Ouai, il a planché sur le problème comme on peut le voir sur #cemu-dev et #ez80-dev. Ce n'est pas encore réglé, et pour autant que je sache, il y a des trucs louches qui se passent.
Quelques extraits récents :
07:21:21 <@MateoC> So, the 2048 bug also causes major glitches in CEmu
07:21:32 <@MateoC> Like, this is the mother of all bugs
07:21:49 <@MateoC> You press the step button
07:21:54 <@MateoC> PC doesn't change
07:22:07 <@MateoC> But the values in HL,BC,DE,and other registers do
07:26:28 <@jacobly> I've seen that before
07:26:38 <@jacobly> that's a bug in the stepping code
07:26:50 <@jacobly> I've seen where the bug is, but not why
07:27:10 <@jacobly> cpu_write_byte looks like it definitely contains a bug
07:27:38 <@jacobly> err
07:27:44 <@jacobly> cpu_read_byte rather
07:28:32 <@jacobly> err, nvm, that is only the step over/step next bugs
07:43:10 <@MateoC> Well, hm
07:43:16 <@MateoC> The calc crashes in
07:43:30 <@MateoC> code_026999
07:43:35 <@MateoC> Which is....
07:43:38 <@MateoC> Something
07:46:37 <@MateoC> It fails rebuilding the VAT?
07:46:38 <@MateoC> idk
07:47:34 <@MateoC> Something is seriously wrong
07:51:53 — Unicorn| hands MateoC power to fix seriously wrong things
07:55:41 <@MateoC> Okay, I did find the root of the issue
07:55:54 — @MateoC It is in _SetArchiveStatus
07:57:24 <@MateoC> UH
07:57:27 <@MateoC> Gr
08:22:49 <@MateoC> I can't even make a test case that causes it
08:22:52 <@MateoC> Hm
08:23:00 <@MateoC> I wonder if TI messed up...
08:23:22 <@MateoC> Maybe the archived 2048 appvar just so happens to cross a page boundry
06:33:14 <@MateoC> Okay, here's a really easy way to do it in CEmu and on hardware:
06:33:25 <@MateoC> 1) Clean reset
06:33:32 <@MateoC> 2) Start 2048
06:33:38 <@MateoC> 3) Press clear
06:33:44 <@MateoC> 4) Start 2048
06:33:50 <@MateoC> 5) Press Clear
06:33:55 <@MateoC> 6) Reset Calc
06:33:59 <@MateoC> And boom, dead
06:37:23 <@DrDnar> Are you still certain it's an issue with the file routines?
06:38:02 <@MateoC> Probably...
06:38:16 <@MateoC> Let me see if removing some of them fixes it
06:41:06 <@MateoC> Yes, if the file is not archived it works fine