Date:	Thu, 7 May 1998 16:59:45 +0300 (IDT)
From:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To:	djgpp AT delorie DOT com
Subject:	Re: DJGPP 2.01 spawnvp problem......
In-Reply-To:	<6irudc$34i$1@nnrp1.dejanews.com>
Message-ID:	<Pine.SUN.3.91.980507164610.6703D-100000@is>
MIME-Version:	1.0

On Thu, 7 May 1998 leetonks AT hotmail DOT com wrote:

> The program runs another DJGPP program using a call to spawnvp, but this call
> just doesn't want to work for some reason.
> 
> The line is :
> result=spawnvp(P_WAIT,prog,par);

Did you remember to make prog the first element of par as well?  This is 
by far the most frequent problem with using spawnXX functions.

> where prog is a null-terminated char array and par is an array of pointers to
> the parameters to be passed.

par[0] should be the same string as prog, or it won't work.  Also, the 
last member of par[] should be a NULL pointer.

If the above hints don't help, post the fragment of your code where you 
invoke spawnvp, showing the strings you pass in prog and par[] variables.

> error is the same either way. Under DOS, an extra error value is reported on
> the screen which says it is error 4. According to the DJGPP docs this is
> EACCES - Permission Denied. What does this mean? :-(

I think you are mixing the error code associated with the SIGSEGV 
exception and error code of file operations.  When SIGSEGV prints error 
code 4, it means that you are accessing memory using illegal pointer or 
address.  The DJGPP FAQ explains how to convert the crash traceback into 
human-readable info in section 9.2 (the FAQ is available as 
v2/faq210b.zip from the same place you get DJGPP).  After this 
conversion, you should be able to see where in your code does it blow up, 
which makes debugging easier.

Also, please note that for efficient help, it is advisable to post the 
entire stack traceback printed when your program crashes (after passing 
it through `symify', see the above-mentioned section of the FAQ).

> I'm 99% confident there isn't a problem in the code as it worked just fine
> under Borland.

Borland, as the rest of real-mode compilers, are much more tolerant to 
illegal memory accesses than DJGPP programs which run in protected mode.

> I've also tried recompiling libc to allow 50 open files as
> opposed to 20 (as mentioned in a similar but different problem on the DJGPP
> bug database) and that hasn't done any good either.

DJGPP supports up to 255 open files, limited only by the value of FILES=
directive on your CONFIG.SYS.  You don't need to recompile the library,
this feature is enabled by default.  Whoever submitted that report to the 
bug-tracker was dead wrong.

- Raw text -