Floating point exceptions on Sun Workstations

[Paul Tomblin]

In article <427 at gestetner.oz> mykel at gestetner.oz.au writes:
>[about his problem with:]
>	psi: USER COPROCESSOR PROTOCOL ERROR
>	trap address 0x34, pid 3445, pc = 9e65e, sr = 4, stkfmt 9, context 5
>	D0-D7  3 fdb62455 44d007db 0 44b239e7 44b291a3 44b1e349 44b29251
>	A0-A7  efff84c d384c 0 d3fcc d384c efff9b8 efff83c
>
>What I want to know is, is there anyone else doing development on Sun's in
>this manner that has come across this problem, and if so, what has been
>done as a work around?  It is making it very difficult to debug anything if
>the machine constantly crashes.
>
I got exactly the same error in a piece of code that had two routines (call
them 'a' and 'b') where 'a' passed 'b' a pointer to a structure, which 'b'
then dereferenced a float member of.  When 'a' passed a NULL pointer, this
is what we got.  Sun was very little help trying to debug it, since they
claimed it wasn't their message.  I had to run strings on vmunix to prove to
them it was their message.  The workaround is DONT DO IT!!!.  Check your
pointers.  Note that this error comes from the 68881 exception handler, so
it only happens with floats and doubles.

I hope this helps.  Sorry about speaking 'C', but I don't speak assembler.
-- 
Paul Tomblin nrcaer!cognos!geovision!pt or uunet!geovision!pt
Life: Loath it or ignore it, you can't like it.  (Marvin)
(My employer may not agree with my opinions, especially my .signature)