I get those errors sometimes when certain apps change resolution, most notably Wine, though with one version of a video driver it would do that when I tried to put dosbox in fullscreen mode. Sounds like an error with X.
But I didn't get it with moving the mouse.
Don't really know what to do.
These are all shots in the dark.
-----
If you use proprietary drivers you could try a different one.
Perhaps try running Wing Commander in a window as opposed to full screen?
Seeing if you can change any mouse settings for WC.
Or perhaps change some of the mouse settings under X?
Perhaps trying a different version of dosbox?
-----
Perhaps you could see what messages you get in the X logs. Should be either /var/log/XFree86.0.log or /archive/var/log/Xorg.0.log
Looking at dmesg or /var/log/messages might help too.

i've downloaded and compiled dosbox version 0.65. the error is gone with this version so i am perfectly happy. thanks for your reply.

seems like it had been an dosbox error.

My experience with 0.70 thus far is that it appears to be a bit more buggy than 0.65 was. It's too bad 0.65 isn't fully functioning on an Intel Mac or I would've started using that myself. I wonder whether or not your problem in 0.70 is indicative of the same dynamic core bug I've been getting nailed with on 0.70 in OSX and Windows. Whenever I play Darklands of Star Trek: 25th Anniversary (CD-ROM version), the game will lock up and die at certain times. Switching to the normal CPU core, solves the problem.

Of course for a game like WCIII, switching to the normal core is likely not an option given how performance intensive the game is.

Hmmmmm.... I just tried the latest CVS tarball code and I get this error after running 25th Anniversary with the dynamic core for a few minutes:

Exit to error: CacheBlock overrun 2 written 8464 size 8240

Perhaps that will help?

Look into core_dyn_x86.cpp and change

1#define CACHE_MAXSIZE	(4096*2)

to 4096*3 so the margin is a bit larger. But the game might still crash then
instead of dosbox generating that error.

That appears to have resolved the issue. I've got to make a 3 hour drive though... so I was only able to play for 15 - 20 minutes. Though normally on the dynamic core the game will not last even close to that long. I'll attempt to verify this better when I get home.

Some functions are larger due to the inlined memory access,
so the generated code seems to not fit in this special case.
Maybe they have a lot pusha/popa calls or so.

Either way the above fix is safe, but it has nothing to do with
the problem of the original poster. Which sounds like bad
gfx drivers or something.

wd,

Will that Dynamic Core fix be included in the next version?

Yes, the only "bad" effect it has is that 4k more memory are allocated.
The best solution of course would be generating more compact code 😀

😀

wd, thanks for the quick response. Unforunately I haven't gotten a chance to fully test the fix, but my initial impression that it appeared to fix the problem still stands. I plan on playing some more 25th anniversary tonight and see how things go.

It appears to resolve the issue. I played for over thirty minutes this morning with no problems whatsoever. This patch combined with the timer precision patch have made DOSBox usable on my Mac!

Note: The latest DOSBox CVS appears to also fix a bug present in the 0.70 source that kept DOSBox from compiling on OS X if the SDL_net libraries and headers are installed on my system. So I can compile a fully functional OSX DOSBox binary now!

Qbix fixed a bunch of configure things, merged some patches from sourceforge and stuff.
Nice that it compiles straight now.