On Fri, 2018-10-19 at 09:06 +0200, Hans Verkuil wrote: > On 10/18/2018 09:23 PM, Ezequiel Garcia wrote: > > Hi everyone, > > > > In Collabora, and as part of our kernelci work, we are doing > > research on kernel functional testing with kernelci. > > > > For those new to kernelci, see https://github.com/kernelci/kernelci-doc/wiki/KernelCI > > and https://kernelci.org/. > > > > The goal is to lay down the infrastructure required to make > > automated test coverage an integral part of our feature > > and bugfix development process. > > > > So, as a first attempt, we've decided to extend kernelci test > > v4l2 plan support, leading the way to extending > > other subsystems' test plans. > > > > Currently, kernelci looks for a list of branches every hour and > > see if anything changed. For any branch that has changed, it triggers > > builds, boots, tests and reports for each branch that had some changes > > since last time it ran. > > > > For this pilot, we've decided to target just a few devices: > > qemu with vivid, rk3399-gru-kevin and rk3288-veyron-jaq > > with uvc. > > It's running v4l2-compliance, right? > Exactly. > Looking at the test cases, they appear in the reverse order that v4l2-compliance > performs them, that's a bit odd. > That's something to check in the parser. I'm sure it's fixable if you find it annoying. > And if we include uvc in the testing, then I need to prioritize the work I > started for uvc to remove the last FAILs. > Well, we can run anything. We decided to go for uvc and vivid, just too pick two popular examples, but there are really no restrictions. Thanks, Ezequiel
