[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [GSoC 2013] MIPS N32/N64 ABI port project

On 04/11/2013 08:47 PM, Aron Xu wrote:
On Fri, Apr 12, 2013 at 10:58 AM, Matt Turner <mattst88@gmail.com> wrote:


My experience porting Gentoo to N32 was bootstrapping my an N32 base
system, finding and fixing bugs, and finally building installation
media. I don't feel like there was enough actual development in what I
did to warrant a GSoC project.

But I'm not a Debian developer or user, so maybe I don't know what is
required for Debian. What would your deliverables be?

Bootstrapping a working environment is easy, but we'd like to make use
of Multiarch cross-build approach, which is also a verification of
existing GSoC works in past years. As said in the projects page this
project's primary target is to make N64 port, and N32 being an
optional item. This means at least two ports will get bootstrapped
(n64 el/non-el), and it can grow up to four (+ n32 el/non-el) if the
we (or rather, the student) can manage to do that. Deliverable in this
project is written in the wiki page and can be updated when there are
good ideas.

I think n32 may not be so important.  Here's why:

With a couple of trivial toolchain patches (one for the linker script and one to change some default GCC options) We can build n64 binaries with -msym32. This gives us the same code size as n32 (along with the more efficient PLT base function calling sequence) as n32, but with no restrictions on the size of the virtual address space.


Aron Xu

Reply to: