[Buildroot] svn commit: trunk/buildroot/target/device/Atme l: atngw100/target_skeleton atngw100/target etc...

[Buildroot] svn commit: trunk/buildroot/target/device/Atme l: atngw100/target_skeleton atngw100/target etc...

[Ulf Samuelsson]

> Added:
>    trunk/buildroot/target/device/Atmel/atngw100/target_skeleton/bin/

What is up with all the directories? Is not defining them
in .../<board>/device_table.txt enough? I find it far better defining
them there since the access permissions can be defined as well.


Some stuff are needed
during the build and
the device table is only
used when you convert
$(TARGET-DIR) to
rootfs-avr32.$(fs) which
is too late.

It can probably be done
better

Best Regards
Ulf Samuelsson

[Buildroot] svn commit: trunk/buildroot/target/device/Atme l: atngw100/target_skeleton atngw100/target etc...

[Hans-Christian Egtvedt]

On Tue, 2007-08-28 at 10:10 +0200, Ulf Samuelsson wrote:
> > Added:
> >    trunk/buildroot/target/device/Atmel/atngw100/target_skeleton/bin/
> 
> What is up with all the directories? Is not defining them
> in .../<board>/device_table.txt enough? I find it far better defining
> them there since the access permissions can be defined as well.
> 
> 
> Some stuff are needed
> during the build and
> the device table is only
> used when you convert
> $(TARGET-DIR) to
> rootfs-avr32.$(fs) which
> is too late.
> 
> It can probably be done
> better

Indeed, the packages failing with an "empty" skeleton should have some
mkdir -p actions in them IMHO. That way not all targets needs to be
fixed, but just one package.

-- 
With kind regards,
Hans-Christian Egtvedt, Applications Engineer

[Buildroot] svn commit: trunk/buildroot/target/device/Atme l: atngw100/target_skeleton atngw100/target etc...

[Bernhard Fischer]

On Tue, Aug 28, 2007 at 10:21:19AM +0200, Hans-Christian Egtvedt wrote:
>
>On Tue, 2007-08-28 at 10:10 +0200, Ulf Samuelsson wrote:
>> > Added:
>> >    trunk/buildroot/target/device/Atmel/atngw100/target_skeleton/bin/
>> 
>> What is up with all the directories? Is not defining them
>> in .../<board>/device_table.txt enough? I find it far better defining
>> them there since the access permissions can be defined as well.
>> 
>> 
>> Some stuff are needed
>> during the build and
>> the device table is only
>> used when you convert
>> $(TARGET-DIR) to
>> rootfs-avr32.$(fs) which
>> is too late.
>> 
>> It can probably be done
>> better
>
>Indeed, the packages failing with an "empty" skeleton should have some
>mkdir -p actions in them IMHO. That way not all targets needs to be
>fixed, but just one package.

Yes, that's the proper thing to do, agree.

[Buildroot] svn commit: trunk/buildroot/target/device/Atme l: atngw100/target_skeleton atngw100/target etc...

[Ulf Samuelsson]

> Some stuff are needed
> during the build and
> the device table is only
> used when you convert
> $(TARGET-DIR) to
> rootfs-avr32.$(fs) which
> is too late.
> 
> It can probably be done
> better

Indeed, the packages failing with an "empty" skeleton should have some
mkdir -p actions in them IMHO. That way not all targets needs to be
fixed, but just one package.

=> Feel free to provide
   those patches.

Best Regards
Ulf Samuelsson

[Buildroot] svn commit: trunk/buildroot/target/device/Atme l: atngw100/target_skeleton atngw100/target etc...

[Hans-Christian Egtvedt]

On Tue, 2007-08-28 at 14:55 +0200, Ulf Samuelsson wrote:
> > Some stuff are needed
> > during the build and
> > the device table is only
> > used when you convert
> > $(TARGET-DIR) to
> > rootfs-avr32.$(fs) which
> > is too late.
> > 
> > It can probably be done
> > better
> 
> Indeed, the packages failing with an "empty" skeleton should have some
> mkdir -p actions in them IMHO. That way not all targets needs to be
> fixed, but just one package.
> 
> => Feel free to provide
>    those patches.

I do whenever I come across them (-:

Are there any specific packages which fails today because of missing
directories?

-- 
With kind regards,
Hans-Christian Egtvedt, Applications Engineer