#67 RZX desync

[Fredrick Meunier]

Originally reported by jweatherley as bug [ 1653344 ] on Fuse for Mac OS X:
Down load the Auf Wiedersehen Monty rzx from the rzx archive:
http://www.rzxarchive.co.uk/download.aspx?id=11&dl=672&name=aufmonty.rzx

It start off fine, but five to ten minutes in the recording stops with the following error:

Fuse - Error
libspectrum: libspectrum_rzx_playback_frame: wrong number of INs in frame 69648: expected 8, got 0

It always stops in the same place, it's during the flight out of Sweden. I don't know if it works in the non-OS X version or not.

[Philip Kendall]

Logged In: YES
user_id=29214
Originator: NO

Using this bug to merge the three open bugs on this topic. See also #995895 (Head over Heels) and #1383664 (Dark Side, Lightforce, Cauldron II).

To that list, I've also found:

* Abu Simbel Profanation: wrong number of INs in frame 13704: expected 8, got 0 (and then segfaults. See next bug)
* Action Farce II: wrong number of INs in frame 0: expected 16, got 8
* Airwolf: wrong number of INs in frame 5581: expected 17, got 9
* Avalon: wrong number of INs in frame 135835: expected 6, got 0

[Fredrick Meunier]

Logged In: YES
user_id=11017
Originator: YES

I tried the Auf Wiedersehen Monty rzx on a few Windows emulators and got the following results:
Spectaculator 5.3.0371 plays till end
RealSpectrum R32 0.97.26 crashes during flight
Spin 0.61 immediate error
Spin 0.504 error in frame 3360

Not sure what to make of it TBH.

[Philip Kendall]

Logged In: YES
user_id=29214
Originator: NO

SPIN 0.61 is known to be completely screwed with respect to RZX playback, so there's nothing to be gained from that datum.

[Philip Kendall]

Logged In: YES
user_id=29214
Originator: NO

Action Farce II fixed: the snap->beta_paged field wasn't being initalised when loading a .z80 file, so we could incorrectly page in the Betadisk ROM.

[Philip Kendall]

Logged In: YES
user_id=29214
Originator: NO

Auf Wiedersehen Monty: what happens here is that the last instruction at the end of frame 69647 is "EI". Interrupts will not be accepted after an EI, so it should occur after the next instruction, which is "HALT", and a 1 instruction frame should be written to the RZX file. However, the RZX was created by a buggy version of SPIN which did not correctly implement the RZX spec with respect to retriggered interrupts and we get a "full" frame in the file. I don't know what Spectaculator's doing to work around this, but whatever it is, Fuse is doing the correct thing, so I'm not going to regard this as critical for 0.8.

[Fredrick Meunier]

Logged In: YES
user_id=11017
Originator: YES

Extra information on Airwolf.rzx, the windows emulators give the following results:
Spin 0.622: Plays with no errors
Spectaculator 5.30.0371 (I don't have 6.x): Plays with no errors
RS32 0.97.26: Aborts with "RZX ERROR: FRAME 1617: IN=8 EXPECTED=17"