[gentoo-soc] [GSoC] native-clang: weekly report #2

[gentoo-soc] [GSoC] native-clang: weekly report #2

[Lei Zhang]

Hi there!

Progress this week:
- be able to bootstrap a freestanding clang and make it work on a musl
based system
- experiment with NetBSD's crtbegin/end files and they appear to work correctly
- send the first set of patches to clang/LLVM, which enables clang to
build binaries agains musl

Plan for next week:
- figure out a reasonable patch which allows clang/LLVM to be built against musl
- think about how to package the clang + musl based toolchain
- try to prepare a GNU-free stage3 system


Lei

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Anthony G. Basile]

On 6/11/16 7:23 AM, Lei Zhang wrote:
> Hi there!
> 
> Progress this week:
> - be able to bootstrap a freestanding clang and make it work on a musl
> based system
> - experiment with NetBSD's crtbegin/end files and they appear to work correctly
> - send the first set of patches to clang/LLVM, which enables clang to
> build binaries agains musl
> 
> Plan for next week:
> - figure out a reasonable patch which allows clang/LLVM to be built against musl
> - think about how to package the clang + musl based toolchain
> - try to prepare a GNU-free stage3 system

I'm not sure what you mean by "package the clang + musl based toolchain"
but the stage3 is how we distribute working a toolchain in Gentoo.

> 
> 
> Lei
> 


-- 
Anthony G. Basile, Ph.D.
Gentoo Linux Developer [Hardened]
E-Mail    : blueness@gentoo.org
GnuPG FP  : 1FED FAD9 D82C 52A5 3BAB  DC79 9384 FA6E F52D 4BBA
GnuPG ID  : F52D4BBA

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Luca Barbato]

On 11/06/16 13:46, Anthony G. Basile wrote:
> On 6/11/16 7:23 AM, Lei Zhang wrote:
>> Hi there!
>>
>> Progress this week:
>> - be able to bootstrap a freestanding clang and make it work on a musl
>> based system
>> - experiment with NetBSD's crtbegin/end files and they appear to work correctly
>> - send the first set of patches to clang/LLVM, which enables clang to
>> build binaries agains musl
>>
>> Plan for next week:
>> - figure out a reasonable patch which allows clang/LLVM to be built against musl
>> - think about how to package the clang + musl based toolchain
>> - try to prepare a GNU-free stage3 system
> 
> I'm not sure what you mean by "package the clang + musl based toolchain"
> but the stage3 is how we distribute working a toolchain in Gentoo.

At very least:

- package the crt from netbsd
- fix the dependencies accordingly

lu

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Lei Zhang]

2016-06-11 19:58 GMT+08:00 Luca Barbato <lu_zero@gentoo.org>:
> On 11/06/16 13:46, Anthony G. Basile wrote:
>> On 6/11/16 7:23 AM, Lei Zhang wrote:
>>> Hi there!
>>>
>>> Progress this week:
>>> - be able to bootstrap a freestanding clang and make it work on a musl
>>> based system
>>> - experiment with NetBSD's crtbegin/end files and they appear to work correctly
>>> - send the first set of patches to clang/LLVM, which enables clang to
>>> build binaries agains musl
>>>
>>> Plan for next week:
>>> - figure out a reasonable patch which allows clang/LLVM to be built against musl
>>> - think about how to package the clang + musl based toolchain
>>> - try to prepare a GNU-free stage3 system
>>
>> I'm not sure what you mean by "package the clang + musl based toolchain"
>> but the stage3 is how we distribute working a toolchain in Gentoo.
>
> At very least:
>
> - package the crt from netbsd
> - fix the dependencies accordingly

Also package compiler-rt and libunwind, which will replace libgcc and
libgcc_eh respectively. I'm not sure yet if I should replace libcxxrt
with libcxxabi.

Lei

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Luca Barbato]

On 11/06/16 14:35, Lei Zhang wrote:
> 2016-06-11 19:58 GMT+08:00 Luca Barbato <lu_zero@gentoo.org>:
>> On 11/06/16 13:46, Anthony G. Basile wrote:
>>> On 6/11/16 7:23 AM, Lei Zhang wrote:
>>>> Hi there!
>>>>
>>>> Progress this week:
>>>> - be able to bootstrap a freestanding clang and make it work on a musl
>>>> based system
>>>> - experiment with NetBSD's crtbegin/end files and they appear to work correctly
>>>> - send the first set of patches to clang/LLVM, which enables clang to
>>>> build binaries agains musl
>>>>
>>>> Plan for next week:
>>>> - figure out a reasonable patch which allows clang/LLVM to be built against musl
>>>> - think about how to package the clang + musl based toolchain
>>>> - try to prepare a GNU-free stage3 system
>>>
>>> I'm not sure what you mean by "package the clang + musl based toolchain"
>>> but the stage3 is how we distribute working a toolchain in Gentoo.
>>
>> At very least:
>>
>> - package the crt from netbsd
>> - fix the dependencies accordingly
> 
> Also package compiler-rt and libunwind, which will replace libgcc and
> libgcc_eh respectively. I'm not sure yet if I should replace libcxxrt
> with libcxxabi.

Would be nice to have, but first the basics =)

lu

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Benda Xu]

Hi,

Lei Zhang <zhanglei.april@gmail.com> writes:

> - send the first set of patches to clang/LLVM, which enables clang to
> build binaries agains musl

That's interesting.  Could you give a reference of your patch?

Benda

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Lei Zhang]

2016-06-11 22:50 GMT+08:00 Benda Xu <heroxbd@gmail.com>:
> Hi,
>
> Lei Zhang <zhanglei.april@gmail.com> writes:
>
>> - send the first set of patches to clang/LLVM, which enables clang to
>> build binaries agains musl
>
> That's interesting.  Could you give a reference of your patch?

Here: https://github.com/zzlei/musl-clang

They are not accepted yet :)


Lei

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Anthony G. Basile]

On 6/11/16 11:08 AM, Lei Zhang wrote:
> 2016-06-11 22:50 GMT+08:00 Benda Xu <heroxbd@gmail.com>:
>> Hi,
>>
>> Lei Zhang <zhanglei.april@gmail.com> writes:
>>
>>> - send the first set of patches to clang/LLVM, which enables clang to
>>> build binaries agains musl
>>
>> That's interesting.  Could you give a reference of your patch?
> 
> Here: https://github.com/zzlei/musl-clang
> 
> They are not accepted yet :)
> 
> 
> Lei
> 

those are approaximately the equivalent of the gcc patches for musl at

https://gitweb.gentoo.org/proj/musl.git/tree/sys-devel/gcc/files/gcc-4.8.3-musl-linker-path.patch


-- 
Anthony G. Basile, Ph.D.
Gentoo Linux Developer [Hardened]
E-Mail    : blueness@gentoo.org
GnuPG FP  : 1FED FAD9 D82C 52A5 3BAB  DC79 9384 FA6E F52D 4BBA
GnuPG ID  : F52D4BBA

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Anthony G. Basile]

On 6/11/16 11:12 AM, Anthony G. Basile wrote:
> On 6/11/16 11:08 AM, Lei Zhang wrote:
>> 2016-06-11 22:50 GMT+08:00 Benda Xu <heroxbd@gmail.com>:
>>> Hi,
>>>
>>> Lei Zhang <zhanglei.april@gmail.com> writes:
>>>
>>>> - send the first set of patches to clang/LLVM, which enables clang to
>>>> build binaries agains musl
>>>
>>> That's interesting.  Could you give a reference of your patch?
>>
>> Here: https://github.com/zzlei/musl-clang
>>
>> They are not accepted yet :)
>>
>>
>> Lei
>>
> 
> those are approaximately the equivalent of the gcc patches for musl at
> 
> https://gitweb.gentoo.org/proj/musl.git/tree/sys-devel/gcc/files/gcc-4.8.3-musl-linker-path.patch
> 
> 

also, Lei, do you mind cc-ing gentoo-musl@lists.gentoo.org when giving
your gsoc reports and other stuff.  for archival reasons.  thanks and
keep up the good work.

-- 
Anthony G. Basile, Ph.D.
Gentoo Linux Developer [Hardened]
E-Mail    : blueness@gentoo.org
GnuPG FP  : 1FED FAD9 D82C 52A5 3BAB  DC79 9384 FA6E F52D 4BBA
GnuPG ID  : F52D4BBA

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Lei Zhang]

2016-06-11 23:12 GMT+08:00 Anthony G. Basile <blueness@gentoo.org>:
> On 6/11/16 11:08 AM, Lei Zhang wrote:
>> 2016-06-11 22:50 GMT+08:00 Benda Xu <heroxbd@gmail.com>:
>>> Hi,
>>>
>>> Lei Zhang <zhanglei.april@gmail.com> writes:
>>>
>>>> - send the first set of patches to clang/LLVM, which enables clang to
>>>> build binaries agains musl
>>>
>>> That's interesting.  Could you give a reference of your patch?
>>
>> Here: https://github.com/zzlei/musl-clang
>>
>> They are not accepted yet :)
>>
>>
>> Lei
>>
>
> those are approaximately the equivalent of the gcc patches for musl at
>
> https://gitweb.gentoo.org/proj/musl.git/tree/sys-devel/gcc/files/gcc-4.8.3-musl-linker-path.patch

I heard GCC 6 has official support for musl, but haven't checked it out myself.


Lei

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Anthony G. Basile]

On 6/11/16 11:17 AM, Lei Zhang wrote:
> 2016-06-11 23:12 GMT+08:00 Anthony G. Basile <blueness@gentoo.org>:
>> On 6/11/16 11:08 AM, Lei Zhang wrote:
>>> 2016-06-11 22:50 GMT+08:00 Benda Xu <heroxbd@gmail.com>:
>>>> Hi,
>>>>
>>>> Lei Zhang <zhanglei.april@gmail.com> writes:
>>>>
>>>>> - send the first set of patches to clang/LLVM, which enables clang to
>>>>> build binaries agains musl
>>>>
>>>> That's interesting.  Could you give a reference of your patch?
>>>
>>> Here: https://github.com/zzlei/musl-clang
>>>
>>> They are not accepted yet :)
>>>
>>>
>>> Lei
>>>
>>
>> those are approaximately the equivalent of the gcc patches for musl at
>>
>> https://gitweb.gentoo.org/proj/musl.git/tree/sys-devel/gcc/files/gcc-4.8.3-musl-linker-path.patch
> 
> I heard GCC 6 has official support for musl, but haven't checked it out myself.
> 
> 
> Lei
> 

it does, i have checked, but there is some c++ symbol poisoning that
needs to be worked out.


-- 
Anthony G. Basile, Ph.D.
Gentoo Linux Developer [Hardened]
E-Mail    : blueness@gentoo.org
GnuPG FP  : 1FED FAD9 D82C 52A5 3BAB  DC79 9384 FA6E F52D 4BBA
GnuPG ID  : F52D4BBA

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Lei Zhang]

2016-06-11 23:13 GMT+08:00 Anthony G. Basile <blueness@gentoo.org>:
> On 6/11/16 11:12 AM, Anthony G. Basile wrote:
>> On 6/11/16 11:08 AM, Lei Zhang wrote:
>>> 2016-06-11 22:50 GMT+08:00 Benda Xu <heroxbd@gmail.com>:
>>>> Hi,
>>>>
>>>> Lei Zhang <zhanglei.april@gmail.com> writes:
>>>>
>>>>> - send the first set of patches to clang/LLVM, which enables clang to
>>>>> build binaries agains musl
>>>>
>>>> That's interesting.  Could you give a reference of your patch?
>>>
>>> Here: https://github.com/zzlei/musl-clang
>>>
>>> They are not accepted yet :)
>>>
>>>
>>> Lei
>>>
>>
>> those are approaximately the equivalent of the gcc patches for musl at
>>
>> https://gitweb.gentoo.org/proj/musl.git/tree/sys-devel/gcc/files/gcc-4.8.3-musl-linker-path.patch
>>
>>
>
> also, Lei, do you mind cc-ing gentoo-musl@lists.gentoo.org when giving
> your gsoc reports and other stuff.  for archival reasons.  thanks and
> keep up the good work.

No problem :D


Lei

Re: [gentoo-soc] [GSoC] native-clang: weekly report #2

[Lei Zhang]

2016-06-11 23:20 GMT+08:00 Anthony G. Basile <blueness@gentoo.org>:
> On 6/11/16 11:17 AM, Lei Zhang wrote:
>>
>> I heard GCC 6 has official support for musl, but haven't checked it out myself.
>
> it does, i have checked, but there is some c++ symbol poisoning that
> needs to be worked out.

Does it work the same way as other libc, i.e. with a single option
like "-mmusl"?

I've been curious why the need to specify libc name in an option,
since it's already embedded in the target triplet.


Lei