[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241003211829.2750436-1-nico@fluxnic.net>
Date: Thu, 3 Oct 2024 17:16:12 -0400
From: Nicolas Pitre <nico@...xnic.net>
To: Arnd Bergmann <arnd@...db.de>,
Russell King <linux@...linux.org.uk>
Cc: Nicolas Pitre <npitre@...libre.com>,
linux-arch@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: [PATCH v4 0/4] simplify do_div() with constant divisor
While working on mul_u64_u64_div_u64() improvements I realized that there
is a better way to perform a 64x64->128 bits multiplication with overflow
handling.
Change from v3:
- Added timings to commit log of patch #4.
Link to v3: https://lore.kernel.org/lkml/20240708012749.2098373-2-nico@fluxnic.net/T/
Change from v2:
- Fix last minute edit screw-up (missing one function return type).
Link to v2: https://lore.kernel.org/lkml/20240707171919.1951895-1-nico@fluxnic.net/
Changes from v1:
- Formalize condition for when overflow handling can be skipped.
- Make this condition apply only if it can be determined at compile time
(guard against the compiler not always inling code).
- Keep the ARM assembly but apply the above changes to it as well.
- Force __always_inline when optimizing for performance.
- Augment test_div64.c with important edge cases.
Link to v1: https://lore.kernel.org/lkml/20240705022334.1378363-1-nico@fluxnic.net/
The diffstat is:
arch/arm/include/asm/div64.h | 13 +++-
include/asm-generic/div64.h | 121 ++++++++++++-----------------------
lib/math/test_div64.c | 85 +++++++++++++++++++++++-
3 files changed, 134 insertions(+), 85 deletions(-)
Powered by blists - more mailing lists