RE: [Ltib] how does one generate a combined initramfs.cpio.gz image with an uImage under ltib

[Newell, James C.]

Hi Stuart,

Thanks for the response.  How about a .spec file that modifies the kernel 
.config to point to the ltib generated initramfs cpio file and reruns the 
kernel build to produce the monolithic uImage?  Does that break the dependency 
chain in ltib?

Jim

>-----Original Message-----
>From: Stuart Hughes [mailto:address@hidden
>Sent: Wednesday, October 14, 2009 9:15 AM
>To: Newell, James C.
>Cc: address@hidden
>Subject: Re: [Ltib] how does one generate a combined initramfs.cpio.gz
>image with an uImage under ltib
>
>Hi Jim,
>
>I understand the problem.  Unfortunately I've not had time to look into
>this though.
>
>The problem is that in a "typical situation" the kernel wants it
>filesystem blown out and available to turn info a cpio image and then
>linked in.  However in a situation where you want the rootfs to be the
>initrams, this is not possible as you have not yet built it.
>
>I believe LTIB correctly builds the initramfs cpio image.  The remaining
>problem is to get the kernel to recognise this when it boots.  I had
>hoped you could load and pass this to the bootloader (u-boot) and have
>it picked up in the same way that ext2.gz ramdisks are, but this doesn't
>  seem to work.  Alternatively maybe it's possible to take the raw
>kernel pieces and "link" the together to build the composite monolithic
>image.
>
>Anyhow I've not much experience with this, so these are just ideas.  It
>would be great to hear from anyone who has good a better idea of how to
>do this.
>
>Regards, Stuart
>
>Newell, James C. wrote:
>> Hello,
>>
>> Any suggestions/hints on how to handle generating a combined
>> initramfs.cpio image with a kernel uImage under ltib.  The kernel
>> build system links the initramfs.cpio file but ltib executes the
>> kernel spec first then builds the initramfs file.  It's a classic
>> chicken/egg problem.
>>
>> Regards,
>>
>> Jim
>>