From: Bhavana Nagendra <bnagendr@redhat.com>
Subject: Re: RHEL5.1 [PATCH]: C-state divisor not functioning correctly on  Greyhound
Date: Mon, 21 May 2007 13:06:06 -0400
Bugzilla: 235404
Message-Id: <4651D17E.5040705@redhat.com>
Changelog: [x86_64] C-state divisor not functioning correctly 


>>BZ 235404
>>
>>Legacy AMD processors support C1 HLT state that allows Power Now! to 
>>function efficiently.
>>With the latest processors from AMD, MWAIT instruction was added.   The 
>>Linux kernel prefers
>>MWAIT for idle wait, but on the previous AMD platforms this was not an 
>>issue.  Testing has
>>shown that the system idles in MWAIT (with no possibility of power 
>>savings) and does not go
>>to HLT state for efficient power savings.
>>

Index: latest/arch/x86_64/kernel/process.c
===================================================================
--- latest.orig/arch/x86_64/kernel/process.c
+++ latest/arch/x86_64/kernel/process.c
@@ -269,10 +269,13 @@ void __cpuinit select_idle_routine(const
 
 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;
Index: latest/arch/x86_64/kernel/setup.c
===================================================================
--- latest.orig/arch/x86_64/kernel/setup.c
+++ latest/arch/x86_64/kernel/setup.c
@@ -89,6 +89,8 @@ int bootloader_type;
 
 unsigned long saved_video_mode;
 
+int force_mwait __initdata;
+
 /* 
  * Early DMI memory
  */
@@ -897,6 +899,10 @@ static void __init init_amd(struct cpuin
 		num_cache_leaves = 4;
 	else
 		num_cache_leaves = 3;
+
+	/* 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)
Index: latest/arch/i386/kernel/process.c
===================================================================
--- latest.orig/arch/i386/kernel/process.c
+++ latest/arch/i386/kernel/process.c
@@ -264,19 +264,19 @@ void __devinit select_idle_routine(const
 	}
 }
 
-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)) {
-		printk("using halt in idle threads.\n");
-		pm_idle = default_idle;
-	}
+	} else if (!strcmp(str, "mwait"))
+		force_mwait = 1;
+	else
+		return -1;
 
 	boot_option_idle_override = 1;
 	return 1;
Index: latest/arch/i386/kernel/cpu/amd.c
===================================================================
--- latest.orig/arch/i386/kernel/cpu/amd.c
+++ latest/arch/i386/kernel/cpu/amd.c
@@ -22,6 +22,8 @@
 extern void vide(void);
 __asm__(".align 4\nvide: ret");
 
+int force_mwait __initdata;
+
 static void __init init_amd(struct cpuinfo_x86 *c)
 {
 	u32 l, h;
@@ -249,6 +251,9 @@ static void __init init_amd(struct cpuin
 		else
 			num_cache_leaves = 3;
 	}
+
+	if ((c->x86 == 0x10 || c->x86 == 0x11) && !force_mwait)
+		clear_bit(X86_FEATURE_MWAIT, &c->x86_capability);
 }
 
 static unsigned int amd_size_cache(struct cpuinfo_x86 * c, unsigned int size)
Index: latest/include/asm-i386/processor.h
===================================================================
--- latest.orig/include/asm-i386/processor.h
+++ latest/include/asm-i386/processor.h
@@ -737,4 +737,6 @@ extern unsigned long boot_option_idle_ov
 extern void enable_sep_cpu(void);
 extern int sysenter_setup(void);
 
+extern int force_mwait;
+
 #endif /* __ASM_I386_PROCESSOR_H */
Index: latest/include/asm-x86_64/proto.h
===================================================================
--- latest.orig/include/asm-x86_64/proto.h
+++ latest/include/asm-x86_64/proto.h
@@ -135,6 +135,8 @@ extern int setup_additional_cpus(char *)
 
 extern void smp_local_timer_interrupt(struct pt_regs * regs);
 
+extern int force_mwait;
+
 long do_arch_prctl(struct task_struct *task, int code, unsigned long addr);
 
 #define round_up(x,y) (((x) + (y) - 1) & ~((y)-1))
Index: latest/Documentation/kernel-parameters.txt
===================================================================
--- latest.orig/Documentation/kernel-parameters.txt
+++ latest/Documentation/kernel-parameters.txt
@@ -651,8 +651,17 @@ running once the system is up.
 	idebus=		[HW] (E)IDE subsystem - VLB/PCI bus speed
 			See Documentation/ide.txt.
 
-	idle=		[HW]
-			Format: idle=poll or idle=halt
+	idle=		[X86]
+			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 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.
 
 	ihash_entries=	[KNL]
 			Set number of hash buckets for inode cache.