Very Computer

by **Richard D. Yent** » Tue, 21 Sep 1993 01:41:12

I am trying to port magic (a VLSI design tool) to Linux. I got the
source from gatekeeper.dec.com. I followed the directions to compile
by doing make config. Here is my session:

Magic works with the following window configurations. X11 is
preferred.
1) X11
2) X10
3) SunView
4) Both X11 or SunView, selectable at run-time
5) AED graphics terminal
I picked 1) X11

The following operating systems are known:

1) Berkeley Unix compatibles (Ultrix, pre-4.0 SunOS, RISCos4.0, etc..)
2) SunOS 4.0 and later
3) Some System V Unix compatibles (A/UX, HPUX)
I picked 1) Berkeley (although I also tried 2) SunOS

Some machines need special compile-time switches:

1) Nothing special, not one of the ones below
2) HP 68000-based workstations
3) An Apple MacII
4) A MIPS workstation running RISCos4.0 or greater (not DecStations)
I picked 1) Nothing special.

Then I checked out src/magic/misc/CFLAGS and changed it to
"-O2 -m486 -I/usr/include/bsd" and set CC to gcc in the main
Makefile.

Next I did a "make force" (as instructed). I saved all of the errors
in a file and then tryed going through the list makeing minor
corrections. (doing new "make force" every once in a while.) This
went ok until I got to the file src/magic/utils/getrect.c. This file
"depends on the structure of a standard I/O library (FILE *)." And it
uses "_ptr" and "_cnt" as members of the struct FILE. These members
don't seem to be in Linux's stdio.h.

I have two questions. One is did I get the "make config" correct?
The other is what do I do about this FILE type problem? The function
is fairly long and complicated. Is don't know enough about this level
of IO to fix it without some advice (or help!).

Thanks,

Richard Yentis, Jr.
---

by **JASON POWELL RHO** » Tue, 21 Sep 1993 23:11:31

> Newsgroups: comp.os.linux.help
> Path: taco!gatech!europa.eng.gtefsd.com!darwin.sura.net!seas.gwu.edu!ryentis

> Subject: porint magic (VLSI CAD); need advice (or help)

> Keywords: magic, VLSI, CAD

> Organization: George Washington University
> Date: Sun, 19 Sep 93 12:41:12 GMT+5:00
> Lines: 49

> I am trying to port magic (a VLSI design tool) to Linux. I got the
> source from gatekeeper.dec.com. I followed the directions to compile
> by doing make config. Here is my session:

> Magic works with the following window configurations. X11 is
> preferred.
> 1) X11
> 2) X10
> 3) SunView
> 4) Both X11 or SunView, selectable at run-time
> 5) AED graphics terminal
> I picked 1) X11

> The following operating systems are known:

> 1) Berkeley Unix compatibles (Ultrix, pre-4.0 SunOS, RISCos4.0, etc..)
> 2) SunOS 4.0 and later
> 3) Some System V Unix compatibles (A/UX, HPUX)
> I picked 1) Berkeley (although I also tried 2) SunOS

> Some machines need special compile-time switches:

> 1) Nothing special, not one of the ones below
> 2) HP 68000-based workstations
> 3) An Apple MacII
> 4) A MIPS workstation running RISCos4.0 or greater (not DecStations)
> I picked 1) Nothing special.

> Then I checked out src/magic/misc/CFLAGS and changed it to
> "-O2 -m486 -I/usr/include/bsd" and set CC to gcc in the main
> Makefile.

> Next I did a "make force" (as instructed). I saved all of the errors
> in a file and then tryed going through the list makeing minor
> corrections. (doing new "make force" every once in a while.) This
> went ok until I got to the file src/magic/utils/getrect.c. This file
> "depends on the structure of a standard I/O library (FILE *)." And it
> uses "_ptr" and "_cnt" as members of the struct FILE. These members
> don't seem to be in Linux's stdio.h.

> I have two questions. One is did I get the "make config" correct?
> The other is what do I do about this FILE type problem? The function
> is fairly long and complicated. Is don't know enough about this level
> of IO to fix it without some advice (or help!).

> Thanks,

> Richard Yentis, Jr.
> ---

Hi. You don't want to port magic- It is really hard and has already been done.
Do an xarchie on "magic6." Binaries are available in several places and they
installed on my machine without a hitch. According to the person who compiled
them, on a 486-66 it takes about 45 seconds to do a 3000 transistor circuit
extraction, which is pretty quick. It is fairly memory hungry though.
Good luck.

Jason