search

libnxz / power-gzip

POWER NX zlib compliant library
23 stars 18 forks source link

Test case to recreate kernel oops #10

Closed rzinsly closed 4 years ago

rzinsly commented 4 years ago

There was a bug in kernel that caused a kernel oops if THP was enabled. Set csb_poll_max to 200, test_stress will trigger kernel oops if the kernel doesn't have fix.

tuliom commented 4 years ago

I'm not sure if it's a good idea to cause a kernel oops when running make check. Can we detect this issue and report the failure without causing the kernel oops? I don't oppose having potentially dangerous tests. But I think they shouldn't be executed unless the user understands the implication.

rzinsly commented 4 years ago

I'm not sure if it's a good idea to cause a kernel oops when running make check. Can we detect this issue and report the failure without causing the kernel oops?

I don't know how we can detect this without the failure.

I don't oppose having potentially dangerous tests. But I think they shouldn't be executed unless the user understands the implication.

It make sense to not execute tests like these with make check, how about a separated script to run this kind of tests?

tuliom commented 4 years ago

It make sense to not execute tests like these with make check, how about a separated script to run this kind of tests?

Yep. Maybe also an extra target, e.g. make unsafe-check