linux-doc.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Dave Martin <Dave.Martin@arm.com>
To: Devarsh Thakkar <devarsht@ti.com>
Cc: mchehab@kernel.org, hverkuil-cisco@xs4all.nl,
	linux-media@vger.kernel.org, linux-kernel@vger.kernel.org,
	sebastian.fricke@collabora.com, linux-doc@vger.kernel.org,
	praneeth@ti.com, nm@ti.com, vigneshr@ti.com, s-jain1@ti.com,
	r-donadkar@ti.com, b-brnich@ti.com, detheridge@ti.com,
	p-mantena@ti.com, vijayp@ti.com, andi.shyti@linux.intel.com,
	nicolas@ndufresne.ca, andriy.shevchenko@linux.intel.com,
	jirislaby@kernel.org, davidgow@google.com, dlatypov@google.com,
	corbet@lwn.net, broonie@kernel.org, jani.nikula@intel.com,
	rdunlap@infradead.org, nik.borisov@suse.com
Subject: Re: [PATCH v2 5/6] lib: math_kunit: Add tests for new macros related to rounding to nearest value
Date: Fri, 30 Aug 2024 15:07:31 +0100	[thread overview]
Message-ID: <ZtHSIyDbljwrkWBE@e133380.arm.com> (raw)
In-Reply-To: <20240826150822.4057164-6-devarsht@ti.com>

On Mon, Aug 26, 2024 at 08:38:21PM +0530, Devarsh Thakkar wrote:
> Add tests for round_closest_up/down and roundclosest macros which round
> to nearest multiple of specified argument. These are tested with kunit
> tool as shared here [1] :

The tests here don't seem to be well targeted at the actual edge cases
where incrementing or decremting the dividend by 1 is expected to
change the result, or when round_closest_up() and round_closest_down()
are expected to differ.

There's also no coverage of:

	* negative inputs

	* types other than int

	* inputs close to type limits

(I've highlighted a few specific issues below, though there seems to be
a more general lack of coverage here.)

> 
> Link: https://gist.github.com/devarsht/3f9042825be3da4e133b8f4eda067876 [1]
> Signed-off-by: Devarsh Thakkar <devarsht@ti.com>
> Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> ---
> V2: No change
> ---
>  lib/math/math_kunit.c | 35 +++++++++++++++++++++++++++++++++++
>  1 file changed, 35 insertions(+)
> 
> diff --git a/lib/math/math_kunit.c b/lib/math/math_kunit.c
> index be27f2afb8e4..05022f010be6 100644
> --- a/lib/math/math_kunit.c
> +++ b/lib/math/math_kunit.c
> @@ -70,6 +70,26 @@ static void round_down_test(struct kunit *test)
>  	KUNIT_EXPECT_EQ(test, round_down((1 << 30) - 1, 1 << 29), 1 << 29);
>  }
>  
> +static void round_closest_up_test(struct kunit *test)
> +{
> +	KUNIT_EXPECT_EQ(test, round_closest_up(17, 4), 16);
> +	KUNIT_EXPECT_EQ(test, round_closest_up(15, 4), 16);
> +	KUNIT_EXPECT_EQ(test, round_closest_up(14, 4), 16);
> +	KUNIT_EXPECT_EQ(test, round_closest_up((1 << 30) - 1, 1 << 30), 1 << 30);
> +	KUNIT_EXPECT_EQ(test, round_closest_up((1 << 30) + 1, 1 << 30), 1 << 30);

These miss the edge case where the result is expected to change; could
we also check:

	round_closest_up(1 << 29, 1 << 30)


> +	KUNIT_EXPECT_EQ(test, round_closest_up((1 << 30) - 1, 2), 1 << 30);
> +}
> +
> +static void round_closest_down_test(struct kunit *test)
> +{
> +	KUNIT_EXPECT_EQ(test, round_closest_down(17, 4), 16);
> +	KUNIT_EXPECT_EQ(test, round_closest_down(15, 4), 16);
> +	KUNIT_EXPECT_EQ(test, round_closest_down(14, 4), 12);
> +	KUNIT_EXPECT_EQ(test, round_closest_down((1 << 30) - 1, 1 << 30), 1 << 30);
> +	KUNIT_EXPECT_EQ(test, round_closest_down((1 << 30) + 1, 1 << 30), 1 << 30);
> +	KUNIT_EXPECT_EQ(test, round_closest_down((1 << 30) - 1, 2), (1 << 30) - 2);
> +}
> +
>  /* These versions can round to numbers that aren't a power of two */
>  static void roundup_test(struct kunit *test)
>  {
> @@ -95,6 +115,18 @@ static void rounddown_test(struct kunit *test)
>  	KUNIT_EXPECT_EQ(test, rounddown(4, 3), 3);
>  }
>  
> +static void roundclosest_test(struct kunit *test)
> +{
> +	KUNIT_EXPECT_EQ(test, roundclosest(21, 5), 20);
> +	KUNIT_EXPECT_EQ(test, roundclosest(19, 5), 20);

This seems to miss the edge cases (e.g., roundclosest(20 * N + 10, 20),
roundclosest(20 * N +/- 9, 20) for some N.

> +	KUNIT_EXPECT_EQ(test, roundclosest(17, 5), 15);
> +	KUNIT_EXPECT_EQ(test, roundclosest((1 << 30), 3), (1 << 30) - 1);
> +	KUNIT_EXPECT_EQ(test, roundclosest((1 << 30) - 1, 1 << 29), 1 << 30);
> +
> +	KUNIT_EXPECT_EQ(test, roundclosest(4, 3), 3);
> +	KUNIT_EXPECT_EQ(test, roundclosest(5, 3), 6);
> +}
> +

[...]

Cheers
---Dave

  reply	other threads:[~2024-08-30 14:07 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-26 15:08 [PATCH v2 0/6] Add rounding macros and enable KUnit tests Devarsh Thakkar
2024-08-26 15:08 ` [PATCH v2 1/6] math.h: Add macros for rounding to the closest value Devarsh Thakkar
2024-08-26 17:44   ` Andy Shevchenko
2024-08-27 11:27     ` Devarsh Thakkar
2024-08-27 12:40       ` Jani Nikula
2024-08-27 15:14         ` Devarsh Thakkar
2024-08-29  9:19           ` Jani Nikula
2024-08-29  9:54             ` Jiri Slaby
2024-08-30 16:09               ` Devarsh Thakkar
2024-09-30  1:18               ` Randy Dunlap
2024-08-30 13:45   ` Dave Martin
2024-08-26 15:08 ` [PATCH v2 2/6] math.h: Use kernel-doc syntax for division macros Devarsh Thakkar
2024-08-26 15:08 ` [PATCH v2 3/6] Documentation: core-api: Add math.h macros and functions Devarsh Thakkar
2024-08-26 15:08 ` [PATCH v2 4/6] lib: Add basic KUnit test for lib/math Devarsh Thakkar
2024-08-26 15:08 ` [PATCH v2 5/6] lib: math_kunit: Add tests for new macros related to rounding to nearest value Devarsh Thakkar
2024-08-30 14:07   ` Dave Martin [this message]
2024-08-26 15:08 ` [PATCH v2 6/6] media: imagination: Round to closest multiple for cropping region Devarsh Thakkar

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ZtHSIyDbljwrkWBE@e133380.arm.com \
    --to=dave.martin@arm.com \
    --cc=andi.shyti@linux.intel.com \
    --cc=andriy.shevchenko@linux.intel.com \
    --cc=b-brnich@ti.com \
    --cc=broonie@kernel.org \
    --cc=corbet@lwn.net \
    --cc=davidgow@google.com \
    --cc=detheridge@ti.com \
    --cc=devarsht@ti.com \
    --cc=dlatypov@google.com \
    --cc=hverkuil-cisco@xs4all.nl \
    --cc=jani.nikula@intel.com \
    --cc=jirislaby@kernel.org \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=mchehab@kernel.org \
    --cc=nicolas@ndufresne.ca \
    --cc=nik.borisov@suse.com \
    --cc=nm@ti.com \
    --cc=p-mantena@ti.com \
    --cc=praneeth@ti.com \
    --cc=r-donadkar@ti.com \
    --cc=rdunlap@infradead.org \
    --cc=s-jain1@ti.com \
    --cc=sebastian.fricke@collabora.com \
    --cc=vigneshr@ti.com \
    --cc=vijayp@ti.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).