Re: [PATCH] test: regression test for a kernel panic on running ARS for BTTs

On Thu, Apr 27, 2017 at 3:39 PM, Vishal Verma <vishal.l.verma(a)intel.co m> wrote: > Runing an Address Range Scrub on a region with a BTT namespace would > cause a kernel null pointer dereference. This tests we don't regress > that. > > Cc: Toshi Kani <toshi.kani(a)hpe.com&gt; > Cc: Dan Williams <dan.j.williams(a)intel.com&gt; > Signed-off-by: Vishal Verma <vishal.l.verma(a)intel.com&gt; > --- >  test/Makefile.am      |  3 +- >  test/btt-ars-panic.sh | 93 > +++++++++++++++++++++++++++++++++++++++++++++++++++ >  2 files changed, 95 insertions(+), 1 deletion(-) >  create mode 100755 test/btt-ars-panic.sh >

> +test_pass() > +{ > +       # If the test failed, there would have been a kernel null > pointer > +       # dereference, and a subsequent ndctl disable-region would > hang > +       $ndctl disable-region all & Test looks good. This line makes me think we need something like "ndctl wait-scrub-bus".