x86, hweight: Use a 32-bit popcnt for __arch_hweight32()
authorH. Peter Anvin <hpa@linux.intel.com>
Mon, 17 May 2010 22:13:23 +0000 (15:13 -0700)
committerH. Peter Anvin <hpa@linux.intel.com>
Mon, 17 May 2010 22:17:16 +0000 (15:17 -0700)
commitc59bd5688299cddb71183e156e7a3c1409b90df2
treed7b99407367b30aa7766219083ad0262c8edbab8
parent4677d4a53e0d565742277e8913e91c821453e63e
x86, hweight: Use a 32-bit popcnt for __arch_hweight32()

Use a 32-bit popcnt instruction for __arch_hweight32(), even on
x86-64.  Even though the input register will *usually* be
zero-extended due to the standard operation of the hardware, it isn't
necessarily so if the input value was the result of truncating a
64-bit operation.

Note: the POPCNT32 variant used on x86-64 has a technically
unnecessary REX prefix to make it five bytes long, the same as a CALL
instruction, therefore avoiding an unnecessary NOP.

Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Cc: Borislav Petkov <borislav.petkov@amd.com>
LKML-Reference: <alpine.LFD.2.00.1005171443060.4195@i5.linux-foundation.org>
arch/x86/include/asm/arch_hweight.h