[PATCH 00/14] Removal of or32/openrisc and introduction of or1k

[Joel Sherrill]

On 4/23/2014 4:38 PM, Hesham Moustafa wrote:
> On Wed, Apr 23, 2014 at 10:45 PM, Christian Svensson <christian at cmd.nu> wrote:
>> On Wed, Apr 23, 2014 at 9:38 PM, Hesham Moustafa
>> <heshamelmatary at gmail.com> wrote:
>>> One issue, building gcc for RTEMS will have to discard some libraries
>>> that are linked by default (like or1ksim) in or1k-gcc when compiling
>>> programs,
>> That sounds strange, GCC should not depend on or1ksim AFAIK - where is this?
> Here [1] line 76. Please correct me if I misunderstood that. When I
> remove this line, a helloworld.c program does not work on gdb/or1ksim.
> I know the library itself is part of libgloss.
>
> [1] https://github.com/openrisc/or1k-gcc/blob/or1k/gcc/config/or1k/or1k.h
You are reading that right. Normally dependencies on something like this
come in through libgloss. There is a pattern to specifying a board or
simulator.

I don't understand the "Override the previous... Linux" comment. The typical
pattern is that the common or1k stuff would be in one file, then you would
Linux, ELF, and RTEMS specific stuff in other files. The file gcc/config.gcc
then lists the individual pieces and parts as either the tmake_file list
for Makefile pieces or tm_file for target .h file pieces. 

See about line 1881 for how m68k-*-rtems* is done.  It is built up from
m68k w/NO OS dependencies, then ELF, more, and then finally the
m68k rtems specific file.

This can be confusing but it shares as much as possible and highlights
the common parts of individual targets triples within an architecture.
Plus it ensures that ABI related items are common.

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel.sherrill at OARcorp.com        On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
Support Available                (256) 722-9985