Hi Matthew Wilcox and Andrew Morton,

On 2025/5/7 00:22, Matthew Wilcox wrote:
> 1. This has nothing to do with the compiler version; the type-checking
> is built into min().

Thank you for pointing that out! My previous statement was incorrect.

The error is actually from the __types_ok check within the 
__careful_cmp_once macro failing, which triggered BUILD_BUG_ON.

But then, why do newer compilers compile this without error? Is it 
perhaps because they consider 4U - 1 to be signed?

> 2. We have min_t for a reason
Yes, using min_t instead of min is a better approach.
> 3. Why is a signed min the right answer instead of an unsigned min?
>
That is indeed false. Just as Andrew Morton said, "negative tier numbers 
are nonsensical".

Thank you for everyone's corrections. I'll submit a v2 patch.

-- 
WangYuli