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

Re: A92: ANOUNCEMENT: System Stablizer v0.7 & Error v0.1





> I detected this error, when I programed some kind of stabilizer library
> for
> programming with Fargo, before there was a 256K - compatible debugger.
> It was able, to stabilize a program with a lot of errors on a normal
> calc, but
> hang up on the tiger-emu. I expected the normal stackframe, 8 bytes more
>
> on the stack, than normaly (actually I got out this, just by printing
> the
> stackpointer-value, then I realized, that I allways had to add #8 onto
> a7, to get the exception to return properly (now I know, that this
> is normal, and which information are put onto the stack))
>


   Yes, there's lots of complexities in stablizing the system.  Bus and Address
Errors on the 68K are 8 bytes longer than usual, and also the return address
may not point to the next instruction, but rather some point in between (for
instance a 6-byte instruction which bus errored may be pointing to the last two
bytes of the instruction rather than the next instruction itself).  Also,
A-Line and F-Line instructions' return addr point to the instruction which
caused the exception, so the return PC has to be incremented.

--Bryan
bcturner@eos.ncsu.edu



References: