[hatari-devel] Maybe a problem into Hatari (from Thomas Bergstrom)

[Laurent Sallafranque <laurent.sallafranque@xxxxxxx>]

Hi,

Thomas Bergstrom sent me what is probably an issue into hatari.
I ask here in case this rings a bell to someone.
Laurent


First mail from him :


Ive stubled accross another issue but I don't know if it's Hatari as I 
havnt been able to test on real HW yet.

I have this code:

    move.b    (a2,save_ccr),ccr
    move.l    (a2,save_state),d1
    move.l    #$FDDE0000,d0
    move.w    #length/8-1,d7
..loop:
    REPT    8
    move.w    (a0,d1.w*2),(a1)+
    addx.w    d0,d1
    ENDR
    dbf    d7,.loop

    move.l    d1,(a2,save_state)
    move.b    ccr,(a2,save_ccr)

For some reason the first two value moved to a1 is always the same, I 
get the feeling that addx doesnt care about the x flag the first time it 
is executed. Like I said, I havn't tested on real hw yet so it could be 
something in my code or something I've missunderstood.


Second mail from him :

I've narrowed the problem down to this:

Doesn't work:

    move.b    (a6,save_ccr),ccr

Works:

    move.b    (a6,save_ccr),d2
    move.b    d2,ccr

I still havn't checked how a real falcon behaves but I think both should 
work.