lynx-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: lynx-dev possible bug in lynx 2.8.1 -- GridText.C


From: Klaus Weide
Subject: Re: lynx-dev possible bug in lynx 2.8.1 -- GridText.C
Date: Tue, 29 Dec 1998 13:48:17 -0600 (CST)

On Tue, 29 Dec 1998, Seth Theriault wrote:

> I have someone here who is a heavy Lynx user -- really heavy. Ever since
> our upgrade to 2.8.1rel.2, he has been experiencing frequent, random
> crashes. The crashes follow no particular pattern or specific URL.
> 
> I have traced the problem to our friend GridText.C, in particular line
> 1154. Lynx sometimes aborts with the error:
> 
> ./GridText.c: 1154 failed assertion 'line !=NULL'
> 
> while it often gives the "Fatal error detected; now exiting" message.
> 
> I checked out GridText.C and it seems that scrollback handling is once
> again an issue. Here is the piece of offending code:
[ ... ] 
> 1145        if (line) {
> 1146            char *data;
> 1147            int offset, HitOffset, LenNeeded;
> 1148            for (i = 0; i < (display_lines); i++)  {
> 1149                /*
> 1151                 */
> 1152    #ifndef VMS
> 1153                if (!LYNoCore) {
> 1154                    assert(line != NULL);
> 1155                } else if (line == NULL) {
> 1156                    if (enable_scrollback) {
> 1157                        addch('*');
> 1158                        refresh();
> 1159                        clear();
> 1160                    }
> 1161                    addstr("\n\nError drawing page!\nBad HText 
> structure!\n" );
> 1162                    refresh();
> 1163                    nlinks = 0;  /* set number of links to 0 */
> 1164                    return;
> 1165                }
[ ... ]
> System info: Digital Unix 4.0b, lynx 2.8.1rel.2 (compiled with slang
> libs 1.2.2, and default-colors enabled)

I see no indication that this has something to do with enable_scrollback,
other than that it's in the same general vicinity.

I think I have never seen that assert fail.  (What happpens if -core or
NO_FORCED_CORE_DUMP is used?)

Since you say lynx often aborts differently, there may be some memory
corruption that happens somewhere else.

Does "really heavy" mean that the process may just ask for too much
memory?  (Are you using ulimit or similar?)  [Although if that's the
problem the message should normally indicate it.]

If this is a 64 bit architecture, could that create a problem?

I suggest you try whether applying the patches in

        URL: ftp://www.slcc.edu/pub/lynx/release2-8-1/patches/

or using the development code

        URL: ftp://www.slcc.edu/pub/lynx/current/

makes a difference.


   Klaus

reply via email to

[Prev in Thread] Current Thread [Next in Thread]