[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <F1D1D3D4-CFB0-47A8-87F0-E1D8A6B5904F@toblux.com>
Date: Fri, 3 May 2024 16:49:05 +0200
From: Thorsten Blum <thorsten.blum@...lux.com>
To: Arnd Bergmann <arnd@...db.de>
Cc: Amadeusz Sławiński <amadeuszx.slawinski@...ux.intel.com>,
Xiao W Wang <xiao.w.wang@...el.com>,
Palmer Dabbelt <palmer@...osinc.com>,
Charlie Jenkins <charlie@...osinc.com>,
Namhyung Kim <namhyung@...nel.org>,
Huacai Chen <chenhuacai@...nel.org>,
Youling Tang <tangyouling@...ngson.cn>,
Tiezhu Yang <yangtiezhu@...ngson.cn>,
Jinyang He <hejinyang@...ngson.cn>,
Linux-Arch <linux-arch@...r.kernel.org>,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] bitops: Change function return types from long to int
On 22. Apr 2024, at 17:55, Arnd Bergmann <arnd@...db.de> wrote:
>>
>> I can generally merge such a series with architecture specific
>> changes through the asm-generic tree, with the appropriate Acks
>> from the maintainers.
What would it take for this patch (with only generic type changes) to be
applied?
I did some further investigations and disassembled my test kernel images.
The patch reduced the number of ARM instructions by 872 with GCC 13 and by
2,354 with GCC 14. Other architectures that rely on the generic bitops
functions will most likely also benefit from this patch.
Tests were done with base commit 9d1ddab261f3e2af7c384dc02238784ce0cf9f98.
Thanks,
Thorsten
Powered by blists - more mailing lists