3.9-rc1: hz.bc compilation breakage

3.9-rc1: hz.bc compilation breakage

[Maxime Ripard]

Hi,

I'm compiling the kernel for an iMX28(ARM) board, and since 3.9-rc1, I
get the following error at compilation time:

make[3]: *** No rule to make target
`/home/max/Work/2012/crystalfontz/rewrite/linux-dt/include/config/hz.h',
needed by `kernel/hz.bc'.  Stop.

I'm doing a clean out-of-tree build, with the following configuration
file: http://code.bulix.org/m4y7se-83102?raw

Reverting commit 1b66e0fd ("kernel: Replace timeconst.pl with a bc
script") make the compilation go on as usual.

Thanks,
Maxime

-- 
Maxime Ripard, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

Re: 3.9-rc1: hz.bc compilation breakage

[H. Peter Anvin]

On 03/06/2013 06:36 AM, Maxime Ripard wrote:
> Hi,
> 
> I'm compiling the kernel for an iMX28(ARM) board, and since 3.9-rc1, I
> get the following error at compilation time:
> 
> make[3]: *** No rule to make target
> `/home/max/Work/2012/crystalfontz/rewrite/linux-dt/include/config/hz.h',
> needed by `kernel/hz.bc'.  Stop.
> 
> I'm doing a clean out-of-tree build, with the following configuration
> file: http://code.bulix.org/m4y7se-83102?raw
> 
> Reverting commit 1b66e0fd ("kernel: Replace timeconst.pl with a bc
> script") make the compilation go on as usual.
> 

That file should have been created by "make *config", and I just
verified that it does indeed do so.  That file is a sentinel for
configuration changes; the entire build system depends on it.

I am not sure what you mean with "clean out-of-tree", however.

kbuild people: is there anything wrong with how this dependency is
specified?

	-hpa

Re: 3.9-rc1: hz.bc compilation breakage

[Maxime Ripard]

Hi,

Le 07/03/2013 04:41, H. Peter Anvin a écrit :
> On 03/06/2013 06:36 AM, Maxime Ripard wrote:
>> I'm compiling the kernel for an iMX28(ARM) board, and since 3.9-rc1, I
>> get the following error at compilation time:
>>
>> make[3]: *** No rule to make target
>> `/home/max/Work/2012/crystalfontz/rewrite/linux-dt/include/config/hz.h',
>> needed by `kernel/hz.bc'.  Stop.
>>
>> I'm doing a clean out-of-tree build, with the following configuration
>> file: http://code.bulix.org/m4y7se-83102?raw
>>
>> Reverting commit 1b66e0fd ("kernel: Replace timeconst.pl with a bc
>> script") make the compilation go on as usual.
>>
> 
> That file should have been created by "make *config", and I just
> verified that it does indeed do so.  That file is a sentinel for
> configuration changes; the entire build system depends on it.

I triggered this issue when switching to 3.9-rc1, and oldconfig runs at
that time, and now, running menuconfig doesn't generates it either.

> I am not sure what you mean with "clean out-of-tree", however.

make O=$(pwd) -C /path/to/linux/source some_configs
make

Thanks,
Maxime

-- 
Maxime Ripard, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

Re: 3.9-rc1: hz.bc compilation breakage

[H. Peter Anvin]

On 03/07/2013 02:42 AM, Maxime Ripard wrote:
> 
> I triggered this issue when switching to 3.9-rc1, and oldconfig runs at
> that time, and now, running menuconfig doesn't generates it either.
> 
>> I am not sure what you mean with "clean out-of-tree", however.
> 
> make O=$(pwd) -C /path/to/linux/source some_configs
> make
> 

Can you give the exact set of commands you are running when you see
this?  I cannot reproduce this no matter what I do, and it doesn't seem
like it is being a common problem.

In the example above the first "make" command does indeed not create any
files in include/config, but they get created immediately on the second
"make"...

	-hpa

-- 
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.

Re: 3.9-rc1: hz.bc compilation breakage

[Maxime Ripard]

Le 07/03/2013 16:06, H. Peter Anvin a écrit :
> On 03/07/2013 02:42 AM, Maxime Ripard wrote:
>>
>> I triggered this issue when switching to 3.9-rc1, and oldconfig runs at
>> that time, and now, running menuconfig doesn't generates it either.
>>
>>> I am not sure what you mean with "clean out-of-tree", however.
>>
>> make O=$(pwd) -C /path/to/linux/source some_configs
>> make
>>
> 
> Can you give the exact set of commands you are running when you see
> this?  I cannot reproduce this no matter what I do, and it doesn't seem
> like it is being a common problem.
> 
> In the example above the first "make" command does indeed not create any
> files in include/config, but they get created immediately on the second
> "make"...

Hmmm, It looks indeed that it can happen only on already existing
directories, I can't reproduce neither with a new build, and not
everyone seems to be impacted by that bug neither, since some colleagues
did the same thing without any trouble (with a completely different
architecture though).

So it's not that bad.

However, when I do:

In my source dir
git checkout v3.9-rc1

In my build dir
ARCH=arm make clean
ARCH=arm CROSS_COMPILE=arm-linux-gnueabi- make zImage

And then it breaks.

So I suspect it's something not cleaned properly, or not generated
properly. Here is what's left after the make clean

http://code.bulix.org/eh1e6c-83108?raw

Thanks,
Maxime

-- 
Maxime Ripard, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com