Re: [PATCH bpf-next v3 09/11] selftests/bpf: Test global functions returning arena pointers by value

[Yonghong Song <yonghong.song@linux.dev>]

On 8/27/26 12:04 AM, bot+bpf-ci@kernel.org wrote:
>> Cover the by-value struct returns a global function may now make: two
>> arena pointers filling R0:R2, an arena pointer beside a scalar, an array
>> of them, and an eight byte struct returned in R0 alone. The existing
>> cases for a struct and a union carrying a plain pointer stay rejected.
> This isn't a bug, but would it be worth adding the new union-of-arena-
> pointer case to the enumeration? The patch adds union arena_upair /
> global_ret_arena_union() / aggregate_ret_global_arena_union() as a
> sixteen-byte union of an arena pointer and a __u64[2]. Since the next
> sentence says "cases for a struct and a union carrying a plain pointer
> stay rejected", a reader might think the only union touched is a rejected
> one, when the patch actually adds a newly-accepted union case as well.

There is no need to have too many union cases. The existing one
should be enough.

>
>> diff --git a/tools/testing/selftests/bpf/progs/verifier_arena.c b/tools/testing/selftests/bpf/progs/verifier_arena.c
>> index 815f342eb4b05..672a3aadd9cb2 100644
>> --- a/tools/testing/selftests/bpf/progs/verifier_arena.c
>> +++ b/tools/testing/selftests/bpf/progs/verifier_arena.c
>> @@ -734,4 +734,52 @@ int check_arena_arg_ret(void *ctx)
>>   	return 0;
>>   }
>>
>> +#if defined(__clang_major__) && __clang_major__ >= 23
>> +
>> +struct arena_page_pair {
>> +	u32 __arena *first;
>> +	u32 __arena *second;
>> +};
>> +
>> +__weak struct arena_page_pair split_arena_page(u32 __arena *page)
>> +{
>> +	struct arena_page_pair pair;
>> +
>> +	pair.first = page;
>> +	pair.second = page + 1;
>> +
>> +	return pair;
>> +}
> This isn't a bug, but since the returned pointers are the first two u32
> slots of a single page rather than two halves of it (page and page + 1
> are offsets 0 and 4 within a single 4096-byte allocation), would a name
> like arena_word_pair() / struct arena_word_pair describe the helper more
> directly? The neighbouring arena return tests in the same file name
> themselves after what they exercise rather than after an imagined data
> layout (check_arena_arg_ret, check_arena_arg_quals1/2).

Okay, I will use arena_work_pair() then.

>
>
> ---
> AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
> See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md
>
> CI run summary: https://github.com/kernel-patches/bpf/actions/runs/33045482525