[Mesa-dev] Resend [PATCH] glx: Handle out-of-sequence swap completion events correctly.

[Mario Kleiner]

Hi all,

a respin of the bugfix for INTEL_swap_events + DRI3/Present.

The code in the patch itself is identical to the one queued
for Mesa 10.5.1, the one apparently nobody wants to review
- I hate wraparound handling too...

The only difference is a much longer commit message which
explains why we need this for DRI3/Present, and why this new
wraparound handling is supposed to work.

If nobody wants to review this, or can come up with a better
solution, i could also submit a patch that simply removes the
broken wraparound handling of current Mesa, which would still
be an improvement over what we have now, given that few OpenGL
clients will submit over 2^32 swapbuffers in a single session
for a single drawable, as that would require a runtime without
interruption of over 13 months at 120 fps redraw rate.

thanks,
-mario