Re: How can I add S+core backend code into GCC package

[Mark Mitchell <mark at codesourcery dot com>]

liqin@sunnorth.com.cn wrote:

> Can you kindly add S+core backend patch to the mainline development trunk 
> of GCC?

Thank you for patiently working through the process.

I have looked over the code.  I am confused by the mention of MIPS in 
some of the comments.  For example:

+#err           "must specify MIPS endian!"

+/* Return a LO_SUM expression for ADDR.  TEMP is as for 
mips_force_temporary

+/* When using assembler macros, keep track of all of small-data externs
+   so that mips_file_end can emit the appropriate declarations for them.

There are some others too.  Perhaps you copied some of this code from 
the MIPS back end, and forgot to update the comments?  If so, would you 
please correct the comments?  Please review each one to see whether you 
should just replace "mips" with "s+core", or whether more extensive 
changes are required.

Has the config.sub patch been approved upstream?  I believe that we're 
supposed to import that file from upstream -- but it might also be that 
we can make local changes.  I'm sure one of the build-system maintainers 
will know.

As David has reviewed the patch, and as it looks reasonable to me 
(though I am not an expert on back ends, and would therefore be happy 
for anyone who is an expert to offer comments!), the patch is OK once 
the issues above are fixed.  If we do need to wait on config.sub, you 
can still check in the other parts.

> And I want to know, Can we maintain S+core backend code by ourself?

In fact, we have a rule that all new substantial contributions come with 
a maintainer.  So, while you can of course stop at any time, once the 
port goes in, we would expect you to maintain it going forward.  Is that 
OK with you?

Thanks,

--
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713