home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!usc!sdd.hp.com!spool.mu.edu!darwin.sura.net!sgiblab!bdcsys!news
- From: sgiblab!doe.carleton.ca!rjg (Richard Griffith)
- Newsgroups: comp.sys.concurrent
- Subject: gcc, etc for masscomp 5500 and ilk (68020 based)
- Message-ID: <rjg.721581083@crusher>
- Date: Thu, 12 Nov 1992 15:11:23 GMT
- Sender: news@bdcsys.suvl.ca.us
- Organization: ncurrent-request
- Approved: concurrent-request@bdcsys.suvl.ca.us
- Lines: 40
-
- I have just about finished compiling gcc on a masscomp 5500-PEP
- running 4.1a with c compiler version 1.4
- Ok it could still be a while, but it looks good.
-
- GCC, stage1 produced assembler code last night but with instructions of the
- form 'movel' rather than the desired 'move.l'. This is a simple change
- in the main header but everything has to be recompiled now.
-
- I have had to hack up the source in a number of places where the concurrent
- c compiler hemorrhaged. ( F(999): codegen ... something about access to
- an unallocated RHS ..., in a section of nonreachable code!!!).
-
- Anyone interested in gcc?
-
- A number of nonreachable code sections had to be commented out of the source
- making the result hard to merge into the standard release. My preferred
- form of release would involve the following:
- 1) tar file of original code with modification to accept masscomp
- configure options.
- 2) tar file of original code modified for concurrent compiler.
- 3) tar file with executable code.
- This allows the use of 3) and 1) to make your own, or use 2) to make your own,
- or just take 3). These would be made available at an FTP site.
-
- How many people are still using the 68020 based hardware and would be
- interested in this? Send email. Mention if you want to try the alpha version,
- it may take a while to get a clean release. Support for 68881, but not the
- other floating point hardware.
-
- Notes:
- 1) The concurrent C compiler does not support bit fields with
- enumerated types.
- 2) A number of overuse of register warning messages were generated.
- 3) Warnings on = and == used between enumerated types and ints.
- 4) Configuration created from a combination of others may include
- some no applicable options.
-
- No promises but it looks like it will all work, real soon now.
-
- If anyone else has done this already, can I see what you have done.
-