Date: Thu, 22 Mar 2001 10:20:49 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: "Nimrod A. Abing" <n_abing AT ns DOT roxas-online DOT net DOT ph>
cc: djgpp-workers AT delorie DOT com, sandmann AT clio DOT rice DOT edu
Subject: Re: Core dumping for DJGPP programs
In-Reply-To: <3.0.1.32.20010322141215.00706d80@wingate>
Message-ID: <Pine.SUN.3.91.1010322100902.7564M-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk


On Thu, 22 Mar 2001, Nimrod A. Abing wrote:

>   if (__djgpp_exception_state == fake_exception
>       && __djgpp_exception_state -> __cs == _my_cs()
>       && (unsigned)__djgpp_exception_state -> __ebp
>       >= __djgpp_exception_state -> __esp
>       && (unsigned *)__djgpp_exception_state -> __ebp >= &end
>       && (unsigned *)__djgpp_exception_state -> __ebp
>       < (unsigned *)__djgpp_selector_limit)
>     _writei4 (corefile, *((unsigned *)__djgpp_exception_state -> __ebp + 1));
>   else
>     _writei4 (corefile, __djgpp_exception_state -> __eip);
> 
> which is basically what the code in dpmiexcp does, except the value of eip
> is written to an open file. The eip is correctly written for signals other
> than the fake signal SIGABRT (well for SIGSEGV anyway, eip is correctly
> written). I also noticed discrepancies in the values of the registers eax,
> ecx, edx, esi, edi, ebp, and esp. There's a big difference between the ones
> generated by the traceback and the ones stored in the core file, this
> apparently happens for fake signal SIGABRT, doesn't seem to happen for
> SIGSEGV.

IIRC, SIGABRT doesn't bother to save the registers at the point of 
abort()'s invocation.  If that's true, there's no surprise that you get 
different register values.  For true exceptions, such as SIGSEGV, the 
exception handler saves the registers on the stack and then copies them 
into the __djgpp_exception_state struct.  But I don't think anything 
similar is done for fake exceptions.  Perhaps the simple-minded way of 
hacking together the fake_exception variable can be changed to do it 
better.

In any case, since a call to abort() is a voluntary action by the 
program's code, I doubt that the registers' values would be of great 
importance.  You need the registers to find out which variable had 
invalid value (e.g., a garbled pointer which caused SIGSEGV), but that 
problem is irrelevant for a code which called abort().  A good debugger, 
such as GDB, should be able to find the registers saved on the stack, and 
doing so involves lots of non-trivial code that I don't think should be 
put into a core dumper.  (For an idea what's involved, see the function 
i386_frame_init_saved_regs on i386-tdep.c in the GDB distribution.)

> Again, if you need the sources, I'll be more than happy to send them to you.

I suggest that you upload the code to the /incoming directory on DJ's 
server, so that it could be put in some beta directory on the ftp server 
where people could download it.