Re: [PATCH 2/2] [CI][DONOT-REVIEW] gpu/tests/buddy: Add range-bias pruning and free-order tracking tests

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] kunit: flaky test due to absolute time bound in gpu_test_buddy_range_bias_pruning
--

commit c5175de7da719ada087bdec652e790f13b219a19
Author: Sanjay Yadav <sanjay.kumar.yadav@intel.com>

gpu/tests/buddy: Add range-bias pruning and free-order tracking tests

This commit adds two KUnit cases for subtree_max_free_order and the pruning
in __alloc_range_bias(). It introduces the gpu_test_buddy_range_bias_pruning
test to reproduce the small-BAR pattern and a free-order test to validate
metadata after various block operations.

> diff --git a/drivers/gpu/tests/gpu_buddy_test.c b/drivers/gpu/tests/gpu_buddy_test.c
> index b75d32ca6ca0a..bdf480d782beb 100644
> --- a/drivers/gpu/tests/gpu_buddy_test.c
> +++ b/drivers/gpu/tests/gpu_buddy_test.c
[ ... ]
> +static void gpu_test_buddy_range_bias_pruning(struct kunit *test)
> +{
[ ... ]
> +	for (i = 0; i < nr; i++) {
> +		ktime_t start = ktime_get();
> +		u64 elapsed;
> +		int err;
> +
> +		err = gpu_buddy_alloc_blocks(&mm, 0, range_end, chunk, chunk,
> +					     &allocated, GPU_BUDDY_RANGE_ALLOCATION);
> +		elapsed = ktime_to_ns(ktime_sub(ktime_get(), start));

[Severity: Low]
Could this absolute time measurement cause spurious test failures under heavy
system load?

KUnit tests execute in environments with preemption, interrupts, and
potential scheduling delays. Measuring wall-clock time here includes all
preemption and interrupt overhead that might occur during the allocation.

[ ... ]
> +	/* The exhausted range advertises no free space, so it is pruned */
> +	block = gpu_test_find_block(&mm, 0, ilog2(range_end / chunk));
> +	KUNIT_ASSERT_NOT_NULL(test, block);
> +	KUNIT_EXPECT_EQ(test, block->subtree_max_free_order, GPU_BUDDY_NO_FREE_ORDER);
> +	gpu_test_check_free_order(test, &mm);
> +
> +	/* Per-allocation cost must not grow with the number of allocated blocks */
> +	KUNIT_EXPECT_LE(test, last_ns, 8 * first_ns + 50 * NSEC_PER_MSEC);

[Severity: Low]
Does this strict time bound make the test flaky?

Asserting on absolute wall-clock time differences (adding a strict 50ms
allowance) is prone to spurious failures when the system is under load or
when the test thread is preempted between the start and end of the timing
windows.

[ ... ]

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261011112117.1170859-3-sanjay.kumar.yadav@intel.com?part=2