savestate.cpp => segmentation fault

[john4p]

Tried to create a savestate with UAE4All using your savestate.cpp-source, Toni.

But it always crashes in the function save_state(...) with a "segmentation fault".

The first call to save_chunk(...) from inside save_state(...) works:

save_chunk(f, header, dst-header, "ASF ");

doesn't crash.

But then, after "dst = save_cpu(&len);" the next call to save_chunk() crashes:

save_chunk(f, dst, len, "CPU ");

The segmentation fault exactly happens when save_chunk(...) now tries to write the "chunk data":

fwrite(chunk, 1, len, f); => CRASH


In the first call to save_chunk(..."ASF ") which works the values for chunk and len were:
chunk: 0
len: 28

In the call to save_chunk(..."CPU ") which crashes the values for chunk and len were:
chunk: 172
len: 239020


Do you have any idea what could be the reason for this segmentation fault?

[Toni Wilen]

I don't know, I don't want to do that kind of debugging (without source level debugger)

But you can't simply use statestate.cpp and expect it to work without all other related routines.

[john4p]

Thanks for the reply.

Just took all the savestate-related stuff from the Nokia N900-port with Cyclone and put it into the Pandora-port of UAE4All. Everything built correct (after fiddling here and there) but resulted in the segmentation fault when trying to create a savestate.
Now I just built the N900-port itself on Pandora and tried to create a savestate => also segmentation fault.

Found out the "implementation" of save_cpu() for Cyclone doesn't seem to save all that much:

static __inline__ uae_u8 *save_cpu (int *len)
{
	return (uae_u8 *)len;
}

Now built the N900 port with "UAE M68k-core" (and without Cyclone) and savestates work.
So for Cyclone the possibility to save the CPU's state just hasn't been implemented.