Re: [PATCH v16 15/16] barrier: timeout validity checks for smp_cond_load_relaxed_timeout()

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] kselftest: barrier: incorrectly skips invalid timeout tests on ARM64
--

commit bda3f35055835da4a210dfc722306737806214d9
Author: Ankur Arora <ankur.a.arora@oracle.com>

barrier: timeout validity checks for smp_cond_load_relaxed_timeout()

This commit adds timeout tests for smp_cond_load_relaxed_timeout() to verify
that the implementation returns early on invalid timeout values and handles
edge cases sanely.

> diff --git a/lib/tests/barrier-timeout-test.c b/lib/tests/barrier-timeout-test.c
> index 9c12cd36e7d29..88b374ceb78b4 100644
> --- a/lib/tests/barrier-timeout-test.c
> +++ b/lib/tests/barrier-timeout-test.c
[ ... ]
> +static char *expiry_skip_test(const struct smp_cond_expiry_params *p)
> +{
> +#ifdef CONFIG_ARM64
> +
> +	/*
> +	 * WFXT waits for the full p->timeout_ns so limit it to a reasonable
> +	 * duration.
> +	 */
> +	if (alternative_has_cap_unlikely(ARM64_HAS_WFXT) &&
> +	    p->timeout_ns >= NSEC_PER_SEC) {

[Severity: Low]
Does this skip the invalid timeout test cases on ARM64 by mistake?

The expiry_params_list includes intentional negative values (like -1LL or
~0ULL) to verify error handling, but p->timeout_ns is defined as a u64.
When these invalid inputs are compared as unsigned, they evaluate to numbers
larger than NSEC_PER_SEC, causing them to be skipped with the reason
"runtime duration too long".

Because the smp_cond_load_relaxed_timeout() macro casts timeouts to s64 and
returns instantly for negative values, these cases never actually execute a
WFXT sleep. Skipping them deprives ARM64 of test coverage for the
early-return fast path.

Could this regression be resolved by checking if
(s64)p->timeout_ns >= NSEC_PER_SEC, or by avoiding the skip when
p->maxiters == 0?

> +
> +		return "runtime duration too long";
> +	}
> +#endif
> +	return NULL;
> +}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261016070837.440748-1-ankur.a.arora@oracle.com?part=15