|Re: [hatari-devel] Hatari segfault on cold boot with MMU + EmuTOS|
[ Thread Index |
| More lists.tuxfamily.org/hatari-devel Archives
On 11/1/18 12:09 AM, Nicolas Pomarède wrote:
Le 31/10/2018 à 22:53, Eero Tamminen a écrit :
Did you have 256k or 512k version of EmuTOS v0.9.7?
(192k version doesn't support MMU and can't trigger the bug.)
I'm using 512k version, md5sum is cee17f68e8dc0bd3a35816650e44fe90
Can you try after removing hatari.cfg to be sure we have the same
defaults to do this test ?
Segfaults with defaults when enabling 030 & MMU.
I removed hatari.cfg and then ran :
./hatari --machine tt --tos ~/Emul/ST/tos/emutos097.img -s 8 --ttram
0 --mmu on --addr24 off
I then press alt+c at the emutos boot page where it says to press shift
to pause the screen, but hatari never crashes for me.
Often I wait until EmuTOS boots to the desktop, then it's always been
enough to press AltGr+C 1-2 times (>90% of time, once is enough, and
second press can be done before welcome screen).
Now that I re-tried doing it right at welcome screen, there was one
time where I had to use AltGr+C a dozen times before segfault. Maybe
EmuTOS initializes MMU right after welcome screen, or at least does
some extra MMU settings after it?
Maybe this crash is created by some specific optimisation flags / gcc
version ? That would be the 1st time it happens, but who knows (I'm
comiling for 64 bit architecture, are you using 32 bit maybe ?)
32bit version, gcc v6.3.
So, I can't reproduce your issue :(
As it seems to be somewhat timing related, could others on the list try
it after EmuTOS has booted to desktop and then press AltGr+C at least
dozen times at few sec intervals?
(hatari --cpulevel 3 --mmu on --tos etos512k.img --cpu-exact on