Re: [PATCH bpf-next v2] bpf: Speed up htab lookups for u32/u64 keys

[Mykyta Yatsenko <mykyta.yatsenko5@gmail.com>]

On 9/22/26 2:40 AM, Alexei Starovoitov wrote:
> On Mon, Sep 21, 2026 at 02:28 PM Mykyta Yatsenko <mykyta.yatsenko5@gmail.com> wrote:
>> +	/*
>> +	 * Fold the jhash constant and key length into hashrnd once for the
>> +	 * fixed-size fast paths instead of doing it on every lookup.
>> +	 */
>> +	if (htab->map.key_size == sizeof(u32) ||
>> +	    htab->map.key_size == sizeof(u64))
>> +		htab->hashrnd += JHASH_INITVAL + htab->map.key_size;
> 
> This optimization is probably not needed anymore?
> 
I'll double check with benchmark and remove if not needed in v3.>> -static void *__htab_map_lookup_elem(struct bpf_map *map, void *key)
>> +static __always_inline void *__htab_lookup(struct bpf_map *map, void *key, u32 key_size)
> 
> __htab_lookup() is __always_inline, but htab_map_hash() and
> lookup_nulls_elem_raw() that consume key_size are not.
> Whether memcmp() becomes a single compare in _u32/_u64 variants
> is up to the compiler. Make both __always_inline?
> 
Make sense, thanks.