linux-kernel - Re: [PATCH v2 1/3] rust: kunit: add KUnit case and suite macros

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CABVgOS=tp1SiZeQYgDX2ET3twJ8-xCZTu8ofsWabjsQZQbyMCg@mail.gmail.com>
Date: Wed, 30 Oct 2024 12:59:18 +0800
From: David Gow <davidgow@...gle.com>
To: Alice Ryhl <aliceryhl@...gle.com>
Cc: Miguel Ojeda <ojeda@...nel.org>, José Expósito <jose.exposito89@...il.com>, 
	Brendan Higgins <brendan.higgins@...ux.dev>, Rae Moar <rmoar@...gle.com>, 
	Boqun Feng <boqun.feng@...il.com>, Alex Gaynor <alex.gaynor@...il.com>, 
	Gary Guo <gary@...yguo.net>, Benno Lossin <benno.lossin@...ton.me>, 
	Björn Roy Baron <bjorn3_gh@...tonmail.com>, 
	Matt Gilbride <mattgilbride@...gle.com>, kunit-dev@...glegroups.com, 
	linux-kselftest@...r.kernel.org, rust-for-linux@...r.kernel.org, 
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 1/3] rust: kunit: add KUnit case and suite macros

On Tue, 29 Oct 2024 at 20:08, Alice Ryhl <aliceryhl@...gle.com> wrote:
>
> On Tue, Oct 29, 2024 at 10:24 AM David Gow <davidgow@...gle.com> wrote:
> >
> > From: José Expósito <jose.exposito89@...il.com>
> >
> > Add a couple of Rust const functions and macros to allow to develop
> > KUnit tests without relying on generated C code:
> >
> >  - The `kunit_unsafe_test_suite!` Rust macro is similar to the
> >    `kunit_test_suite` C macro. It requires a NULL-terminated array of
> >    test cases (see below).
> >  - The `kunit_case` Rust function is similar to the `KUNIT_CASE` C macro.
> >    It generates as case from the name and function.
> >  - The `kunit_case_null` Rust function generates a NULL test case, which
> >    is to be used as delimiter in `kunit_test_suite!`.
> >
> > While these functions and macros can be used on their own, a future
> > patch will introduce another macro to create KUnit tests using a
> > user-space like syntax.
> >
> > Signed-off-by: José Expósito <jose.exposito89@...il.com>
> > Co-developed-by: Matt Gilbride <mattgilbride@...gle.com>
> > Signed-off-by: Matt Gilbride <mattgilbride@...gle.com>
> > Co-developed-by: David Gow <davidgow@...gle.com>
> > Signed-off-by: David Gow <davidgow@...gle.com>
> > ---
> >
> > Changes since v1:
> > https://lore.kernel.org/lkml/20230720-rustbind-v1-1-c80db349e3b5@google.com/
> > - Rebase on top of rust-next
> > - As a result, KUnit attributes are new set. These are hardcoded to the
> >   defaults of "normal" speed and no module name.
> > - Split the kunit_case!() macro into two const functions, kunit_case()
> >   and kunit_case_null() (for the NULL terminator).
> >
> > ---
> >  rust/kernel/kunit.rs | 108 +++++++++++++++++++++++++++++++++++++++++++
> >  rust/kernel/lib.rs   |   1 +
> >  2 files changed, 109 insertions(+)
> >
> > diff --git a/rust/kernel/kunit.rs b/rust/kernel/kunit.rs
> > index 824da0e9738a..fc2d259db458 100644
> > --- a/rust/kernel/kunit.rs
> > +++ b/rust/kernel/kunit.rs
> > @@ -161,3 +161,111 @@ macro_rules! kunit_assert_eq {
> >          $crate::kunit_assert!($name, $file, $diff, $left == $right);
> >      }};
> >  }
> > +
> > +/// Represents an individual test case.
> > +///
> > +/// The test case should have the signature
> > +/// `unsafe extern "C" fn test_case(test: *mut crate::bindings::kunit)`.
> > +///
> > +/// The `kunit_unsafe_test_suite!` macro expects a NULL-terminated list of test cases.
> > +/// Use `kunit_case_null` to generate such a delimeter.
> > +const fn kunit_case(name: &kernel::str::CStr, run_case: unsafe extern "C" fn(*mut kernel::bindings::kunit)) -> kernel::bindings::kunit_case {
>
> This should probably say `name: &'static CStr` to require that the
> name lives forever.

Fixed in v3, thanks.

> > +/// Registers a KUnit test suite.
> > +///
> > +/// # Safety
> > +///
> > +/// `test_cases` must be a NULL terminated array of test cases.
> > +///
> > +/// # Examples
> > +///
> > +/// ```ignore
> > +/// unsafe extern "C" fn test_fn(_test: *mut crate::bindings::kunit) {
> > +///     let actual = 1 + 1;
> > +///     let expected = 2;
> > +///     assert_eq!(actual, expected);
> > +/// }
> > +///
> > +/// static mut KUNIT_TEST_CASE: crate::bindings::kunit_case = crate::kunit_case(name, test_fn);
> > +/// static mut KUNIT_NULL_CASE: crate::bindings::kunit_case = crate::kunit_case_null();
> > +/// static mut KUNIT_TEST_CASES: &mut[crate::bindings::kunit_case] = unsafe {
> > +///     &mut[KUNIT_TEST_CASE, KUNIT_NULL_CASE]
> > +/// };
> > +/// crate::kunit_unsafe_test_suite!(suite_name, KUNIT_TEST_CASES);
> > +/// ```
> > +#[macro_export]
> > +macro_rules! kunit_unsafe_test_suite {
> > +    ($name:ident, $test_cases:ident) => {
> > +        const _: () = {
> > +            static KUNIT_TEST_SUITE_NAME: [i8; 256] = {
> > +                let name_u8 = core::stringify!($name).as_bytes();
> > +                let mut ret = [0; 256];
> > +
> > +                let mut i = 0;
> > +                while i < name_u8.len() {
> > +                    ret[i] = name_u8[i] as i8;
> > +                    i += 1;
> > +                }
>
> I assume the name must be zero-terminated? If so, you probably need to
> enforce that somehow, e.g. by failing if `name_u8` is longer than 255
> bytes.

Nice catch. I'm not sure how to nicely throw a compile time error in
this function, so I'm truncating it here and doing a compile error in
the macro in patch #2. This isn't ideal, but seems to work.

> > +
> > +            static mut KUNIT_TEST_SUITE: core::cell::UnsafeCell<$crate::bindings::kunit_suite> =
>
> You don't actually need UnsafeCell here since it's already `static mut`.
>

Thanks, this works.

> > +                core::cell::UnsafeCell::new($crate::bindings::kunit_suite {
> > +                    name: KUNIT_TEST_SUITE_NAME,
> > +                    // SAFETY: User is expected to pass a correct `test_cases`, hence this macro
> > +                    // named 'unsafe'.
> > +                    test_cases: unsafe { $test_cases.as_mut_ptr() },
> > +                    suite_init: None,
> > +                    suite_exit: None,
> > +                    init: None,
> > +                    exit: None,
> > +                    attr: $crate::bindings::kunit_attributes {
> > +                        speed: $crate::bindings::kunit_speed_KUNIT_SPEED_NORMAL,
> > +                    },
> > +                    status_comment: [0; 256usize],
> > +                    debugfs: core::ptr::null_mut(),
> > +                    log: core::ptr::null_mut(),
> > +                    suite_init_err: 0,
> > +                    is_init: false,
> > +                });
> > +
> > +            #[used]
> > +            #[link_section = ".kunit_test_suites"]
> > +            static mut KUNIT_TEST_SUITE_ENTRY: *const $crate::bindings::kunit_suite =
> > +                // SAFETY: `KUNIT_TEST_SUITE` is static.
> > +                unsafe { KUNIT_TEST_SUITE.get() };
>
> If you change `KUNIT_TEST_SUITE` to not be an UnsafeCell, then you'll
> want to do `core::ptr::addr_of_mut!(KUNIT_TEST_SUITE` instead.
>
> > diff --git a/rust/kernel/lib.rs b/rust/kernel/lib.rs
> > index b62451f64f6e..d662aa37070f 100644
> > --- a/rust/kernel/lib.rs
> > +++ b/rust/kernel/lib.rs
> > @@ -18,6 +18,7 @@
> >  #![feature(inline_const)]
> >  #![feature(lint_reasons)]
> >  #![feature(unsize)]
> > +#![feature(const_mut_refs)]
>
> Is this still needed with the above changes?
>

Yes, this seems to still be needed for the kunit_unsafe_test_suite!() macro.

Thanks,
-- David

Download attachment "smime.p7s" of type "application/pkcs7-signature" (5294 bytes)