[#5952] X4 9.00 build 611726 – reproducible crash at approximately 95% while loading a save

[lordadb]

Hi guys,

the attached original quicksave consistently crashes at approximately 95% during loading. Older saves from the same game load correctly.
The gzip archive passes integrity testing and the XML is well formed. Steam file verification completed successfully.

Two edited copies were also tested:

• one with the temporary playerbomb shoot controller removed;
• one with that controller removed and the player spacesuit relocated approximately 5 km away with zero velocity.

Both edited saves produce exactly the same crash signature as the original:

Exception: 0xC0000005
Thread: Movement worker
Address: X4.exe+0x7805EC
Failed read from address 0x38
RAX = 0

The player was in a spacesuit near a microchip production station when the original quicksave was created, preparing to deploy an EMP.

Attached: original quicksave and both crash dumps.

https://www.dropbox.com/scl/fi/hgjpbqwr ... dnb6r&dl=0
https://www.dropbox.com/scl/fi/sc97st62 ... jfwqd&dl=0
https://www.dropbox.com/scl/fi/6wv1os2z ... hn3jc&dl=0

Details:
Platform: Steam
Operating system: Windows 10
Game version: 9.00
Third-party mods: None
Official extensions/DLC: Enabled
Steam file verification: Completed successfully
Steam Cloud: Disabled after the problem occurred
Older save loads correctly
Crash point: approximately 95%
Original situation: player in spacesuit near a microchip production module, preparing to deploy an EMP

This save represents several hours of progress and is currently my only recent save. I would therefore greatly appreciate any attempt to recover it or confirm whether it may become loadable after a fix. For the moment, I will avoid continuing the game from the older save while waiting for your assessment.

Thank you in advance.

[CBJ]

Could you provide your DXDiag please, as requested in the forum rules? It may be relevant.

[lordadb]

Thanks. I am currently away from the laptop on which the crash occurred, so unfortunately I will not be able to provide its DXDiag for a few days. I will upload it as soon as I return.
In the meantime, I also intend to test the same original, unmodified quicksave on a second and considerably more powerful PC, using the same X4 version and the same official extensions. If the crash is reproduced at approximately 95% on that system as well, I will report the result and provide its crash dump and DXDiag as additional comparison data.
The DXDiag from the original laptop will follow as soon as I have access to it.

[lordadb]

Here are the requested diagnostic files from the laptop:

DXDiag:
https://www.dropbox.com/scl/fi/s5ey4fqm ... zuifo&dl=0
VulkanInfo:
https://www.dropbox.com/scl/fi/ctxxe4ka ... ptj40&dl=0

Please let me know if any additional information is required.

[CBJ]

That laptop doesn't meet the minimum requirements for the game. It doesn't have a proper dedicated graphics device, and the device it does have shares system memory. If it's worked in the past then it's more by luck than anything else, and it's quite possible that the savegame you're trying it with now simply has too much going on for it to handle. If the savegame works on your other PC then that's almost certainly what's happening.

[lordadb]

I have now reproduced the crash using the same original quicksave on a second PC:

AMD Ryzen 7 5800X
AMD Radeon RX 580 2048SP with approximately 8 GB of dedicated VRAM
16 GB RAM
Windows 11 Pro

VulkanInfo identifies the GPU as PHYSICAL_DEVICE_TYPE_DISCRETE_GPU, and DxDiag reports no display-device problems.
The save crashes reproducibly at approximately 85% on this system as well, while an older save from the same game loads correctly.
I have attached the second PC’s DXDiag and VulkanInfo. I will also attach the crash dump generated on this system.
I have noticed that the OBS Vulkan hook was present during this test, so I will repeat the loading attempt once more with OBS completely closed. However, OBS was not present on the original laptop, where the same save also crashes.
Since the same individual save fails on two substantially different systems, this appears to be related to the saved game state or its loading process rather than solely to the original laptop’s integrated graphics.

I have now analysed the crash dump generated on the second PC.
The crash signature is identical to the one produced on the original laptop:

Exception: 0xC0000005 read access violation
Thread: Movement worker
Address: X4.exe+0x7805EC
Instruction attempts to read [RAX+0x38]
RAX = 0, resulting in an attempted read from address 0x38

The second PC uses a Ryzen 7 5800X and a discrete Radeon RX 580 2048SP with approximately 8 GB of dedicated VRAM. VulkanInfo identifies it as PHYSICAL_DEVICE_TYPE_DISCRETE_GPU, and DxDiag reports no display-device problems.
The same original save therefore produces the same null-pointer crash inside X4.exe on two different computers. An older save from the same game loads correctly on both systems.
This appears to rule out the original laptop’s lack of a dedicated GPU as the cause of this particular save-loading failure.

DxDiag: https://www.dropbox.com/scl/fi/w4dzqdow ... 20lq7&dl=0
VulkanInfo: https://www.dropbox.com/scl/fi/mjuw289t ... k41xu&dl=0
Dump file: https://www.dropbox.com/scl/fi/nrsu9oig ... lug7n&dl=0

[Alan Phipps]

Hi, the AMD graphics driver (and hence Vulkan version) on that second PC is over a year old. You also have several sound drivers going at once and one is over 4 years old.

In your Windows Error Reports on that PC, the first two AppCrashes are for bad_module_info. Microsoft has this to say about it:

"The "bad_module_info" error usually pops up when running high-intensity applications, indicating that the system's physical memory or video memory is insufficient. During shutdown, the system tries to close all running processes and release their resources. However, some programs might still be holding onto memory and video memory, causing this error when the system can't release those resources properly."

I downloaded your save and loaded it in my vanilla 9.00 game and it crashed X4 (CTD) at 95% loaded so I would say that the save is incomplete or corrupted. Are your other good saves also around 48 MB in size? (My mid-game saves are about 90 MB.)

[lordadb]

Thank you. I understand the concerns regarding the drivers and the system configuration, and I can carry out the suggested updates and repeat the test.

However, since the exact same original save crashes reproducibly on two different computers while the older saves load correctly, would it be possible for someone at Egosoft to try loading the save on a known-good system that fully meets the requirements?

That test would establish immediately whether the crash is system-specific or reproducible from the saved game itself. If it loads correctly on your system, I will know that further troubleshooting of my PCs is necessary. If it also crashes there, the save and dump should provide a reproducible test case.

[Alan Phipps]

As above, I just did that and it crashed.

[lordadb]

Thank you for testing it. That confirms the crash is reproducible on a supported vanilla system.

My other working saves from the same campaign are also approximately 46-48 MB, so the 48 MB size does not appear unusual for this particular game.
The quicksave also passes a gzip integrity test and the XML is well formed, so it may be logically inconsistent rather than physically truncated. I can upload one of the working saves from the same campaign for comparison if that would help.

Is there any possibility of identifying the invalid object/state or repairing the save?

[Alan Phipps]

The devs have given this a tracking number and (if they can get round to it amongst their other work and priorities) they will try to find out what went wrong in the save. Unfortunately the save cannot realistically be restored for gameplay purposes and so I would suggest that you forget that one and play on from an earlier good one.

That said, please post here again if the problem reoccurs (on the PC at least rather than on a struggling laptop which might well have played a part in the save issue arising).