[U-Boot] Linux hangs when uncompressing kernel image: linux-2.6.26

[U-Boot] Linux hangs when uncompressing kernel image: linux-2.6.26

["Marco Antônio Possamai"]

Hello

I've been working on building a 2.6.26 linux kernel to the mpc5200Lite
PowerPc board. However, I am unable to boot the system using the
generated image.

I am using what seems to be the latest U-Boot for the job. Got it from
www.denx.de/git/u-boot.git.

I'm using the device tree compiler dtc-v1.2.0 to compile the device tree
source into a device tree blob. Got it at:
http://jdl.com/software/dtc-latest.tgz

For the device, my design procedure is the following:

1. In the first step I generate the uImage file ('make ARCH=powerpc
zImage'), the corresponding lite5200b.dts and compile it to
lite5200b.dtb.
2. Afterwards, I do the U-Boot configs. I define 0x00600000 as the device
tree address and 0x00200000 as the uImage's.
3. Finally, I transfer them all (using tftp for each, of course) and try
to boot: 'bootm 0x00200000 - 0x00600000'

The result is that the console hangs...

Does anyone have any idea of why it happens? I checked a similar problem
at the
http://www.denx.de/wiki/view/DULG/LinuxHangsAfterUncompressingKernel, and
I'm almost convinced that it is probably a bd_info problem, concerning
the definitions of the kernel's IMAP_ADDR and the CFG_IMMR of U-Boot to
be unmatching.

If that is the case, how/where can I correct it? Which files should I
look for?

Thanks for the attention.



Marco Ant?nio Possamai

Universidade Federal de Santa Catarina

-- 
It's News. It's Reviews. It's Interviews. It's Free. What Are You Waiting For?
www.movieline.com

[U-Boot] Linux hangs when uncompressing kernel image: linux-2.6.26

[Jerry Van Baren]

Marco Ant?nio Possamai wrote:
> Hello
> 
> I've been working on building a 2.6.26 linux kernel to the mpc5200Lite
> PowerPc board. However, I am unable to boot the system using the
> generated image.
> 
> I am using what seems to be the latest U-Boot for the job. Got it from
> www.denx.de/git/u-boot.git.

Good.  :-)

> I'm using the device tree compiler dtc-v1.2.0 to compile the device tree
> source into a device tree blob. Got it at:
> http://jdl.com/software/dtc-latest.tgz

Good, but *what* did you compile?

If this were /., I would insert an automobile analogy about you telling 
us all about the wrenches you used, how they were the best Binford-2000 
lifetime guaranteed wrenches with more power, but never told us what 
part you were adjusting with the wrenches.
   <http://en.wikipedia.org/wiki/Home_Improvement>

> For the device, my design procedure is the following:
> 
> 1. In the first step I generate the uImage file ('make ARCH=powerpc
> zImage'), the corresponding lite5200b.dts and compile it to
> lite5200b.dtb.
> 2. Afterwards, I do the U-Boot configs. I define 0x00600000 as the device
> tree address and 0x00200000 as the uImage's.
> 3. Finally, I transfer them all (using tftp for each, of course) and try
> to boot: 'bootm 0x00200000 - 0x00600000'
> 
> The result is that the console hangs...

Is linux hung or is there no serial output?  From the outside, the two 
symptoms are indistinguishable.  If linux is actually running, it means 
your console/serial configuration in your dtb is not right.  Use your 
debugger to halt the processor and see what/where it is executing.  Dump 
the linux log buffer and see if there is anything interesting in there 
that wanted to go out the serial channel but couldn't.

> Does anyone have any idea of why it happens? I checked a similar problem
> at the
> http://www.denx.de/wiki/view/DULG/LinuxHangsAfterUncompressingKernel, and
> I'm almost convinced that it is probably a bd_info problem, concerning
> the definitions of the kernel's IMAP_ADDR and the CFG_IMMR of U-Boot to
> be unmatching.

The flattened device tree replaces the bd_info structure as the way to 
communicate configuration info from u-boot to linux.

> If that is the case, how/where can I correct it? Which files should I
> look for?

What are you using for your device tree source (.dts)?  Odds are pretty 
good that is the problem.  If that is OK, second best is if your console 
simply isn't configured properly.  If neither of those are the case, you 
have some serious debugging ahead.

> Thanks for the attention.
> 
> Marco Ant?nio Possamai
> 
> Universidade Federal de Santa Catarina

Good luck,
gvb

[U-Boot] Linux hangs when uncompressing kernel image: linux-2.6.26

[Jerry Van Baren]

Hi Marco,

Marco Ant?nio Possamai wrote:
> Thank you for your quick reply!
> 
> Sorry for the missing information. The .dts I'm using is the 
> lite5200b.dts that comes along with the linux-2.6-denx kernel (got it at 
> git://git.denx.de/linux-2.6-denx.git). What other .dts should I try? I'm 
> using the mpc5200b board so it seems fitting that I'm working with 
> lite5200b.dts.
> 
> There's a lite5200.dts source tree that also comes along with the 
> kernel. I've tried with it and got the same results.

1) Please don't top post.

2) Please CC: the u-boot list.  By replying directly to me, you excluded 
all the people with real knowledge.

The .dts has to match to board: it is the description of the hardware. 
If I understand your message, your board *is* the standard mpc5200b 
board (note that I have no idea what a mpc5200b board is, see point #2), 
in which case the standard .dts should work.

What debug capability do you have?  Can you dump the linux log buffer 
(from memory) after a failed boot?
   <http://www.denx.de/wiki/view/DULG/LinuxPostMortemAnalysis>

I'm out of easy answers.

Good luck,
gvb


>     ----- Original Message -----
>     From: "Jerry Van Baren"
>     To: "Marco Ant?nio Possamai"
>     Cc: "U-Boot"
>     Subject: Re: [U-Boot] Linux hangs when uncompressing kernel image:
>     linux-2.6.26
>     Date: Thu, 16 Apr 2009 23:37:45 -0400
> 
> 
>     Marco Ant?nio Possamai wrote:
>      > Hello
>      >
>      > I've been working on building a 2.6.26 linux kernel to the
>     mpc5200Lite
>      > PowerPc board. However, I am unable to boot the system using the
>      > generated image.
>      >
>      > I am using what seems to be the latest U-Boot for the job. Got it
>     from
>      > www.denx.de/git/u-boot.git.
> 
>     Good. :-)
> 
>      > I'm using the device tree compiler dtc-v1.2.0 to compile the
>     device tree
>      > source into a device tree blob. Got it at:
>      > http://jdl.com/software/dtc-latest.tgz
> 
>     Good, but *what* did you compile?
> 
>     If this were /., I would insert an automobile analogy about you
>     telling us all about the wrenches you used, how they were the best
>     Binford-2000 lifetime guaranteed wrenches with more power, but
>     never told us what part you were adjusting with the wrenches.
> 
> 
>      > For the device, my design procedure is the following:
>      >
>      > 1. In the first step I generate the uImage file ('make ARCH=powerpc
>      > zImage'), the corresponding lite5200b.dts and compile it to
>      > lite5200b.dtb.
>      > 2. Afterwards, I do the U-Boot configs. I define 0x00600000 as
>     the device
>      > tree address and 0x00200000 as the uImage's.
>      > 3. Finally, I transfer them all (using tftp for each, of course)
>     and try
>      > to boot: 'bootm 0x00200000 - 0x00600000'
>      >
>      > The result is that the console hangs...
> 
>     Is linux hung or is there no serial output? From the outside, the
>     two symptoms are indistinguishable. If linux is actually running,
>     it means your console/serial configuration in your dtb is not
>     right. Use your debugger to halt the processor and see what/where
>     it is executing. Dump the linux log buffer and see if there is
>     anything interesting in there that wanted to go out the serial
>     channel but couldn't.
> 
>      > Does anyone have any idea of why it happens? I checked a similar
>     problem
>      > at the
>      >
>     http://www.denx.de/wiki/view/DULG/LinuxHangsAfterUncompressingKernel,
>     and
>      > I'm almost convinced that it is probably a bd_info problem,
>     concerning
>      > the definitions of the kernel's IMAP_ADDR and the CFG_IMMR of
>     U-Boot to
>      > be unmatching.
> 
>     The flattened device tree replaces the bd_info structure as the way
>     to communicate configuration info from u-boot to linux.
> 
>      > If that is the case, how/where can I correct it? Which files should I
>      > look for?
> 
>     What are you using for your device tree source (.dts)? Odds are
>     pretty good that is the problem. If that is OK, second best is if
>     your console simply isn't configured properly. If neither of those
>     are the case, you have some serious debugging ahead.
> 
>      > Thanks for the attention.
>      >
>      > Marco Ant?nio Possamai
>      >
>      > Universidade Federal de Santa Catarina
> 
>     Good luck,
>     gvb
> 
> 
> 
> Marco Ant?nio Possamai
> 
> Universidade Federal de Santa Catarina
> 
> 
> -- 
> It's News. It's Reviews. It's Interviews. It's Free. What Are You 
> Waiting For?
> www.movieline.com <http://www.movieline.com >!