Re: linux-next: build failure after merge of the net-next tree

From: Masahiro Yamada
Date: Wed Jun 19 2019 - 00:20:07 EST


On Wed, Jun 19, 2019 at 1:02 PM Masahiro Yamada
<yamada.masahiro@xxxxxxxxxxxxx> wrote:
>
> Hi.
>
>
> On Wed, Jun 19, 2019 at 12:23 PM Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
> >
> > Hi all,
> >
> > After merging the net-next tree, today's linux-next build (x86_64
> > allmodconfig) failed like this:
> >
> > In file included from usr/include/linux/tc_act/tc_ctinfo.hdrtest.c:1:
> > ./usr/include/linux/tc_act/tc_ctinfo.h:30:21: error: implicit declaration of function 'BIT' [-Werror=implicit-function-declaration]
> > CTINFO_MODE_DSCP = BIT(0),
> > ^~~
> > ./usr/include/linux/tc_act/tc_ctinfo.h:30:2: error: enumerator value for 'CTINFO_MODE_DSCP' is not an integer constant
> > CTINFO_MODE_DSCP = BIT(0),
> > ^~~~~~~~~~~~~~~~
> > ./usr/include/linux/tc_act/tc_ctinfo.h:32:1: error: enumerator value for 'CTINFO_MODE_CPMARK' is not an integer constant
> > };
> > ^
> >
> > Caused by commit
> >
> > 24ec483cec98 ("net: sched: Introduce act_ctinfo action")
> >
> > Presumably exposed by commit
> >
> > b91976b7c0e3 ("kbuild: compile-test UAPI headers to ensure they are self-contained")
> >
> > from the kbuild tree.
>
>
> My commit correctly blocked the broken UAPI header, Hooray!
>
> People export more and more headers that
> are never able to compile in user-space.
>
> We must block new breakages from coming in.
>
>
> BIT() is not exported to user-space
> since it is not prefixed with underscore.
>
>
> You can use _BITUL() in user-space,
> which is available in include/uapi/linux/const.h
>
>


I just took a look at
include/uapi/linux/tc_act/tc_ctinfo.h


I just wondered why the following can be compiled:

struct tc_ctinfo {
tc_gen;
};


Then, I found 'tc_gen' is a macro.

#define tc_gen \
__u32 index; \
__u32 capab; \
int action; \
int refcnt; \
int bindcnt



What a hell.



--
Best Regards
Masahiro Yamada