[Buildroot] [PATCH V3 RESEND 0/5] Add ILP32 support in aarch64

Romain Naour romain.naour at gmail.com
Wed Feb 3 21:11:38 UTC 2016


Hello Bamvor, All,

Le 02/02/2016 12:16, Arnout Vandecappelle a écrit :
> On 02-02-16 11:49, Zhangjian (Bamvor) wrote:
>> Hi, Romain
>>
>> On 2016/2/2 17:04, Romain Naour wrote:
>>> Hi Bamvor,
>>>
>>> Le 02/02/2016 03:29, Zhangjian (Bamvor) a écrit :
> [snip]
>>>> You could this glibc[1]. IIRC, It is for patch v6 rfc2. We are planning to
>>>> update it for patch v6 rfc5.
>>>
>>> Thanks for this link.

The git tree from github doesn't build:

../sysdeps/aarch64/multiarch/memset.c: In function '__libc_memset_ifunc':
../sysdeps/aarch64/multiarch/memset.c:35:1: error: implicit declaration of
function '__getauxval' [-Werror=implicit-function-declaration]
libc_ifunc (__libc_memset,

>>>
>>> To be honest, I don't think this series will be merged as is until a stable
>>> toolchain release is made.
>> What's your mean stable toolchain? A toolchain release from a organization or
>> company?
> 
>  It means a release made by someone, not a random version from a git tree
> somewhere. Basically something where people know where to send bug reports to.
> This indeed usually means it's done by an organisation.

At the beginning of this review I expected that it was easy to rebuild an IPL32
toolchain using the abe script, but it's not the case.

> 
> 
>> Do you think it will be accepted after patch of ilp32 of kernel and glibc
>> upstream?
> 
>  It doesn't necessarily mean that upstream has to have accepted it already. We
> just prefer to have a release tarball.

See as example the toolchains provided by Imagination Technologies [1] or
Synopsys [2]. All their patches they are using may not upstream yet, but they
provide a working toolchain which is pretty well tested.

[1] http://codescape-mips-sdk.imgtec.com/components/toolchain/2015.06-05/
[2] https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/releases/

> 
>  Regards,
>  Arnout
> 
>> I am trying to improve the glibc support for ilp32 base on the exist one

I understand, that's great.

Best regards,
Romain

> 
>>> When it's done, can you respin your series on top of
>>> master since the toolchain-external code has been updated recently ?
>> Yes, of course.
>>
> 
> [snip]
> 
> 


More information about the buildroot mailing list