[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Crash in GC during save-application



Byron Davies-R writes:
> Has anyone encountered -- and hopefully resolved -- a problem with a crash 
> occuring in the GC that takes place in save-application?
> 
> We build big applications (30-40Mb application sizes typically running in 65-
> 110Mb image sizes on Quadra 650s, 800s and 840s, under System 7.5 and 7.5.1).  
> We generally use lots of RAM and run without virtual memory.  We have recently 
> run into the nagging crash problem described above.  After the Listener has been 
> closed and during the final GC, the Mac just freezes up and must be restarted, 
> even if MacsBug is turned on.  All state is lost, so we have no way to get 
> additional information about what happened.  Zapping the PRAM seems to have some 
> preventive value, but not always.  There does seem to be some sensitivity to 
> image size.  For example, we'll occasionally be able to save from a 65Mb image, 
> but not from a 110Mb image -- with both built using the same script, software, 
> and data.

I have encountered a similar tendency to crash in save-application, but
under slightly different circumstances.

I am saving a 3.5-5.5MB image on a MacIIsi with 17MB, System 7.1,
virtual memory off, and Connectix RAM Doubler installed.  My crashes do
not occur during the GC phase of the save, but instead after a disk file
with image data has been created.  When the finder returns to the
foreground, the desktop and menus blank out, the finder is reloaded from
disk, then blanks out again, and so on indefinitely.  After restarting
the machine, the image file does not contain 5.5MB, but only a few
hundred kB, and the saved image is lost.  This happens during most uses
of save-application, but not always; it is almost guaranteed to occur
after doing a lot of things in MCL before saving the image.

In my case, the problem disappeared after RAM Doubler was removed.
However, no other application I use has shown any incompatibility with
that extension.

Tom Herbig
Princeton University
Physics Department