One unclear point is where these tests should be run. They are huge (1.5G) so probably adding them to ptest images isn't the best idea. ltp perhaps? Another investigation is what the performance is under software renderer or under virgl, and if either is feasible.
It may make sense to copy the ltp module and add a specific target into the autobuilder which builds the tests into a special purpose image and then runs them.
I ran the 'quick' set under software renderer in qemu: time 1h5m total 41428 skip 15373 pass 26591 warn 4 fail 287 crash 73 Next I'd like to try virgl.
A new runtime test for Piglit to enable test execution would be fairly simple, albeit slow so it shouldn't be enabled by default!
Bulk move from 4.99 or 0.00 to 5.99
Revised position: I think maybe piglit should inherit ptest. The run-ptest should be able to be configured to run any tests but have a sensible default, and transform the test report so that the ptest framework can parse it. This way we get test execution easily (bitbake core-image-ptest-piglit -c testimage) and people who care can run that often and use resulttool to get regression reports.
Not really a newcomer bug.