[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAK7LNATg+cnvs9P5Ud6FPAoqQUozrGif5SJm+Pw7cs5qJ1C3gw@mail.gmail.com>
Date: Mon, 25 Sep 2023 23:24:11 +0900
From: Masahiro Yamada <masahiroy@...nel.org>
To: Fangrui Song <maskray@...gle.com>
Cc: Jack Brennen <jbrennen@...gle.com>,
Nathan Chancellor <nathan@...nel.org>,
Nick Desaulniers <ndesaulniers@...gle.com>,
Nicolas Schier <nicolas@...sle.eu>, Tom Rix <trix@...hat.com>,
linux-kernel@...r.kernel.org, linux-kbuild@...r.kernel.org,
llvm@...ts.linux.dev
Subject: Re: [PATCH] modpost: Optimize symbol search from linear to binary search
On Mon, Sep 25, 2023 at 7:20 AM Fangrui Song <maskray@...gle.com> wrote:
> However, the `- 1` part in `unsigned int hi = elf->symsearch->table_size - 1;` can be improved.
> I'd prefer an implementation similar to typical C++ https://en.cppreference.com/w/cpp/algorithm/upper_bound implementation.
>
> lo = 0;
> hi = n; // or replace hi with count
> while (lo < hi) {
> mid = (lo + hi) / 2; // we don't care about (lo+hi) overflow
> if (less_or_eq(&table[mid], &target))
> lo = mid+1;
> else
> hi = mid;
> }
>
> // lo == hi: the index of the first element that is > target
> // if elements equal to target are present, they are on the left of lo
Ah, it is much more elegant!
Thanks.
--
Best Regards
Masahiro Yamada
Powered by blists - more mailing lists