Help with PPC440GX + MontaVista + uBoot

Help with PPC440GX + MontaVista + uBoot

[Howard, Marc]

Hi,

I'm trying to boot Linux on the IBM (now AMCC) 440GX eval board.  Using
the latest MontaVista 3.1 toolset + 440GX eval board package (Linux
kernel =3D 2.4.20) I can create a ramdisk-only Linux glob and boot it
successfully using the IBM supplied PIBS bios.

PIBS isn't what we want to use long term, we'd rather use uBoot.  I've
downloaded and built the latest version of uBoot (1.1.2).  I then
created a multi-file image with mkimage.  I can download it via TFTP but
when I start to run it strange/bad things happen:

	Kernel command line: ip=3Don console=3DttyS0,115200 root=3D/dev/ram rw
	Warning: real time clock seems stuck!
	Calibrating delay loop... 37.27 BogoMIPS

...now the RTC is know to work and the mips calculation is seriously
off.  Then...

	IBM ZMII: EMAC 0: SMII mode
	OCP Reset timeout #2!
	eth0: Got bad Phy Read (repeats)
	.....
	eth0: No PHY device found.
	removing net dev

	.....

	cramfs: wrong magic
	Kernel panic: VFS: Unable to mount root fs on 01:00
 	<0>Rebooting in 180 seconds..

It actually reboots in about 10 seconds.

I realize that ocotea.c will have to be modified to grab command line
options from NVRAM (uBoot) instead of FLASH (PIBS) but that doesn't seem
to explain why the internal timer is way off and the ethernet PHYs can't
be found.

If anyone on the list has done this I'd appreciate hearing what else
needs to be modified to make this configuration work.

Thanks,


Marc W. Howard

Re: Help with PPC440GX + MontaVista + uBoot

[Wolfgang Denk]

In message <91B22F93A880FA48879475E134D6F0BE316B8C@CA1EXCLV02.adcorp.kla-tencor.com> you wrote:
> 
> PIBS isn't what we want to use long term, we'd rather use uBoot.  I've
> downloaded and built the latest version of uBoot (1.1.2).  I then
> created a multi-file image with mkimage.  I can download it via TFTP but
> when I start to run it strange/bad things happen:
> 
> 	Kernel command line: ip=3Don console=3DttyS0,115200 root=3D/dev/ram rw
> 	Warning: real time clock seems stuck!
> 	Calibrating delay loop... 37.27 BogoMIPS
> 
> ...now the RTC is know to work and the mips calculation is seriously
> off.  Then...

That's a FAQ. Please see the part about "Bad definition of the
bd_info structure" in
http://www.denx.de/twiki/bin/view/DULG/LinuxHangsAfterUncompressingKernel

> I realize that ocotea.c will have to be modified to grab command line
> options from NVRAM (uBoot) instead of FLASH (PIBS) but that doesn't seem

Wrong. U-Boot (please mind the spelling) does not pass  command  line
options in NVRAM. They get passed in RAM.

> to explain why the internal timer is way off and the ethernet PHYs can't
> be found.

Because you use a wrong definition of bd_info.  Delete  the  existing
definition and include asm/ppcboot.h in your board config file.


Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de
Bradley's Bromide: If computers get too  powerful,  we  can  organize
them into a committee - that will do them in.

RE: Help with PPC440GX + MontaVista + uBoot

[Matt Sheahan]

Check ocotea_calibrate_decr.  It is using a hard coded clock value of 25
MHz.  When I ported the Ocotea code to my custom board, used the value
passed from U-Boot (bi_intfreq).

I think the Ocotea code in 2.4.x might have been developed using a rev. B
440GX and there are some differences in how the timers and decrementers
derive their clock on the rev. C parts.

Also, using the bd_info definition in asm/ppcboot.h did not work perfectly
for me.  Be sure to check the conditional compiles that are being set in
U-Boot against Linux for the bd_info structure.  The 440GX structures did
not align properly and I had to tweak them and make sure U-Boot was filing
them in correctly (lib_ppc/board.c).


-----Original Message-----
From: linuxppc-embedded-bounces@ozlabs.org
[mailto:linuxppc-embedded-bounces@ozlabs.org] On Behalf Of Wolfgang Denk
Sent: Wednesday, January 12, 2005 6:37 AM
To: Howard, Marc
Cc: linuxppc-embedded@ozlabs.org
Subject: Re: Help with PPC440GX + MontaVista + uBoot 

In message
<91B22F93A880FA48879475E134D6F0BE316B8C@CA1EXCLV02.adcorp.kla-tencor.com>
you wrote:
> 
> PIBS isn't what we want to use long term, we'd rather use uBoot.  I've
> downloaded and built the latest version of uBoot (1.1.2).  I then
> created a multi-file image with mkimage.  I can download it via TFTP but
> when I start to run it strange/bad things happen:
> 
> 	Kernel command line: ip=3Don console=3DttyS0,115200 root=3D/dev/ram
rw
> 	Warning: real time clock seems stuck!
> 	Calibrating delay loop... 37.27 BogoMIPS
> 
> ...now the RTC is know to work and the mips calculation is seriously
> off.  Then...

That's a FAQ. Please see the part about "Bad definition of the
bd_info structure" in
http://www.denx.de/twiki/bin/view/DULG/LinuxHangsAfterUncompressingKernel

> I realize that ocotea.c will have to be modified to grab command line
> options from NVRAM (uBoot) instead of FLASH (PIBS) but that doesn't seem

Wrong. U-Boot (please mind the spelling) does not pass  command  line
options in NVRAM. They get passed in RAM.

> to explain why the internal timer is way off and the ethernet PHYs can't
> be found.

Because you use a wrong definition of bd_info.  Delete  the  existing
definition and include asm/ppcboot.h in your board config file.


Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de
Bradley's Bromide: If computers get too  powerful,  we  can  organize
them into a committee - that will do them in.
_______________________________________________
Linuxppc-embedded mailing list
Linuxppc-embedded@ozlabs.org
https://ozlabs.org/mailman/listinfo/linuxppc-embedded