well alderon, i'm afraid i have good news and bad news pertaining to this
problem... the good news is that i know exactly the problem that you're talking
about, i experienced it myself when i wiped all areas in my mud to start from
scratch, and one of my imms had a container on them that no longer existed...
the mud would hang when they tried to login, and it was from that container
object sending the mud into an infinite loop... the bad news is that i was
never able to find the fix for it, and the easiest way to fix it was just to
delete the trouble pfile and let him recreate, and hopefully not have to delete
any more areas :) i can point you to where i left off at in trying to
troubleshoot this, though... and you can maybe find it in the archives by now,
i'm not sure how often the archives are updated, but i'd imagine it's in there
by now... go do a search for corrupt rgObjNest, because that's where i was
tracking the problem to... for whatever reason, my code was doing this, which i
didn't realize until i stepped through it in gdb...
it would attempt to get_obj_index and fail because the index data didn't exist,
so it would then load a dummy object to start creating the object out of...
it would set rgObjNest[iNest] = obj just in case there were other items that
were nested inside of it
it would load all data it had about the obj, set strings, item type, etc., and
increment to the next obj in the pfile, which happened to be a content of it...
the content had a higher nest value, so the code would go to set the
rgObjNest[iNest] = obj, only for some reason, it would be remembering "obj" as
the container from the previous loop iteration, instead of the current obj, so
it would then log (in real index values) rgObjNest[0] and [1] both as obj, and
later on it would go into an infinite loop basically trying to execute the
equivalent of the following:
for ( obj2 = obj->contents; obj != NULL; obj = obj->contents )
because it would see itself as the first content item and try to go to the next
one over and over and over, each time just reassigning back to itself... i was
told to check that my "obj" variable was initialized, and that nothing was
trampling anything else's place in memory, and everything checked out on my
end... this seems to be a stock bug now that i hear you mention the same
problem under the same circumstances (container object types of deleted
items)... hopefully someone can help come up with a fix for this, because i'd
love to go apply it to my own code in case i ever have to delete something
again :) but you can start in the archives by looking for rgobjnest
wavewave
Richard Lindsey.
-----Original Message-----
From: Alderon [mailto:[EMAIL PROTECTED]
Sent: Mon 10/18/2004 2:02 PM
To: [email protected]
Cc:
Subject: Character Loading Problem
Ok,
I've recently subscribed to the list, and I did some searching before
posting this. In all reality, I could just be blind. I'll ask anyway:
I've decided to clean up my area list. Mostly to get rid of overpowered
eq
that previous immortals have created.
When I remove areas, and there are players who have eq that is from that
area, the mud would freeze until I could
kill the process and reboot. It would throw out the "Fread_obj: bad vnum
%d." error in the logs. I have fixed this
problem when it comes to individual pieces of eq by editing my save.c to
look like:
if (get_obj_index(vnum) == NULL) { /*
Changed to fix crash when player
logged in with */
bug("Fread_obj: bad vnum %d.", vnum);
/* non-existant eq. -Alderon */
obj = new_obj();
obj->name = str_dup("");
obj->short_descr = str_dup("");
obj->description = str_dup("");
} else {
obj = create_object(get_obj_index(vnum), -1);
new_format = TRUE;
}
The problem is that if the piece of eq is a container (like a backpack)
that
is full of eq. The mud will freeze with the
"Fread_obj: incomplete object." error message in the logs. From what I
can
gather, the mud is removing the illegal
container, but when it comes to an object that is "nested" in the
container
it has no clue what to do and just stops
doing everything. The piece of code is:
if (!str_cmp(word, "End")) {
// if (!fNest || !fVnum || obj->pIndexData == NULL) {
if ( !fNest || ( fVnum && obj->pIndexData == NULL ) ){
bug("Fread_obj: incomplete object.", 0);
free_obj(obj);
return;
} else
{
if ( !fVnum )
{
free_obj( obj );
obj = create_object( get_obj_index(
OBJ_VNUM_DUMMY ), 0
);
}
I'm sorry that the note is too long...I'll try to do better on my next
posting.
Thank you,
Alderon
AnimeMUD 2 anime2.ipupdater.net:7000
--
ROM mailing list
[email protected]
http://www.rom.org/cgi-bin/mailman/listinfo/rom
