Re: [hatari-devel] IDE IO register range access commit

[ Thread Index | Date Index | More Archives ]


On 9/24/20 3:26 PM, Uwe Seimet wrote:
Looks as if there is still an issue with the IDE handling. I don't know
whether this is a new issue or has always been there > In Falcon030 mode Hatari crashes when no IDE images are assigned:

Thread 1 "hatari" received signal SIGSEGV, Segmentation fault.
0x00005555558f7258 in Ide_Mem_bget ()
(gdb) bt
#0  0x00005555558f7258 in Ide_Mem_bget ()
#1  0x00005555559c5a99 in op_4a39_0_ff ()
#2  0x000055555597d092 in m68k_run_2_020 ()
#3  0x00005555559758c2 in m68k_go ()
#4  0x00005555558ba703 in main ()

In TT mode everything is fine, i.e. the IDE interface is not present.

I can reproduce this just with:
	$ hatari --machine falcon --dsp none

Thread 1 "hatari" received signal SIGSEGV, Segmentation fault.
Ide_Mem_bget (addr=15728669) at src/ide.c:107
107			retval = ide_ioport_read(opaque_ide_if, ideport);
(gdb) print opaque_ide_if
$2 = (struct IDEState *) 0x0

It's due to Nicolas' Ide_MmioIsAvailable()
function change claiming that IDE is available
(on Falcon) even when there are no IDE devices

That (previously internal) function is used by IDE
data access functions which chrash when there's no
actual IDE device...

	- Eero

Mail converted by MHonArc 2.6.19+