[LEDE-DEV] [PATCH] iproute2: fix ip monitor can't work when NET_NS is not enabled

[Syrone Wong]

My point is to respect musl-libc changes and patch iproute2 as how we
patch kernel.

I take two patches[0] from generic kernel patch dir and change glibc
check to libc check.

If you have any interest, you can check mine[1], it is based on iproute2 4.7.0

[0]: 271-uapi-libc-compat.h-do-not-rely-on-__GLIBC__.patch and
272-uapi-if_ether.h-prevent-redefinition-of-struct-ethhd.patch
[1]: https://github.com/wongsyrone/lede-1/commit/85f2c5dde96fc90e7984182529c7cb72b88adbf9


Best Regards,
Syrone Wong


On Sun, Mar 19, 2017 at 4:30 PM, Yousong Zhou <yszhou4tech at gmail.com> wrote:
> On 19 March 2017 at 14:19, Russell Senior <russell at personaltelco.net> wrote:
>>>>>>> "Yousong" == Yousong Zhou <yszhou4tech at gmail.com> writes:
>>
>> Florian> On 03/17/2017 09:51 PM, Yousong Zhou wrote:
>>>>>> The bug appeared in v4.1.0 and was fixed since v4.8.0
>>>>>>
>>>>>> Fixes FS#620
>>>>>>
>>>>>> Signed-off-by: Yousong Zhou <yszhou4tech at gmail.com>
>>
>> Florian> Is there a reason not to upgrade to a newer iproute2 then?
>> Florian> Maybe not for the 17.01 branch but for the master branch would
>> Florian> not that be preferred?  -- Florian
>>
>>>> There are some unresolved (at least I haven't resolved them) UAPI
>>>> problems.  I have taken a few runs at it, thus far unsuccessful.  Any
>>>> help on that would be welcome.
>>
>> Yousong> Fixing the issue at hand was my intention.
>>
>> Yousong> To be honest, I did not think much about bumping the version ;)
>> Yousong> but can you elaborate what's the UAPI issue with that?
>>
>> This is the build log from my last run at the problem:
>>
>>   http://sprunge.us/ceIL
>>
>> It was a build for alix2 on the 4.9 kernel using iproute2-4.10, with
>> updated patches.
>>
>
> It seems that those __UAPI_DEF_xx are private macros that should be
> defined and used by the kernel uapi headers and are not intended to be
> defined by others to tailor uapi headers' behaviour.
>
> Looking through the history, the issue should be already there since
> the following musl commit: "make netinet/in.h suppress clashing
> definitions from kernel headers"
>
>     http://git.musl-libc.org/cgit/musl/commit/?id=04983f2272382af92eb8f8838964ff944fbb8258
>
> But it reveals itself only after the following kernel commit: "uapi
> glibc compat: fix compile errors when glibc net/if.h included before
> linux/if.h"
>
>     https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=4a91cb61bb995e5571098188092e296192309c77
>
> I would suggest reverting the musl change and see if it will work for
> iproute2 and how bad it will affect other packages...
>
> A related discussion: http://www.openwall.com/lists/musl/2016/03/03/6
>
> Cheers,
>                yousong
>
> _______________________________________________
> Lede-dev mailing list
> Lede-dev at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/lede-dev