Crash research

[Nyerguds 100]

umm... if I see the "Select Affinity" option in my task manager and can select the cores I do think I can be pretty sure it's dual core.

 

Not like it matters. I saw it set the game to run on only one, and that was the whole point here.

[Jonttu 0]

Alright, Nyerguds i have a new one.

 

My game froze. It got stuck into the briefing screen when i clicked OK (to launch the game).

The map would be the first funpark mission of the GDI side of the campaign.

________________________________________________________________________________

_______________

Error note:

 

The instruction at 004dadb7 referenced memory at 00000140

The memory could not be read from

________________________________________________________________________________

_______________

 

Haven't been around on cncnet for awhile. Instead I've had a c&c95 singleplayer marathon for about 1 and a half weeks. Just finished my commanding & conquering and went through both of the campaigns and played through every remaining alternative mission choiche in the campaigns that I had not yet played before. Also completed the funpark missions with both NOD & GDI, completed every PSX and N64 special ops missions, and of course every covert ops mission. So in other words I've now finished every single singleplayer map in c&c95, covert ops and ps1 and n64 versions of c&c95.

And never had to resort to cheating nor sandbagging. Sandbagging as in intentionally exploiting the AI's inability to destroy those walls when blocking the ai-opponent's movement INTO certain areas, or inflitrating enemy bases etc. Never had to resort to either of the two, not in Deceit or Cloak and Dagger nor any other map. Speed was second fastest (at least) always.

Also I managed to beat Hell's Fury without the chinook. Yea it's possible just had to come up with a real strategy for maybe the first time instead of just massing something, countering something or microing.

 

EDIT: Also did you fix it or do you know anything about the error dappygit got? the dade1 error? Shouldn't u update the original post and put it in there?

[Nyerguds 100]

Alright, Nyerguds i have a new one.

 

My game froze. It got stuck into the briefing screen when i clicked OK (to launch the game).

The map would be the first funpark mission of the GDI side of the campaign.

________________________________________________________________________________

_______________

Error note:

 

The instruction at 004dadb7 referenced memory at 00000140

The memory could not be read from

Hmm, so then that's after loading the mission, when actually displaying it. Odd. The error is in the same function as the 004dadde / 004dade1, which is a part of the memory freeing function _nfree_. The problem is that crashes in this function can come from anywhere; this is nothing but a memory freeing function. The problem is usually that the object reference it's supposed to free is already gone or corrupted, and with the information from the error message there's no way to find out where the _nfree_ itself was called.

 

 

Haven't been around on cncnet for awhile. Instead I've had a c&c95 singleplayer marathon for about 1 and a half weeks. Just finished my commanding & conquering and went through both of the campaigns and played through every remaining alternative mission choiche in the campaigns that I had not yet played before. Also completed the funpark missions with both NOD & GDI, completed every PSX and N64 special ops missions, and of course every covert ops mission. So in other words I've now finished every single singleplayer map in c&c95, covert ops and ps1 and n64 versions of c&c95.

And never had to resort to cheating nor sandbagging. Sandbagging as in intentionally exploiting the AI's inability to destroy those walls when blocking the ai-opponent's movement INTO certain areas, or inflitrating enemy bases etc. Never had to resort to either of the two, not in Deceit or Cloak and Dagger nor any other map. Speed was second fastest (at least) always.

Also I managed to beat Hell's Fury without the chinook. Yea it's possible just had to come up with a real strategy for maybe the first time instead of just massing something, countering something or microing.

lol, neat. Hell's Fury without the chinook is kinda silly though, since that chinook maneuvre is what the mission programming is all about. Still, nicely done.

 

I once beat Nod Death Squad without losing any units. Can you match that?

 

EDIT: Also did you fix it or do you know anything about the error dappygit got? the dade1 error? Shouldn't u update the original post and put it in there?

As I just explained, it's pretty much impossible to trace where these memory freeing errors come from unless you're actually playing with a debugger loaded.

[Jonttu 0]

Oh ok i see.

Sorry for the late reply.

 

And yea Nod death squad is pretty easy.

Just sneak the stealthtanks and the commando past the southeast corner, kill the enemy commando with yours since he's gonna follow you, destroy a few pieces of concrete wall and a sandbag, move in with stealth tanks, move in with commando and blow up the adv comm while distracting the adv guard tower with stealthtank fire. The probability of finishing the mission with no losses is fairly high.

 

It's also possible to destroy all enemy units and buildings too. Just have to make sure when sniping infantry or shelling vehicles that they're not standing next to a building. However they can be lured away from the building and after you by 'exploiting' the splash when firing next to the unit, the same way enemy units would be alarmed when you inflict splash damage to buildings. Just set up an ambush and take everything out one by one. I think neutralizing the mlrs's also alarmed the enemy, that's why they should be taken out later on.

[pichorra 4]

If you need someone to debug the game to get errors, i'm here

 

Anyway, what's the "recommended" way to debug it? Win32Dasm have a debugger, but it is pretty simple, there's also the MS-Debug tools. Can IDA debug running applications?

[Nyerguds 100]

Can IDA debug running applications?

Yes, obviously. It either starts the program itself, or can be attached to a running program (if it corresponds to the opened database anyway).

[Chimas 7]

Crash Report

 

playing SINGLE Player mission: Big Land (by Kohlrus)

 

IA - 004dbc37 @ 0001917f

 

Events prior to the crash:

In this mission, there's a "vertical" NOD base at NORTHEAST

I forced SAMs to expose its positions by making them shoot at my ORCAS

They flew over the "border" of their air space, from my base at SE to NW.

The 1st SAM shoot, but right after the second sent the missiles (in the middle of the base), I got the crash.

Might be important to know that these wasn't the first SAMs to counterattack within this mission.

Rather, it was problably near the end of the SAMs.

 

Info:

Windows XP 2002 SP3

Intel Pentium 4

CPU 3,00GHz

0,99GB RAM

Intel 82945G Express Chipset Family

 

Another important thing to comment, I've been taking a lot of crashes recently in SP missions and the only different recent event was

downloading the netcode version of C&C95.EXE at www.cnc-comm.com

 

I'll start to take notes from now on, whenever it happens.

Edited May 7, 2011 by Solo

[Nyerguds 100]

The crash address is in a function that's used in the mouse drawing/hiding/showing functions. You probably don't know exactly what your mouse was pointing at when the crash occurred?

[pichorra 4]

The crash address is in a function that's used in the mouse drawing/hiding/showing functions. You probably don't know exactly what your mouse was pointing at when the crash occurred?

 

I've got a wreid bug when i was playing in my Old computer (the Cyrix one), i was scrolling for the North, then the Scrolling cursor stuck in the "Option" button and the Game crashed... When i was trying to get the error code, Meh, the Windows 98 frozen, a BSOD shows and i restarted my computer

 

crappy 9x systems

[Nyerguds 100]

lol, kay.

[Chimas 7]

Crash Report #02

 

playing SINGLE Player mission: A long War (by Kohlrus)

 

IA 00465ec2 @ 0ee07000

 

Events prior to the crash:

After an hour playing this mission, I sold a comm-center and then the game crashed.

 

Info:

Windows XP 2002 SP3

Intel Pentium 4

CPU 3,00GHz

0,99GB RAM

Intel 82945G Express Chipset Family

 

[Nyerguds 100]

Hmm, I got that marked as reported before, but no idea what causes it. The function is "load_into_buffer", and the spot is marked as "sometimes crashes because address in esi is wrong?"

...which is very odd, by the way, because esi is the game's central buffer for loading files into memory, and that buffer is created at one spot at the game startup, and that spot is never changed after that.

 

Anyway, it's related to SHP drawing, so I would assume it's about drawing the radar logo frames after radar shutdown; most likely the call at address 0049531C.

 

About the mission... which side do you play, and do you have that radar from the start? I can imagine this happens if you play with a multi-team which has no radar logo.

 

If that's the problem though, it's not hard to fix; I'll just have to add a blank radar logo to use for the multi-teams.

[Kamikave 0]

The game crashed yesterday, but I forgot to note the error message ^^' (I'm useless, I know).

It happened when I clicked on the 'Save' button, in the middle of a covert ops mission (Blindsided). After restarting from an older savegame, it went well.

Since it's the first time it happened (I went through both campaigns and almost every covert ops missions) I guess it was just a random error.

Too bad there is no log for these crash. I was playing with rev 1, by the way.

[Nyerguds 100]

Eh, so did it crash when you clicked to open the list to save, or when actually saving the game?

[Kamikave 0]

When saving. But it didn't save anything. However, after restart, when I tried again, on the same savegame, it worked.

[Nyerguds 100]

So you tried overwriting an existing savegame, and it was unchanged when you tried loading it?

 

I see. Probably the saving preparation stuff then, before actually writing anything. Still, not much I can do without the actual address.

 

Thanks for telling, anyway, could be useful information if it ever pops up again.

[Kamikave 0]

Ok, bad news is, the "saving" bug happened again, so I guess it's not that random anymore :/

Good news is, this time I've something for you to work with.

 

Here is the message :

 

"Application Error C&C95.exe

The instruction at 0045af8a referenced memory at 00000008

The memory could not be read from

 

Click ok to terminate the application"

 

 

Just when the fun began, it crashed when I clicked on "Save", like last time T_T

 

My specs, just in case :

motherboard Gigabyte GA-890FXA-UD5

AMD Phenom II X6 1100T

ATI 6970 2Go

8 Go RAM

Win7 64 pro

SSD hard disk for Windows

Edited June 17, 2011 by Kamikave

[Nyerguds 100]

It's indeed in Heap_Save_MouseClass, the first called function in a long list of items that get saved into the save game. I think it saves the actual map data, ,but I'm not sure, since it's apparently "mouse"class.

It's really hard to figure out what exactly is wrong though, since it's an object stored in an object stored in an array stored in an object stored in the map object.

 

I've narrowed it down to this, with 'i' being a counter going over all map cells (0 to 4096)

[[{[MouseClass Map]+21h*i}+0Ch]+8] = null

 

Basically, this means some property of some map cell has a null pointer. I know exactly which property gives the null error (the +8 bytes property of the cell's +12 bytes property), but seeing as I don't know the structure of the map cell objects, I have no idea what it actually is

 

 

Is it in one certain mission, or loaded from a certain savegame? That kind of info is important. If I can't reproduce it while debugging I usually can't fix it. And I think this is more likely due to a corruption in a previous savegame or on a map than an actual problem in the saving logic.

[Kamikave 0]

Neither. The mission was different (though it still was a covert ops mission), and so was the savegame. Although I only use 3 savegames, that I overwrite one after another during a game. I never create new ones. Maybe it has to do with multiple overwrites...

And no, it wasn't loaded from a savegame, it was a new game.

Edited June 19, 2011 by Kamikave

[Nyerguds 100]

No, save games are 100% rewritten when you overwrite them, so that can't cause stuff like that.

 

It's possible this is another game object that somehow got corupted internally but only crashes when saved though... I've had that before, with a tree. There was a corruption of a terrain cell then too, but I fixed it right away, and it never caused problems, so it shouldn't be that one.

 

Can you tell me exactly which missions it happened on?

[Kamikave 0]

Let's see... Blindsided and Twist of Fate, the two last GDI missions of covert ops.

[Chimas 7]

 

Crash Report

 

playing SINGLE Player mission: The fanatics (by SID) - see attached

 

IA - 004b411e @ 00000014

 

Events prior to the crash:

None, crashed on start up.

Must be a BIN or INI problem.

Tried to open in XCC and couldn't,

even when I've erased almost all the INI file leaving just the [MAP] section

XCC crashed anyway. So, I think the problem is the MAP.

 

Info:

Windows XP 2002 SP3

Intel Pentium 4

CPU 3,00GHz

0,99GB RAM

Intel 82945G Express Chipset Family

SCG180EA.zip

[Sonarpulse 1]

No, save games are 100% rewritten when you overwrite them, so that can't cause stuff like that.

I didn't read the whole thread so this is probably been said, but with UAC on the saves will go elsewhere, and Windows doesn't always swap out the files correctly when you try to load them again.

[Nyerguds 100]

umm... saves will never go "somewhere else". If the game tries saving files in its own folder but can't, it'll most likely freeze, stuck in an endless loop waiting for permission to write there.

 

The obvious solution is to take ownership of the folder after installing.

 

 

Solo: Actually, that crash address is in the scanning of a trigger string.

 

More specifically, it seems to be the trigger of a Terrain object. Not sure if those are ever used, but that's where it crashes anyway. I can't see anything wrong on the ini info it crashes on though... really odd.

 

[edit]

 

Oh, ouch. It appears that C&C has a maximum of 300 terrain objects it can put on a map. Any more, and it doesn't create an object, and then crashes when it tries to set the associated trigger in the non-existent object. I found the function that has the problem (at 004B3928), but such limits seem odd in C&C, so I'm not sure if it's intended or an error.

 

Anyway, if you remove everything after "264=" under the [Terrain] header it works.

 

Could you stress test that when you got some time, solo?

 

 

Btw, I haven't been able to open the map in XCC; probably desert remains outside the borders of the visible map; that happens a lot, especially since ccmap never shows those outer borders for editing. I'll check it out when I get home. Can't test it here though; easiest way to see that is by using Mix Manager to view the map, and I can't run that on this 64-bit Windows. Not without installing DOSBox anyway.

[Sonarpulse 1]

umm... saves will never go "somewhere else". If the game tries saving files in its own folder but can't, it'll most likely freeze, stuck in an endless loop waiting for permission to write there.

Nah I just tried it. Windows clones the files system as needed in %SystemDrive%\Users\%username%\AppData\Local\VirtualStore , and then tries to silently merge the files here when called.