Re: [hatari-devel] ide.c segfault when config changed to Falcon

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Le 07/12/2020 à 20:46, Thomas Huth a écrit :
> Am Mon, 7 Dec 2020 15:28:13 +0100
> schrieb Nicolas Pomarède <npomarede@xxxxxxxxxxxx>:
>
> > Le 06/12/2020 à 17:59, Eero Tamminen a écrit :
> > > Hi,
> > >
> > > I've pushed doc updates for all the fixes.
> > >
> > >
> > > Unless there are objections, I was thinking of
> > > pushing also my fix for the issue with double
> > > usage of same image through ACSI/SCSI/IDE
> > > (the one Uwe reported).
> >
> > Hi
> >
> >
> > I'm not against it, but on my side I don't use ACSI/SCSI/IDE, so I'm
> > not able to test if these changes can have any side effects.
> >
> > More testers would be needed for this fix to ensure all cases are
> > handled.
>
> FWIW, the patches look pretty straight-forward to me, so I think it
> should be safe to commit them. I currently don't have much spare time
> for testing, though, sorry.
>
>   Thomas

Hi

unfortunately, I found another case where Hatari might crash because of 
IDE IO.

start hatari in falcon mode (with --tos tos404.img) then do a memory 
save while falcon mode is booting (atari logo displayed in top left).

exit hatari and restart hatari with --memstate and the name of the save 
file. hatari should start, complete the floppy test and then core dump 
when it comes to checking IDE state.

(gdb) bt
#0  0x00000000005f6980 in Ide_Mem_bget ()
#1  0x0000000000691d34 in wait_cpu_cycle_read_ce020 ()
#2  0x0000000000bd1070 in op_4a39_24_ff ()
#3  0x000000000066fca8 in m68k_run_3ce ()
#4  0x000000000066a852 in m68k_go ()
#5  0x00000000005bc9a3 in main ()

In my case I don't use any IDE drive, so it's likely to be some null 
pointers in IDE part as we saw before.

Nicolas