Kdump tests can be invided into 2 categories: Tests of dumping of vmcore (crash- dump- tests) Tests of analyzing vmcore (analyse-* tests)
Vmcore analyzing tests
It's very simple to execute a Kdump test. What you need to do is to change directory to each test case directory and execute
./runtest.sh with root privilege, either manually or by restraint.
Before running a dumping test, make sure there is enough disk space on dump target to save vmcore. Usually it is the size of physical memory.
It may requires to update kernel boot cmdline (e.g. update crashkernel=<xx>M in Fedora) to start kdump service. And kdump panic handling will reboot system after collecting vmcore. So in general, system may reboot up to 2 times in each dumping test.
Make sure the test framework you use to execute Kdump test is able to handle system reboot. Otherwise, you need to re-run
runtest.sh after each reboot until the test is done.
This is the typical workflow in a dumping test + vmcore analyzing test
Install kexec-tools and configure kernel boot cmdline if needed.
Modify kdump configuration file at /etc/kdump.conf.
Trigger system crash.
Install dependent packages for analyzing vmcore by crash.
Analyze and validate vmcore by crash utility.
For bugs of this test suite, feel free to report it in issue page.
For bugs of kdump on Fedora, please file the report in Bugzilla with product "Fedora".
For bugs of kdump on Red Hat Enterprise Linux, please file the report in Red Hat Bugzilla.
If you want to contribute in code, feel free to send your pull requests.