tools/testing/nvdimm: fix nfit_test shutdown crash
authorDan Williams <dan.j.williams@intel.com>
Fri, 14 Apr 2017 06:14:34 +0000 (23:14 -0700)
committerDan Williams <dan.j.williams@intel.com>
Tue, 18 Apr 2017 17:26:09 +0000 (10:26 -0700)
commit8b06b884cd98f7ec8b5028680b99fabfb7b3e192
treedde4af2e717639c5681a8855f7bad11b72291340
parent9ccaed4bfd4f186e8350ffc7d8f188f8d2991fd9
tools/testing/nvdimm: fix nfit_test shutdown crash

Keep the nfit_test instances alive until after nfit_test_teardown(), as
we may be doing resource lookups until the final un-registrations have
completed. This fixes crashes of the form.

 BUG: unable to handle kernel NULL pointer dereference at 0000000000000038
 IP: __release_resource+0x12/0x90
 Call Trace:
  remove_resource+0x23/0x40
  __wrap_remove_resource+0x29/0x30 [nfit_test_iomap]
  acpi_nfit_remove_resource+0xe/0x10 [nfit]
  devm_action_release+0xf/0x20
  release_nodes+0x16d/0x2b0
  devres_release_all+0x3c/0x60
  device_release+0x21/0x90
  kobject_release+0x6a/0x170
  kobject_put+0x2f/0x60
  put_device+0x17/0x20
  platform_device_unregister+0x20/0x30
  nfit_test_exit+0x36/0x960 [nfit_test]

Reported-by: Linda Knippers <linda.knippers@hpe.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
tools/testing/nvdimm/test/nfit.c