Re: [PATCH v10 00/11] uapi: export all headers under uapi directories

[Nicolas Dichtel]

Hi Masahiro,

Le 27/03/2017 Ã 07:26, Masahiro Yamada a Ãcrit :
> Hi Nocolas,
> 
> 
> 2017-03-24 18:03 GMT+09:00 Nicolas Dichtel <nicolas.dichtel@xxxxxxxxx>:
[snip]
> 
> 
> As a whole, this series is amazing.  Thanks for your great work!
Thank you. And thank you for taking time to review it.

> 
> 
> I added some comments, but they are trivial.
> 
> 
> 
> 
> I wanted to leave comments/questions on 10/11,
> but I could not find 10/11 in my mailbox.  I do not know why.
Note that you can download the mail from the kbuild patchwork, open it with your
email client and do a reply ;-)

> 
> 
> I am leaving comments on the cover-letter,
> the following are related to 10/11.
> 
> 
> 
> [1]
> 
>> mandatory-y += $(foreach hdr,$(opt-header), \
>>              $(if \
>>                $(wildcard \
>>                        $(srctree)/arch/$(SRCARCH)/include/uapi/asm/$(hdr) \
>>                        $(srctree)/arch/$(SRCARCH)/include/asm/$(hdr) \
>>                ), \
>>                $(hdr) \
>>                ))
> 
> What is this actually checking?
> 
> If ARCH has its own (uapi/)asm/{kvm.h,kvm_para.h,a.out.h},
> they are added to mandatory-y, then they are checked if they exist.
> But, we know they exist.
Yes, you're right. With english words : 'those files are mandatory only if they
exist', thus they are not mandatory at all :)

> 
> 
> This check reminds us only when we added asm/*.h
> but forgot to add uapi/asm/*.h
> 
> $(srctree)/arch/$(SRCARCH)/include/uapi/asm/$(hdr) seems unneeded at least.
> (perhaps, the whole hunk might be unneeded.)
I think we can remove the whole hunk (see also [2]).

> 
> 
> 
> [2]
> 
>> ifeq ($(wildcard $(srctree)/arch/$(SRCARCH)/include/uapi/asm/a.out.h \
>>                 $(srctree)/arch/$(SRCARCH)/include/asm/a.out.h),)
>> header-n += a.out.h
>> endif
>>
>> ifeq ($(wildcard $(srctree)/arch/$(SRCARCH)/include/uapi/asm/kvm.h \
>>                 $(srctree)/arch/$(SRCARCH)/include/asm/kvm.h),)
>> header-n += kvm.h
>> endif
>>
>> ifeq ($(wildcard $(srctree)/arch/$(SRCARCH)/include/uapi/asm/kvm_para.h \
>>                 $(srctree)/arch/$(SRCARCH)/include/asm/kvm_para.h),)
>> header-n += kvm_para.h
>> endif
> 
> This series intends all headers are exported from uapi/, correct?
> Do we still need to check $(srctree)/arch/$(SRCARCH)/include/asm/*.h ?
> (related to [1])
No you're right, uapi/asm/*.h is enough. Those files should be exported only if
the uapi/asm/ counterpart exists.

> 
> 
> 
> [3]
> 
>> --- 7.1 header-n
>>
>> header-n is essentially used by include/uapi/linux/Kbuild to avoid
>> exporting specific headers (e.g. kvm.h) on architectures that do not
>> support it. It should be avoided as much as possible.
> 
> 
> Going forward, header-y will be never used
> because uapi/ is exported by default.
> 
> So, I wonder if we could rename this into something clearer.
> 
> Kbuild supports "no-clean-files".
> (Please see ./Kbuild for its usage)
> I guess this notation seems clearer
> when we want to negate the default behavior.
> 
> Can you consider "no-export", "no-export-files", "no-export-headers"
> or whatever you like?
No problem, let's use no-export-headers.


Thank you,
Nicolas