Falling through surface bug can cause all saves in a campaign to be unloadable (with work-around to fix)

[Steve7893]

Over in another thread, a discussion arose about an issue I was having, and I was asked to repost this here in the bugs thread.

My goal in testing was to do a redezvous mission.   Land 5 kerbals on Minmus in a single mission, and then join them with another 5 kerbals in a subsequent mission.  One hundred percent of the time, when you get close enough to the landed ship, close enough for the mechanics to engage, the land ship "falls through the surface of the world" on a straight line trajectory towards the center of that planet.   Once it reaches the center, it insta-translates somewhere along that line with an extremely high rate of speed, something like 56,000 m/s or so.

This part you all know already as it is posted in another thread.   Here is the extra part:

If you quickly go to the tracking station after the fall-through, you can "Control" the wayward fall-through ship.   Do that, and then gradually accelerate timewarp for a reasonable long time.   Long enough to far leave the kerbol system.    Choose "Exit Game and Save"

When you reload, none of the saves in that campaign will load.   Here is a cut-paste from the other thread:

I just got around to recreating the campaign that won't load.   Tried your suggestion after seeing the list of saves.   Sadly none of the saves will load.

Using Microsoft Procmon, focused on KSP2_x64.exe, the load appears to be stuck in an endless loop reading the file at:

AppData\LocalLow\Intercept Games\Kerbal Space Program 2\backtrace\bt-breadcrumbs-0

over and over from start to end of file over and over.   So I do get a chance to load another save from the campaign, but it doesn't matter.   Even attempting to load the saves near the bottom of that list results the same.

task manager show 7 gigs of free memory with 8.7 in use.  KSP2 is using 2.3 gig.  Percent CPU usage is only 7%.   The reading of that file is only taking 0 to 1% of the SSD's throughput.

Using the Analyze Wait Chain function indicates it is running normally, not blocked on anything.

Rebooting does not affect it.

New campaign works normally.

System is a 20 core I9 with 32 gigs overall.  GeForce RTX graphics.  OS is Windows 10 Pro.

 

At this point a fellow gamer offered to "take a look" at the save files.   At that time I had some difficulty understanding what he meant because I had never looked over there before.

Actually, while I was waiting to hear from you I was able to 

1. Zip up the files at AppData\LocalLow\Intercept Games in the original broken state in case you wanted it.

2. Rename the folder.

3. Start the game, reaccept EULA, save and exit.

4. Using Beyond Compare on the two folders, copy files, except anything in the backtrace folder or either of the Player logs.

Then upon starting the game, I noticed that the top 3 saves now had a tag "VESSEL UNAVAILABLE".   Skipping those 3, the next one loaded fine, and in fact all ships in orbit are still in orbit, all kerbals are alive and well, and the game is running smoothly again.

Let me know if you want the original zip file I made, its only about 6.5 meg or so.

And then later I realized this:

It would be interesting to know what the vessel was that was unavailable.   Reloading the save that worked, all of the vessels are accounted for.   There were no others.

oh gee, yes I know.   The unavailable vessel was the vessel that disappeared that I used to create the situation in the first place.   The game somehow still had a memory of it even if I could not see it anymore.

Please understand that I am a Principle Developer at a large software company, and so I have some experience with Procmon, Windbg and so forth.   Just nothing specifically KSP2 to work with, so in the dark pooking around and guessing.   But I am posting this for what it might be worth to you.   The "wayward" vessel had eventually disappeared from anything I could detect on the UI, once it was far away and I transited into the Tracking Station.   But that vessel was still remembered and the saves somehow caused that bt_bitcrumbs loop to engage regardless of the save I choose until I recopied the whole folder tree leaving that part out.