> Ok, this is definitely wrong. > > I very strongly suspect that this is part of smsq_atari_hwinit_asm, > which should however go : > > at_mdloop > move.w (a0)+,(a1)+ (NOT moveq #$41,d5) > cmpa.l a2,a1 > blt.s at_mdlopp > rts >
That's exactly what I thought about. > You can check this, hopefully, by checking whether a few bytes before > these instructions you find the string "Use sprite for cursor". > Yep I found that : > disasm $12b40 $012b40 : 6052 bra.s $12b94 $012b42 : 7a4b moveq #$4b,d5 $012b44 : 6002 bra.s $12b48 $012b46 : 7a41 moveq #$41,d5 $012b48 : b3ca cmpa.l a2,a1 $012b4a : 6dfa blt.s $12b46 $012b4c : 4e75 rts $012b4e : 6012 bra.s $12b62 This is right after the string. > If this is really there, then somehow the instructions get corrupted, > making the move.w (a0)+,(a1)+ into a moveq. > > Could you disassemble the code as of the address contained in A6, which > is a few bytes before the loop? > See above. > Is there really no way you can insert a breakpoint into the code which > would call up your debugger? > No. I can set a breakpoint at a address. As I explained in the beginning of this thread it somehow relates to the TOS. The same SMSQ.PRG work with emutos but not with 1.62. I guess I can patch the binary. But if this (the bad moveq ) fails there are probably more places that gets wrong. _______________________________________________ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm