Re: [PATCH 1/4] hashmap: introduce memhash_feed to access the internals of FNV-1 hash

From: Stefan Beller <sbeller@google.com>
Date: Oct 24, 2017, 20:48 UTC
Message-ID: <CAGZ79kaQA90t6o2bn=H4Xrf2RiBOHc=gBE8XuKb+iHB6oEQVqA@mail.gmail.com>
In-Reply-To: <06c950e8-92c6-4926-ccac-2827d480f373@web.de>

>> +unsigned int memhash_feed(unsigned int hash_seed, const unsigned char next)
>
> Why is the second parameter const and the first one isn't?  (We tend
> not to bother with const for value types.)

will do.

Show 7 quoted lines

>> +             hash = memhash_feed(hash, c);
>
> I guess compilers inline a copy of the function here with -O2.  My
> knee-jerk reaction, however, is horror in the face of adding a function
> call to the inner loop of a hash function.  Do you have performance
> test results, ideally also with -O0?  And why not make memhash_feed()
> an inline function or macro to sidestep that issue?

My gut feeling was similar, but then I assumed compilers of today would be smart.

As per the discussion on a later patch, we could migrate all memhash* functions to fnv32.c except for the _feed, which will be static in its header fnv32.h

Thanks, Stefan