Re: [hatari-devel] Hatari reboots with "Love vs darkness"

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

On maanantai 26 tammikuu 2015, Laurent Sallafranque wrote:
> I've tested with the 1.8 version of hatari (newcpu core) and it doesn't
> reboot.

WinUAE version from end of November, just before Nicolas updated to
newer WinUAE core works fine.


> It runs all the intro and then stops

Above version doesn't stop there.  Some graphics flicker badly though,
I guess this is Videl emulation issue.  It bus errors some time later in 
demo.


> WARNING: have to clip 13 chars from 'main.binend_part.bin' extension!
> WARNING: have to clip 13 chars from 'main.binend_part.bin' extension!
> M68000 Bus Error reading at address $fffe4f.
> Exception 2 (c8322) at c832a -> b569e!
> Fixed by ' extension! 13 chars from 'end_part.bin
> Fixed by ' extension! 13 chars from 'end_part.bin
> M68000 Bus Error writing at address $0.
> Exception 2 (84938) at 8493a -> 84702!
> 
> I've traced a little the reboot in the latest version of hatari, it
> seems that the code detects something and then jumps to $e00000, where
> the 4th instruction is reset (which explains the reboot).


		- Eero

PS. If you want to set breakpoint on specific instructions,
here's little program that tells you suitable Hatari
breakpoint for given (68000) instruction:
http://koti.mbnet.fi/tammat/hatari/sources/instructions.py


> Laurent
> 
> Le 25/01/2015 23:50, Laurent Sallafranque a écrit :
> > Same behaviour with MMU emulation.
> > 
> > Le 25/01/2015 23:47, Laurent Sallafranque a écrit :
> >> In complement, when it breaks at address $e02cde, I've dumped the
> >> next instructions.
> >> Then I do "c 1" to execute the next instruction, and it seems to me
> >> that there's one instruction "jumped" :
> >> 
> >> c 1 executes the       tst.b     $ffff8e09.w
> >> 
> >> Then the PC is at address $e02ce6, (the move.w    #3,d0  is not
> >> executed).
> >> It seems that this address $ffff8e09.w generates a problem.
> >> 
> >> I'm going to test with the mmu to be sure.
> >> 
> >> CPU=$e02cde, VBL=7710, FrameCycles=106936, HBL=210, LineCycles=256,
> >> DSP=$51
> >> $00e02cde : 4a38 8e09                          tst.b $ffff8e09.w
> >> 
> >> > d
> >> 
> >> $00e02cde : 4a38 8e09                          tst.b $ffff8e09.w
> >> $00e02ce2 : 303c 0003                          move.w    #3,d0
> >> $00e02ce6 : 2e49                               movea.l   a1,sp
> >> $00e02ce8 : 21ca 0008                          move.l a2,$0008.w
> >> $00e02cec : 31c0 11ca                          move.w d0,$11ca.w
> >> $00e02cf0 : 43fa fef8                          lea $e02bea(pc),a1
> >> $00e02cf4 : 303c 0011                          move.w    #$11,d0
> >> $00e02cf8 : 20d9                               move.l (a1)+,(a0)+
> >> 
> >> $00e02cfa : 51c8 fffc                          dbra d0,$e02cf8
> >> $00e02cfe : 31fc 0007 11cc                     move.w #7,$11cc.w
> >> $00e02d04 : 43f9 00e0 2c02                     lea $e02c02,a1
> >> $00e02d0a : 41f8 0522                          lea $0522.w,a0
> >> $00e02d0e : 2099                               move.l (a1)+,(a0)
> >> $00e02d10 : 2159 0020                          move.l (a1)+,$20(a0)
> >> $00e02d14 : 2159 0040                          move.l (a1)+,$40(a0)
> >> $00e02d18 : 2159 0060                          move.l (a1)+,$60(a0)
> >> 
> >> > c 1
> >> 
> >> Returning to emulation for 1 CPU instructions...
> >> Exception 2 (e02cde) at e02cde -> e02ce6!
> >> 
> >> Le 25/01/2015 23:39, Laurent Sallafranque a écrit :
> >>> Hi all,
> >>> 
> >>> I've tested the hydroxyd demo : "love vs darkness" under hatari and
> >>> hatari reboots just after the first choice menu.
> >>> 
> >>> Link is :  http://demozoo.org/productions/131613/
> >>> 
> >>> In the trace, I can see :
> >>> 
> >>> $00e0268a : 21fc 00e0 3920 1136                move.l
> >>> #$e03920,$1136.w $00e02692 : 21fc 00e0 32ac 1152               
> >>> move.l #$e032ac,$1152.w $00e0269a : 21fc 00e0 32ba 1156             
> >>>   move.l #$e032ba,$1156.w $00e026a2 : 21fc 00e0 340a 1132           
> >>>     move.l #$e0340a,$1132.w $00e026aa : 11fc 0003 fc04              
> >>>       move.b #3,$fffffc04.w $00e026b0 : 11fc 0095 fc04              
> >>>       move.b #$95,$fffffc04.w $00e026b6 : 11fc 0007 0484            
> >>>         move.b #7,$0484.w $00e026bc : 21fc 00e0 225c 114a           
> >>>     move.l #$e0225c,$114a.w $00e026c4 : 203c 00e0 2750              
> >>>       move.l #$e02750,d0 $00e026ca : 21c0 1142                      
> >>>    move.l d0,$1142.w $00e026ce : 21c0 1146                         
> >>> move.l d0,$1146.w $00e026d2 : 21c0 114e                         
> >>> move.l d0,$114e.w $00e026d6 : 6100 05e8                          bsr
> >>> $e02cc0
> >>> $00e02cc0 : 41f8 11d6                          lea $11d6.w,a0
> >>> $00e02cc4 : 21c8 11c6                          move.l a0,$11c6.w
> >>> $00e02cc8 : 31fc 0003 11ca                     move.w #3,$11ca.w
> >>> $00e02cce : 7001                               moveq     #1,d0
> >>> $00e02cd0 : 224f                               movea.l   sp,a1
> >>> $00e02cd2 : 2478 0008                          movea.l $0008.w,a2
> >>> $00e02cd6 : 21fc 00e0 2ce6 0008                move.l
> >>> #$e02ce6,$0008.w Debugger: *CPU breakpoint*
> >>> 
> >>> My breakpoint was :  > b pc=$e02cde
> >>> 
> >>> It reboots as soon as I use "continue".
> >>> 
> >>> Does this write to address $8.w generates a problem or should I dig
> >>> deeper at instructions after $e02cde ?
> >>> 
> >>> Regards
> >>> 
> >>> Laurent