Hi! I have cobbled together a bunch of scripts to test the atomic updates API, based around kms++; see here: https://github.com/uli/kmsxx/tree/autotest Instructions on how to set up the environment can be found here: http://elinux.org/User:Uli/Tests:KMS Currently, tests try various sensible and nonsensical combinations of source and destination properties and check if the driver behaves reasonably. Test results look like this: 1. Lager: # bash run_tests.sh aaa_sanity.py Testing connector VGA-1 PASS atomic_flip.py PASS atomic_set_plane.py PASS atomic_small_dest.py PASS atomic_small_fb.py PASS fb_large.py PASS fb_small.py PASS mode_invalid.py PASS mode_reset.py PASS mode_set_all.py PASS 10 passed, 0 failed, 0 skipped 2. Salvator-X: # bash run_tests.sh aaa_sanity.py Testing connector VGA-1 PASS atomic_flip.py PASS atomic_set_plane.py FAIL atomic_small_dest.py PASS atomic_small_fb.py PASS fb_large.py PASS fb_small.py PASS mode_invalid.py PASS mode_reset.py PASS mode_set_all.py PASS 9 passed, 1 failed, 0 skipped I have not yet looked at the FAIL on Salvator-X (might well be that the test is bogus) because there is a bigger issue there. After running the tests a couple of times, framebuffer allocation starts to fail: [ 956.562285] rcar-du feb00000.display: failed to allocate buffer with size 3145728 [ 957.130012] rcar-du feb00000.display: failed to allocate buffer with size 3145728 [ 957.689012] rcar-du feb00000.display: failed to allocate buffer with size 3194880 [ 959.926459] rcar-du feb00000.display: failed to allocate buffer with size 1921024 [...] This does not happen on Lager, and I use the same 32-bit userspace for both platforms. I guess there must be a memory leak in the driver somewhere. Is this a known issue? CU Uli
