[Open-graphics] Inconveniences with the video controller

[Timothy Normand Miller]

On 6/25/07, Paul Brook <paul at codesourcery.com> wrote:
> On Monday 25 June 2007, Timothy Normand Miller wrote:
> > On 6/25/07, Paul Brook <paul at codesourcery.com> wrote:
> > > Is this worth the hassle compared to just rerunning the VC program
> > > generator?
> >
> > Well... that's hard to say.  Keep in mind that the program is running
> > while this is happening.  Generally speaking, as long as nothing
> > structural is changed (moving program-control or changing loop
> > counts), it's safe to change it while it's running.
>
> Isn't there still a race condition that can result in a frame of garbage?
> We need to avoid changing the values in between execution of the ADDR and INC
> commands.

The interrupt flag is at the end of the last active scanline.  The
ADDR and INC instructions are the scanline before the first active
line.  There should be ample time to change both instructions before
the potential race condition happens.

Also, it's more than just a race condition.  We're writing to the
program memory in one clock domain and executing it in another.  If an
instruction were read too soon after being written, there could be
problems with metastability (although that's a very low-probability
situation).

-- 
Timothy Normand Miller
http://www.cse.ohio-state.edu/~millerti
Open Graphics Project