devsummit:201708:testing

https://hackmd.io/KYJhGMA4CMFYBMC0BDAbARgMyICwAZ1IV151FSRJUBOScWCHIA==?both

Test all the things! http://goo.gl/Xcocrp

Last Working Group: Testing

  • Automated analysis of fuzzing results?
  • Ideas to generate an analysis data structure (Crash Graph ideas)
  • QEMU fuzzing?
  • real hardware testing (arm/mips/… boards)
  • unit testing of kernel functions/modules/units (requires framework)?
  • user-space versions of kernel code (examples: SCTP library, TCP library, etc.)?
  • Create “culture of testing”?
    • Should reviewers ask for tests for big features just like they ask for man pages?
  • criteria of “good to ship”
    • no regression
    • feature
      • experimental feature should be marked explicitly
  • Have we converted all things under tools/regression?
  • setup performance testing cluster
  • location? NYI
  • hardware requirements? (brd@ / gnn@)
  • generalize freebsd-ci codes for downstream adoption
  • a package to setup everything?
  • Some form of automated testing for security mitigation deployment (W^X, ASLR, PAN, …) (brooks@, Domagoj)
  • Talk to Backtrace (gnn@)
  • devsummit/201708/testing.txt
  • Last modified: 2017/09/18 12:58
  • by KubilayKocak