Hello Phil, the fix looks fine to me.
Thanks, Andrew On 10/10/2014 8:20 PM, Phil Race wrote:
This looks like its a driver bug. We use a scratch surface of size 1x1 when iterating through testing all the gl capable visuals and some of them get a SIGFPE: FPE_INTDIV when as we try to make the context current. Since setting the scratch surface to 4x4 makes the crash go away it seems that some cases in the driver code must round 1 down to 0 and then divide by it .. the workaround seems worth doing since its easy enough and harmless. http://cr.openjdk.java.net/~prr/8059848/ https://bugs.openjdk.java.net/browse/JDK-8059848 -phil.