Registered protocol family 1
NET: Registered protocol family 17
powernow-k8: Processor cpuid 5a2 not supported
BUG: unable to handle kernel NULL pointer dereference at virtual address
0000000
0
  printing eip:
c041041c
*pde = 00000000
Oops: 0000 [#1]
SMP
last sysfs file:
Modules linked in:
CPU:    0
EIP:    0060:[<c041041c>]    Not tainted VLI
EFLAGS: 00010202   (2.6.18-92.1.1.el5 #1)
EIP is at powernowk8_init+0x5e/0x1c2
eax: 00000000   ebx: 00000000   ecx: 0000000e   edx: 00000020
esi: 00000000   edi: c06242c3   ebp: 00000000   esp: c147cfa0
ds: 007b   es: 007b   ss: 0068
Process swapper (pid: 1, ti=c147c000 task=c147daa0 task.ti=c147c000)
Stack: 00000000 c071bbe8 00000000 c06ec5a8 c06e7fd8 c0404dee 00000202
c06ec42b
        00000000 00000000 00000000 00000000 00000000 00000000 c06ec42b
00000000
        00000000 c0405c3b 00000000 00000000 00000000 00000000 00000000
00000000
Call Trace:
  [<c06ec5a8>] init+0x17d/0x24a
  [<c0404dee>] ret_from_fork+0x6/0x1c
  [<c06ec42b>] init+0x0/0x24a
  [<c06ec42b>] init+0x0/0x24a
  [<c0405c3b>] kernel_thread_helper+0x7/0x10
  =======================
Code: 83 3d 20 41 67 c0 01 75 40 83 3d 84 d4 76 c0 00 75 37 b8 01 00 00
00 bf c3
  42 62 c0 e8 0a 15 19 00 b9 0f 00 00 00 89 c6 49 78 08 <ac> ae 75 08 84
c0 75 f5
  31 c0 eb 04 19 c0 0c 01 85 c0 75 0a c7
EIP: [<c041041c>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:c147cfa0
  <0>Kernel panic - not syncing: Fatal exception

############################################################################


I recompiled kernel from redhat .src.rpm (with rpmbuild, with all
patches ...), with Geode GX/LX CPU. I included support for AMD
powernow-k8. I tried to boot and I received similar as before:

############################################################################
Linux version 2.6.18-SOEKRIS (root@vivaldi) (gcc version 4.1.2 20071124
(Red Hat
  4.1.2-42)) #1 SMP Fri Jun 20 07:58:23 CEST 2008

[...]

TCP bic registered
Initializing IPsec netlink socket
NET: Registered protocol family 1
NET: Registered protocol family 17
powernow-k8: Processor cpuid 5a2 not supported
BUG: unable to handle kernel NULL pointer dereference at virtual address
0000000
0
  printing eip:
c04104f5
*pde = 00000000
Oops: 0000 [#1]
SMP
last sysfs file:
Modules linked in:
CPU:    0
EIP:    0060:[<c04104f5>]    Not tainted VLI
EFLAGS: 00010202   (2.6.18-SOEKRIS #1)
EIP is at powernowk8_init+0x5e/0x1c2
eax: 00000000   ebx: 00000000   ecx: 0000000e   edx: c06e7fe4
esi: 00000000   edi: c0626270   ebp: 00000000   esp: dffe9fa0
ds: 007b   es: 007b   ss: 0068
Process swapper (pid: 1, ti=dffe9000 task=dffe8aa0 task.ti=dffe9000)
Stack: 00000000 c071da88 00000000 c06ee5a9 c06e9fd8 c0404dce 00000202
c06ee42c
        00000000 00000000 00000000 00000000 00000000 00000000 c06ee42c
00000000
        00000000 c0405c1b 00000000 00000000 00000000 00000000 00000000
00000000
Call Trace:
  [<c06ee5a9>] init+0x17d/0x24a
  [<c0404dce>] ret_from_fork+0x6/0x1c
  [<c06ee42c>] init+0x0/0x24a
  [<c06ee42c>] init+0x0/0x24a
  [<c0405c1b>] kernel_thread_helper+0x7/0x10
  =======================
Code: 83 3d 20 61 67 c0 01 75 40 83 3d 84 f4 76 c0 00 75 37 b8 01 00 00
00 bf 70
  62 62 c0 e8 81 29 19 00 b9 0f 00 00 00 89 c6 49 78 08 <ac> ae 75 08 84
c0 75 f5
  31 c0 eb 04 19 c0 0c 01 85 c0 75 0a c7
EIP: [<c04104f5>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:dffe9fa0
  <0>Kernel panic - not syncing: Fatal exception

#######


I know, I lame :), but what should I do run to RHEL5.2 with net5501-70?

1) Should I recompile kernel? (For me is important to have kernel
version and patchlevel exactly as RHEL5.2, changing arch is acceptable)

2) For best performance and stability what should I choose in
menuconfig. Which arch and CPU?

PS.
I checked that over PXE I could boot kernel from RHEL4.6 ...

thanks.







_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech

[Otfried Geffert-2]

>> EIP: [<c041041c>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:c147cfa0

Looks like there is a kernel module called "powernowk8"
The crash is while loading this module.

Get rid of the kernel module in the config.

OR: Maybe the module is part of the initrd RAM disk.
You can make your own with
# mkdinitrd  your_rd   kernel_version

Otfried

----- Original Message -----
From: "Przemyslaw Maszenda" <p.maszenda@...>
To: <soekris-tech@...>
Sent: Friday, June 20, 2008 9:32 AM
Subject: [Soekris] net 5501 and RHEL5.2

> Hi
>
> I try to boot and install RedHat Enterprise Linux 5.2 on net5501-70
> (with sandisk CF but this doesn't matter, I think), but without success.
>
> With original RHEL5.2 kernel booting from CF (CF was installed on
> another PC) or booting over PXE I received:
>
> ############################################################################
> Linux version 2.6.18-92.1.1.el5
> (brewbuilder@...) (gcc v
> ersion 4.1.2 20071124 (Red Hat 4.1.2-41)) #1 SMP Thu May 22 09:01:29 EDT
> 2008
> BIOS-provided physical RAM map:
>
> [...]
>
> TCP bic registered
> Initializing IPsec netlink socket
> NET: Registered protocol family 1
> NET: Registered protocol family 17
> powernow-k8: Processor cpuid 5a2 not supported
> BUG: unable to handle kernel NULL pointer dereference at virtual address
> 0000000
> 0
>  printing eip:
> c041041c
> *pde = 00000000
> Oops: 0000 [#1]
> SMP
> last sysfs file:
> Modules linked in:
> CPU:    0
> EIP:    0060:[<c041041c>]    Not tainted VLI
> EFLAGS: 00010202   (2.6.18-92.1.1.el5 #1)
> EIP is at powernowk8_init+0x5e/0x1c2
> eax: 00000000   ebx: 00000000   ecx: 0000000e   edx: 00000020
> esi: 00000000   edi: c06242c3   ebp: 00000000   esp: c147cfa0
> ds: 007b   es: 007b   ss: 0068
> Process swapper (pid: 1, ti=c147c000 task=c147daa0 task.ti=c147c000)
> Stack: 00000000 c071bbe8 00000000 c06ec5a8 c06e7fd8 c0404dee 00000202
> c06ec42b
>        00000000 00000000 00000000 00000000 00000000 00000000 c06ec42b
> 00000000
>        00000000 c0405c3b 00000000 00000000 00000000 00000000 00000000
> 00000000
> Call Trace:
>  [<c06ec5a8>] init+0x17d/0x24a
>  [<c0404dee>] ret_from_fork+0x6/0x1c
>  [<c06ec42b>] init+0x0/0x24a
>  [<c06ec42b>] init+0x0/0x24a
>  [<c0405c3b>] kernel_thread_helper+0x7/0x10
>  =======================
> Code: 83 3d 20 41 67 c0 01 75 40 83 3d 84 d4 76 c0 00 75 37 b8 01 00 00
> 00 bf c3
>  42 62 c0 e8 0a 15 19 00 b9 0f 00 00 00 89 c6 49 78 08 <ac> ae 75 08 84
> c0 75 f5
>  31 c0 eb 04 19 c0 0c 01 85 c0 75 0a c7
> EIP: [<c041041c>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:c147cfa0
>  <0>Kernel panic - not syncing: Fatal exception
>
> ############################################################################
>
>
> I recompiled kernel from redhat .src.rpm (with rpmbuild, with all
> patches ...), with Geode GX/LX CPU. I included support for AMD
> powernow-k8. I tried to boot and I received similar as before:
>
> ############################################################################
> Linux version 2.6.18-SOEKRIS (root@vivaldi) (gcc version 4.1.2 20071124
> (Red Hat
>  4.1.2-42)) #1 SMP Fri Jun 20 07:58:23 CEST 2008
>
> [...]
>
> TCP bic registered
> Initializing IPsec netlink socket
> NET: Registered protocol family 1
> NET: Registered protocol family 17
> powernow-k8: Processor cpuid 5a2 not supported
> BUG: unable to handle kernel NULL pointer dereference at virtual address
> 0000000
> 0
>  printing eip:
> c04104f5
> *pde = 00000000
> Oops: 0000 [#1]
> SMP
> last sysfs file:
> Modules linked in:
> CPU:    0
> EIP:    0060:[<c04104f5>]    Not tainted VLI
> EFLAGS: 00010202   (2.6.18-SOEKRIS #1)
> EIP is at powernowk8_init+0x5e/0x1c2
> eax: 00000000   ebx: 00000000   ecx: 0000000e   edx: c06e7fe4
> esi: 00000000   edi: c0626270   ebp: 00000000   esp: dffe9fa0
> ds: 007b   es: 007b   ss: 0068
> Process swapper (pid: 1, ti=dffe9000 task=dffe8aa0 task.ti=dffe9000)
> Stack: 00000000 c071da88 00000000 c06ee5a9 c06e9fd8 c0404dce 00000202
> c06ee42c
>        00000000 00000000 00000000 00000000 00000000 00000000 c06ee42c
> 00000000
>        00000000 c0405c1b 00000000 00000000 00000000 00000000 00000000
> 00000000
> Call Trace:
>  [<c06ee5a9>] init+0x17d/0x24a
>  [<c0404dce>] ret_from_fork+0x6/0x1c
>  [<c06ee42c>] init+0x0/0x24a
>  [<c06ee42c>] init+0x0/0x24a
>  [<c0405c1b>] kernel_thread_helper+0x7/0x10
>  =======================
> Code: 83 3d 20 61 67 c0 01 75 40 83 3d 84 f4 76 c0 00 75 37 b8 01 00 00
> 00 bf 70
>  62 62 c0 e8 81 29 19 00 b9 0f 00 00 00 89 c6 49 78 08 <ac> ae 75 08 84
> c0 75 f5
>  31 c0 eb 04 19 c0 0c 01 85 c0 75 0a c7
> EIP: [<c04104f5>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:dffe9fa0
>  <0>Kernel panic - not syncing: Fatal exception
>
> #######
>
>
> I know, I lame :), but what should I do run to RHEL5.2 with net5501-70?
>
> 1) Should I recompile kernel? (For me is important to have kernel
> version and patchlevel exactly as RHEL5.2, changing arch is acceptable)
>
> 2) For best performance and stability what should I choose in
> menuconfig. Which arch and CPU?
>
> PS.
> I checked that over PXE I could boot kernel from RHEL4.6 ...
>
> thanks.
>
>
>
>
>
>
>
> _______________________________________________
> Soekris-tech mailing list
> Soekris-tech@...
> http://lists.soekris.com/mailman/listinfo/soekris-tech 

_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech

[Sargun Dhillon]

You can't run RHEL 5.X on the Soekris. The 5501 has a '586' CPU, whereas RHEL5 expects a 686 CPU. You're simply going to come across a lot of crashing, bugs, and other issues. This is a bad idea.

Go back to Debian, and debootstrap your machine together.


Sargun Dhillon
925.202.9485
deCarta
sdhillon@...
www.decarta.com





-----Original Message-----
From: soekris-tech-bounces@... on behalf of Otfried Geffert
Sent: Fri 6/20/2008 1:53 AM
To: Przemyslaw Maszenda; soekris-tech@...
Subject: Re: [Soekris] net 5501 and RHEL5.2
 
>> EIP: [<c041041c>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:c147cfa0

Looks like there is a kernel module called "powernowk8"
The crash is while loading this module.

Get rid of the kernel module in the config.

OR: Maybe the module is part of the initrd RAM disk.
You can make your own with
# mkdinitrd  your_rd   kernel_version

Otfried

----- Original Message -----
From: "Przemyslaw Maszenda" <p.maszenda@...>
To: <soekris-tech@...>
Sent: Friday, June 20, 2008 9:32 AM
Subject: [Soekris] net 5501 and RHEL5.2

> Hi
>
> I try to boot and install RedHat Enterprise Linux 5.2 on net5501-70
> (with sandisk CF but this doesn't matter, I think), but without success.
>
> With original RHEL5.2 kernel booting from CF (CF was installed on
> another PC) or booting over PXE I received:
>
> ############################################################################
> Linux version 2.6.18-92.1.1.el5
> (brewbuilder@...) (gcc v
> ersion 4.1.2 20071124 (Red Hat 4.1.2-41)) #1 SMP Thu May 22 09:01:29 EDT
> 2008
> BIOS-provided physical RAM map:
>
> [...]
>
> TCP bic registered
> Initializing IPsec netlink socket
> NET: Registered protocol family 1
> NET: Registered protocol family 17
> powernow-k8: Processor cpuid 5a2 not supported
> BUG: unable to handle kernel NULL pointer dereference at virtual address
> 0000000
> 0
>  printing eip:
> c041041c
> *pde = 00000000
> Oops: 0000 [#1]
> SMP
> last sysfs file:
> Modules linked in:
> CPU:    0
> EIP:    0060:[<c041041c>]    Not tainted VLI
> EFLAGS: 00010202   (2.6.18-92.1.1.el5 #1)
> EIP is at powernowk8_init+0x5e/0x1c2
> eax: 00000000   ebx: 00000000   ecx: 0000000e   edx: 00000020
> esi: 00000000   edi: c06242c3   ebp: 00000000   esp: c147cfa0
> ds: 007b   es: 007b   ss: 0068
> Process swapper (pid: 1, ti=c147c000 task=c147daa0 task.ti=c147c000)
> Stack: 00000000 c071bbe8 00000000 c06ec5a8 c06e7fd8 c0404dee 00000202
> c06ec42b
>        00000000 00000000 00000000 00000000 00000000 00000000 c06ec42b
> 00000000
>        00000000 c0405c3b 00000000 00000000 00000000 00000000 00000000
> 00000000
> Call Trace:
>  [<c06ec5a8>] init+0x17d/0x24a
>  [<c0404dee>] ret_from_fork+0x6/0x1c
>  [<c06ec42b>] init+0x0/0x24a
>  [<c06ec42b>] init+0x0/0x24a
>  [<c0405c3b>] kernel_thread_helper+0x7/0x10
>  =======================
> Code: 83 3d 20 41 67 c0 01 75 40 83 3d 84 d4 76 c0 00 75 37 b8 01 00 00
> 00 bf c3
>  42 62 c0 e8 0a 15 19 00 b9 0f 00 00 00 89 c6 49 78 08 <ac> ae 75 08 84
> c0 75 f5
>  31 c0 eb 04 19 c0 0c 01 85 c0 75 0a c7
> EIP: [<c041041c>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:c147cfa0
>  <0>Kernel panic - not syncing: Fatal exception
>
> ############################################################################
>
>
> I recompiled kernel from redhat .src.rpm (with rpmbuild, with all
> patches ...), with Geode GX/LX CPU. I included support for AMD
> powernow-k8. I tried to boot and I received similar as before:
>
> ############################################################################
> Linux version 2.6.18-SOEKRIS (root@vivaldi) (gcc version 4.1.2 20071124
> (Red Hat
>  4.1.2-42)) #1 SMP Fri Jun 20 07:58:23 CEST 2008
>
> [...]
>
> TCP bic registered
> Initializing IPsec netlink socket
> NET: Registered protocol family 1
> NET: Registered protocol family 17
> powernow-k8: Processor cpuid 5a2 not supported
> BUG: unable to handle kernel NULL pointer dereference at virtual address
> 0000000
> 0
>  printing eip:
> c04104f5
> *pde = 00000000
> Oops: 0000 [#1]
> SMP
> last sysfs file:
> Modules linked in:
> CPU:    0
> EIP:    0060:[<c04104f5>]    Not tainted VLI
> EFLAGS: 00010202   (2.6.18-SOEKRIS #1)
> EIP is at powernowk8_init+0x5e/0x1c2
> eax: 00000000   ebx: 00000000   ecx: 0000000e   edx: c06e7fe4
> esi: 00000000   edi: c0626270   ebp: 00000000   esp: dffe9fa0
> ds: 007b   es: 007b   ss: 0068
> Process swapper (pid: 1, ti=dffe9000 task=dffe8aa0 task.ti=dffe9000)
> Stack: 00000000 c071da88 00000000 c06ee5a9 c06e9fd8 c0404dce 00000202
> c06ee42c
>        00000000 00000000 00000000 00000000 00000000 00000000 c06ee42c
> 00000000
>        00000000 c0405c1b 00000000 00000000 00000000 00000000 00000000
> 00000000
> Call Trace:
>  [<c06ee5a9>] init+0x17d/0x24a
>  [<c0404dce>] ret_from_fork+0x6/0x1c
>  [<c06ee42c>] init+0x0/0x24a
>  [<c06ee42c>] init+0x0/0x24a
>  [<c0405c1b>] kernel_thread_helper+0x7/0x10
>  =======================
> Code: 83 3d 20 61 67 c0 01 75 40 83 3d 84 f4 76 c0 00 75 37 b8 01 00 00
> 00 bf 70
>  62 62 c0 e8 81 29 19 00 b9 0f 00 00 00 89 c6 49 78 08 <ac> ae 75 08 84
> c0 75 f5
>  31 c0 eb 04 19 c0 0c 01 85 c0 75 0a c7
> EIP: [<c04104f5>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:dffe9fa0
>  <0>Kernel panic - not syncing: Fatal exception
>
> #######
>
>
> I know, I lame :), but what should I do run to RHEL5.2 with net5501-70?
>
> 1) Should I recompile kernel? (For me is important to have kernel
> version and patchlevel exactly as RHEL5.2, changing arch is acceptable)
>
> 2) For best performance and stability what should I choose in
> menuconfig. Which arch and CPU?
>
> PS.
> I checked that over PXE I could boot kernel from RHEL4.6 ...
>
> thanks.
>
>
>
>
>
>
>
> _______________________________________________
> Soekris-tech mailing list
> Soekris-tech@...
> http://lists.soekris.com/mailman/listinfo/soekris-tech 

_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech

_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech

[Mike Chirico]

On Fri, Jun 20, 2008 at 05:25:40AM -0700, Sargun Dhillon wrote:
> You can't run RHEL 5.X on the Soekris. The 5501 has a '586' CPU, whereas RHEL5 expects a 686 CPU. You're simply going to come across a lot of crashing, bugs, and other issues. This is a bad idea.
>

I'm running Fedora (8,9 and rawhide) on three separate Soekris
5501-70's. I'm not sure the above statement is true, since I'm
inclined to believe that if Fedora works, RHEL 5.x will work on can
work with a correctly configured kernel.

> Go back to Debian, and debootstrap your machine together.
>

I'd recommend taking a closer look at your kernel config. The
following is my config, which include support for SELINUX,
the NATSEMI (additional 4 port ethernet card), and temperature
sensors.

 http://chirico.googlecode.com/svn/trunk/soekris/configs/2.6.25.4/config

I've got a lot of the Fedora development tools, and I haven't had
any problems. I'm using 2G, 4G, and 8G SandDisk.  

Regards,

Mike Chirico
_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech

[Chris Babcock-3]

Przemyslaw Maszenda wrote:
> Hi
>
> I try to boot and install RedHat Enterprise Linux 5.2 on net5501-70
> (with sandisk CF but this doesn't matter, I think), but without success.
>
<snip>
>
>
> I know, I lame :), but what should I do run to RHEL5.2 with net5501-70?
>
> 1) Should I recompile kernel? (For me is important to have kernel
> version and patchlevel exactly as RHEL5.2, changing arch is acceptable)
>
RedHat does not distribute a RHEL5.2 binary kernel that supports
anything other than a i686 or greater CPU, which means that out of the
box RHEL5.2 WILL NOT RUN on a Soekris.  It is completely possible build
your own kernel, and use a different CPU type.  Unfortunately, in all of
my past experiences with vendors and RHEL, you will be expected to run
the BINARY kernel as distributed by RedHat.  I have NEVER run into a
situation where a vendor would support custom built kernels on RHEL,
even built from RedHat's source packages.  In addition, RedHat HEAVILY
modifies their kernels, and if they are building their kernels _only_
for i686 you have no assurance whatsoever that there aren't RedHat
specific patches that have been added that depend on using i686 only
constructs, regardless of how you build the kernel.

If this is being done as a personal or academic exercise this can be
"made to work".  The easiest way is to use a kernel downloaded from
kernel.org, and tweaked to have similar settings to a RHEL kernel, but
set for a i586 CPU (or lower).  Yes, some things will not work, but most
userspace programs will work.

If this is for a *production* system, it's critically important that you
either choose a different distro (that supports Soekris class hardware)
or use hardware certified for RHEL.  In all honesty, I have had
HORRIBLE, CATASTROPHIC DATA LOSS on multiple occasions using RHEL on
non-RHEL certified hardware.  The patches that RedHat applies to their
kernels are NOT community tested, they are ONLY TESTED ON HARDWARE THEY
HAVE CERTIFIED.  As a result, catastrophic bugs and regressions can and
sometimes *DO* occur on hardware they haven't tested.

The only reliable, safe way to run RHEL, is to run it on certified
hardware.  In fact, if you are running an app that is certified on RHEL,
the certification only applies if all of the hardware is also RHEL
certified.

> 2) For best performance and stability what should I choose in
> menuconfig. Which arch and CPU?
>
> PS.
> I checked that over PXE I could boot kernel from RHEL4.6 ...
>
> thanks.
_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech

[Przemyslaw Maszenda]

It's not true.

RHEL5.2 with 'i686' CPU kernel 2.6.18-92.1.1 work very well.
uname -a just shows:
Linux my.host.domain.net 2.6.18-92.1.1.el5.SOEKRIS #1 SMP Sat Jun 21
23:04:59 CEST 2008 i586 i586 i386 GNU/Linux

I recompiled kernel from .src.rpm. The problem was powernow-k8.
Now Net5501 work with all 8 ETH, CF and IDE HDD.

RHEL5.2 has powernow-k8 compiled into kernel (not module) and OOPS on
it. RHEL4.6 has powernow-k8 as a module and work well.

I recompiled kernel with powernow-k8 as a module or with disabled
powernow (safer methods because if I load module powernow-k8.ko with
modprobe/insmod I receive OOPS and kernel crash).

Why powernow-k8 crash? Becuase of one of two theoretic reasons (I not sure):
1) bug in linux kernel in powernow-k8 (in original kernel or in RedHat
patch)
2) bug in SOEKRIS bios (in linux kernel in Documentation for powernow:
"amd-powernow.txt" is written that this module uses bios:

"
BIOS supplied data, for powernow-k7 and for powernow-k8, may be
from either the PSB table or from ACPI objects. The ACPI support
is only available if the kernel config sets CONFIG_ACPI_PROCESSOR.
The powernow-k8 driver will attempt to use ACPI if so configured,
and fall back to PST if that fails." *

Below I describe for all RedHat / RHEL maniac steps how to recompile
original/updated RHEL5.2 kernel to work with net5501 SOEKRIS:


1) download rpm with source of last kernel
"kernel-2.6.18-92.1.1.el5.src.rpm" from rhn.redhat.com

2) "install" this package:
     rpm -ivh kernel-2.6.18-92.1.1.el5.src.rpm

We receive source in /usr/src/redhat:
kernel-2.6.spec in SPECS
linux-2.6.18.tar.bzp and a lot of patches and configs in SOURCE

3) go to SOURCE and change kernel config files:
kernel-2.6.18-i686.config (AND OTHER relevant ...).
We search string: "CONFIG-X86-POWERNOW_K8=y" and change to
"CONFIG-X86-POWERNOW_K8=m" or
"CONFIG-X86-POWERNOW_K8=n" (safer)

save file, quite editor

4) go to SPECS

cd /usr/src/redhat/SPECS
edit kernel-2.6.spec
look for commented with # line "define build" and change it in:
"%define build .SOEKRIS".
It's give us special suffix for our kernel and RPMs with it.
save file, quite editor

5) build kernel in rpms:
rpmbuild -ba --target=i686 kernel-2.6.spec

6) go sleep or ride your bike or go walk or something else: it take long
time to compile all this kernels/rpms. It took about 1,5 h on Intel Core
2 Due 1,8 Ghz. Don't try that on SOEKRIS !!! :-).

7) We need a lot of disk space: about 2,5-3 GB in /usr/src/redhat and a
lot of in /var/tmp ...

8) When our CPU finished work we received 13rpm with kernel in
/usr/src/redhat/RPMS/i686. Most important is of course:
kernel-2.6.18-92.1.1.el5.SOEKRIS.i686.rpm


9) We transfer this RPM file to our CF or HDD for net5501 and install
that (from VMWware or from computer other than net5501):
rpm -ivh kernel-2.6.18-92.1.1.el5.SOEKRIS.i686.rpm

10) change /boot/grub/grub.conf:
add all staff for console support:

My grub.conf looks like (You can don't used LVM and have root=LABEL=/ or
root=/dev/hda1 or something similar instant of root=/dev/VolGroup00/LogVol00
default=0
timeout=5
#splashimage=(hd0,0)/grub/splash.xpm.gz
#hiddenmenu
#
serial --unit=0 --speed=57600 --word=8 --parity=no --stop=1
terminal --timeout=5 console serial
#
title Red Hat Enterprise Linux Server (2.6.18-92.1.1.el5.SOEKRIS)
         root (hd0,0)
         kernel /vmlinuz-2.6.18-92.1.1.el5.SOEKRIS ro
root=/dev/VolGroup00/LogVol00 console=ttyS0,57600n8


remember to change /etc/inittab:
we need this line:
cp:2345:respawn:/sbin/agetty ttyS0 57600 vt100-nav

11) shutdown and mount CF/Hdd in your SOEKRIS net5501 and fireup it.
If your SOEKRIS have F@#%@ something with power (my have :() you have
try power on few times. Now your SOEKRIS should boot without powernow-k8
problem.

12) You can compile kernel in faster manner (only one rpm). Please see
kernel-2.6.spec and use rpmbuild with additional parameters/options


And last three advice for RHEL/linux:

1)
For Four Eth onboard RHEL/linux uses via_rhine + mii module as a driver.
For Four additional Eth (PCI card lan1641 sold by soekris) RHEL/linux
uses natsemi + mii module as a driver. If kudzu didn't do it for you
just prepare /etc/modprobe.conf. Add this line in it:

alias eth0 via-rhine
alias eth1 via-rhine
alias eth2 via-rhine
alias eth3 via-rhine
alias eth4 natsemi
alias eth5 natsemi
alias eth6 natsemi
alias eth7 natsemi

2)
My net5501 and RHEL5.2 change order of interfaces: eth0-eth3 on PCI,
eth5-eth7 onboard. To correct this check, write-down and use mac-addresses:

in /etc/sysconfig/network-scripts/ change MAC in line HWADDR in each
ifcfg-eth? file as you want. I prefer names of eth with conformance of
description on net5501 box :).

3) You can compile kernel not as i686 but as MGEODE_LX.
CONFIG_M686 is not set
CONFIG_MGODE_LX=y


P.S.
I asked about RedHat / RHEL and please dont write me about debian. This
is tech list, not advocacy .... Additionally you write not true...


* cat amd-powernow.txt
###################################################################
PowerNow! and Cool'n'Quiet are AMD names for frequency
management capabilities in AMD processors. As the hardware
implementation changes in new generations of the processors,
there is a different cpu-freq driver for each generation.

Note that the driver's will not load on the "wrong" hardware,
so it is safe to try each driver in turn when in doubt as to
which is the correct driver.

Note that the functionality to change frequency (and voltage)
is not available in all processors. The drivers will refuse
to load on processors without this capability. The capability
is detected with the cpuid instruction.

The drivers use BIOS supplied tables to obtain frequency and
voltage information appropriate for a particular platform.
Frequency transitions will be unavailable if the BIOS does
not supply these tables.

6th Generation: powernow-k6

7th Generation: powernow-k7: Athlon, Duron, Geode.

8th Generation: powernow-k8: Athlon, Athlon 64, Opteron, Sempron.
Documentation on this functionality in 8th generation processors
is available in the "BIOS and Kernel Developer's Guide", publication
26094, in chapter 9, available for download from www.amd.com.

BIOS supplied data, for powernow-k7 and for powernow-k8, may be
from either the PSB table or from ACPI objects. The ACPI support
is only available if the kernel config sets CONFIG_ACPI_PROCESSOR.
The powernow-k8 driver will attempt to use ACPI if so configured,
and fall back to PST if that fails.
The powernow-k7 driver will try to use the PSB support first, and
fall back to ACPI if the PSB support fails. A module parameter,
acpi_force, is provided to force ACPI support to be used instead
of PSB support.

###################################################################


Sargun Dhillon wrote:

> You can't run RHEL 5.X on the Soekris. The 5501 has a '586' CPU, whereas RHEL5 expects a 686 CPU. You're simply going to come across a lot of crashing, bugs, and other issues. This is a bad idea.
>
> Go back to Debian, and debootstrap your machine together.
>
>
> Sargun Dhillon
> 925.202.9485
> deCarta
> sdhillon@...
> www.decarta.com
>
>
>
>
>
> -----Original Message-----
> From: soekris-tech-bounces@... on behalf of Otfried Geffert
> Sent: Fri 6/20/2008 1:53 AM
> To: Przemyslaw Maszenda; soekris-tech@...
> Subject: Re: [Soekris] net 5501 and RHEL5.2
>  
>>> EIP: [<c041041c>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:c147cfa0
>
> Looks like there is a kernel module called "powernowk8"
> The crash is while loading this module.
>
> Get rid of the kernel module in the config.
>
> OR: Maybe the module is part of the initrd RAM disk.
> You can make your own with
> # mkdinitrd  your_rd   kernel_version
>
> Otfried
>
> ----- Original Message -----
> From: "Przemyslaw Maszenda" <p.maszenda@...>
> To: <soekris-tech@...>
> Sent: Friday, June 20, 2008 9:32 AM
> Subject: [Soekris] net 5501 and RHEL5.2
>
>
>> Hi
>>
>> I try to boot and install RedHat Enterprise Linux 5.2 on net5501-70
>> (with sandisk CF but this doesn't matter, I think), but without success.
>>
>> With original RHEL5.2 kernel booting from CF (CF was installed on
>> another PC) or booting over PXE I received:
>>
>> ############################################################################
>> Linux version 2.6.18-92.1.1.el5
>> (brewbuilder@...) (gcc v
>> ersion 4.1.2 20071124 (Red Hat 4.1.2-41)) #1 SMP Thu May 22 09:01:29 EDT
>> 2008
>> BIOS-provided physical RAM map:
>>
>> [...]
>>
>> TCP bic registered
>> Initializing IPsec netlink socket
>> NET: Registered protocol family 1
>> NET: Registered protocol family 17
>> powernow-k8: Processor cpuid 5a2 not supported
>> BUG: unable to handle kernel NULL pointer dereference at virtual address
>> 0000000
>> 0
>>  printing eip:
>> c041041c
>> *pde = 00000000
>> Oops: 0000 [#1]
>> SMP
>> last sysfs file:
>> Modules linked in:
>> CPU:    0
>> EIP:    0060:[<c041041c>]    Not tainted VLI
>> EFLAGS: 00010202   (2.6.18-92.1.1.el5 #1)
>> EIP is at powernowk8_init+0x5e/0x1c2
>> eax: 00000000   ebx: 00000000   ecx: 0000000e   edx: 00000020
>> esi: 00000000   edi: c06242c3   ebp: 00000000   esp: c147cfa0
>> ds: 007b   es: 007b   ss: 0068
>> Process swapper (pid: 1, ti=c147c000 task=c147daa0 task.ti=c147c000)
>> Stack: 00000000 c071bbe8 00000000 c06ec5a8 c06e7fd8 c0404dee 00000202
>> c06ec42b
>>        00000000 00000000 00000000 00000000 00000000 00000000 c06ec42b
>> 00000000
>>        00000000 c0405c3b 00000000 00000000 00000000 00000000 00000000
>> 00000000
>> Call Trace:
>>  [<c06ec5a8>] init+0x17d/0x24a
>>  [<c0404dee>] ret_from_fork+0x6/0x1c
>>  [<c06ec42b>] init+0x0/0x24a
>>  [<c06ec42b>] init+0x0/0x24a
>>  [<c0405c3b>] kernel_thread_helper+0x7/0x10
>>  =======================
>> Code: 83 3d 20 41 67 c0 01 75 40 83 3d 84 d4 76 c0 00 75 37 b8 01 00 00
>> 00 bf c3
>>  42 62 c0 e8 0a 15 19 00 b9 0f 00 00 00 89 c6 49 78 08 <ac> ae 75 08 84
>> c0 75 f5
>>  31 c0 eb 04 19 c0 0c 01 85 c0 75 0a c7
>> EIP: [<c041041c>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:c147cfa0
>>  <0>Kernel panic - not syncing: Fatal exception
>>
>> ############################################################################
>>
>>
>> I recompiled kernel from redhat .src.rpm (with rpmbuild, with all
>> patches ...), with Geode GX/LX CPU. I included support for AMD
>> powernow-k8. I tried to boot and I received similar as before:
>>
>> ############################################################################
>> Linux version 2.6.18-SOEKRIS (root@vivaldi) (gcc version 4.1.2 20071124
>> (Red Hat
>>  4.1.2-42)) #1 SMP Fri Jun 20 07:58:23 CEST 2008
>>
>> [...]
>>
>> TCP bic registered
>> Initializing IPsec netlink socket
>> NET: Registered protocol family 1
>> NET: Registered protocol family 17
>> powernow-k8: Processor cpuid 5a2 not supported
>> BUG: unable to handle kernel NULL pointer dereference at virtual address
>> 0000000
>> 0
>>  printing eip:
>> c04104f5
>> *pde = 00000000
>> Oops: 0000 [#1]
>> SMP
>> last sysfs file:
>> Modules linked in:
>> CPU:    0
>> EIP:    0060:[<c04104f5>]    Not tainted VLI
>> EFLAGS: 00010202   (2.6.18-SOEKRIS #1)
>> EIP is at powernowk8_init+0x5e/0x1c2
>> eax: 00000000   ebx: 00000000   ecx: 0000000e   edx: c06e7fe4
>> esi: 00000000   edi: c0626270   ebp: 00000000   esp: dffe9fa0
>> ds: 007b   es: 007b   ss: 0068
>> Process swapper (pid: 1, ti=dffe9000 task=dffe8aa0 task.ti=dffe9000)
>> Stack: 00000000 c071da88 00000000 c06ee5a9 c06e9fd8 c0404dce 00000202
>> c06ee42c
>>        00000000 00000000 00000000 00000000 00000000 00000000 c06ee42c
>> 00000000
>>        00000000 c0405c1b 00000000 00000000 00000000 00000000 00000000
>> 00000000
>> Call Trace:
>>  [<c06ee5a9>] init+0x17d/0x24a
>>  [<c0404dce>] ret_from_fork+0x6/0x1c
>>  [<c06ee42c>] init+0x0/0x24a
>>  [<c06ee42c>] init+0x0/0x24a
>>  [<c0405c1b>] kernel_thread_helper+0x7/0x10
>>  =======================
>> Code: 83 3d 20 61 67 c0 01 75 40 83 3d 84 f4 76 c0 00 75 37 b8 01 00 00
>> 00 bf 70
>>  62 62 c0 e8 81 29 19 00 b9 0f 00 00 00 89 c6 49 78 08 <ac> ae 75 08 84
>> c0 75 f5
>>  31 c0 eb 04 19 c0 0c 01 85 c0 75 0a c7
>> EIP: [<c04104f5>] powernowk8_init+0x5e/0x1c2 SS:ESP 0068:dffe9fa0
>>  <0>Kernel panic - not syncing: Fatal exception
>>
>> #######
>>
>>
>> I know, I lame :), but what should I do run to RHEL5.2 with net5501-70?
>>
>> 1) Should I recompile kernel? (For me is important to have kernel
>> version and patchlevel exactly as RHEL5.2, changing arch is acceptable)
>>
>> 2) For best performance and stability what should I choose in
>> menuconfig. Which arch and CPU?
>>
>> PS.
>> I checked that over PXE I could boot kernel from RHEL4.6 ...
>>
>> thanks.
>>
>>
>>
>>
>>
>>
>>
>> _______________________________________________
>> Soekris-tech mailing list
>> Soekris-tech@...
>> http://lists.soekris.com/mailman/listinfo/soekris-tech 
>
> _______________________________________________
> Soekris-tech mailing list
> Soekris-tech@...
> http://lists.soekris.com/mailman/listinfo/soekris-tech
>

_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech

[Chris Babcock-3]

Przemyslaw Maszenda wrote:

> It's not true.
>
> RHEL5.2 with 'i686' CPU kernel 2.6.18-92.1.1 work very well.
> uname -a just shows:
> Linux my.host.domain.net 2.6.18-92.1.1.el5.SOEKRIS #1 SMP Sat Jun 21
> 23:04:59 CEST 2008 i586 i586 i386 GNU/Linux
>
> I recompiled kernel from .src.rpm. The problem was powernow-k8.
> Now Net5501 work with all 8 ETH, CF and IDE HDD.
>

Regardless of whether it works or not, by recompiling the kernel and
running it on non-certified hardware, you are now running a
"non-supported" configuration.  If you have issues in production with
this system and need support, you are going to get a "We don't support
doing that, too bad." from RedHat support if you have problems.

Why would you want to run a commercial system in a configuration that
the vendor for the system refuses to support?



_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech

[Mike Chirico]

On Sun, Jun 22, 2008 at 11:02:43PM +0200, Przemyslaw Maszenda wrote:
> RHEL5.2 with 'i686' CPU kernel 2.6.18-92.1.1 work very well.
> uname -a just shows:
> Linux my.host.domain.net 2.6.18-92.1.1.el5.SOEKRIS #1 SMP Sat Jun 21

>From the above line, it looks like CONFIG_SMP may be set to y.

You may want to change this to no.

CONFIG_SMP=n

Reference:
http://kernel.xc.net/html/linux-2.6.18/i386/SMP

"On a singleprocessor machine, the kernel
will run faster if you say N here."

Regards,

Mike
_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech

[Mike Chirico]

On Sun, Jun 22, 2008 at 11:02:43PM +0200, Przemyslaw Maszenda wrote:
> It's not true.
>
> RHEL5.2 with 'i686' CPU kernel 2.6.18-92.1.1 work very well.
> uname -a just shows:
> Linux my.host.domain.net 2.6.18-92.1.1.el5.SOEKRIS #1 SMP Sat Jun 21


>From the above line, it looks like CONFIG_SMP may be set to y.

[Correction to my previous post]
You may want to unset this

# CONFIG_SMP is not set


Reference:
http://kernel.xc.net/html/linux-2.6.18/i386/SMP

"On a singleprocessor machine, the kernel
will run faster if you say N here."

Regards,

Mike
_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech