RE: [Automated-testing] [boot-time] Call for boot time data!!
"Bird, Tim" <[email protected]> Fri, 21 Feb 2025 16:53:47 +0000
| Newsgroups | dev.linux.lists.kernelci |
|---|---|
| Message-ID | <MW5PR13MB563258539CC8345B0550C7CEFDC72@MW5PR13MB5632.namprd13.prod.outlook.com> |
> -----Original Message----- > From: [email protected] <[email protected]> On Behalf Of Gustavo Padovan via > lists.yoctoproject.org > Hello, > > ---- On Thu, 20 Feb 2025 21:41:35 -0300 Tim Bird via lists.yoctoproject.org wrote --- > > > -----Original Message----- > > > From: [email protected] [email protected]> On Behalf Of Tim Bird via > lists.yoctoproject.org > > > Dear Linux testers, > > > > > > I could really use more boot data for the initcall and boot region analysis I am doing. (I'm trying to build > > > a system of reference value files for an initcall duration regression test I'm writing for the Linux kernel). > > > > > > If you have a machine, or particularly any embedded dev boards (like a beagleplay, beaglebone (*), raspberry pi, etc.), it > > > would help me out a lot of you could run this tool: grab-boot-data.sh > > > One can create a KernelCI job with this in the rootfs and grab this data for **all** devices. We'd be happy to support that in KernelCI as we > really want to grow boot time testing capabilities. That's on my list of requests. Benjamin Copeland reported a bug on the server side that made the whole site fall over, based on badly parsed data in one of the boot data files. I fixed that this morning, but it demonstrates that this is not quite production-grade yet. After shaking out some more bugs, I'd really like to investigate getting this into shape so that it can be run automatically in people's labs easily. Right now this is a data gathering tool. I need lots of data in order to refine the test and make reference value files that would support putting this into upstream kselftest. I recognize that my efforts and Laura's efforts in this area (boot-time regression testing) could be aligned better. We should have a discussion and see how we can leverage each other's work and/or collaborate on an upstream test. Thanks, -- Tim