Re: Quick test building a module cross all targets and architectures

On 10/26/15 19:23, Konstantin Belousov wrote:
> On Mon, Oct 26, 2015 at 11:03:07AM -0700, John Baldwin wrote:
>> On Monday, October 26, 2015 10:11:43 AM Hans Petter Selasky wrote:
>>> Hi,
>>>
>>> We have NO_MODULES for building kernel without modules, but no NO_KERNEL
>>> to only build the modules.
>>>
>>> What do you think about the following patch:
>>>
>>>> diff --git a/sys/conf/kern.post.mk b/sys/conf/kern.post.mk
>>>> index ddf828e..f0920df 100644
>>>> --- a/sys/conf/kern.post.mk
>>>> +++ b/sys/conf/kern.post.mk
>>>> _at__at_ -32,7 +32,11 _at__at_ KERN_DEBUGDIR?=      ${DEBUGDIR}
>>>>
>>>>   .for target in all clean cleandepend cleandir clobber depend install \
>>>>       obj reinstall tags
>>>> +.if !defined(NO_KERNEL)
>>>>   ${target}: kernel-${target}
>>>> +.else
>>>> +${target}:
>>>> +.endif
>>>>   .if !defined(MODULES_WITH_WORLD) && !defined(NO_MODULES) && exists($S/modules)
>>>>   ${target}: modules-${target}
>>>>   modules-${target}:
>>>
>>> It allows only a single module with MODULES_OVERRIDE= and NO_KERNEL=YES
>>> to be built with universe in very little time. This can save a lot of
>>> build time when changes are limited to a set of kernel modules.
>>
>> Can you just use something like MODULES_WITH_WORLD instead?
>>
>> make tinderbox MAKE_JUST_WORLDS=yes SUBDIR_OVERRIDE=sys/modules MODULES_OVERRIDE=foo
>>
>> (If it's only 1 module directory you can probably just use SUBDIR_OVERRIDE directly?)
>>
>> make tinderbox MAKE_JUST_WORLDS=yes SUBDIR_OVERRIDE=sys/modules/foo
>>
>
> In any variant, this proposal sounds strange. Almost all in-kernel
> code is compiled both for kernel and for modules. I am only aware of
> exceptions for i915kms, which was done for a reason which is no longer
> valid. In other words, if your goal is to check that the change does not
> break compilation of some kernel code, then it is wrong to not compile
> kernels.
>
> Note that kernel and modules compilation environments are differrent.
>

Hi,

I understand that the compilation environments are different.

How would you suggest to build-test a handful of C-files under a single 
device keyword and associated kernel module cross all kernels we have in 
a 10-minutes time-frame? MODULES_OVERRIDE can be defined from within 
kernel configs aswell, so possibly a MODULES_OVERRIDE_OVERRIDE is needed 
for this kind of feature. How about some kind of KERNCONF_APPEND= 
parameter, which contains instructions for "config" to only emit a 
single device keyword, yet, keeping all options and parameters.

--HPS