[PATCH] x86: Don't use MWAIT on AMD Family 10

It doesn't put the CPU into deeper sleep states, so it's better to use the standard
idle loop to save power. But allow to reenable it anyways for benchmarking.

I also removed the obsolete idle=halt on i386

Cc: andreas.herrmann@amd.com

Signed-off-by: Andi Kleen <ak@suse.de>

Documentation/kernel-parameters.txt

@@ -695,8 +695,15 @@ and is between 256 and 4096 characters. It is defined in the file
 	idebus=		[HW] (E)IDE subsystem - VLB/PCI bus speed
 			See Documentation/ide.txt.
 
-	idle=		[HW]
+	idle=		[X86]
-			Format: idle=poll or idle=halt
+			Format: idle=poll or idle=mwait
+			Poll forces a polling idle loop that can slightly improves the performance
+			of waking up a idle CPU, but will use a lot of power and make the system
+			run hot. Not recommended.
+			idle=mwait. On systems which support MONITOR/MWAIT but the kernel chose
+			to not use it because it doesn't save as much power as a normal idle
+			loop use the MONITOR/MWAIT idle loop anyways. Performance should be the same
+			as idle=poll.
 
 	ignore_loglevel	[KNL]
 			Ignore loglevel setting - this will print /all/

arch/i386/kernel/cpu/amd.c

@@ -53,6 +53,8 @@ static __cpuinit int amd_apic_timer_broken(void)
 	return 0;
 }
 
+int force_mwait __cpuinitdata;
+
 static void __cpuinit init_amd(struct cpuinfo_x86 *c)
 {
 	u32 l, h;
@@ -275,6 +277,9 @@ static void __cpuinit init_amd(struct cpuinfo_x86 *c)
 
 	if (amd_apic_timer_broken())
 		set_bit(X86_FEATURE_LAPIC_TIMER_BROKEN, c->x86_capability);
+
+	if (c->x86 == 0x10 && !force_mwait)
+		clear_bit(X86_FEATURE_MWAIT, c->x86_capability);
 }
 
 static unsigned int __cpuinit amd_size_cache(struct cpuinfo_x86 * c, unsigned int size)

arch/i386/kernel/process.c

@@ -272,25 +272,24 @@ void __devinit select_idle_routine(const struct cpuinfo_x86 *c)
 	}
 }
 
-static int __init idle_setup (char *str)
+static int __init idle_setup(char *str)
 {
-	if (!strncmp(str, "poll", 4)) {
+	if (!strcmp(str, "poll")) {
 		printk("using polling idle threads.\n");
 		pm_idle = poll_idle;
 #ifdef CONFIG_X86_SMP
 		if (smp_num_siblings > 1)
 			printk("WARNING: polling idle and HT enabled, performance may degrade.\n");
 #endif
-	} else if (!strncmp(str, "halt", 4)) {
+	} else if (!strcmp(str, "mwait"))
-		printk("using halt in idle threads.\n");
+		force_mwait = 1;
-		pm_idle = default_idle;
+	else
-	}
+		return -1;
 
 	boot_option_idle_override = 1;
-	return 1;
+	return 0;
 }
-
+early_param("idle", idle_setup);
-__setup("idle=", idle_setup);
 
 void show_regs(struct pt_regs * regs)
 {

arch/x86_64/kernel/process.c

@@ -288,16 +288,18 @@ void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c)
 
 static int __init idle_setup (char *str)
 {
-	if (!strncmp(str, "poll", 4)) {
+	if (!strcmp(str, "poll")) {
 		printk("using polling idle threads.\n");
 		pm_idle = poll_idle;
-	}
+	} else if (!strcmp(str, "mwait"))
+		force_mwait = 1;
+	else
+		return -1;
 
 	boot_option_idle_override = 1;
-	return 1;
+	return 0;
 }
-
+early_param("idle", idle_setup);
-__setup("idle=", idle_setup);
 
 /* Prints also some state that isn't saved in the pt_regs */ 
 void __show_regs(struct pt_regs * regs)

arch/x86_64/kernel/setup.c

@@ -79,6 +79,8 @@ int bootloader_type;
 
 unsigned long saved_video_mode;
 
+int force_mwait __cpuinitdata;
+
 /* 
  * Early DMI memory
  */
@@ -604,6 +606,10 @@ static void __cpuinit init_amd(struct cpuinfo_x86 *c)
 
 	/* RDTSC can be speculated around */
 	clear_bit(X86_FEATURE_SYNC_RDTSC, &c->x86_capability);
+
+	/* Family 10 doesn't support C states in MWAIT so don't use it */
+	if (c->x86 == 0x10 && !force_mwait)
+		clear_bit(X86_FEATURE_MWAIT, &c->x86_capability);
 }
 
 static void __cpuinit detect_ht(struct cpuinfo_x86 *c)

include/asm-i386/processor.h

@@ -779,4 +779,6 @@ extern int sysenter_setup(void);
 extern void cpu_set_gdt(int);
 extern void cpu_init(void);
 
+extern int force_mwait;
+
 #endif /* __ASM_I386_PROCESSOR_H */

include/asm-x86_64/proto.h

@@ -119,6 +119,8 @@ extern int gsi_irq_sharing(int gsi);
 
 extern void smp_local_timer_interrupt(void);
 
+extern int force_mwait;
+
 long do_arch_prctl(struct task_struct *task, int code, unsigned long addr);
 
 void i8254_timer_resume(void);