Re: [RFC v5 1/7] testing/nvdimm: Add test module for non-nfit platforms

[Santosh Sivaraj <santosh@fossix.org>]

Dan Williams <dan.j.williams@intel.com> writes:

> On Mon, Dec 14, 2020 at 2:39 AM Santosh Sivaraj <santosh@fossix.org> wrote:
>>
>> The current test module cannot be used for testing platforms (make check)
>> that do not have support for NFIT. In order to get the ndctl tests working,
>> we need a module which can emulate NVDIMM devices without relying on
>> ACPI/NFIT.
>>
>> The aim of this proposed module is to implement a similar functionality to
>> the existing module but without the ACPI dependencies.
>>
>> This RFC series is split into reviewable and compilable chunks.
>>
>> This patch adds a new driver and registers two nvdimm bus needed for ndctl
>> make check.
>
> I'd like to be able to test either nfit_test or nd_test by environment
> variable from the same build. See the attached patch. Otherwise, if
> the ndctl release process is not constantly testing nd_test it *will*
> regress / bitrot.
>
> So, "make check" should try nfit_test.ko, fallback to nd_test.ko, or
> otherwise be forced to one or the other via an environment variable.
> For example I'd like the release process on x86 to be:
>
> make check
> NVDIMM_TEST_MOD=nd_test make check
>
> ...where the first invocation assumes to test nfit_test.ko.
>
> It needs some fixups to either prevent nfit_test and nd_test from
> being loaded at the same time, or fixups to allow them to coexist.
>
> This rework implies v5.11 is too aggressive a merge target.

Yes I agree. If you feel good about this series, I would like you to take this
and I can send the corresponding ndctl changes incrementally to work with the
patch you sent.

Thanks,
Santosh
_______________________________________________
Linux-nvdimm mailing list -- linux-nvdimm@lists.01.org
To unsubscribe send an email to linux-nvdimm-leave@lists.01.org