From: Kuan-Wei Chiu Date: Tue, 30 Sep 2025 04:30:55 +0000 (+0800) Subject: interconnect: Optimize kbps_to_icc() macro X-Git-Tag: v6.19-rc1~65^2~55^2~7 X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=245f14f5fe283c782b16143280f283bee29dbb5f;p=thirdparty%2Fkernel%2Flinux.git interconnect: Optimize kbps_to_icc() macro The current expansion of kbps_to_icc() introduces unnecessary logic when compiled from a general expression. Rewriting it allows compilers to emit shorter and more efficient code across architectures. For example, with gcc -O2: arm64: old: tst x0, 7 add w1, w0, 7 cset w2, ne cmp w0, 0 csel w0, w1, w0, lt add w0, w2, w0, asr 3 new: add w1, w0, 14 adds w0, w0, 7 csel w0, w1, w0, mi asr w0, w0, 3 x86-64: old: xor eax, eax test dil, 7 lea edx, [rdi+7] setne al test edi, edi cmovns edx, edi sar edx, 3 add eax, edx new: lea eax, [rdi+14] add edi, 7 cmovns eax, edi sar eax, 3 In both cases the old form relies on extra test and compare instructions (tst, test, cmp) combined with conditional moves or sets, while the new form uses fewer instructions by folding the addition and flag update together (adds on arm64, add on x86). This reduces the instruction sequence, prevents multiple evaluations of x when it is an expression or a function call, and keeps the macro simpler. Signed-off-by: Kuan-Wei Chiu Link: https://lore.kernel.org/r/20250930043055.2200322-1-visitorckw@gmail.com Signed-off-by: Georgi Djakov --- diff --git a/include/linux/interconnect.h b/include/linux/interconnect.h index e4b8808823ad8..4b12821528a6b 100644 --- a/include/linux/interconnect.h +++ b/include/linux/interconnect.h @@ -16,7 +16,7 @@ #define MBps_to_icc(x) ((x) * 1000) #define GBps_to_icc(x) ((x) * 1000 * 1000) #define bps_to_icc(x) (1) -#define kbps_to_icc(x) ((x) / 8 + ((x) % 8 ? 1 : 0)) +#define kbps_to_icc(x) (((x) + 7) / 8) #define Mbps_to_icc(x) ((x) * 1000 / 8) #define Gbps_to_icc(x) ((x) * 1000 * 1000 / 8)