home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.sgi
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!sgiblab!sgigate!sgi!quasar.mti.sgi.com!davea
- From: davea@quasar.mti.sgi.com (David B.Anderson)
- Subject: Re: PROBLEMS with CodeVision
- Message-ID: <s8e9vbs@sgi.sgi.com>
- Sender: davea@quasar.mti.sgi.com
- Organization: Silicon Graphics, Inc. Mountain View, CA
- Date: Thu, 12 Nov 1992 00:57:33 GMT
- Lines: 58
-
- In article <1992Nov10.133551.15059@news.tu-graz.ac.at> blah@tu-graz.ac.at (blahowsky) writes:
- >Summary: Problems with CodeVision
- >Keywords:CodeVision
- >
- >
- >We installed this year three Crimson/Elan workstations with
- >the CodeVision (noe CaseVision)debugging tool.
- >
- >After we had some trouble with the FORTRAN compiler we didn't
- >install the last maintainance kit on one maschine in order to
- >produce on one maschine a running executable. At that stage
- >the CodeVision debugger worked fine on that machine.
- >Now we had to bring the 3. machine on the same OS level and
- >therefor installed the last maintaince kit. Since then
- >we have the following cumbersome irregularities with cvd:
- >
- > 1. whan going with step threw a subroutine
- > the program counter jumps quite often
- > to the first line in the subroutine before
- > actually executing the next line
-
- See below.
-
- >The same things happen with edge or dbx so I guess its the
- >fortran compiler or the debugging kernel.
-
- Mail to blah@tu-graz.ac.at bounced, so the only way I can reply is to
- repost something from a few days ago:
-
- If you do the file command on one of your object files, I expect it to
- say something like:
-
- st1.o: mipseb executable not stripped (R4000 fpdiv unchecked) - version 2.40
- (the important item here is the ^^^^^^^^^^^^)
-
- This means you are using the 2.40 compilers. In some releases in the
- 4.0.x series (but not 4.0.1) the assembler inserted nop instructions
- but did not mark them with the "right" line number. This was done so
- the R4000 (even with early (now replaced) chips) would work correctly.
- The result: step/next jumps around disconcertingly.
-
- All fixups needed for the R4K chips in the field are handled
- transparently by the kernel by 4.0.5. So the assembler no longer
- inserts the nop instructions.
-
- I suggest you get the 4.0.5 IRIX release (or later) and the 3.10
- compilers (a seperate IDO update) and recompile. This will fix "next"
- and "step" (*and* get you better application run-time performance
- though, of course, your mileage may vary).
-
- (I have never kept quite straight which compiler releases generated the
- extra nop instructions: They are gone in 4.0.5 IRIX with 3.10 compilers.)
-
- Hope this helps.
- [ David B. Anderson (415)390-4263 davea@sgi.com ]
-
-
-
-