[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <87ilej51x0.fsf@intel.com>
Date: Wed, 29 Mar 2023 12:42:35 +0300
From: Jani Nikula <jani.nikula@...ux.intel.com>
To: David Gow <davidgow@...gle.com>,
Luís Mendes <luis.p.mendes@...il.com>,
Christian König <christian.koenig@....com>,
Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
Maxime Ripard <mripard@...nel.org>,
Thomas Zimmermann <tzimmermann@...e.de>,
David Airlie <airlied@...il.com>,
Daniel Vetter <daniel@...ll.ch>,
Maíra Canal
<mairacanal@...eup.net>, Arthur Grillo <arthurgrillo@...eup.net>
Cc: dri-devel@...ts.freedesktop.org, David Gow <davidgow@...gle.com>,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] drm: test: Fix 32-bit issue in drm_buddy_test
On Wed, 29 Mar 2023, Jani Nikula <jani.nikula@...ux.intel.com> wrote:
> On Wed, 29 Mar 2023, David Gow <davidgow@...gle.com> wrote:
>> The drm_buddy_test KUnit tests verify that returned blocks have sizes
>> which are powers of two using is_power_of_2(). However, is_power_of_2()
>> operations on a 'long', but the block size is a u64. So on systems where
>> long is 32-bit, this can sometimes fail even on correctly sized blocks.
>>
>> This only reproduces randomly, as the parameters passed to the buddy
>> allocator in this test are random. The seed 0xb2e06022 reproduced it
>> fine here.
>>
>> For now, just hardcode an is_power_of_2() implementation using
>> x & (x - 1).
>>
>> Signed-off-by: David Gow <davidgow@...gle.com>
>> ---
>>
>> There are actually a couple of is_power_of_2_u64() implementations
>> already around in:
>> - drivers/gpu/drm/i915/i915_utils.h
>> - fs/btrfs/misc.h (called is_power_of_two_u64)
>>
>> So the ideal thing would be to consolidate these in one place.
>>
>>
>> ---
>> drivers/gpu/drm/tests/drm_buddy_test.c | 3 ++-
>> 1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/tests/drm_buddy_test.c b/drivers/gpu/drm/tests/drm_buddy_test.c
>> index f8ee714df396..09ee6f6af896 100644
>> --- a/drivers/gpu/drm/tests/drm_buddy_test.c
>> +++ b/drivers/gpu/drm/tests/drm_buddy_test.c
>> @@ -89,7 +89,8 @@ static int check_block(struct kunit *test, struct drm_buddy *mm,
>> err = -EINVAL;
>> }
>>
>> - if (!is_power_of_2(block_size)) {
>> + /* We can't use is_power_of_2() for a u64 on 32-bit systems. */
>> + if (block_size & (block_size - 1)) {
>
> Then maybe use is_power_of_2_u64() instead?
*sigh* And as soon as I wrote that I realized it's a local thing in
btrfs. Never mind for now...
...but in the long run would be nice to either fix is_power_of_2() for
u64 or add that u64 version...
BR,
Jani.
>
> BR,
> Jani.
>
>> kunit_err(test, "block size not power of two\n");
>> err = -EINVAL;
>> }
--
Jani Nikula, Intel Open Source Graphics Center
Powered by blists - more mailing lists