Bug 1596

Summary: Automate BSP Testing
Product: [QA/Testing] Runtime Testing Reporter: Richard Purdie <richard.purdie>
Component: generalAssignee: Paul Eggleton <bluelightning>
Status: VERIFIED FIXED QA Contact: Alexandru Georgescu <alexandru.c.georgescu>
Severity: enhancement    
Priority: High CC: adnannabicharmang, alexandru.c.georgescu, elizabeth.flanagan, jessica.zhang, poky.bs.watcher, poky.watcher, ross.burton, sgw
Version: unspecified   
Target Milestone: 1.6   
Hardware: x86   
OS: Multiple   
Whiteboard: Co-Assign with Michael
OS type for building Yocto: --- Type of Regression: ---
Verified: Documentation change: Don't know
Bug Depends on: 5554, 5613, 5614, 5616    
Bug Blocks: 5471    

Description Richard Purdie 2011-10-04 07:33:52 UTC
Build-testing is only half the job of BSP testing, and new BSP development is and will be ramping up even more rapidly soon. We need to start building a framework to make as much of this as automated as possible. Phase I, what we can complete for 1.2, should be the ability to load an image from the autobuilder and make it ready for booting, or actually boot it, on a given target. It would be good to use systems running Yocto images as much as possible for this.

Could we use CATS?
Comment 1 Beth Flanagan 2011-10-31 09:56:13 UTC
Looking at shoeleather for this. Shoeleather needs some ramp up, but I was able to get into it this weekend and poke around.

The idea from my end is this:

1. similar to qemu image test we define how the image is loaded (and to where).
2. we define a test scenario for each board as we do for qemu image test
3. Once the autobuilder has generated the image, we use shoeleather lab to dump the image and run tests on it.


Success here should be an example using one of the boards currently at the lab.
Comment 2 Paul Eggleton 2013-11-29 18:40:08 UTC
I've sent out a plan for 1.6 here:

https://lists.yoctoproject.org/pipermail/yocto/2013-November/017451.html

I've also added individual bugs for the remaining tasks and set them as blockers for this bug.
Comment 3 Paul Eggleton 2014-04-14 13:14:22 UTC
We got to the point of being able to deploy and boot EFI-based machines for 1.6; since this is the 1.6 umbrella bug and we no longer really need it, I'm marking this as resolved. If we have further enhancements we can add additional enhancement bugs to cover them.
Comment 4 Alexandru Georgescu 2014-07-28 07:10:02 UTC
marking this as verified since we are using the HW automation infrastructure.