[hatari-devel] Hatari interrupts still assert, if VDI mode is used

[Eero Tamminen]

Hi,

On Sunday 09 August 2009, Eero Tamminen wrote:
> Latest Hatari version in repo still asserts.  It happens always on
> startup if I do following:
> ./hatari --machine tt --tos etos512k.img -s 14 --vdi-planes 1 --vdi-width
> 800 --vdi-height 600  -d .

Happens also when starting with ST, STE or Falcon emulation if VDI
resolution is used.

With others backtrace had always gone through video.c, with Falcon
the assert came through MFP code:
#0  0xb7be6e95 in abort () from /lib/i686/cmov/libc.so.6
#1  0xb7bde5be in __assert_fail () from /lib/i686/cmov/libc.so.6
#2  0x08063ce6 in Int_AddRelativeInterruptWithOffset (CycleTime=-215498,
    CycleType=1, Handler=INTERRUPT_VIDEO_ENDLINE, CycleOffset=0) at 
int.c:392
#3  0x08063d17 in Int_AddRelativeInterrupt (CycleTime=-215498, CycleType=1,
    Handler=INTERRUPT_VIDEO_ENDLINE) at int.c:352
#4  0x08069bd8 in MFP_ActiveEdge_WriteByte () at mfp.c:1327
#5  0x080643bf in IoMem_bput (addr=16775683, val=0) at ioMem.c:358
#6  0x0808d96f in op_1c8_0_ff (opcode=<value optimized out>) at memory.h:123
#7  0x08081fe2 in m68k_go (may_quit=1) at newcpu.c:1709
#8  0x080667b9 in M68000_Start () at m68000.c:224
#9  0x08067309 in main (argc=15, argv=0x18) at main.c:712
(gdb) up 4
#4  0x08069bd8 in MFP_ActiveEdge_WriteByte () at mfp.c:1327
1327                            Int_AddRelativeInterrupt ( LineTimerBCycle - 
LineCycles + nCyclesPerLine,
(gdb) info locals
FrameCycles = 216542
HblCounterVideo = 1
LineCycles = 216034
LineTimerBCycle_old = 404
(gdb) print nCyclesPerLine
$1 = 508


I got also once interrupt assert without VDI:
	--machine ste --tos etos512k.img -s 4 -m

When Hatari had after use idled in GEM desktop for some time.


	- Eero