Thanks Grant.

After loading Uboot, we will get Uboot command.

I will copy the kernel image to tftpboot directory. And I will pass the command,

run tftp_nfs

to boot the kernel.

My question is after Uboot gets loaded, the control has to be transferred directly to the kernel, without waiting in the UBOOT prompt.

So now after Uboot gets loaded, kernel should be loaded directly, where I have to modify in the Uboot code for this could be done.

Note my question is how to transfer control directly to kernel from Uboot. I should not get Uboot prompt, directly I should get the kernel prompt.

Thanks, Yaso.

-----Original Message----- From: glikely@secretlab.ca [mailto:glikely@secretlab.ca] On Behalf Of Grant Likely Sent: Monday, August 06, 2007 11:03 AM To: Yasothabalan Ramaswamy-TLS,Chennai Cc: u-boot-users@lists.sourceforge.net Subject: Re: [U-Boot-Users] Doubt regarding bootm in Uboot

On 8/5/07, Yasothabalan Ramaswamy-TLS,Chennai yasothabalanr@hcl.in wrote:

Hi all,

I want to know how to transfer the control from Uboot to Linux Kernel.

I am using MPC5200B powerpc and ICECUBE board.

Where exactly the Linux Kernel should be loaded?

Wherever you want to load it. u-boot will relocate it to the correct location after uncompressing the kernel uImage file.

In cmd_bootm.c file,

<snip>

What do these lines signify?

There are 2 different directories in the Linux kernel source tree which can be used to build a powerpc kernel; arch/ppc and arch/powerpc. arch/ppc is obsolete and slowly all board support in arch/ppc is moving over to arch/powerpc (support for mpc5200 is currently in both trees).

BTW, for new designs, you should be using arch/powerpc (by setting ARCH=powerpc in the environment when building the kernel)

One significant difference between arch/ppc and arch/powerpc is how the bootloader communicates with the kernel for embedded platforms. In arch/ppc, u-boot passes data via a static structure which has no standard format as is prone to breakage. In arch/powerpc, the bootloader instead passes a device tree blob which does have a standard format, and is much less prone to breakage.

So, when booting your Linux kernel that was built with ARCH=powerpc, you should pass three parameters to the bootm command. 1. address of the kernel 2. address of the (or '-' if you don't have one) 3. address of the device tree

You can generate a device tree blob (.dtb file) by compiling the dts file for your board with the device tree compiler (dtc). The Lite5200B device tree is in the kernel source tree in the directory arch/powerpc/boot/dts.

Side note: you're email client is sending mail in HTML, not plain ASCII text. Etiquette on this list is for ASCII email only.

Cheers, g.

Hello, U-Boot is already running in my board. I configured the Flash, SDRAM, SRAM, FEC (ethernet) and console. Apparently all is working and I can trace the code just before the call in order to transfer the control to Linux, that is: (* kernel) (kbd, initrd_start, initrd_end, cmd_start, cmd_end); There it is supposed the U-Boot finishes its operation and then the Liinux kernel takes the control, but it hangs. I did the Linux image with entry address and load address 0x0. U-Boot can check the Multi-image correctly. So I don't know where the problem is. I think the problem is in the Linux image, not in U-Boot, I can not assure it. I can't trace the Linux booting because it hans at the beginning, even the console is not initialized.

Could anybody help me?

One question more, I wouldn't have to configure the memories again, I mean, Linux doesn't participate in the configuration of the memories because U-Boot has already done, right?

Thanks,

Raúl Moreno

Hi Detlev,

you were right. I can't debug, but I followed the code writing in a memory buffer and it doesn't hang, but the output display is not working.

I understand now that the problem is in the Linux kernel (version 2.6.15, for U-boot I am using 1.1.6) because the command "bdinfo" shows me the parameters and all is OK. I thought I would need to initialize the console because U-Boot has already done. I know my problem is now out of the reach of this mailing-list, but anyway, I would appreciate if you could tell me where and when I should initialize the console in the Linux kernel (in my case I am using a MPC866 and the console is on the SCM1).

Thanks,

Raúl Moreno

Detlev Zundel dzu@denx.de Detlev Zundel 08/08/2007 18:54

Para: raul.moreno@telvent.abengoa.com

cc: u-boot-users@lists.sourceforge.net

Asunto: Re: [U-Boot-Users] Linux hangs just at the beginning

Hi Raul,

U-Boot is already running in my board. I configured the Flash, SDRAM,

SRAM,

FEC (ethernet) and console. Apparently all is working and I can trace the code just before the call in order to transfer the control to Linux, that is: (* kernel) (kbd, initrd_start, initrd_end, cmd_start, cmd_end);

Congrats.

There it is supposed the U-Boot finishes its operation and then the

Liinux

kernel takes the control, but it hangs.

Are you sure it hangs or is it just that you don't see any output? Are you sure you are passing a correct console parameter to Linux? Search the archives, many people (including myself) already considered a perfectly working Linux kernel to be dead just because they did not see any console output...

I did the Linux image with entry address and load address 0x0. U-Boot can check the Multi-image correctly. So I don't know where the problem is. I think the problem is in the Linux image, not in U-Boot, I can not assure it. I can't trace the Linux booting because it hans at the beginning, even

the

console is not initialized.

Hook up a debugger if at all possible.

One question more, I wouldn't have to configure the memories again, I mean, Linux doesn't participate in the configuration of the memories because U-Boot has already done, right?

If the parameter passing between U-Boot and the kernel works, then yes, this is correct.

By the way, what kernel version on what arch are you trying to boot?

Cheers Detlev

-- "Win32 sucks so hard it could pull matter out of a Black Hole." -- Pohl Longsine -- DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-40 Fax: (+49)-8142-66989-80 Email: dzu@denx.de

Hi Raul,

I understand now that the problem is in the Linux kernel (version 2.6.15, for U-boot I am using 1.1.6) because the command "bdinfo" shows me the parameters and all is OK.

And you are sure that your kernel sees the same info? "bdinfo" shows you the U-Boot side of the story. One *very good* reason of moving to flat device trees is not being bitten by non-matching C header files.

I thought I would need to initialize the console because U-Boot has already done. I know my problem is now out of the reach of this mailing-list, but anyway, I would appreciate if you could tell me where and when I should initialize the console in the Linux kernel (in my case I am using a MPC866 and the console is on the SCM1).

When your kernel correctly supports the SMC1, it is only a question of using "console=ttyCPM1,${baudrate}" (maybe adjust the 1 regarding actual SMC/SCC support in your kernel).

Cheers Detlev