Re: [PATCH] sim: bfin: handle saturation with fract multiplications

[Robin Getz <robin dot getz at analog dot com>]

On Tue 15 Mar 2011 11:38, Joel Brobecker pondered:
> > From: Robin Getz <robin.getz@analog.com>
> > 
> > The saturation behavior with fract modes differs from non-fract modes.
> > 
> > Signed-off-by: Robin Getz <robin.getz@analog.com>
> > Signed-off-by: Mike Frysinger <vapier@gentoo.org>
> > 
> > 2010-03-14  Robin Getz  <robin.getz@analog.com>
> > 
> > 	* bfin-sim.c (extract_mult): Handle M_IU.
> 
> We're going to have to trust you on these ones... 

The offer for hardware for people who want to help verify things is 
always there -- just ask Mike.

> For small patches 
> like these that seem obvious to you , go ahead and commit if you feel 
> confident about them.  If we have any comments, worst case scenario
> is that we end up temporarily reverting, which I don't really expect.

Just some background (now that I'm back from a trip) -- we had tested the gnu 
sim against a known testbench, which was enough to get a kernel booting on it 
(both Linux and RTEMS), and all the random applications we could try.

However, what we found on further inspection was the testbench was, well, 
lacking coverage, and the sim didn't match the hardware all the time -- so we 
now have a framework that now generates pseudorandom code sequences (with 
random register values), runs it on the target hardware to determine the 
expected results, effectively creating a test case, and then re-runs the test 
case on the sim, and extracts any failures. I have been using those failing 
test cases to fix the sim (and extend the testbench).

Failures are still being found, (but it is taking much longer than it used 
to), so I would expect these types of patches to continue to trickle in over 
the next month or so.

-Robin