2022-10-17 04:54:41

by Jason A. Donenfeld

Subject: [PATCH] kcsan: remove rng selftest

The first test of the kcsan selftest appears to test if get_random_u32()
returns two zeros in a row, and requires that it doesn't. This seems
like a bogus critera. Remove it.

Signed-off-by: Jason A. Donenfeld <[email protected]>
---
kernel/kcsan/selftest.c | 8 --------
1 file changed, 8 deletions(-)

diff --git a/kernel/kcsan/selftest.c b/kernel/kcsan/selftest.c
index 00cdf8fa5693..1740ce389e7f 100644
--- a/kernel/kcsan/selftest.c
+++ b/kernel/kcsan/selftest.c
@@ -22,13 +22,6 @@

#define ITERS_PER_TEST 2000

-/* Test requirements. */
-static bool __init test_requires(void)
-{
- /* random should be initialized for the below tests */
- return get_random_u32() + get_random_u32() != 0;
-}
-
/*
* Test watchpoint encode and decode: check that encoding some access's info,
* and then subsequent decode preserves the access's info.
@@ -259,7 +252,6 @@ static int __init kcsan_selftest(void)
pr_err("selftest: " #do_test " failed"); \
} while (0)

- RUN_TEST(test_requires);
RUN_TEST(test_encode_decode);
RUN_TEST(test_matching_access);
RUN_TEST(test_barrier);
--
2.37.3

2022-10-17 05:12:59

by Marco Elver

Subject: Re: [PATCH] kcsan: remove rng selftest

On Sun, 16 Oct 2022 at 21:43, Jason A. Donenfeld <[email protected]> wrote:
>
> The first test of the kcsan selftest appears to test if get_random_u32()
> returns two zeros in a row, and requires that it doesn't. This seems
> like a bogus critera. Remove it.
>
> Signed-off-by: Jason A. Donenfeld <[email protected]>

Acked-by: Marco Elver <[email protected]>

Looks pretty redundant at this point (I think some early version had
it because somehow I managed to run the test too early and wanted to
avoid that accidentally happening again).

> ---
> kernel/kcsan/selftest.c | 8 --------
> 1 file changed, 8 deletions(-)
>
> diff --git a/kernel/kcsan/selftest.c b/kernel/kcsan/selftest.c
> index 00cdf8fa5693..1740ce389e7f 100644
> --- a/kernel/kcsan/selftest.c
> +++ b/kernel/kcsan/selftest.c
> @@ -22,13 +22,6 @@
>
> #define ITERS_PER_TEST 2000
>
> -/* Test requirements. */
> -static bool __init test_requires(void)
> -{
> - /* random should be initialized for the below tests */
> - return get_random_u32() + get_random_u32() != 0;
> -}
> -
> /*
> * Test watchpoint encode and decode: check that encoding some access's info,
> * and then subsequent decode preserves the access's info.
> @@ -259,7 +252,6 @@ static int __init kcsan_selftest(void)
> pr_err("selftest: " #do_test " failed"); \
> } while (0)
>
> - RUN_TEST(test_requires);
> RUN_TEST(test_encode_decode);
> RUN_TEST(test_matching_access);
> RUN_TEST(test_barrier);
> --
> 2.37.3
>

2022-10-17 05:14:54

by Marco Elver

Subject: Re: [PATCH] kcsan: remove rng selftest

On Sun, 16 Oct 2022 at 22:07, Marco Elver <[email protected]> wrote:
>
> On Sun, 16 Oct 2022 at 21:43, Jason A. Donenfeld <[email protected]> wrote:
> >
> > The first test of the kcsan selftest appears to test if get_random_u32()
> > returns two zeros in a row, and requires that it doesn't. This seems
> > like a bogus critera. Remove it.
> >
> > Signed-off-by: Jason A. Donenfeld <[email protected]>
>
> Acked-by: Marco Elver <[email protected]>
>
> Looks pretty redundant at this point (I think some early version had
> it because somehow I managed to run the test too early and wanted to
> avoid that accidentally happening again).
>

And kindly queue it in your tree with all the rng related changes. Thanks!

> > ---
> > kernel/kcsan/selftest.c | 8 --------
> > 1 file changed, 8 deletions(-)
> >
> > diff --git a/kernel/kcsan/selftest.c b/kernel/kcsan/selftest.c
> > index 00cdf8fa5693..1740ce389e7f 100644
> > --- a/kernel/kcsan/selftest.c
> > +++ b/kernel/kcsan/selftest.c
> > @@ -22,13 +22,6 @@
> >
> > #define ITERS_PER_TEST 2000
> >
> > -/* Test requirements. */
> > -static bool __init test_requires(void)
> > -{
> > - /* random should be initialized for the below tests */
> > - return get_random_u32() + get_random_u32() != 0;
> > -}
> > -
> > /*
> > * Test watchpoint encode and decode: check that encoding some access's info,
> > * and then subsequent decode preserves the access's info.
> > @@ -259,7 +252,6 @@ static int __init kcsan_selftest(void)
> > pr_err("selftest: " #do_test " failed"); \
> > } while (0)
> >
> > - RUN_TEST(test_requires);
> > RUN_TEST(test_encode_decode);
> > RUN_TEST(test_matching_access);
> > RUN_TEST(test_barrier);
> > --
> > 2.37.3
> >

2022-10-17 18:23:58

by Jason A. Donenfeld

Subject: Re: [PATCH] kcsan: remove rng selftest

On Sun, Oct 16, 2022 at 11:09 PM Marco Elver <[email protected]> wrote:
>
> On Sun, 16 Oct 2022 at 22:07, Marco Elver <[email protected]> wrote:
> >
> > On Sun, 16 Oct 2022 at 21:43, Jason A. Donenfeld <[email protected]> wrote:
> > >
> > > The first test of the kcsan selftest appears to test if get_random_u32()
> > > returns two zeros in a row, and requires that it doesn't. This seems
> > > like a bogus critera. Remove it.
> > >
> > > Signed-off-by: Jason A. Donenfeld <[email protected]>
> >
> > Acked-by: Marco Elver <[email protected]>
> >
> > Looks pretty redundant at this point (I think some early version had
> > it because somehow I managed to run the test too early and wanted to
> > avoid that accidentally happening again).
> >
>
> And kindly queue it in your tree with all the rng related changes. Thanks!

Okay sure, will do.

Jason