1) image_to_vm aborts because e2fschk requires a terminal to
be connected in order to check whether or not you want to repair the fs. Since
we always want this to be true, set -p
2) Clean up test harness to only verify the number of tests that passed on the
base image are the same on the update back to. This is because older images
have many tests that fail. I leave 10% as a reasonable number to pass.
3) Redirect output from autotest in run_remote_tests to stderr so it gets
logged in a calling script that captures stdout.
Change-Id: If412274353683add20d136747113eb9c2bd41330
BUG=4690, 5533, 7287
TEST=Ran with internal tools for builders.
Review URL: http://codereview.chromium.org/3536018
Change-Id: I913acd95c49b1cf0fd666b49fb5ec1e24f0f83d1
BUG=7285
TEST=Ran through test against with two test images.
Review URL: http://codereview.chromium.org/3619009
Change-Id: I4afa86fa7a6b772124de56114910a1ca5e3d4511
BUG=
TEST=Ran through update harness code and ran loman.
Review URL: http://codereview.chromium.org/3612004
This harness has support for test on vm machines. It takes a base image and target image path, creates an update-enabled vm image from the base image and performs updates using cros_run_vm_update (vm wrapper around image_to_live). It uses --snapshot for updating vm's and keeps them on (like a real machine would) so that the VM image is clean on the next test.
The current check in has two test cases -- updates without clobbering the stateful partition and updates with clobbering the stateful partition.
Change-Id: I5406eadb39fa36601d17f8dd15625bd6a8b38c15
BUG=6911
TEST=Ran through tests.
Review URL: http://codereview.chromium.org/3554002
