Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel

public inbox for linuxppc-dev@ozlabs.org 
 help / color / mirror / Atom feed

* Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel
@ 2005-02-26 22:28 Niklaus Giger
  2005-02-27 15:10 ` Niklaus Giger
  2005-02-27 15:59 ` Matt Porter
  0 siblings, 2 replies; 6+ messages in thread
From: Niklaus Giger @ 2005-02-26 22:28 UTC (permalink / raw)
  To: linuxppc-embedded

Hi

I would like to port Linux to a custom PPC405GPr board. Its hardware runs=20
vxWorks fine for more than a year.=20

Within a few days I had a working U-boot image, which allows me to download=
=20
via TFTP Linux-kernels.

I started with a vanilla 2.6.10 kernel, added the entries in the=20
arch/ppc/platforms/4xx/Makefile and Kconfig, two file hcu3.c and hcu3.h whi=
ch=20
are very minimal. The board has no PCI and I do not want to add support for=
=20
the CFI-Flashs or the onboard CAN-Chips.=20
Enabling the SERIAL_TEXT_DEBUG I got the following output
## Transferring control to Linux (at address 00000000) ...
id mach(): done
MMU:enter
MMU:hw init
MMU:mapin
MMU:setio
hcu3_map_io
MMU:exit
setup_arch: enter ngng
setup_arch: bootmem
ocp: exit
arch: exit
If I attach a BDI, I get a lot of "- TARGET: target has entered debug mode"=
=20
lines and then
DI>i
    Target state      : debug mode
    Debug entry cause : unconditional debug event
    Current PC        : 0xfffffffc
    Current CR        : 0x00000000
    Current MSR       : 0x00000000
    Current LR        : 0xc019a15c
BDI>               =20
The corresponding line from my System.map are
c0199e20 t free_bootmem_core
c0199ecc t __alloc_bootmem_core
c019a278 t free_all_bootmem_core
c019a4c8 T init_bootmem_node

My board has no RTC. Is this a problem?=20

Should I start with another kernel? Can anybody point me to a known good=20
kernel version for a specific PPC405 board?

Any hint would be greatly appreciated.

Best regards

=2D-=20
Niklaus Giger
Wieshoschet 6
CH-8753 Mollis
Tel. ++41 55 612 20 54 (privat)
Tel. ++41 55 618 64 68 (Gesch=E4ft)

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel
  2005-02-26 22:28 Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel Niklaus Giger
@ 2005-02-27 15:10 ` Niklaus Giger
  2005-02-27 16:08   ` Matt Porter
  2005-02-27 15:59 ` Matt Porter
  1 sibling, 1 reply; 6+ messages in thread
From: Niklaus Giger @ 2005-02-27 15:10 UTC (permalink / raw)
  To: linuxppc-embedded

Am Samstag, 26. Februar 2005 23.28 schrieb Niklaus Giger:
> Hi
>
> I would like to port Linux to a custom PPC405GPr board. Its hardware runs
> vxWorks fine for more than a year.
>
I made some more progress. After adding 3 lines for MMU support with the BDI
I can debug the startup up to kernel_start using BDI.=20

Afterwards my console changes the baudrate for still unknown reasons.

Analysing the __log_buf I get the following output
(gdb) x/s &__log_buf
0xc01accfc <ratelimit_lock.9>:   "<4>Linux version 2.6.10=20
(niklaus@ng•ngiger.dyndns.org) (gcc-Version 3.3.5 (Debian 1:3.3.5-8)) #15 S=
un=20
=46eb 27 14:44:27 CET 2005\n<7>On node 0 totalpages: 8240\n<7>  DMA zone: 8=
240=20
pages, LIFO batch:2\n<7>"...
(gdb) x
0xc01acdc4 <ratelimit_lock.9+200>:       "  Normal zone: 0 pages, LIFO=20
batch:1\n<7>  HighMem zone: 0 pages, LIFO batch:1\n<4>Built 1=20
zonelists\n<4>Kernel command line: ip=3D172.25.1.6\n<4>PID hash table entri=
es:=20
256 (order:8, 4096 bytes)\n"
(gdb) x
0xc01ace84 <ratelimit_lock.9+392>:       ""

>My board has no RTC. Is this a problem?=20
>
>Should I start with another kernel? Can anybody point me to a known good=20
>kernel version for a specific PPC405 board?

Any hint would be greatly appreciated.

Best regards

=2D-=20
Niklaus Giger
Wieshoschet 6
CH-8753 Mollis
Tel. ++41 55 612 20 54 (privat)
Tel. ++41 55 618 64 68 (Gesch=E4ft)

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel
  2005-02-27 15:10 ` Niklaus Giger
@ 2005-02-27 16:08   ` Matt Porter
  2005-02-27 17:20     ` Niklaus Giger
  0 siblings, 1 reply; 6+ messages in thread
From: Matt Porter @ 2005-02-27 16:08 UTC (permalink / raw)
  To: Niklaus Giger; +Cc: linuxppc-embedded

On Sun, Feb 27, 2005 at 04:10:30PM +0100, Niklaus Giger wrote:
> Am Samstag, 26. Februar 2005 23.28 schrieb Niklaus Giger:
> > Hi
> >
> > I would like to port Linux to a custom PPC405GPr board. Its hardware runs
> > vxWorks fine for more than a year.
> >
> I made some more progress. After adding 3 lines for MMU support with the BDI
> I can debug the startup up to kernel_start using BDI. 
> 
> Afterwards my console changes the baudrate for still unknown reasons.

How do you know this? Console isn't initialized until after the kernel 
command line printk and a few other facilities are initialized.

> Analysing the __log_buf I get the following output
> (gdb) x/s &__log_buf
> 0xc01accfc <ratelimit_lock.9>:   "<4>Linux version 2.6.10 
> (niklaus@ng•ngiger.dyndns.org) (gcc-Version 3.3.5 (Debian 1:3.3.5-8)) #15 Sun 
> Feb 27 14:44:27 CET 2005\n<7>On node 0 totalpages: 8240\n<7>  DMA zone: 8240 
> pages, LIFO batch:2\n<7>"...
> (gdb) x
> 0xc01acdc4 <ratelimit_lock.9+200>:       "  Normal zone: 0 pages, LIFO 
> batch:1\n<7>  HighMem zone: 0 pages, LIFO batch:1\n<4>Built 1 
> zonelists\n<4>Kernel command line: ip=172.25.1.6\n<4>PID hash table entries: 
> 256 (order:8, 4096 bytes)\n"
> (gdb) x
> 0xc01ace84 <ratelimit_lock.9+392>:       ""

Is this all of the log_buf output? Try setting your kernel cmdline
with "console=ttyS0,115200" where 115200 is you console baudrate
you are using in U-Boot. Otherwise, the kernel 8250 driver has no
idea which baudrate to set for the 8250/console.

-Matt 

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel
  2005-02-27 16:08   ` Matt Porter
@ 2005-02-27 17:20     ` Niklaus Giger
  2005-04-17 20:13       ` Niklaus Giger
  0 siblings, 1 reply; 6+ messages in thread
From: Niklaus Giger @ 2005-02-27 17:20 UTC (permalink / raw)
  To: Matt Porter; +Cc: linuxppc-embedded

Am Sonntag, 27. Februar 2005 17.08 schrieb Matt Porter:
> On Sun, Feb 27, 2005 at 04:10:30PM +0100, Niklaus Giger wrote:
> > Am Samstag, 26. Februar 2005 23.28 schrieb Niklaus Giger:
> > > Hi
> > >
> > > I would like to port Linux to a custom PPC405GPr board. Its hardware
> > > runs vxWorks fine for more than a year.
> >
> > I made some more progress. After adding 3 lines for MMU support with the
> > BDI I can debug the startup up to kernel_start using BDI.
> >
> > Afterwards my console changes the baudrate for still unknown reasons.
> How do you know this? Console isn't initialized until after the kernel
> command line printk and a few other facilities are initialized.
Because I see the following output
arch: exit
*=CB=AB=E3=BE=B5=BA=B3=B9=BF=BE=B6=B9=B3=BC=B1=B5=B3=A0=B6=B7=BE=B7=B9=B7=
=B5=B2=B4=B9=B6=B4=BE=B3=BF=BA=B7=B7=B3=B3=A6=B5=B2=BB=B9=BF=BE=A4=B5=B2=B9=
=B1=BE=A3=B5=BE=A6=B5=B2=A3=A5=A4=A2=BD=B9=BC
=B4=BA=B7=BE=B5=BC=B9=B3=BC=B3=A3=B5=B2=BE=B5=BC=B3=B7=BD=BD=B1=BE=B4=BC=B1=
=BE=B5=B3=B7=BE=BB=BF=BC=B5=B4=BC=B9=AB=B1=B0=A0=A1=A4=B8=B1=B3=B8=BC=B1=B2=
=BC=B5=B5=BE=BC=B2=B1=B5=BB=B7=B2=B4=B5=B2=B2=B9=B4=B5=B3  =20
which for me is typical for a wrong baudrate.=20
Okay, I do not know this. Specially it may not be the "console" device but=
=20
whatever is going out at this moment via the PPC405GPr internal UART0 devic=
e.
> > Analysing the __log_buf I get the following output
> > (gdb) x/s &__log_buf
> > 0xc01accfc <ratelimit_lock.9>:   "<4>Linux version 2.6.10
> > (niklaus@ng•ngiger.dyndns.org) (gcc-Version 3.3.5 (Debian 1:3.3.5-8)) #=
15
> > Sun Feb 27 14:44:27 CET 2005\n<7>On node 0 totalpages: 8240\n<7>  DMA
> > zone: 8240 pages, LIFO batch:2\n<7>"...
> > (gdb) x
> > 0xc01acdc4 <ratelimit_lock.9+200>:       "  Normal zone: 0 pages, LIFO
> > batch:1\n<7>  HighMem zone: 0 pages, LIFO batch:1\n<4>Built 1
> > zonelists\n<4>Kernel command line: ip=3D172.25.1.6\n<4>PID hash table
> > entries: 256 (order:8, 4096 bytes)\n"
> > (gdb) x
> > 0xc01ace84 <ratelimit_lock.9+392>:       ""
>
> Is this all of the log_buf output? Try setting your kernel cmdline
> with "console=3DttyS0,115200" where 115200 is you console baudrate
> you are using in U-Boot. Otherwise, the kernel 8250 driver has no
> idea which baudrate to set for the 8250/console.
Yes, this is all. I tried you suggestion, but I got no other result. The=20
paramter "console=3DttyS0,9600" now is found in the log buffer. Or is ttyS0=
 the=20
wrong device name for the internal PPC405Gpr UART0?
I am using 9600 8 bit no parity which should be anyway the default baudrate=
=20
for the console.

Thanks for your help
=2D-=20
Niklaus Giger
Wieshoschet 6
CH-8753 Mollis
Tel. ++41 55 612 20 54 (privat)
Tel. ++41 55 618 64 68 (Gesch=E4ft)

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel
  2005-02-27 17:20     ` Niklaus Giger
@ 2005-04-17 20:13       ` Niklaus Giger
  0 siblings, 0 replies; 6+ messages in thread
From: Niklaus Giger @ 2005-04-17 20:13 UTC (permalink / raw)
  To: Matt Porter; +Cc: linuxppc-embedded

Hi

Just to document my error I respond to my previous posting.

After having dropped my work on this problem for a few weeks
I figured my problem. I was merging code from the walnut board which was ba=
sed=20
on the IBM boot eprom which has a different layout to pass the information =
to
the Linux kernel.

After switching to the U-Boot memory the kernel booted okay.=20

After disabling the following two lines in drivers/serial/8250.c
	serial_outp(up, UART_DLL, quot & 0xff);		/* LS of divisor */
	serial_outp(up, UART_DLM, quot >> 8);		/* MS of divisor */
my console stays at 9600 baud.=20

Best regards

Am Sonntag, 27. Februar 2005 18.20 schrieb Niklaus Giger:
> Am Sonntag, 27. Februar 2005 17.08 schrieb Matt Porter:
> > On Sun, Feb 27, 2005 at 04:10:30PM +0100, Niklaus Giger wrote:
> > > Am Samstag, 26. Februar 2005 23.28 schrieb Niklaus Giger:
> > > > Hi
> > > >
> > > > I would like to port Linux to a custom PPC405GPr board. Its hardware
> > > > runs vxWorks fine for more than a year.
> > >
> > > I made some more progress. After adding 3 lines for MMU support with
> > > the BDI I can debug the startup up to kernel_start using BDI.
> > >
> > > Afterwards my console changes the baudrate for still unknown reasons.
> >
> > How do you know this? Console isn't initialized until after the kernel
> > command line printk and a few other facilities are initialized.
>
> Because I see the following output
> arch: exit
> *=CB=AB=E3=BE=B5=BA=B3=B9=BF=BE=B6=B9=B3=BC=B1=B5=B3=A0=B6=B7=BE=B7=B9=B7=
=B5=B2=B4=B9=B6=B4=BE=B3=BF=BA=B7=B7=B3=B3=A6=B5=B2=BB=B9=BF=BE=A4=B5=B2=B9=
=B1=BE=A3=B5=BE=A6=B5=B2=A3=A5=A4=A2=BD=B9=BC
> =B4=BA=B7=BE=B5=BC=B9=B3=BC=B3=A3=B5=B2=BE=B5=BC=B3=B7=BD=BD=B1=BE=B4=BC=
=B1=BE=B5=B3=B7=BE=BB=BF=BC=B5=B4=BC=B9=AB=B1=B0=A0=A1=A4=B8=B1=B3=B8=BC=B1=
=B2=BC=B5=B5=BE=BC=B2=B1=B5=BB=B7=B2=B4=B5=B2=B2=B9=B4=B5=B3
> which for me is typical for a wrong baudrate.
> Okay, I do not know this. Specially it may not be the "console" device but
> whatever is going out at this moment via the PPC405GPr internal UART0
> device.
>
> > > Analysing the __log_buf I get the following output
> > > (gdb) x/s &__log_buf
> > > 0xc01accfc <ratelimit_lock.9>:   "<4>Linux version 2.6.10
> > > (niklaus@ng•ngiger.dyndns.org) (gcc-Version 3.3.5 (Debian 1:3.3.5-8))
> > > #15 Sun Feb 27 14:44:27 CET 2005\n<7>On node 0 totalpages: 8240\n<7>=
=20
> > > DMA zone: 8240 pages, LIFO batch:2\n<7>"...
> > > (gdb) x
> > > 0xc01acdc4 <ratelimit_lock.9+200>:       "  Normal zone: 0 pages, LIFO
> > > batch:1\n<7>  HighMem zone: 0 pages, LIFO batch:1\n<4>Built 1
> > > zonelists\n<4>Kernel command line: ip=3D172.25.1.6\n<4>PID hash table
> > > entries: 256 (order:8, 4096 bytes)\n"
> > > (gdb) x
> > > 0xc01ace84 <ratelimit_lock.9+392>:       ""
> >
> > Is this all of the log_buf output? Try setting your kernel cmdline
> > with "console=3DttyS0,115200" where 115200 is you console baudrate
> > you are using in U-Boot. Otherwise, the kernel 8250 driver has no
> > idea which baudrate to set for the 8250/console.
>
> Yes, this is all. I tried you suggestion, but I got no other result. The
> paramter "console=3DttyS0,9600" now is found in the log buffer. Or is tty=
S0
> the wrong device name for the internal PPC405Gpr UART0?
> I am using 9600 8 bit no parity which should be anyway the default baudra=
te
> for the console.
>
> Thanks for your help

=2D-=20
Niklaus Giger
Wieshoschet 6
CH-8753 Mollis
Tel. ++41 55 612 20 54 (privat)
Tel. ++41 55 618 64 68 (Gesch=E4ft)

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel
  2005-02-26 22:28 Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel Niklaus Giger
  2005-02-27 15:10 ` Niklaus Giger
@ 2005-02-27 15:59 ` Matt Porter
  1 sibling, 0 replies; 6+ messages in thread
From: Matt Porter @ 2005-02-27 15:59 UTC (permalink / raw)
  To: Niklaus Giger; +Cc: linuxppc-embedded

On Sat, Feb 26, 2005 at 11:28:23PM +0100, Niklaus Giger wrote:
> Hi
> 
> I would like to port Linux to a custom PPC405GPr board. Its hardware runs 
> vxWorks fine for more than a year. 

Good, gives a good feeling about the hardware.
 
> My board has no RTC. Is this a problem? 

No, not at all.

> Should I start with another kernel? Can anybody point me to a known good 
> kernel version for a specific PPC405 board?

The stock kernel.org 2.6.10 that you seem to be using is fine.
 
> Any hint would be greatly appreciated.

Working out your BDI issues will be very helpful to debugging. It's
only clear that it passes beyond setup_arch() at this point. That's
good, though.

-Matt

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2005-04-17 20:54 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2005-02-26 22:28 Problems porting to a custom PPC405GPr board using a vanilla 2.6.10 kernel Niklaus Giger
2005-02-27 15:10 ` Niklaus Giger
2005-02-27 16:08   ` Matt Porter
2005-02-27 17:20     ` Niklaus Giger
2005-04-17 20:13       ` Niklaus Giger
2005-02-27 15:59 ` Matt Porter

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox