Re: [Patch v3 1/3] dt: OF_UNITTEST make dependency broken, framework fixes

[Frank Rowand <frowand.list@gmail.com>]

On 3/13/2015 6:13 AM, Rob Herring wrote:
> On Thu, Mar 12, 2015 at 8:15 AM, Frank Rowand <frowand.list@gmail.com> wrote:
>> From: Frank Rowand <frank.rowand@sonymobile.com>
>>
>> This 3 patch series is not bisectable.  If CONFIG_OF_UNITTEST=y then the kernel
>> will not build with just patch 1 or just patches 1 and 2 applied.
> 
> If you did 2 patches with the move first, wouldn't that make it
> bisectable. However, I'd rather avoid the move.

Yes, I did not like the move either.  Fortunately you gave me the cluebat I needed
to avoid the move.

> 
>> If CONFIG_OF_UNITTEST=y then a kernel image make will always cause .version to
>> be incremented, even if there are not source changes.  This is caused by
>> a lack of dependency tracking and checking for
>> drivers/of/unittest-data/testcases.dtb.o.  Fixing the problem was made more
>> complicated by the fact that testcases.dtb.o was linked into ../of_unittest.o.
> 
> Couldn't we change that to be 2 modules.

Thanks, that was just the hint I needed to realize that of_unittest.o was
just an artifact of the problem I was trying to solve.  I will send a new
patch that removes the of_unittest.o cruft and does not need to move unittest.c

> 
>> Patch 1 modifies makefiles to move of_unittest.c into unittest-data/ and
>> creates missing dependency tracking for testcases.dtb.o.
>>
>> Patch 2 will move of_unittest.c into unittest-data/
>>
>> Patch 3 will fix an of_unittest.c include path to account for the move.
> 
> This should all be in an intro email, not patch #1.