History and intentions of linux.inc ?

History and intentions of linux.inc ?

[Darren Hart]

I've stumbled over a couple of interesting bits in the existing
recipes-kernel/linux/linux.inc, for example (but not limited to):

> do_install_prepend() {
>         if test -e arch/${ARCH}/boot/Image ; then
>              ln -f arch/${ARCH}/boot/Image arch/${ARCH}/boot/uImage
>         fi

On a recent kernel, the above will overwrite a valid uImage with Image,
resulting in a non-bootable set of files in deploy/images.

> 
>         if test -e arch/${ARCH}/boot/images/uImage ; then
>              ln -f arch/${ARCH}/boot/images/uImage arch/${ARCH}/boot/uImage
>         fi

The arch/arm/boot/images directory does not exist on a current kernel
(2.6.35-linaro anyway).


Does anyone know the historical significance and/or intent of this code?
Are there architectures for which this does something sane? After
reading through linux.inc, I'm inclined to avoid using it for the
meta-linaro kernel, and then going after it with chainsaw.

Anyone have anything to add before I start cutting?

-- 
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel

Re: History and intentions of linux.inc ?

[Gary Thomas]

On 02/08/2011 02:41 PM, Darren Hart wrote:
> I've stumbled over a couple of interesting bits in the existing
> recipes-kernel/linux/linux.inc, for example (but not limited to):
>
>> do_install_prepend() {
>>          if test -e arch/${ARCH}/boot/Image ; then
>>               ln -f arch/${ARCH}/boot/Image arch/${ARCH}/boot/uImage
>>          fi
>
> On a recent kernel, the above will overwrite a valid uImage with Image,
> resulting in a non-bootable set of files in deploy/images.
>
>>
>>          if test -e arch/${ARCH}/boot/images/uImage ; then
>>               ln -f arch/${ARCH}/boot/images/uImage arch/${ARCH}/boot/uImage
>>          fi
>
> The arch/arm/boot/images directory does not exist on a current kernel
> (2.6.35-linaro anyway).
>
>
> Does anyone know the historical significance and/or intent of this code?
> Are there architectures for which this does something sane? After
> reading through linux.inc, I'm inclined to avoid using it for the
> meta-linaro kernel, and then going after it with chainsaw.
>
> Anyone have anything to add before I start cutting?
>

I also found the same odd behaviour.  My [simple] solution was
to use my own linux.inc, much like there is a linux-yocto.inc,
which totally removes do_install_prepend()

I'm sure this is safe to do in the mainline meta/recipes-kernel;
as I said the yocto kernel already has done away with the whole
file.

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

Re: History and intentions of linux.inc ?

[Tom Rini]

On 02/08/2011 02:41 PM, Darren Hart wrote:
> I've stumbled over a couple of interesting bits in the existing
> recipes-kernel/linux/linux.inc, for example (but not limited to):
>
>> do_install_prepend() {
>>          if test -e arch/${ARCH}/boot/Image ; then
>>               ln -f arch/${ARCH}/boot/Image arch/${ARCH}/boot/uImage
>>          fi
>
> On a recent kernel, the above will overwrite a valid uImage with Image,
> resulting in a non-bootable set of files in deploy/images.
>
>>
>>          if test -e arch/${ARCH}/boot/images/uImage ; then
>>               ln -f arch/${ARCH}/boot/images/uImage arch/${ARCH}/boot/uImage
>>          fi
>
> The arch/arm/boot/images directory does not exist on a current kernel
> (2.6.35-linaro anyway).
>
>
> Does anyone know the historical significance and/or intent of this code?
> Are there architectures for which this does something sane? After
> reading through linux.inc, I'm inclined to avoid using it for the
> meta-linaro kernel, and then going after it with chainsaw.
>
> Anyone have anything to add before I start cutting?

This really sounds like older stuff that's not relevant anymore.  That 
second example for example makes sense to me but only in the scope of 
some old arch/ppc stuff.

-- 
Tom Rini
Mentor Graphics Corporation

Re: History and intentions of linux.inc ?

[Darren Hart]

On 02/08/2011 01:49 PM, Gary Thomas wrote:
> On 02/08/2011 02:41 PM, Darren Hart wrote:
>> I've stumbled over a couple of interesting bits in the existing
>> recipes-kernel/linux/linux.inc, for example (but not limited to):
>>
>>> do_install_prepend() {
>>> if test -e arch/${ARCH}/boot/Image ; then
>>> ln -f arch/${ARCH}/boot/Image arch/${ARCH}/boot/uImage
>>> fi
>>
>> On a recent kernel, the above will overwrite a valid uImage with Image,
>> resulting in a non-bootable set of files in deploy/images.
>>
>>>
>>> if test -e arch/${ARCH}/boot/images/uImage ; then
>>> ln -f arch/${ARCH}/boot/images/uImage arch/${ARCH}/boot/uImage
>>> fi
>>
>> The arch/arm/boot/images directory does not exist on a current kernel
>> (2.6.35-linaro anyway).
>>
>>
>> Does anyone know the historical significance and/or intent of this code?
>> Are there architectures for which this does something sane? After
>> reading through linux.inc, I'm inclined to avoid using it for the
>> meta-linaro kernel, and then going after it with chainsaw.
>>
>> Anyone have anything to add before I start cutting?
>>
>
> I also found the same odd behaviour. My [simple] solution was
> to use my own linux.inc, much like there is a linux-yocto.inc,
> which totally removes do_install_prepend()
>
> I'm sure this is safe to do in the mainline meta/recipes-kernel;
> as I said the yocto kernel already has done away with the whole
> file.
>

Right, which was prompting me to start looking into whether or not 
linux.inc should continue to exist in it's current form.

-- 
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel

Re: History and intentions of linux.inc ?

[Bruce Ashfield]

On Tue, Feb 8, 2011 at 4:41 PM, Darren Hart <dvhart@linux.intel.com> wrote:
> I've stumbled over a couple of interesting bits in the existing
> recipes-kernel/linux/linux.inc, for example (but not limited to):
>
>> do_install_prepend() {
>>         if test -e arch/${ARCH}/boot/Image ; then
>>              ln -f arch/${ARCH}/boot/Image arch/${ARCH}/boot/uImage
>>         fi
>
> On a recent kernel, the above will overwrite a valid uImage with Image,
> resulting in a non-bootable set of files in deploy/images.
>
>>
>>         if test -e arch/${ARCH}/boot/images/uImage ; then
>>              ln -f arch/${ARCH}/boot/images/uImage arch/${ARCH}/boot/uImage
>>         fi
>
> The arch/arm/boot/images directory does not exist on a current kernel
> (2.6.35-linaro anyway).
>
>
> Does anyone know the historical significance and/or intent of this code?
> Are there architectures for which this does something sane? After
> reading through linux.inc, I'm inclined to avoid using it for the
> meta-linaro kernel, and then going after it with chainsaw.
>
> Anyone have anything to add before I start cutting?

You already know my opinion (since as Gary pointed out, I side stepped
much of what's in there), but I'll throw a word in here as well.

Although I don't know all the history, you can see how the code was
needed at different points, but as we move forward it isn't used in the
same way, and can definitely use some spring cleaning.

I'm factoring out the dtb code (it's good) into something that can be included
from multiple places, but there are parts of linux.inc that look like they can
be tossed since they modify things like the kernel config in ways that aren't
always appropriate.

The base kernel class also has good that can use some cleaning (the
uImage, etc) and we'll tackle that in good time as well.

Cheers,

Bruce

>
> --
> Darren Hart
> Intel Open Source Technology Center
> Yocto Project - Linux Kernel
> _______________________________________________
> poky mailing list
> poky@yoctoproject.org
> https://lists.yoctoproject.org/listinfo/poky
>



-- 
"Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end"