Re: [PATCH bpf v2 2/2] selftests/bpf: Add test for usdt probe with page-crossing nop10

[Jiayuan Chen <jiayuan.chen@linux.dev>]

On 8/25/26 11:13 PM, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [Medium] The search loop limit `2 * page_sz` for locating the USDT probe can cause the test to falsely fail if the compiler generates a prologue of 11 bytes or larger.
> --
>
> --- Patch [2]: [PATCH bpf v2 2/2] selftests/bpf: Add test for usdt probe with page-crossing nop10 ---
> commit 3ab03ec0febc59ef7c61c7f1449047ef827d5e7b
> Author: Jiayuan Chen <jiayuan.chen@linux.dev>
>
> selftests/bpf: Add test for usdt probe with page-crossing nop10
>
> This commit adds a USDT probe layout where the nop10 crosses a page boundary.
> It verifies that attaching to it succeeds and the probe fires without the
> uprobe being shifted onto the nop10 incorrectly.
>
>> diff --git a/tools/testing/selftests/bpf/prog_tests/usdt.c b/tools/testing/selftests/bpf/prog_tests/usdt.c
>> index 8004c9568ffa5..eff1e57ab13c0 100644
>> --- a/tools/testing/selftests/bpf/prog_tests/usdt.c
>> +++ b/tools/testing/selftests/bpf/prog_tests/usdt.c
> [ ... ]
>> @@ -342,6 +343,57 @@ static void subtest_optimized_attach(void)
>>   	test_usdt__destroy(skel);
>>   }
>>   
>> +/*
>> + * Test attachment to a USDT probe whose nop10 crosses a page boundary.
>> + * The kernel can't optimize such nop10, so libbpf keeps the uprobe on
>> + * the preceding 1-byte nop. Do not assume any particular placement
>> + * here, though: however the probe ends up attached, the attachment
>> + * must succeed and the probe must fire.
>> + */
>> +static void subtest_optimized_attach_cross_page(void)
>> +{
>> +	long page_sz = sysconf(_SC_PAGESIZE);
>> +	struct test_usdt *skel;
>> +	__u8 *addr = NULL;
>> +	long i;
>> +
>> +	/* combo is placed up to a page of padding after the function start */
>> +	for (i = 0; i < 2 * page_sz; i++) {
> [Severity: Medium]
> Will this loop bound of 2 * page_sz (8192) cause the test to falsely fail on
> builds with larger compiler prologues (like ASAN builds or those with frame
> pointers enabled)?
>
> If the compiler generates a prologue of size P, the offset to the probe is
> P + align_padding (up to 4095) + skip_padding (4086) = P + 8181.
>
> If P is 11 bytes or larger, the offset reaches 8192, which would exceed
> this i < 2 * page_sz search bound and cause a spurious ASSERT_OK_PTR
> failure.

I got such review locally, but I want to say that if such a compiler 
existed, the existing tests would break before

this one: find_instr() in subtest_optimized_attach() only scans 10 bytes 
past the function start.

>> +		if (!memcmp((void *)usdt_2_cross_page + i, nop1_nop10_combo, 11)) {
>> +			addr = (void *)usdt_2_cross_page + i;
>> +			break;
>> +		}
>> +	}
>> +	if (!ASSERT_OK_PTR(addr, "find_nop1_nop10_combo"))
>> +		return;