Re: [PATCH v5 00/18] kunit: introduce KUnit, the Linux kernel unit testing framework

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Jun 21, 2019 at 5:54 PM Brendan Higgins
<brendanhiggins@xxxxxxxxxx> wrote:
>
> On Fri, Jun 21, 2019 at 12:21 PM shuah <shuah@xxxxxxxxxx> wrote:
> >
> > On 6/21/19 12:13 PM, Theodore Ts'o wrote:
> > > On Fri, Jun 21, 2019 at 08:59:48AM -0600, shuah wrote:
> > >>>> ### But wait! Doesn't kselftest support in kernel testing?!
> > >>>>
> > >>>> ....
> > >>
> > >> I think I commented on this before. I agree with the statement that
> > >> there is no overlap between Kselftest and KUnit. I would like see this
> > >> removed. Kselftest module support supports use-cases KUnit won't be able
> > >> to. I can build an kernel with Kselftest test modules and use it in the
> > >> filed to load and run tests if I need to debug a problem and get data
> > >> from a system. I can't do that with KUnit.
>
> Sure, I think this point has been brought up a number of times before.
> Maybe I didn't write this section well because, like Frank said, it
> comes across as being critical of the Kselftest module support; that
> wasn't my intention. I was speaking from the perspective that
> Kselftest module support is just a feature of Kselftest, and not a
> full framework like KUnit is (obviously Kselftest itself *is* a
> framework, but a very small part of it is not).
>
> It was obvious to me what Kselftest module support was intended for,
> and it is not intended to cover the use case that KUnit is targeting.
>
> > >> In my mind, I am not viewing this as which is better. Kselftest and
> > >> KUnit both have their place in the kernel development process. It isn't
> > >> productive and/or necessary to comparing Kselftest and KUnit without a
> > >> good understanding of the problem spaces for each of these.
>
> Again, I didn't mean to draw a comparison of which is better than the
> other. I was just trying to point out that Kselftest module support
> doesn't make sense as a stand alone unit testing framework, or really
> a framework of any kind, despite how it might actually be used.
>
> > >> I would strongly recommend not making reference to Kselftest and talk
> > >> about what KUnit offers.
>
> I can see your point. It seems that both you and Frank seem to think
> that I drew a comparison between Kselftest and KUnit, which was
> unintended. I probably should have spent more time editing this
> section, but I can see the point of drawing the comparison itself
> might invite this confusion.
>
> > > Shuah,
> > >
> > > Just to recall the history, this section of the FAQ was added to rebut
> > > the ***very*** strong statements that Frank made that there was
> > > overlap between Kselftest and Kunit, and that having too many ways for
> > > kernel developers to do the identical thing was harmful (he said it
> > > was too much of a burden on a kernel developer) --- and this was an
> > > argument for not including Kunit in the upstream kernel.
>
> I don't think he was actually advocating that we don't include KUnit,
> maybe playing devil's advocate; nevertheless, at the end, Frank seemed
> to agree that there were valuable things that KUnit offered. I thought
> he just wanted to make the point that I hadn't made the distinction
> sufficiently clear in the cover letter, and other reviewers might get
> confused in the future as well.
>
> Additionally, it does look like people were trying to use Kselftest
> module support to cover some things which really were trying to be
> unit tests. I know this isn't really intended - everything looks like
> a nail when you only have a hammer, which I think Frank was pointing
> out furthers the above confusion.
>
> In anycase, it sounds like I have, if anything, only made the
> discussion even more confusing by adding this section; sorry about
> that.
>
> > > If we're past that objection, then perhaps this section can be
> > > dropped, but there's a very good reason why it was there.  I wouldn't
> > > Brendan to be accused of ignoring feedback from those who reviewed his
> > > patches.   :-)
> > >
> >
> > Agreed. I understand that this FAQ probably was needed at one time and
> > Brendan added it to address the concerns.
>
> I don't want to speak for Frank, but I don't think it was an objection
> to KUnit itself, but rather an objection to not sufficiently
> addressing the point about how they differ.
>
> > I think at some point we do need to have a document that outlines when
> > to KUnit and when to use Kselftest modules. I think one concern people
> > have is that if KUnit is perceived as a  replacement for Ksefltest
> > module, Kselftest module will be ignored leaving users without the
> > ability to build and run with Kselftest modules and load them on a need
> > basis to gather data on a systems that aren't dedicated strictly for
> > testing.
>
> I absolutely agree! I posed a suggestion here[1], which after I just
> now searched for a link, I realize for some reason it didn't seem like
> it reached a number of the mailing lists that I sent it to, so I
> should probably resend it.
>
> Anyway, a summary of what I suggested: We should start off by better
> organizing Documentation/dev-tools/ and create a landing page that
> groups the dev-tools by function according to what person is likely to
> use them and for what. Eventually and specifically for Kselftest and
> KUnit, I would like to have a testing guide for the kernel that
> explains what testing procedure should look like and what to use and
> when.
>
> > I am trying to move the conversation forward from KUnit vs. Kselftest
> > modules discussion to which problem areas each one addresses keeping
> > in mind that it is not about which is better. Kselftest and KUnit both
> > have their place in the kernel development process. We just have to be
> > clear on usage as we write tests for each.
>
> I think that is the right long term approach. I think a good place to
> start, like I suggested above, is cleaning up
> Documentation/dev-tools/, but I think that belongs in a (probably
> several) follow-up patchset.
>
> Frank, I believe your objection was mostly related to how KUnit is
> presented specifically in the cover letter, and doesn't necessarily
> deal with the intended use case. So I don't think that doing this,
> especially doing this later, really addresses your concern. I don't
> want to belabor the issue, but I would also rather not put words in
> your mouth, what are your thoughts on the above?
>
> I think my main concern moving forward on this point is that I am not
> sure that I can address the debate that this section covers in a way
> that is both sufficiently concise for a cover letter, but also doesn't
> invite more potential confusion. My inclination at this point is to
> drop it since I think the set of reviewers for this patchset has at
> this point become fixed, and it seems that it will likely cause more
> confusion rather than reduce it; also, I don't really think this will

Since no one has objected to dropping the "### But wait! Doesn't
kselftest support in kernel testing?!" section in the past almost two
weeks, I am just going to continue on without it.

Cheers

> be an issue for end users, especially once we have proper
> documentation in place. Alternatively, I guess I could maybe address
> the point elsewhere and refer to it in the cover letter? Or maybe just
> put it at the end of the cover letter?
>
> [1] https://www.mail-archive.com/kgdb-bugreport@xxxxxxxxxxxxxxxxxxxxx/msg05059.html



[Index of Archives]     [Device Tree Compilter]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux PCI Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Yosemite Backpacking]


  Powered by Linux