Date: Sun, 18 Dec 1994 21:43:02 -0800 (PST) From: John Utz <spaz@u.washington.edu> To: Sean Eric Fagan <sef@kithrup.com> Cc: FreeBSD hackerlist <FreeBSD-hackers@freefall.cdrom.com> Subject: Is this qualify as a complete and utter hack? Re: any #define experts out there? Message-ID: <Pine.OSF.3.91a.941218211139.5070A-100000@saul3.u.washington.edu> In-Reply-To: <199412160856.AAA27938@kithrup.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Thanks Sean and Folks!!!
The guess Sean makes at the bottom is explicitly correct...
On Fri, 16 Dec 1994, Sean Eric Fagan wrote:
> >////usr/local/spice3f3/src/lib/dev/mos3//mos3sset.c < /dev/null
...... much mindless compil-o-blather deleted..
> >////usr/local/spice3f3/src/include/spice.h:29: syntax error before `extern'
>
> It is impossible to say why gcc is behaving the way it appears to without
> knowing how "NUMELENS" is referenced.
>
good point...how come he always nails me on these? :-)
> Looking at the error messages above, I'd say that the problem is something
> completely different than what you think it is.
Thankyou you helpful soul!!!
>
> Since I do not know what the source code you are trying to compile is, I
> cannot help further.
The source code was spice3f3. Dan Andersson and I have been
whacking away at it for a few weeks now. It is built with one of these
nutty convoluted build scripts that seems to make things much harder then
it need be, especially since the damn thing will fail in incomprehensible
ways if you use any other shell besides bash, which i did not posess
until Dan Andersson told me i needed it.
Fixing an unfinished header file fixed this error.
The program compiled. Upon execution it promptly dumps core.
So I sighed, ran util/build clean gcc; Added -g ,Restarted the compile and
went to bed ( this takes !!hours!! on a 386 ). Well, after eating up all my
swap space..(.16 or 33 megs i forget which) a few times, I got it made
with debugging symbols and was pointed to this little wonder:
#ifdef HAS_UNIX_SEGMENT_HACK
#include <signal.h>
#include <setjmp.h>
/*
* baseaddr( ) returns the base address of the data segment on most Unix
* systems. It's an ugly hack for info that should be provided by the OS.
*/
/* Does anyone use a pagesize < 256 bytes?? I'll bet not;
* too small doesn't hurt
*/
#define LOG2_PAGESIZE 8
static jmp_buf env;
static SIGNAL_TYPE
fault( )
{
signal(SIGSEGV, (SIGNAL_FUNCTION) fault); /* SysV style */
longjmp(env, 1);
}
static void *
baseaddr( )
{
char *low, *high, *at;
char *sbrk( );
long x;
SIGNAL_TYPE (*orig_signal)( );
if (getenv("SPICE_NO_DATASEG_CHECK")
return 0;
else
( head for the sphaghetti!!! i was gonna add all this in
to the letter, but u guys did not do anything to deserve it... ). So
basically, the code that caused the core dump tries to figure out how
much memory the system has left, for some reason. Obviously, it breaks
alot, so they provided this helpful spot of code to try and make things
function, but they did not bother to document it or comment it or
anything! Eventually, i should sort this out, but since this is already
1/2way thru break,and i have more, more, more, code to make...
*******************************************************************************
John Utz spaz@stein.u.washington.edu
idiocy is the impulse function in the convolution of life
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.OSF.3.91a.941218211139.5070A-100000>