From 9e79735ac00de56c13a4fc4b1e03148c98a48553 Mon Sep 17 00:00:00 2001 From: leoliu-oc Date: Thu, 13 Jun 2024 16:22:54 +0800 Subject: [PATCH] Add MWAIT Cx support for Zhaoxin CPUs MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit zhaoxin inclusion category: bugfix CVE: NA ----------------- When the processor is idle，low-power idle states (C-states) can be used to save power. For Zhaoxin processors，there are two methods to enter idle states. One is HLT instruction and legacy method of I/O reads from the ACPI-defined register (known as P_LVLx)，the other one is MWAIT instruction with idle states hints. Default for legacy operating system，HLT and P_LVLx I/O reads are used for Zhaoxin Processors to enter idle states, but we have checked on some Zhaoxin platform that MWAIT instruction is more efficient than P_LVLx I/O reads and HLT, so we add MWAIT Cx support for Zhaoxin Processors. Signed-off-by: leoliu-oc --- arch/x86/include/asm/acpi.h | 3 ++- arch/x86/kernel/acpi/cstate.c | 4 +++- 2 files changed, 5 insertions(+), 2 deletions(-) diff --git a/arch/x86/include/asm/acpi.h b/arch/x86/include/asm/acpi.h index f896eed4516c7..1f5fe16484fac 100644 --- a/arch/x86/include/asm/acpi.h +++ b/arch/x86/include/asm/acpi.h @@ -102,7 +102,8 @@ static inline bool arch_has_acpi_pdc(void) { struct cpuinfo_x86 *c = &cpu_data(0); return (c->x86_vendor == X86_VENDOR_INTEL || - c->x86_vendor == X86_VENDOR_CENTAUR); + c->x86_vendor == X86_VENDOR_CENTAUR || + c->x86_vendor == X86_VENDOR_ZHAOXIN); } static inline void arch_acpi_set_proc_cap_bits(u32 *cap) diff --git a/arch/x86/kernel/acpi/cstate.c b/arch/x86/kernel/acpi/cstate.c index 401808b47af3e..90f22148acc79 100644 --- a/arch/x86/kernel/acpi/cstate.c +++ b/arch/x86/kernel/acpi/cstate.c @@ -221,7 +221,9 @@ static int __init ffh_cstate_init(void) if (c->x86_vendor != X86_VENDOR_INTEL && c->x86_vendor != X86_VENDOR_AMD && - c->x86_vendor != X86_VENDOR_HYGON) + c->x86_vendor != X86_VENDOR_HYGON && + c->x86_vendor != X86_VENDOR_CENTAUR && + c->x86_vendor != X86_VENDOR_ZHAOXIN) return -1; cpu_cstate_entry = alloc_percpu(struct cstate_entry);