mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "Nakajima, Jun" <jun.nakajima@intel.com>
To: "Linus Torvalds" <torvalds@osdl.org>
Cc: <linux-kernel@vger.kernel.org>,
	"Saxena, Sunil" <sunil.saxena@intel.com>,
	"Mallick, Asit K" <asit.k.mallick@intel.com>,
	"Pallipadi, Venkatesh" <venkatesh.pallipadi@intel.com>
Subject: [PATCH] idle using PNI monitor/mwait
Date: Tue, 8 Jul 2003 14:23:14 -0700	[thread overview]
Message-ID: <3014AAAC8E0930438FD38EBF6DCEB5640201719C@fmsmsx407.fm.intel.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 4348 bytes --]

Hi Linus,

Attached is a patch that enables PNI (Prescott New Instructions)
monitor/mwait in kernel idle (opcodes are now public). Basically MWAIT
is similar to hlt, but you can avoid IPI to wake up the processor
waiting. A write (by another processor) to the address range specified
by MONITOR would wake up the processor waiting on MWAIT.

Please apply.

Thanks,
Jun

----------------
diff -ur /build/orig/linux-2.5.74/arch/i386/kernel/cpu/intel.c
linux-2.5.74/arch/i386/kernel/cpu/intel.c
--- /build/orig/linux-2.5.74/arch/i386/kernel/cpu/intel.c
2003-07-02 13:43:55.000000000 -0700
+++ linux-2.5.74/arch/i386/kernel/cpu/intel.c	2003-07-08
09:18:28.000000000 -0700
@@ -13,6 +13,7 @@
 
 static int disable_P4_HT __initdata = 0;
 extern int trap_init_f00f_bug(void);
+extern void select_idle_routine(const struct cpuinfo_x86 *c);
 
 #ifdef CONFIG_X86_INTEL_USERCOPY
 /*
@@ -172,7 +173,7 @@
 	}
 #endif
 
-
+	select_idle_routine(c);
 	if (c->cpuid_level > 1) {
 		/* supports eax=2  call */
 		int i, j, n;
diff -ur /build/orig/linux-2.5.74/arch/i386/kernel/process.c
linux-2.5.74/arch/i386/kernel/process.c
--- /build/orig/linux-2.5.74/arch/i386/kernel/process.c	2003-07-02
13:38:40.000000000 -0700
+++ linux-2.5.74/arch/i386/kernel/process.c	2003-07-08
11:52:42.000000000 -0700
@@ -148,11 +148,56 @@
 	}
 }
 
+/*
+ * This uses new MONITOR/MWAIT instructions on P4 processors with PNI, 
+ * which can obviate IPI to trigger checking of need_resched.
+ * We execute MONITOR against need_resched and enter optimized wait
state 
+ * through MWAIT. Whenever someone changes need_resched, we would be
woken 
+ * up from MWAIT (without an IPI).
+ */
+static void mwait_idle (void)
+{
+	local_irq_enable();
+
+	if (!need_resched()) {
+		set_thread_flag(TIF_POLLING_NRFLAG);
+		do {
+			__monitor((void *)&current_thread_info()->flags,
0, 0);
+			if (need_resched())
+				break;
+			__mwait(0, 0);
+		} while (!need_resched());
+		clear_thread_flag(TIF_POLLING_NRFLAG);
+	}
+}
+
+void __init select_idle_routine(const struct cpuinfo_x86 *c)
+{
+	if (cpu_has(c, X86_FEATURE_MWAIT)) {
+		printk("Monitor/Mwait feature present.\n");
+		/* 
+		 * Skip, if setup has overridden idle.
+		 * Also, take care of system with asymmetric CPUs.
+		 * Use, mwait_idle only if all cpus support it.
+		 * If not, we fallback to default_idle()
+		 */
+		if (!pm_idle) {
+			pm_idle = mwait_idle;
+		}
+		return;
+	}
+	pm_idle = default_idle;
+	return;
+}
+
 static int __init idle_setup (char *str)
 {
 	if (!strncmp(str, "poll", 4)) {
 		printk("using polling idle threads.\n");
 		pm_idle = poll_idle;
+	} else if (!strncmp(str, "halt", 4)) {
+		printk("using halt in idle threads.\n");
+		pm_idle = default_idle;
 	}
 
 	return 1;
diff -ur /build/orig/linux-2.5.74/include/asm-i386/cpufeature.h
linux-2.5.74/include/asm-i386/cpufeature.h
--- /build/orig/linux-2.5.74/include/asm-i386/cpufeature.h
2003-07-02 13:51:50.000000000 -0700
+++ linux-2.5.74/include/asm-i386/cpufeature.h	2003-07-08
09:18:28.000000000 -0700
@@ -71,6 +71,8 @@
 
 /* Intel-defined CPU features, CPUID level 0x00000001 (ecx), word 4 */
 #define X86_FEATURE_EST		(4*32+ 7) /* Enhanced SpeedStep
*/
+#define X86_FEATURE_MWAIT	(4*32+ 3) /* Monitor/Mwait support */
+
 
 /* VIA/Cyrix/Centaur-defined CPU features, CPUID level 0xC0000001, word
5 */
 #define X86_FEATURE_XSTORE	(5*32+ 2) /* on-CPU RNG present (xstore
insn) */
diff -ur /build/orig/linux-2.5.74/include/asm-i386/processor.h
linux-2.5.74/include/asm-i386/processor.h
--- /build/orig/linux-2.5.74/include/asm-i386/processor.h
2003-07-02 13:40:24.000000000 -0700
+++ linux-2.5.74/include/asm-i386/processor.h	2003-07-08
09:18:28.000000000 -0700
@@ -272,6 +272,22 @@
 #define pc98 0
 #endif
 
+static __inline__ void __monitor(const void *eax, unsigned long ecx, 
+		unsigned long edx)
+{
+	/* "monitor %eax,%ecx,%edx;" */
+	asm volatile(
+		".byte 0x0f,0x01,0xc8;"
+		: :"a" (eax), "c" (ecx), "d"(edx));
+}
+
+static __inline__ void __mwait(unsigned long eax, unsigned long ecx)
+{
+	/* "mwait %eax,%ecx;" */
+	asm volatile(
+		".byte 0x0f,0x01,0xc9;"
+		: :"a" (eax), "c" (ecx));
+}
 
 /* from system description table in BIOS.  Mostly for MCA use, but
 others may find it useful. */



[-- Attachment #2: mwait-2.5.74.patch --]
[-- Type: application/octet-stream, Size: 3807 bytes --]

diff -ur /build/orig/linux-2.5.74/arch/i386/kernel/cpu/intel.c linux-2.5.74/arch/i386/kernel/cpu/intel.c
--- /build/orig/linux-2.5.74/arch/i386/kernel/cpu/intel.c	2003-07-02 13:43:55.000000000 -0700
+++ linux-2.5.74/arch/i386/kernel/cpu/intel.c	2003-07-08 09:18:28.000000000 -0700
@@ -13,6 +13,7 @@
 
 static int disable_P4_HT __initdata = 0;
 extern int trap_init_f00f_bug(void);
+extern void select_idle_routine(const struct cpuinfo_x86 *c);
 
 #ifdef CONFIG_X86_INTEL_USERCOPY
 /*
@@ -172,7 +173,7 @@
 	}
 #endif
 
-
+	select_idle_routine(c);
 	if (c->cpuid_level > 1) {
 		/* supports eax=2  call */
 		int i, j, n;
diff -ur /build/orig/linux-2.5.74/arch/i386/kernel/process.c linux-2.5.74/arch/i386/kernel/process.c
--- /build/orig/linux-2.5.74/arch/i386/kernel/process.c	2003-07-02 13:38:40.000000000 -0700
+++ linux-2.5.74/arch/i386/kernel/process.c	2003-07-08 11:52:42.000000000 -0700
@@ -148,11 +148,56 @@
 	}
 }
 
+/*
+ * This uses new MONITOR/MWAIT instructions on P4 processors with PNI, 
+ * which can obviate IPI to trigger checking of need_resched.
+ * We execute MONITOR against need_resched and enter optimized wait state 
+ * through MWAIT. Whenever someone changes need_resched, we would be woken 
+ * up from MWAIT (without an IPI).
+ */
+static void mwait_idle (void)
+{
+	local_irq_enable();
+
+	if (!need_resched()) {
+		set_thread_flag(TIF_POLLING_NRFLAG);
+		do {
+			__monitor((void *)&current_thread_info()->flags, 0, 0);
+			if (need_resched())
+				break;
+			__mwait(0, 0);
+		} while (!need_resched());
+		clear_thread_flag(TIF_POLLING_NRFLAG);
+	}
+}
+
+void __init select_idle_routine(const struct cpuinfo_x86 *c)
+{
+	if (cpu_has(c, X86_FEATURE_MWAIT)) {
+		printk("Monitor/Mwait feature present.\n");
+		/* 
+		 * Skip, if setup has overridden idle.
+		 * Also, take care of system with asymmetric CPUs.
+		 * Use, mwait_idle only if all cpus support it.
+		 * If not, we fallback to default_idle()
+		 */
+		if (!pm_idle) {
+			pm_idle = mwait_idle;
+		}
+		return;
+	}
+	pm_idle = default_idle;
+	return;
+}
+
 static int __init idle_setup (char *str)
 {
 	if (!strncmp(str, "poll", 4)) {
 		printk("using polling idle threads.\n");
 		pm_idle = poll_idle;
+	} else if (!strncmp(str, "halt", 4)) {
+		printk("using halt in idle threads.\n");
+		pm_idle = default_idle;
 	}
 
 	return 1;
diff -ur /build/orig/linux-2.5.74/include/asm-i386/cpufeature.h linux-2.5.74/include/asm-i386/cpufeature.h
--- /build/orig/linux-2.5.74/include/asm-i386/cpufeature.h	2003-07-02 13:51:50.000000000 -0700
+++ linux-2.5.74/include/asm-i386/cpufeature.h	2003-07-08 09:18:28.000000000 -0700
@@ -71,6 +71,8 @@
 
 /* Intel-defined CPU features, CPUID level 0x00000001 (ecx), word 4 */
 #define X86_FEATURE_EST		(4*32+ 7) /* Enhanced SpeedStep */
+#define X86_FEATURE_MWAIT	(4*32+ 3) /* Monitor/Mwait support */
+
 
 /* VIA/Cyrix/Centaur-defined CPU features, CPUID level 0xC0000001, word 5 */
 #define X86_FEATURE_XSTORE	(5*32+ 2) /* on-CPU RNG present (xstore insn) */
diff -ur /build/orig/linux-2.5.74/include/asm-i386/processor.h linux-2.5.74/include/asm-i386/processor.h
--- /build/orig/linux-2.5.74/include/asm-i386/processor.h	2003-07-02 13:40:24.000000000 -0700
+++ linux-2.5.74/include/asm-i386/processor.h	2003-07-08 09:18:28.000000000 -0700
@@ -272,6 +272,22 @@
 #define pc98 0
 #endif
 
+static __inline__ void __monitor(const void *eax, unsigned long ecx, 
+		unsigned long edx)
+{
+	/* "monitor %eax,%ecx,%edx;" */
+	asm volatile(
+		".byte 0x0f,0x01,0xc8;"
+		: :"a" (eax), "c" (ecx), "d"(edx));
+}
+
+static __inline__ void __mwait(unsigned long eax, unsigned long ecx)
+{
+	/* "mwait %eax,%ecx;" */
+	asm volatile(
+		".byte 0x0f,0x01,0xc9;"
+		: :"a" (eax), "c" (ecx));
+}
 
 /* from system description table in BIOS.  Mostly for MCA use, but
 others may find it useful. */

             reply	other threads:[~2003-07-08 22:30 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-07-08 21:23 Nakajima, Jun [this message]
2003-07-08 23:34 ` Linus Torvalds
2003-07-09 10:59 ` Alan Cox
2003-07-09  0:35 Nakajima, Jun
2003-07-09  6:41 ` Zwane Mwaikambo
2003-07-09 16:39 Mallick, Asit K
2003-07-09 17:01 Mallick, Asit K
2003-07-10  1:17 Saxena, Sunil

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3014AAAC8E0930438FD38EBF6DCEB5640201719C@fmsmsx407.fm.intel.com \
    --to=jun.nakajima@intel.com \
    --cc=asit.k.mallick@intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=sunil.saxena@intel.com \
    --cc=torvalds@osdl.org \
    --cc=venkatesh.pallipadi@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®