LinuxPostMortemAnalysis only smells like DEADBEEF

U-Boot-Users Clemens Koller 2 participants 4 messages Nov 13, 2007

Share & tools

Clemens Koller

author

2007-11-13 19:09 UTC #0 lore raw

Hi!

I'm using an kernel 2.6.23.1 cuImage.85xx target and an u-boot from last week's git (which seems to work fine) on an mpc8540ads compatible board.

I just tried to follow:
http://www.denx.de/wiki/view/DULG/LinuxPostMortemAnalysis
because of no console output.

But no matter what address I try to figure out to dump __log_buf, it seems that u-boot wipes my memory during POST, so I cannot get anything @ __log_buf.

md 003b2ea8 (where my __log_buf points) 003b2ea8: deadbeef deadbeef deadbeef ........

I am pretty sure that only my console got stuck since my harddisks (on PCI) seems to get initialized (some noise and led activity from the HDDs) and the system reboots automatically after approx. 180 seconds due to the usual minor problems.

So, there are some questions:

Is u-boot wiping my memory during POST?

How can I disable that?

What are the default console settings to get a console on ttyS0?

Regards,

Signature · 10 lines

Clemens Koller
__________________________________
R&D Imaging Devices
Anagramm GmbH
Rupert-Mayer-Stra?e 45/1
Linhof Werksgel?nde
D-81379 M?nchen
Tel.089-741518-50
Fax 089-741518-19
http://www.anagramm-technology.com

robert lazarski

2007-11-13 19:36 UTC #1 lore raw

Quoted 29 lines · Clemens Koller

Hi!
I'm using an kernel 2.6.23.1 cuImage.85xx target and an u-boot from last week's git (which seems to work fine) on an mpc8540ads compatible board.
I just tried to follow:
http://www.denx.de/wiki/view/DULG/LinuxPostMortemAnalysis
because of no console output.
But no matter what address I try to figure out to dump __log_buf, it seems that u-boot wipes my memory during POST, so I cannot get anything @ __log_buf.
md 003b2ea8 (where my __log_buf points) 003b2ea8: deadbeef deadbeef deadbeef ........
I am pretty sure that only my console got stuck since my harddisks (on PCI) seems to get initialized (some noise and led activity from the HDDs) and the system reboots automatically after approx. 180 seconds due to the usual minor problems.
So, there are some questions:
Is u-boot wiping my memory during POST?
How can I disable that?

Quoted 7 lines

What are the default console settings to get a console on ttyS0?
Regards,
--
Clemens Koller
__________________________________

Hi Clemens, maybe I can help you this time ;-) . I just so happen to be running 85xx / 2.6.23 / latest u-boot also have serial port problems - though in my case its intermittent. I've been able to get output once every 20 times or so. However, I can always get the messages from "__log_buf" . In my case, my log buf is 2a0f24 . So I execute:

=> md 2a0f24 700

deadbeef is the magic value from u-boot that is used to init memory, and 003b2ea8 doesn't seem right considering I'm using the same kernel.

As for my problems, I'm waiting for my hardware guys to give me a second serial port to try out kgdb. I've been using just printk's for now, but I just got my bdi back so I might know more soon. Maybe we could help each other.

HTH,
Robert

Clemens Koller

author

2007-11-13 20:08 UTC #2 lore raw

Quoted 10 lines · robert lazarski

Hi Clemens, maybe I can help you this time ;-) . I just so happen to be running 85xx / 2.6.23 / latest u-boot also have serial port problems - though in my case its intermittent. I've been able to get output once every 20 times or so. However, I can always get the messages from "__log_buf" . In my case, my log buf is 2a0f24 . So I execute:
=> md 2a0f24 700
deadbeef is the magic value from u-boot that is used to init memory, and 003b2ea8 doesn't seem right considering I'm using the same kernel.

I'm using 2.6.23.1 and would expect the __log_buf address to change with varying kernel configuration.

Can you please send me your kernel .config to verify some stuff. (off-list, I think) I am currently building the latest paulus.git to check my mileage with that tree...

The next step is to kick u-boot's memory wipe out of the way. (AFAIK it sets up a DMA channel to wipe the available RAM...) ...and to setup the PRAM command (Protect RAM) which might be useful here.

Quoted 4 lines

As for my problems, I'm waiting for my hardware guys to give me a second serial port to try out kgdb. I've been using just printk's for now, but I just got my bdi back so I might know more soon. Maybe we could help each other.

I would be interested in the kgdb stuff as well... once it's working 8-)

Well, talk to you tomorrow.
Regards,

Clemens Koller
__________________________________
R&D Imaging Devices
Anagramm GmbH
Rupert-Mayer-Stra?e 45/1
Linhof Werksgel?nde
D-81379 M?nchen
Tel.089-741518-50
Fax 089-741518-19
http://www.anagramm-technology.com

Clemens Koller

author

2007-11-13 20:26 UTC #3 lore raw

[U-Boot-Users] OT: solved: Re: LinuxPostMortemAnalysis only smells like DEADBEEF

Pah!

Quoted 3 lines · Clemens Koller

Can you please send me your kernel .config to verify some stuff. (off-list, I think) I am currently building the latest paulus.git to check my mileage with that tree...

paulus.git's mpc8540_ads_defconfig seems to work. You propably want to give it a try:
git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc.git

I just added:
CONFIG_WANT_DEVICE_TREE=y
CONFIG_DEVICE_TREE="mpc8540ads.dts"
and did a

$ make zImage

So, mainline or at least it's 'defconfig is broken again... :-( If I'm in the right mood I'll do a git diff to linus' master tomorrow.

Regards,