Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220502093625.GA23225@kernel.org> <CAGS_qxoc=AnkzUtaFpJvF08Z_Z15h9sZxYNQT5-S9C7aM4poEg@mail.gmail.com>
 <CANiq72m70q+zviHVNSV_AEwOByVBiMuSQL5vyo2UMMpD-vd+_Q@mail.gmail.com>
 <CABVgOSmkZ9EXijBRx_pvS=Opizb1z2632B5rVsd1WRNOL5o8GA@mail.gmail.com> <CANiq72kVLzOu_3UjoUiSp5y+H0tA+SQYNQ=fsuhtg=i6MQe=Fg@mail.gmail.com>
In-Reply-To: <CANiq72kVLzOu_3UjoUiSp5y+H0tA+SQYNQ=fsuhtg=i6MQe=Fg@mail.gmail.com>
From:   David Gow <davidgow@google.com>
Date:   Thu, 5 May 2022 07:47:03 +0800
Message-ID: <CABVgOS=77WFFV1Xu5vQD6DPJODRr0f+qxJTRgJWS0qobv6vDaA@mail.gmail.com>
Subject: Re: [PATCH] kunit: take `kunit_assert` as `const`
To:     Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
Cc:     Daniel Latypov <dlatypov@google.com>,
        Miguel Ojeda <ojeda@kernel.org>,
        Brendan Higgins <brendanhiggins@google.com>,
        "open list:KERNEL SELFTEST FRAMEWORK" 
        <linux-kselftest@vger.kernel.org>,
        KUnit Development <kunit-dev@googlegroups.com>,
        linux-kernel <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, May 5, 2022 at 3:26 AM Miguel Ojeda
<miguel.ojeda.sandonis@gmail.com> wrote:
>
> Hi David,
>
> On Wed, May 4, 2022 at 4:05 PM David Gow <davidgow@google.com> wrote:
> >
> > I definitely agree here -- I can't recall any particular plan that
> > would require this to be non-const, and we can always change it back
> > if we really need to.
>
> That is good to know, thanks! Out-of-tree users can always be a surprise... :)
>
> > Very exciting! I assume that's the PR here:
> > https://github.com/Rust-for-Linux/linux/pull/757
>
> Indeed! I hope you like it -- we are taking the documentation tests in
> Rust (which are a very lightweight way of writing examples which
> double as tests) and generating KUnit test cases on the fly. For the
> moment it is just for the `kernel` crate, but the idea is to
> generalize it for modules etc.
>
> By the way, since you saw the PR... do you know if KUnit relies (or
> will rely) on "stack-dumping" functions like `longjmp`?

I don't think so -- though there's no fundamental individual tests
couldn't use them if it made sense for them.

KUnit spins off a new kthread per test, and uses
kthread_complete_and_exit() to unwind when an assertion fails. See
lib/kunit/try-catch.c for the actual implementation. The only really
dodgy bit is the test timeout support, which attempts to stop a thread
with kthread_stop(), and IIRC has some problems.

Hope that helps!
-- David