[Regression] Tales of Xillia crashes right before the game loads from the 2/26 update and on

[Daniel-Katz]

Quick summary

New versions of RPCS3 cause Tales of Xillia to crash

Details

I don't know what the issue with the 2/26 update is, but from that version and on (double checked a couple of versions including the latest one as of this moment) Tales of Xillia causes RPCS3 to crash before the game "starts". The pre-game loading bars all do their thing, a black screen shows for a few seconds and then instead of Tales of Xillia starting, RPCS3 just closes. No error message or anything, just closes.

When I reverted to the 2/24 update the game resumed working normally.

For the record, no, I'm not talking about how the game freezes if you try to skip past all the pre-credits. THAT happens even with the working versions.

Build with regression

0.0.30-16146

Attach two log files

I made copies of the 2 versions of the log, named the one that was a log for the 2/24 version that worked "RPCS3 - Good" and the one that was a log for the 2/26 version that crashed "RPCS3 - Bad"

RPCS3.-.Good.log
RPCS3.-.Bad.log

Attach capture files for visual issues

No response

System configuration

No response

Other details

No response

[Megamouse]

@kd-11 I've briefly looked into this and it turns out that image_to_flip2 is nullptr in GLGSRender::flip, which either way leads to a simple_array of size 2 and then images.map(FN...). It then tries to call xform with the dereferenced nullptr in simple_array::map.

So there's 3 potential bugs here:

1. deref nullptr in map
2. using a simple_array of size 2 even though image_to_flip2 is nullptr
3. having a nullptr in image_to_flip2 in the first place

And probably the simple solution being to push a default initialized object if "it" is nullptr, since the run function checks for GL_NONE