Xref: news2.mv.net comp.os.linux.development.apps:21130 comp.os.msdos.djgpp:7551
Newsgroups: comp.os.msdos.djgpp,comp.os.linux.development.apps
From: dolittle AT weniac DOT math DOT toronto DOT edu (Ed Doolittle)
Subject: Linux libgrx 1.0.4, g++ segfaults solved
Message-ID: <slrn517la9.1a7.dolittle@weniac.math.toronto.edu>
Sender: news AT utcc DOT utoronto DOT ca (News)
Reply-To: dolittle AT math DOT toronto DOT edu
Organization: UTCC Campus Access
Date: Fri, 16 Aug 1996 01:52:03 GMT
Lines: 63
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

I posted previously on the subject of segfaults in g++ programs linked
with libgrx under Linux. It took me quite a while to track the problem
down, but thanks to the advice of a number of kind people on these
newsgroups I have found what seems to be a number of errors in Linux
libgrx 1.0.4.

In order to do the job, I had to do the following.

	* recompile the library with debugging information
	* link the applications with ElectricFence
	* run them under gdb on a second terminal attached
	  to my computer with standard error redirected to the
	  currently displayed virtual console, e.g.,

		gdb testshape 2> /dev/tty1

The latter step was necessary because variables and pointers would
get corrupted when gdb forced the console back into text mode after a
segfault, rendering the backtrace information almost worthless.

Three off-by-one errors turned up in lnxmode.c. I suspect there may be
more, but someone who knows more about the code ought to have a look at
it to determine whether the errors I have identified really are errors,
and check that I have fixed them properly, and make sure that there are
no others. I'm all set up to debug the library now :) so if someone
wants to send me code to test, I would be more than happy.

Ed

--- lnxmode.c.orig	Thu Aug 15 20:19:38 1996
+++ lnxmode.c	Thu Aug 15 20:21:48 1996
@@ -199,7 +199,7 @@
         return;
 
 
-    for ( i = start; i <= end; i++)
+    for ( i = start; i < end; i++)
     {
         if (Dirty[i].flag == TRUE)
         {
@@ -234,7 +234,7 @@
     int i;
   
     if (Accel == TRUE) return;
-    for (i = 0; i <= _GrScreenY; i++)
+    for (i = 0; i < _GrScreenY; i++)
     {
         Dirty[i].flag =TRUE;
         Dirty[i].starting = 0;
@@ -425,7 +425,7 @@
             info = info_tmp;
         }
     }
-    memset(&GM[modes+1], 0, sizeof(GR_DRIVER_MODE_ENTRY));
+    memset(&GM[modes], 0, sizeof(GR_DRIVER_MODE_ENTRY));
 
     /* Third sort table */
     for (i=modes-1; i>0; --i)

--
Ed Doolittle <dolittle AT math DOT toronto DOT edu>
 "Being on time is like being on fire."