LLVM/clang status?

[Chris Johns]

On 17/9/17 4:41 am, Joel Sherrill wrote:
> On Fri, Sep 15, 2017 at 4:53 PM, Gedare Bloom <gedare at rtems.org
> <mailto:gedare at rtems.org>> wrote:
> 
>     Ah, thanks for the update Daniel!
> 
>     I recently saw someone on newlib mailing list say that the newlib can
>     be compiled by clang for the ARM processor now.
> 
> 
> llvm/clang's disadvantage vs gcc is its support for as many architectures
> and older CPU variants. Given it a try for the arm but I would be surprised
> if it supports all of the CPU models we have BSPs for.

Yes, any support we can add with clang would be great and of course gcc is still
critical for us.

> 
> Any idea on PowerPC status? I haven't checked that in a few years but
> they had a lack of maintainer issue then.
> 
> I am really curious to find out how this works out. It would be nice to
> have RSB entries for the targets that work so we can at least try it
> for analysis and make it easier for us all to bang on it. Otherwise,
> it won't get used.
> 

I agree we need to package it. I thought llvm built all backends so a standard
build will include arm, powerpc or what ever it contains.

How does the threading model map to RTEMS?
How does newlib get built?

Does anyone have a recipe for using clang with RTEMS?

Chris