The current implementation requires up to 3 buffers to properly
implement page flipping without losing frames: one is configured
before the video stream is started, one just after that and page
flipping is synchronized to the frame done interrupt. The comment in
the code mentions that "CSI will lookup the next dma buffer for next
frame before the current frame done IRQ triggered".

Based on observations of the CSI unit behavior, it seems that the
buffer DMA address is sampled when the frame scan begins (in addition
to starting the stream), which corresponds to the vblank interrupt
that hits just before the frame-done interrupt of the previous frame.

As a result, the address configured at the frame done interrupt is not
actually used for the next frame but for the one after that.

This proposal changes the page flipping sync point to the vsync
interrupt, which allows the DMA address to be sampled for the next
frame instead and allows operating with only two buffers.

In addition to the change in the sync point, the code is refactored
to introduce a notion of state that clarifies tracking of the buffers
with tidy functions.
Signed-off-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>