[GSoC] Porting musl libc to RISC-V

[GSoC] Porting musl libc to RISC-V

[Masanori Ogino]

Hello,

I'm a master's student in Japan majoring in informatics. In this year,
my proposal for lowRISC has been accepted as part of Google Summer of
Code, so I'd like to introduce myself and the project I will work on.
You may have already discussed with me on the mailing lists, though
:-)

The goal of my project is completing a RISC-V port of musl
(http://www.musl-libc.org/), mentored by Rich Felker. I hope the
project could benefit the RISC-V community. Since musl has been used
by some lightweight/embedded Linux distributions including Alpine
Linux and OpenWrt, the project will be attractive especially for
embedded system developers, I guess.

The project aims not only adding RV-specific part of the libc, but
also integrating it into the toolchain. I also consider some
improvements to the kernel side interacting with libc (e.g. migrating
compare-and-swap emulation from system calls to vDSO functions) if
possible.

For details, you can get my proposal on GitHub:
https://github.com/omasanori/gsoc2016-proposal

I'm definitely new to the community though I have had interest in both
RISC-V and musl for years. Nice to meet you!

Any comments would be appreciated. I hope you will get excited at this
project too.

-- 
Masanori Ogino

Re: [GSoC] Porting musl libc to RISC-V

[Rich Felker]

On Wed, Apr 27, 2016 at 11:06:55AM +0900, Masanori Ogino wrote:
> Hello,
> 
> I'm a master's student in Japan majoring in informatics. In this year,
> my proposal for lowRISC has been accepted as part of Google Summer of
> Code, so I'd like to introduce myself and the project I will work on.
> You may have already discussed with me on the mailing lists, though
> :-)
> 
> The goal of my project is completing a RISC-V port of musl
> (http://www.musl-libc.org/), mentored by Rich Felker. I hope the
> project could benefit the RISC-V community. Since musl has been used
> by some lightweight/embedded Linux distributions including Alpine
> Linux and OpenWrt, the project will be attractive especially for
> embedded system developers, I guess.
> 
> The project aims not only adding RV-specific part of the libc, but
> also integrating it into the toolchain. I also consider some
> improvements to the kernel side interacting with libc (e.g. migrating
> compare-and-swap emulation from system calls to vDSO functions) if
> possible.
> 
> For details, you can get my proposal on GitHub:
> https://github.com/omasanori/gsoc2016-proposal
> 
> I'm definitely new to the community though I have had interest in both
> RISC-V and musl for years. Nice to meet you!
> 
> Any comments would be appreciated. I hope you will get excited at this
> project too.

Congratulations, and welcome to the project. I know you've already
been involved on the mailing list, including in topics not directly
related to your proposal, which is great and very much in the spirit
of the "community bonding period" of GSoC that's starting now. I
believe some of those threads never reached proper conclusion and you
may still be waiting for action from me or someone else, so between
now and the start of the GSoC coding period would be a good time to
follow up.

While the focus of your project is musl, your proposal extends into
some other areas (kernel, testing, etc.) and it would also be nice to
identify people you might work with in those areas. Alexander Monakov,
who also works on GCC and is involved in our community, has expressed
interest in helping with anything related to the GCC side of your
work. And Szabolcs Nagy maintains libc-test, the test system it will
make sense to submit any new tests to. I'm also involved in kernel
development now (as arch/sh maintainer) so I can assist on things
related to kernel side (vdso) and procedures for submitting patches if
you're not familiar, but the riscv stuff might still be mostly
out-of-tree anyway.

It's not mandatory for being involved in musl development, but myself
and a lot of our contributors use IRC, #musl on the Freenode network,
as part of development discussion. You're welcome to join us there if
you like.

Rich

Re: [GSoC] Porting musl libc to RISC-V

[Masanori Ogino]

Hello,

2016-04-29 13:11 GMT+09:00 Rich Felker <dalias@libc.org>:
> On Wed, Apr 27, 2016 at 11:06:55AM +0900, Masanori Ogino wrote:
>> Hello,
>>
>> I'm a master's student in Japan majoring in informatics. In this year,
>> my proposal for lowRISC has been accepted as part of Google Summer of
>> Code, so I'd like to introduce myself and the project I will work on.
>> You may have already discussed with me on the mailing lists, though
>> :-)
>>
>> The goal of my project is completing a RISC-V port of musl
>> (http://www.musl-libc.org/), mentored by Rich Felker. I hope the
>> project could benefit the RISC-V community. Since musl has been used
>> by some lightweight/embedded Linux distributions including Alpine
>> Linux and OpenWrt, the project will be attractive especially for
>> embedded system developers, I guess.
>>
>> The project aims not only adding RV-specific part of the libc, but
>> also integrating it into the toolchain. I also consider some
>> improvements to the kernel side interacting with libc (e.g. migrating
>> compare-and-swap emulation from system calls to vDSO functions) if
>> possible.
>>
>> For details, you can get my proposal on GitHub:
>> https://github.com/omasanori/gsoc2016-proposal
>>
>> I'm definitely new to the community though I have had interest in both
>> RISC-V and musl for years. Nice to meet you!
>>
>> Any comments would be appreciated. I hope you will get excited at this
>> project too.
>
> Congratulations, and welcome to the project. I know you've already
> been involved on the mailing list, including in topics not directly
> related to your proposal, which is great and very much in the spirit
> of the "community bonding period" of GSoC that's starting now. I
> believe some of those threads never reached proper conclusion and you
> may still be waiting for action from me or someone else, so between
> now and the start of the GSoC coding period would be a good time to
> follow up.

Sure.

> While the focus of your project is musl, your proposal extends into
> some other areas (kernel, testing, etc.) and it would also be nice to
> identify people you might work with in those areas. Alexander Monakov,
> who also works on GCC and is involved in our community, has expressed
> interest in helping with anything related to the GCC side of your
> work. And Szabolcs Nagy maintains libc-test, the test system it will
> make sense to submit any new tests to. I'm also involved in kernel
> development now (as arch/sh maintainer) so I can assist on things
> related to kernel side (vdso) and procedures for submitting patches if
> you're not familiar, but the riscv stuff might still be mostly
> out-of-tree anyway.

Thank you. I will definitely need help from experts to archive achieve the goal.

> It's not mandatory for being involved in musl development, but myself
> and a lot of our contributors use IRC, #musl on the Freenode network,
> as part of development discussion. You're welcome to join us there if
> you like.

OK, I will try it later.

-- 
Masanori Ogino