[PATCH 1/6] fastboot: Asynchronous function calls to speed up kernel boot

[PATCH 1/6] fastboot: Asynchronous function calls to speed up kernel boot

[Arjan van de Ven]

>From 15815a54b95e6866ff9532dead9cca4d6a298b54 Mon Sep 17 00:00:00 2001
From: Arjan van de Ven <arjan@linux.intel.com>
Date: Sun, 4 Jan 2009 05:32:28 -0800
Subject: [PATCH] fastboot: Asynchronous function calls to speed up kernel boot

Right now, most of the kernel boot is strictly synchronous, such that
various hardware delays are done sequentially.

In order to make the kernel boot faster, this patch introduces
infrastructure to allow doing some of the initialization steps
asynchronously, which will hide significant portions of the hardware delays
in practice.

In order to not change device order and other similar observables, this
patch does NOT do full parallel initialization.

Rather, it operates more in the way an out of order CPU does; the work may
be done out of order and asynchronous, but the observable effects
(instruction retiring for the CPU) are still done in the original sequence.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
---
 include/linux/async.h  |   21 ++++
 init/do_mounts.c       |    2 +
 init/main.c            |    5 +-
 kernel/Makefile        |    3 +-
 kernel/async.c         |  307 ++++++++++++++++++++++++++++++++++++++++++++++++
 kernel/irq/autoprobe.c |    5 +
 kernel/module.c        |    2 +
 7 files changed, 343 insertions(+), 2 deletions(-)
 create mode 100644 include/linux/async.h
 create mode 100644 kernel/async.c

diff --git a/include/linux/async.h b/include/linux/async.h
new file mode 100644
index 0000000..678d4fd
--- /dev/null
+++ b/include/linux/async.h
@@ -0,0 +1,21 @@
+/*
+ * async.h: Asynchronous function calls for boot performance
+ *
+ * (C) Copyright 2009 Intel Corporation
+ * Author: Arjan van de Ven <arjan@linux.intel.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2
+ * of the License.
+ */
+
+#include <linux/types.h>
+
+typedef u64 async_cookie_t;
+typedef void (async_func_ptr) (void *data, async_cookie_t cookie);
+
+extern async_cookie_t async_schedule(async_func_ptr *ptr, void *data);
+extern void async_synchronize_full(void);
+extern void async_synchronize_cookie(async_cookie_t cookie);
+
diff --git a/init/do_mounts.c b/init/do_mounts.c
index d055b19..b9df336 100644
--- a/init/do_mounts.c
+++ b/init/do_mounts.c
@@ -13,6 +13,7 @@
 #include <linux/init.h>
 #include <linux/fs.h>
 #include <linux/initrd.h>
+#include <linux/async.h>
 
 #include <linux/nfs_fs.h>
 #include <linux/nfs_fs_sb.h>
@@ -372,6 +373,7 @@ void __init prepare_namespace(void)
 	/* wait for the known devices to complete their probing */
 	while (driver_probe_done() != 0)
 		msleep(100);
+	async_synchronize_full();
 
 	md_run_setup();
 
diff --git a/init/main.c b/init/main.c
index cd168eb..40d5373 100644
--- a/init/main.c
+++ b/init/main.c
@@ -63,6 +63,7 @@
 #include <linux/signal.h>
 #include <linux/idr.h>
 #include <linux/ftrace.h>
+#include <linux/async.h>
 #include <trace/boot.h>
 
 #include <asm/io.h>
@@ -687,7 +688,7 @@ asmlinkage void __init start_kernel(void)
 	rest_init();
 }
 
-static int initcall_debug;
+int initcall_debug;
 core_param(initcall_debug, initcall_debug, bool, 0644);
 
 int do_one_initcall(initcall_t fn)
@@ -788,6 +789,8 @@ static void run_init_process(char *init_filename)
  */
 static int noinline init_post(void)
 {
+	/* need to finish all async __init code before freeing the memory */
+	async_synchronize_full();
 	free_initmem();
 	unlock_kernel();
 	mark_rodata_ro();
diff --git a/kernel/Makefile b/kernel/Makefile
index e1c5bf3..2921d90 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -9,7 +9,8 @@ obj-y     = sched.o fork.o exec_domain.o panic.o printk.o \
 	    rcupdate.o extable.o params.o posix-timers.o \
 	    kthread.o wait.o kfifo.o sys_ni.o posix-cpu-timers.o mutex.o \
 	    hrtimer.o rwsem.o nsproxy.o srcu.o semaphore.o \
-	    notifier.o ksysfs.o pm_qos_params.o sched_clock.o cred.o
+	    notifier.o ksysfs.o pm_qos_params.o sched_clock.o cred.o \
+	    async.o
 
 ifdef CONFIG_FUNCTION_TRACER
 # Do not trace debug files and internal ftrace files
diff --git a/kernel/async.c b/kernel/async.c
new file mode 100644
index 0000000..8ecebf5
--- /dev/null
+++ b/kernel/async.c
@@ -0,0 +1,307 @@
+/*
+ * async.c: Asynchronous function calls for boot performance
+ *
+ * (C) Copyright 2009 Intel Corporation
+ * Author: Arjan van de Ven <arjan@linux.intel.com>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2
+ * of the License.
+ */
+
+
+/*
+
+Goals and Theory of Operation
+
+The primary goal of this feature is to reduce the kernel boot time,
+by doing various independent hardware delays and discovery operations
+decoupled and not strictly serialized.
+
+More specifically, the asynchronous function call concept allows
+certain operations (primarily during system boot) to happen
+asynchronously, out of order, while these operations still
+have their externally visible parts happen sequentially and in-order.
+(not unlike how out-of-order CPUs retire their instructions in order)
+
+Key to the asynchronous function call implementation is the concept of
+a "sequence cookie" (which, although it has an abstracted type, can be
+thought of as a monotonically incrementing number).
+
+The async core will assign each scheduled event such a sequence cookie and
+pass this to the called functions.
+
+The asynchronously called function should before doing a globally visible
+operation, such as registering device numbers, call the
+async_synchronize_cookie() function and pass in its own cookie. The
+async_synchronize_cookie() function will make sure that all asynchronous
+operations that were scheduled prior to the operation corresponding with the
+cookie have completed.
+
+Subsystem/driver initialization code that scheduled asynchronous probe
+functions, but which shares global resources with other drivers/subsystems
+that do not use the asynchronous call feature, need to do a full
+synchronization with the async_synchronize_full() function, before returning
+from their init function. This is to maintain strict ordering between the
+asynchronous and synchronous parts of the kernel.
+
+*/
+
+#include <linux/async.h>
+#include <linux/module.h>
+#include <linux/wait.h>
+#include <linux/sched.h>
+#include <linux/init.h>
+#include <linux/kthread.h>
+#include <asm/atomic.h>
+
+static async_cookie_t next_cookie = 1;
+static async_cookie_t lowest_in_progress = 1;
+
+
+static LIST_HEAD(async_pending);
+static LIST_HEAD(async_running);
+static DEFINE_SPINLOCK(async_lock);
+
+struct async_entry {
+	struct list_head list;
+	async_cookie_t   cookie;
+	async_func_ptr	 *func;
+	void             *data;
+};
+
+static DECLARE_WAIT_QUEUE_HEAD(async_done);
+static DECLARE_WAIT_QUEUE_HEAD(async_new);
+
+static atomic_t entry_count;
+static atomic_t thread_count;
+
+extern int initcall_debug;
+
+
+/*
+ * MUST be called with the lock held!
+ */
+static void __recalc_lowest_in_progress(void)
+{
+	struct async_entry *entry;
+	if (!list_empty(&async_pending)) {
+		entry = list_first_entry(&async_pending,
+			struct async_entry, list);
+		lowest_in_progress = entry->cookie;
+	} else if (!list_empty(&async_running)) {
+		entry = list_first_entry(&async_running,
+			struct async_entry, list);
+		lowest_in_progress = entry->cookie;
+	} else {
+		/* nothing in progress... next_cookie is "infinity" */
+		lowest_in_progress = next_cookie;
+	}
+
+}
+/*
+ * pick the first pending entry and run it
+ */
+static void run_one_entry(void)
+{
+	unsigned long flags;
+	struct async_entry *entry;
+	ktime_t calltime, delta, rettime;
+
+	/* 1) pick one task from the pending queue */
+
+	spin_lock_irqsave(&async_lock, flags);
+	if (list_empty(&async_pending))
+		goto out;
+	entry = list_first_entry(&async_pending, struct async_entry, list);
+
+	/* 2) move it to the running queue */
+	list_del(&entry->list);
+	list_add_tail(&entry->list, &async_running);
+	spin_unlock_irqrestore(&async_lock, flags);
+
+	/* 3) run it (and print duration)*/
+	if (initcall_debug) {
+		printk("calling  %lli_%pF @ %i\n", entry->cookie, entry->func, task_pid_nr(current));
+		calltime = ktime_get();
+	}
+	entry->func(entry->data, entry->cookie);
+	if (initcall_debug) {
+		rettime = ktime_get();
+		delta = ktime_sub(rettime, calltime);
+		printk("initcall %lli_%pF returned 0 after %lld usecs\n", entry->cookie,
+			entry->func, ktime_to_ns(delta) >> 10);
+	}
+
+	/* 4) remove it from the running queue */
+	spin_lock_irqsave(&async_lock, flags);
+	list_del(&entry->list);
+
+	/* 5) free the entry  */
+	kfree(entry);
+	atomic_dec(&entry_count);
+
+	/* 6) update the lowest_in_progress value */
+	__recalc_lowest_in_progress();
+
+	spin_unlock_irqrestore(&async_lock, flags);
+
+	/* 7) wake up any waiters. */
+	wake_up(&async_done);
+	return;
+
+out:
+	spin_unlock_irqrestore(&async_lock, flags);
+}
+
+
+async_cookie_t async_schedule(async_func_ptr *ptr, void *data)
+{
+	struct async_entry *entry;
+	unsigned long flags;
+	async_cookie_t newcookie;
+	
+
+	/* allow irq-off callers */
+	entry = kzalloc(sizeof(struct async_entry), GFP_ATOMIC);
+	if (!entry) {
+		spin_lock_irqsave(&async_lock, flags);
+		newcookie = next_cookie++;
+		spin_unlock_irqrestore(&async_lock, flags);
+
+		/* low on memory.. run synchronously */
+		ptr(data, newcookie);
+		return newcookie;
+	}
+	entry->func = ptr;
+	entry->data = data;
+
+	spin_lock_irqsave(&async_lock, flags);
+	newcookie = entry->cookie = next_cookie++;
+	list_add_tail(&entry->list, &async_pending);
+	atomic_inc(&entry_count);
+	spin_unlock_irqrestore(&async_lock, flags);
+	wake_up(&async_new);
+	return newcookie;
+}
+EXPORT_SYMBOL_GPL(async_schedule);
+
+void async_synchronize_full(void)
+{
+	async_synchronize_cookie(next_cookie);
+}
+EXPORT_SYMBOL_GPL(async_synchronize_full);
+
+void async_synchronize_cookie(async_cookie_t cookie)
+{
+	ktime_t starttime, delta, endtime;
+
+	if (initcall_debug) {
+		printk("async_waiting @ %i\n", task_pid_nr(current));
+		starttime = ktime_get();
+	}
+
+	wait_event(async_done, lowest_in_progress >= cookie);
+
+	if (initcall_debug) {
+		endtime = ktime_get();
+		delta = ktime_sub(endtime, starttime);
+
+		printk("async_continuing @ %i after %lli usec\n",
+			task_pid_nr(current), ktime_to_ns(delta) >> 10);
+	}
+}
+EXPORT_SYMBOL_GPL(async_synchronize_cookie);
+
+
+static int async_thread(void *unused)
+{
+	DECLARE_WAITQUEUE(wq, current);
+	add_wait_queue(&async_new, &wq);
+
+	while (!kthread_should_stop()) {
+		int ret = HZ;
+		set_current_state(TASK_INTERRUPTIBLE);
+		/*
+		 * check the list head without lock.. false positives
+		 * are dealt with inside run_one_entry() while holding
+		 * the lock.
+		 */
+		rmb();
+		if (!list_empty(&async_pending))
+			run_one_entry();
+		else
+			ret = schedule_timeout(HZ);
+
+		if (ret == 0) {
+			/*
+			 * we timed out, this means we as thread are redundant.
+			 * we sign off and die, but we to avoid any races there
+			 * is a last-straw check to see if work snuck in.
+			 */
+			atomic_dec(&thread_count);
+			wmb(); /* manager must see our departure first */
+			if (list_empty(&async_pending))
+				break;
+			/*
+			 * woops work came in between us timing out and us
+			 * signing off; we need to stay alive and keep working.
+			 */
+			atomic_inc(&thread_count);
+		}
+	}
+	remove_wait_queue(&async_new, &wq);
+
+	return 0;
+}
+
+static int async_manager_thread(void *unused)
+{
+	int max_threads;
+	DECLARE_WAITQUEUE(wq, current);
+	add_wait_queue(&async_new, &wq);
+
+	while (!kthread_should_stop()) {
+		int tc, ec;
+
+		/*
+		 * Maximum number of worker threads.
+		 * Even on the smallest machine we want 8
+		 * Scaling by 4 per logical CPU
+		 * But no more than 256 to not overload stuff too much
+		 * (and yes these are magic numbers that might need tuning)
+		 * Calculated dynamic because early on the nr of online cpus
+		 * is 1...
+		 */
+		max_threads = 4 + 4 * num_online_cpus();
+		if (max_threads > 256)
+			max_threads = 256;
+
+
+		set_current_state(TASK_INTERRUPTIBLE);
+
+		tc = atomic_read(&thread_count);
+		rmb();
+		ec = atomic_read(&entry_count);
+
+		while (tc < ec && tc < max_threads) {
+			kthread_run(async_thread, NULL, "async/%i", tc);
+			atomic_inc(&thread_count);
+			tc++;
+		}
+
+		schedule();
+	}
+	remove_wait_queue(&async_new, &wq);
+
+	return 0;
+}
+
+static int __init async_init(void)
+{
+	kthread_run(async_manager_thread, NULL, "async/mgr");
+	return 0;
+}
+
+core_initcall(async_init);
diff --git a/kernel/irq/autoprobe.c b/kernel/irq/autoprobe.c
index cc0f732..1de9700 100644
--- a/kernel/irq/autoprobe.c
+++ b/kernel/irq/autoprobe.c
@@ -10,6 +10,7 @@
 #include <linux/module.h>
 #include <linux/interrupt.h>
 #include <linux/delay.h>
+#include <linux/async.h>
 
 #include "internals.h"
 
@@ -34,6 +35,10 @@ unsigned long probe_irq_on(void)
 	unsigned int status;
 	int i;
 
+	/*
+	 * quiesce the kernel, or at least the asynchronous portion
+	 */
+	async_synchronize_full();
 	mutex_lock(&probing_active);
 	/*
 	 * something may have generated an irq long ago and we want to
diff --git a/kernel/module.c b/kernel/module.c
index dd2a541..da3ad68 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -51,6 +51,7 @@
 #include <asm/sections.h>
 #include <linux/tracepoint.h>
 #include <linux/ftrace.h>
+#include <linux/async.h>
 
 #if 0
 #define DEBUGP printk
@@ -809,6 +810,7 @@ sys_delete_module(const char __user *name_user, unsigned int flags)
 		mod->exit();
 	blocking_notifier_call_chain(&module_notify_list,
 				     MODULE_STATE_GOING, mod);
+	async_synchronize_full();
 	mutex_lock(&module_mutex);
 	/* Store the name of the last unloaded module for diagnostic purposes */
 	strlcpy(last_unloaded_module, mod->name, sizeof(last_unloaded_module));
-- 
1.6.0.6



-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

[PATCH 2/6] fastboot: make scsi probes asynchronous

[Arjan van de Ven]

>From 19617cfb1840abc69cc2fe74433007fc1daa1cb7 Mon Sep 17 00:00:00 2001
From: Arjan van de Ven <arjan@linux.intel.com>
Date: Sun, 4 Jan 2009 05:32:28 -0800
Subject: [PATCH] fastboot: make scsi probes asynchronous

This patch makes part of the scsi probe (which is mostly device spin up and the
partition scan) asynchronous. Only the part that runs after getting the device
number allocated is asynchronous, ensuring that device numbering remains stable.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
---
 drivers/scsi/scsi_scan.c |    3 +
 drivers/scsi/sd.c        |  109 ++++++++++++++++++++++++++++------------------
 2 files changed, 70 insertions(+), 42 deletions(-)

diff --git a/drivers/scsi/scsi_scan.c b/drivers/scsi/scsi_scan.c
index 18486b5..17914a3 100644
--- a/drivers/scsi/scsi_scan.c
+++ b/drivers/scsi/scsi_scan.c
@@ -32,6 +32,7 @@
 #include <linux/delay.h>
 #include <linux/kthread.h>
 #include <linux/spinlock.h>
+#include <linux/async.h>
 
 #include <scsi/scsi.h>
 #include <scsi/scsi_cmnd.h>
@@ -179,6 +180,8 @@ int scsi_complete_async_scans(void)
 	spin_unlock(&async_scan_lock);
 
 	kfree(data);
+	/* Synchronize async operations globally */
+	async_synchronize_full();
 	return 0;
 }
 
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index 62b28d5..e035c11 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -48,6 +48,7 @@
 #include <linux/delay.h>
 #include <linux/mutex.h>
 #include <linux/string_helpers.h>
+#include <linux/async.h>
 #include <asm/uaccess.h>
 
 #include <scsi/scsi.h>
@@ -1802,6 +1803,71 @@ static int sd_format_disk_name(char *prefix, int index, char *buf, int buflen)
 	return 0;
 }
 
+/*
+ * The asynchronous part of sd_probe
+ */
+static void sd_probe_async(void *data, async_cookie_t cookie)
+{
+	struct scsi_disk *sdkp = data;
+	struct scsi_device *sdp;
+	struct gendisk *gd;
+	u32 index;
+	struct device *dev;
+
+	sdp = sdkp->device;
+	gd = sdkp->disk;
+	index = sdkp->index;
+	dev = &sdp->sdev_gendev;
+
+	if (!sdp->request_queue->rq_timeout) {
+		if (sdp->type != TYPE_MOD)
+			blk_queue_rq_timeout(sdp->request_queue, SD_TIMEOUT);
+		else
+			blk_queue_rq_timeout(sdp->request_queue,
+					     SD_MOD_TIMEOUT);
+	}
+
+	device_initialize(&sdkp->dev);
+	sdkp->dev.parent = &sdp->sdev_gendev;
+	sdkp->dev.class = &sd_disk_class;
+	strncpy(sdkp->dev.bus_id, sdp->sdev_gendev.bus_id, BUS_ID_SIZE);
+
+	if (device_add(&sdkp->dev))
+		goto out_free_index;
+
+	get_device(&sdp->sdev_gendev);
+
+	if (index < SD_MAX_DISKS) {
+		gd->major = sd_major((index & 0xf0) >> 4);
+		gd->first_minor = ((index & 0xf) << 4) | (index & 0xfff00);
+		gd->minors = SD_MINORS;
+	}
+	gd->fops = &sd_fops;
+	gd->private_data = &sdkp->driver;
+	gd->queue = sdkp->device->request_queue;
+
+	sd_revalidate_disk(gd);
+
+	blk_queue_prep_rq(sdp->request_queue, sd_prep_fn);
+
+	gd->driverfs_dev = &sdp->sdev_gendev;
+	gd->flags = GENHD_FL_EXT_DEVT | GENHD_FL_DRIVERFS;
+	if (sdp->removable)
+		gd->flags |= GENHD_FL_REMOVABLE;
+
+	dev_set_drvdata(dev, sdkp);
+	add_disk(gd);
+	sd_dif_config_host(sdkp);
+
+	sd_printk(KERN_NOTICE, sdkp, "Attached SCSI %sdisk\n",
+		  sdp->removable ? "removable " : "");
+
+	return;
+
+ out_free_index:
+	ida_remove(&sd_index_ida, index);
+}
+
 /**
  *	sd_probe - called during driver initialization and whenever a
  *	new scsi device is attached to the system. It is called once
@@ -1865,48 +1931,7 @@ static int sd_probe(struct device *dev)
 	sdkp->openers = 0;
 	sdkp->previous_state = 1;
 
-	if (!sdp->request_queue->rq_timeout) {
-		if (sdp->type != TYPE_MOD)
-			blk_queue_rq_timeout(sdp->request_queue, SD_TIMEOUT);
-		else
-			blk_queue_rq_timeout(sdp->request_queue,
-					     SD_MOD_TIMEOUT);
-	}
-
-	device_initialize(&sdkp->dev);
-	sdkp->dev.parent = &sdp->sdev_gendev;
-	sdkp->dev.class = &sd_disk_class;
-	strncpy(sdkp->dev.bus_id, sdp->sdev_gendev.bus_id, BUS_ID_SIZE);
-
-	if (device_add(&sdkp->dev))
-		goto out_free_index;
-
-	get_device(&sdp->sdev_gendev);
-
-	if (index < SD_MAX_DISKS) {
-		gd->major = sd_major((index & 0xf0) >> 4);
-		gd->first_minor = ((index & 0xf) << 4) | (index & 0xfff00);
-		gd->minors = SD_MINORS;
-	}
-	gd->fops = &sd_fops;
-	gd->private_data = &sdkp->driver;
-	gd->queue = sdkp->device->request_queue;
-
-	sd_revalidate_disk(gd);
-
-	blk_queue_prep_rq(sdp->request_queue, sd_prep_fn);
-
-	gd->driverfs_dev = &sdp->sdev_gendev;
-	gd->flags = GENHD_FL_EXT_DEVT | GENHD_FL_DRIVERFS;
-	if (sdp->removable)
-		gd->flags |= GENHD_FL_REMOVABLE;
-
-	dev_set_drvdata(dev, sdkp);
-	add_disk(gd);
-	sd_dif_config_host(sdkp);
-
-	sd_printk(KERN_NOTICE, sdkp, "Attached SCSI %sdisk\n",
-		  sdp->removable ? "removable " : "");
+	async_schedule(sd_probe_async, sdkp);
 
 	return 0;
 
-- 
1.6.0.6




-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

[PATCH 3/6] fastboot: make the libata port scan asynchronous

[Arjan van de Ven]

>From 3637ab5b7e9fa4dd3e831f76979fb67314d058e2 Mon Sep 17 00:00:00 2001
From: Arjan van de Ven <arjan@linux.intel.com>
Date: Sun, 4 Jan 2009 05:32:28 -0800
Subject: [PATCH] fastboot: make the libata port scan asynchronous

This patch makes the libata port scanning asynchronous (per device).
There is a synchronization point before doing the actual disk scan
so that device ordering is not affected.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
---
 drivers/ata/libata-core.c |   84 ++++++++++++++++++++++++--------------------
 1 files changed, 46 insertions(+), 38 deletions(-)

diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c
index fecca42..fe2c208 100644
--- a/drivers/ata/libata-core.c
+++ b/drivers/ata/libata-core.c
@@ -56,6 +56,7 @@
 #include <linux/workqueue.h>
 #include <linux/scatterlist.h>
 #include <linux/io.h>
+#include <linux/async.h>
 #include <scsi/scsi.h>
 #include <scsi/scsi_cmnd.h>
 #include <scsi/scsi_host.h>
@@ -5909,6 +5910,48 @@ void ata_host_init(struct ata_host *host, struct device *dev,
 	host->ops = ops;
 }
 
+
+void async_port_probe(void *data, async_cookie_t cookie)
+{
+	int rc;
+	struct ata_port *ap = data;
+	/* probe */
+	if (ap->ops->error_handler) {
+		struct ata_eh_info *ehi = &ap->link.eh_info;
+		unsigned long flags;
+
+		ata_port_probe(ap);
+
+		/* kick EH for boot probing */
+		spin_lock_irqsave(ap->lock, flags);
+
+		ehi->probe_mask |= ATA_ALL_DEVICES;
+		ehi->action |= ATA_EH_RESET | ATA_EH_LPM;
+		ehi->flags |= ATA_EHI_NO_AUTOPSY | ATA_EHI_QUIET;
+
+		ap->pflags &= ~ATA_PFLAG_INITIALIZING;
+		ap->pflags |= ATA_PFLAG_LOADING;
+		ata_port_schedule_eh(ap);
+
+		spin_unlock_irqrestore(ap->lock, flags);
+
+		/* wait for EH to finish */
+		ata_port_wait_eh(ap);
+	} else {
+		DPRINTK("ata%u: bus probe begin\n", ap->print_id);
+		rc = ata_bus_probe(ap);
+		DPRINTK("ata%u: bus probe end\n", ap->print_id);
+
+		if (rc) {
+			/* FIXME: do something useful here?
+			 * Current libata behavior will
+			 * tear down everything when
+			 * the module is removed
+			 * or the h/w is unplugged.
+			 */
+		}
+	}
+}
 /**
  *	ata_host_register - register initialized ATA host
  *	@host: ATA host to register
@@ -5988,45 +6031,9 @@ int ata_host_register(struct ata_host *host, struct scsi_host_template *sht)
 	DPRINTK("probe begin\n");
 	for (i = 0; i < host->n_ports; i++) {
 		struct ata_port *ap = host->ports[i];
-
-		/* probe */
-		if (ap->ops->error_handler) {
-			struct ata_eh_info *ehi = &ap->link.eh_info;
-			unsigned long flags;
-
-			ata_port_probe(ap);
-
-			/* kick EH for boot probing */
-			spin_lock_irqsave(ap->lock, flags);
-
-			ehi->probe_mask |= ATA_ALL_DEVICES;
-			ehi->action |= ATA_EH_RESET | ATA_EH_LPM;
-			ehi->flags |= ATA_EHI_NO_AUTOPSY | ATA_EHI_QUIET;
-
-			ap->pflags &= ~ATA_PFLAG_INITIALIZING;
-			ap->pflags |= ATA_PFLAG_LOADING;
-			ata_port_schedule_eh(ap);
-
-			spin_unlock_irqrestore(ap->lock, flags);
-
-			/* wait for EH to finish */
-			ata_port_wait_eh(ap);
-		} else {
-			DPRINTK("ata%u: bus probe begin\n", ap->print_id);
-			rc = ata_bus_probe(ap);
-			DPRINTK("ata%u: bus probe end\n", ap->print_id);
-
-			if (rc) {
-				/* FIXME: do something useful here?
-				 * Current libata behavior will
-				 * tear down everything when
-				 * the module is removed
-				 * or the h/w is unplugged.
-				 */
-			}
-		}
+		async_schedule(async_port_probe, ap);
 	}
-
+	async_synchronize_full();
 	/* probes are done, now scan each port's disk(s) */
 	DPRINTK("host probe begin\n");
 	for (i = 0; i < host->n_ports; i++) {
@@ -6034,6 +6041,7 @@ int ata_host_register(struct ata_host *host, struct scsi_host_template *sht)
 
 		ata_scsi_scan_host(ap, 1);
 	}
+	DPRINTK("host probe end\n");
 
 	return 0;
 }
-- 
1.6.0.6


-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

[PATCH 4/6] fastboot: make ACPI bus drivers probe asynchronous

[Arjan van de Ven]

>From 34c47dd38f49e29eb7cd7fa2078efc6b8b258bb9 Mon Sep 17 00:00:00 2001
From: Arjan van de Ven <arjan@linux.intel.com>
Date: Sun, 4 Jan 2009 05:32:28 -0800
Subject: [PATCH] fastboot: make ACPI bus drivers probe asynchronous

the various ACPI bus drivers have non-overlapping devices and can
each be run asynchronous. Some of the ACPI drivers (especially the
battery one, but others as well) can take quite a long time to probe.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
---
 drivers/acpi/scan.c |   23 ++++++++++++++++-------
 1 files changed, 16 insertions(+), 7 deletions(-)

diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c
index 39b7233..a9e542d 100644
--- a/drivers/acpi/scan.c
+++ b/drivers/acpi/scan.c
@@ -8,6 +8,7 @@
 #include <linux/acpi.h>
 #include <linux/signal.h>
 #include <linux/kthread.h>
+#include <linux/async.h>
 
 #include <acpi/acpi_drivers.h>
 #include <acpi/acinterp.h>	/* for acpi_ex_eisa_id_to_string() */
@@ -578,6 +579,19 @@ static int acpi_start_single_object(struct acpi_device *device)
 	return result;
 }
 
+static void acpi_bus_register_async(void *data, async_cookie_t cookie)
+{
+	int ret;
+	struct acpi_driver *driver = data;
+	driver->drv.name = driver->name;
+	driver->drv.bus = &acpi_bus_type;
+	driver->drv.owner = driver->owner;
+
+	async_synchronize_cookie(cookie);
+
+	ret = driver_register(&driver->drv);
+	WARN_ON(ret != 0);
+}
 /**
  * acpi_bus_register_driver - register a driver with the ACPI bus
  * @driver: driver being registered
@@ -588,16 +602,11 @@ static int acpi_start_single_object(struct acpi_device *device)
  */
 int acpi_bus_register_driver(struct acpi_driver *driver)
 {
-	int ret;
 
 	if (acpi_disabled)
 		return -ENODEV;
-	driver->drv.name = driver->name;
-	driver->drv.bus = &acpi_bus_type;
-	driver->drv.owner = driver->owner;
-
-	ret = driver_register(&driver->drv);
-	return ret;
+	async_schedule(acpi_bus_register_async, driver);
+	return 0;
 }
 
 EXPORT_SYMBOL(acpi_bus_register_driver);
-- 
1.6.0.6



-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

[PATCH 5/6] bootchart: improve output

[Arjan van de Ven]

>From ec57f2c6755c5439dcb273f9529f0575b6084d02 Mon Sep 17 00:00:00 2001
From: Arjan van de Ven <arjan@linux.intel.com>
Date: Sun, 4 Jan 2009 10:59:18 -0800
Subject: [PATCH] bootchart: improve output based on Dave Jones' feedback

Dave Jones, in his blog, had some feedback about the bootchart script:
Primarily his complaint was that shorter delays weren't visualized.

The reason for that was that too small delays will have their labels
mixed up in the graph in an unreadable mess.

This patch has a fix for this; for one, it makes the output wider,
so more will fit.
The second part is that smaller delays are now shown with a
much smaller font for the label; while this isn't per se
readable at a 1:1 zoom, at least you can zoom in with most SVG
viewing applications and see what it is you are looking at.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
---
 scripts/bootgraph.pl |   16 +++++++++++-----
 1 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/scripts/bootgraph.pl b/scripts/bootgraph.pl
index f0af9aa..0a498e3 100644
--- a/scripts/bootgraph.pl
+++ b/scripts/bootgraph.pl
@@ -88,7 +88,7 @@ END
 }
 
 print "<?xml version=\"1.0\" standalone=\"no\"?> \n";
-print "<svg width=\"1000\" height=\"100%\" version=\"1.1\" xmlns=\"http://www.w3.org/2000/svg\">\n";
+print "<svg width=\"2000\" height=\"100%\" version=\"1.1\" xmlns=\"http://www.w3.org/2000/svg\">\n";
 
 my @styles;
 
@@ -105,8 +105,9 @@ $styles[9] = "fill:rgb(255,255,128);fill-opacity:0.5;stroke-width:1;stroke:rgb(0
 $styles[10] = "fill:rgb(255,128,255);fill-opacity:0.5;stroke-width:1;stroke:rgb(0,0,0)";
 $styles[11] = "fill:rgb(128,255,255);fill-opacity:0.5;stroke-width:1;stroke:rgb(0,0,0)";
 
-my $mult = 950.0 / ($maxtime - $firsttime);
-my $threshold = ($maxtime - $firsttime) / 60.0;
+my $mult = 1950.0 / ($maxtime - $firsttime);
+my $threshold2 = ($maxtime - $firsttime) / 120.0;
+my $threshold = $threshold2/10;
 my $stylecounter = 0;
 my %rows;
 my $rowscount = 1;
@@ -116,7 +117,7 @@ foreach my $key (@initcalls) {
 	my $duration = $end{$key} - $start{$key};
 
 	if ($duration >= $threshold) {
-		my ($s, $s2, $e, $w, $y, $y2, $style);
+		my ($s, $s2, $s3, $e, $w, $y, $y2, $style);
 		my $pid = $pids{$key};
 
 		if (!defined($rows{$pid})) {
@@ -125,6 +126,7 @@ foreach my $key (@initcalls) {
 		}
 		$s = ($start{$key} - $firsttime) * $mult;
 		$s2 = $s + 6;
+		$s3 = $s + 1;
 		$e = ($end{$key} - $firsttime) * $mult;
 		$w = $e - $s;
 
@@ -138,7 +140,11 @@ foreach my $key (@initcalls) {
 		};
 
 		print "<rect x=\"$s\" width=\"$w\" y=\"$y\" height=\"145\" style=\"$style\"/>\n";
-		print "<text transform=\"translate($s2,$y2) rotate(90)\">$key</text>\n";
+		if ($duration >= $threshold2) {
+			print "<text transform=\"translate($s2,$y2) rotate(90)\">$key</text>\n";
+		} else {
+			print "<text transform=\"translate($s3,$y2) rotate(90)\" font-size=\"3pt\">$key</text>\n";
+		}
 	}
 }
 
-- 
1.6.0.6




-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

[PATCH 6/6] fastboot: Make libata initialization even more async

[Arjan van de Ven]

>From d9edec388ac06b74f190ed9ea929158003b915ee Mon Sep 17 00:00:00 2001
From: Arjan van de Ven <arjan@linux.intel.com>
Date: Mon, 5 Jan 2009 15:07:07 -0800
Subject: [PATCH] fastboot: Make libata initialization even more async

As suggested by Linus: Don't do the libata init in 2 separate
steps with a global sync inbetween, but do it as one async step,
with a local sync before registering the device.

This cuts the boottime on my machine with 2 sata controllers down
significantly, and it seems to work. Would be nice if the libata
folks take a good look at this patch though..

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
---
 drivers/ata/libata-core.c |   16 +++++++---------
 1 files changed, 7 insertions(+), 9 deletions(-)

diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c
index fe2c208..5d4f7e5 100644
--- a/drivers/ata/libata-core.c
+++ b/drivers/ata/libata-core.c
@@ -5951,6 +5951,12 @@ void async_port_probe(void *data, async_cookie_t cookie)
 			 */
 		}
 	}
+
+	/* in order to keep device order, we need to synchronize at this point */
+	async_synchronize_cookie(cookie);
+
+	ata_scsi_scan_host(ap, 1);
+
 }
 /**
  *	ata_host_register - register initialized ATA host
@@ -6033,15 +6039,7 @@ int ata_host_register(struct ata_host *host, struct scsi_host_template *sht)
 		struct ata_port *ap = host->ports[i];
 		async_schedule(async_port_probe, ap);
 	}
-	async_synchronize_full();
-	/* probes are done, now scan each port's disk(s) */
-	DPRINTK("host probe begin\n");
-	for (i = 0; i < host->n_ports; i++) {
-		struct ata_port *ap = host->ports[i];
-
-		ata_scsi_scan_host(ap, 1);
-	}
-	DPRINTK("host probe end\n");
+	DPRINTK("probe end\n");
 
 	return 0;
 }
-- 
1.6.0.6




-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

Re: [PATCH 1/6] fastboot: Asynchronous function calls to speed up kernel boot

[Andrew Morton]

On Mon, 5 Jan 2009 20:10:41 -0800 Arjan van de Ven <arjan@infradead.org> wrote:

>  kernel/async.c         |  307 ++++++++++++++++++++++++++++++++++++++++++++++++

Am still wondering if this is unacceptably duplicative of dhowells's
slow-work infrastructure: http://lkml.org/lkml/2008/11/20/193

Re: [PATCH 1/6] fastboot: Asynchronous function calls to speed up kernel boot

[Arjan van de Ven]

On Mon, 5 Jan 2009 20:25:14 -0800
Andrew Morton <akpm@linux-foundation.org> wrote:

> On Mon, 5 Jan 2009 20:10:41 -0800 Arjan van de Ven
> <arjan@infradead.org> wrote:
> 
> >  kernel/async.c         |  307
> > ++++++++++++++++++++++++++++++++++++++++++++++++
> 
> Am still wondering if this is unacceptably duplicative of dhowells's
> slow-work infrastructure: http://lkml.org/lkml/2008/11/20/193
> 
> 

I talked to David about this today (as I wrote in 0/6). 
Based on that discussion we got to the conclusion we have incompatible
requirements. He needs priorities to be honored (which means rather out
of order execution of the tasks) while I need strict ordering (for the
synchronization). While it's not entirely impossible to combine those
two into one system, the resulting complexity isn't really worth it yet.

Of the 307 lines, only 100 are actual thread pool code (the rest is
synchronization and admin code), and about half of those 100 lines are
comments.


-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

Re: [PATCH 1/6] fastboot: Asynchronous function calls to speed up kernel boot

[Arjan van de Ven]

On Mon, 5 Jan 2009 20:25:14 -0800
Andrew Morton <akpm@linux-foundation.org> wrote:

> On Mon, 5 Jan 2009 20:10:41 -0800 Arjan van de Ven
> <arjan@infradead.org> wrote:
> 
> >  kernel/async.c         |  307
> > ++++++++++++++++++++++++++++++++++++++++++++++++
> 
> Am still wondering if this is unacceptably duplicative of dhowells's
> slow-work infrastructure: http://lkml.org/lkml/2008/11/20/193

ok having looked at it a lot more; I can add priorities to the async
function calls, at which point they are a superset to the slow-work
infrastructure basically. I'll need to check with David to make 10)%
sure the solution I have in mind will work for him, but I suspect it
will.


-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

Re: [PATCH 3/6] fastboot: make the libata port scan asynchronous

[Bert Wesarg]

Hi,

On Tue, Jan 6, 2009 at 05:12, Arjan van de Ven <arjan@infradead.org> wrote:
> diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c
> index fecca42..fe2c208 100644
> --- a/drivers/ata/libata-core.c
> +++ b/drivers/ata/libata-core.c
> @@ -56,6 +56,7 @@
>  #include <linux/workqueue.h>
>  #include <linux/scatterlist.h>
>  #include <linux/io.h>
> +#include <linux/async.h>
>  #include <scsi/scsi.h>
>  #include <scsi/scsi_cmnd.h>
>  #include <scsi/scsi_host.h>
> @@ -5909,6 +5910,48 @@ void ata_host_init(struct ata_host *host, struct device *dev,
>        host->ops = ops;
>  }
>
> +
> +void async_port_probe(void *data, async_cookie_t cookie)
static?

Bert

Re: [PATCH 4/6] fastboot: make ACPI bus drivers probe asynchronous

[Yong Wang]

On Mon, Jan 05, 2009 at 08:12:39PM -0800, Arjan van de Ven wrote:
> 
> >From 34c47dd38f49e29eb7cd7fa2078efc6b8b258bb9 Mon Sep 17 00:00:00 2001
> From: Arjan van de Ven <arjan@linux.intel.com>
> Date: Sun, 4 Jan 2009 05:32:28 -0800
> Subject: [PATCH] fastboot: make ACPI bus drivers probe asynchronous
> 
> the various ACPI bus drivers have non-overlapping devices and can
> each be run asynchronous. Some of the ACPI drivers (especially the
> battery one, but others as well) can take quite a long time to probe.
> 
> Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
> ---
>  drivers/acpi/scan.c |   23 ++++++++++++++++-------
>  1 files changed, 16 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c
> index 39b7233..a9e542d 100644
> --- a/drivers/acpi/scan.c
> +++ b/drivers/acpi/scan.c
> @@ -8,6 +8,7 @@
>  #include <linux/acpi.h>
>  #include <linux/signal.h>
>  #include <linux/kthread.h>
> +#include <linux/async.h>
>  
>  #include <acpi/acpi_drivers.h>
>  #include <acpi/acinterp.h>	/* for acpi_ex_eisa_id_to_string() */
> @@ -578,6 +579,19 @@ static int acpi_start_single_object(struct acpi_device *device)
>  	return result;
>  }
>  
> +static void acpi_bus_register_async(void *data, async_cookie_t cookie)
> +{
> +	int ret;
> +	struct acpi_driver *driver = data;
> +	driver->drv.name = driver->name;
> +	driver->drv.bus = &acpi_bus_type;
> +	driver->drv.owner = driver->owner;
> +
> +	async_synchronize_cookie(cookie);
> +
> +	ret = driver_register(&driver->drv);
> +	WARN_ON(ret != 0);
> +}
>  /**
>   * acpi_bus_register_driver - register a driver with the ACPI bus
>   * @driver: driver being registered
> @@ -588,16 +602,11 @@ static int acpi_start_single_object(struct acpi_device *device)
>   */
>  int acpi_bus_register_driver(struct acpi_driver *driver)
>  {
> -	int ret;
>  
>  	if (acpi_disabled)
>  		return -ENODEV;
> -	driver->drv.name = driver->name;
> -	driver->drv.bus = &acpi_bus_type;
> -	driver->drv.owner = driver->owner;
> -
> -	ret = driver_register(&driver->drv);
> -	return ret;
> +	async_schedule(acpi_bus_register_async, driver);
> +	return 0;
>  }
>  
>  EXPORT_SYMBOL(acpi_bus_register_driver);

For ACPI battery, button and thermal drivers whose init funtion immediately ends
right after calling acpi_bus_register_driver, it might be OK. But for other drivers
whose init function has more work to do and depends on acpi_bus_register_driver
having *really* done its job, it is certainly not OK. In the latter case, the
'return 0' in acpi_bus_register_driver is effectively cheating those drivers that
acpi_bus_register_driver has already done its job which is unfortunately not
the case. Example drivers include eeepc_laptop and asus_laptop. At least I am
seeing a kernel oops on eeepc because of that.

-Yong

Re: [PATCH 4/6] fastboot: make ACPI bus drivers probe asynchronous

[Arjan van de Ven]

On Wed, 7 Jan 2009 22:12:56 +0800
Yong Wang <yong.y.wang@linux.intel.com> wrote:

> On Mon, Jan 05, 2009 at 08:12:39PM -0800, Arjan van de Ven wrote:
> > 
> > >From 34c47dd38f49e29eb7cd7fa2078efc6b8b258bb9 Mon Sep 17 00:00:00
> > >2001
> > From: Arjan van de Ven <arjan@linux.intel.com>
> > Date: Sun, 4 Jan 2009 05:32:28 -0800
> > Subject: [PATCH] fastboot: make ACPI bus drivers probe asynchronous
> > 
> > the various ACPI bus drivers have non-overlapping devices and can
> > each be run asynchronous. Some of the ACPI drivers (especially the
> > battery one, but others as well) can take quite a long time to
> > probe.
> > 
> > Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
> > ---
> >  drivers/acpi/scan.c |   23 ++++++++++++++++-------
> >  1 files changed, 16 insertions(+), 7 deletions(-)
> > 
> > diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c
> > index 39b7233..a9e542d 100644
> > --- a/drivers/acpi/scan.c
> > +++ b/drivers/acpi/scan.c
> > @@ -8,6 +8,7 @@
> >  #include <linux/acpi.h>
> >  #include <linux/signal.h>
> >  #include <linux/kthread.h>
> > +#include <linux/async.h>
> >  
> >  #include <acpi/acpi_drivers.h>
> >  #include <acpi/acinterp.h>	/* for
> > acpi_ex_eisa_id_to_string() */ @@ -578,6 +579,19 @@ static int
> > acpi_start_single_object(struct acpi_device *device) return result;
> >  }
> >  
> > +static void acpi_bus_register_async(void *data, async_cookie_t
> > cookie) +{
> > +	int ret;
> > +	struct acpi_driver *driver = data;
> > +	driver->drv.name = driver->name;
> > +	driver->drv.bus = &acpi_bus_type;
> > +	driver->drv.owner = driver->owner;
> > +
> > +	async_synchronize_cookie(cookie);
> > +
> > +	ret = driver_register(&driver->drv);
> > +	WARN_ON(ret != 0);
> > +}
> >  /**
> >   * acpi_bus_register_driver - register a driver with the ACPI bus
> >   * @driver: driver being registered
> > @@ -588,16 +602,11 @@ static int acpi_start_single_object(struct
> > acpi_device *device) */
> >  int acpi_bus_register_driver(struct acpi_driver *driver)
> >  {
> > -	int ret;
> >  
> >  	if (acpi_disabled)
> >  		return -ENODEV;
> > -	driver->drv.name = driver->name;
> > -	driver->drv.bus = &acpi_bus_type;
> > -	driver->drv.owner = driver->owner;
> > -
> > -	ret = driver_register(&driver->drv);
> > -	return ret;
> > +	async_schedule(acpi_bus_register_async, driver);
> > +	return 0;
> >  }
> >  
> >  EXPORT_SYMBOL(acpi_bus_register_driver);
> 
> For ACPI battery, button and thermal drivers whose init funtion
> immediately ends right after calling acpi_bus_register_driver, it
> might be OK. But for other drivers whose init function has more work
> to do and depends on acpi_bus_register_driver having *really* done
> its job, it is certainly not OK. In the latter case, the 'return 0'
> in acpi_bus_register_driver is effectively cheating those drivers
> that acpi_bus_register_driver has already done its job which is
> unfortunately not the case. Example drivers include eeepc_laptop and
> asus_laptop. At least I am seeing a kernel oops on eeepc because of
> that.

ok I've dropped the ACPI part for now; I'll work with you and Len to
see how we can solve the ACPI side, but for now it seems to have too
many issues

> 
> -Yong


-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

Re: [PATCH 2/6] fastboot: make scsi probes asynchronous

[Mark Lord]

Arjan van de Ven wrote:
>>From 19617cfb1840abc69cc2fe74433007fc1daa1cb7 Mon Sep 17 00:00:00 2001
> From: Arjan van de Ven <arjan@linux.intel.com>
> Date: Sun, 4 Jan 2009 05:32:28 -0800
> Subject: [PATCH] fastboot: make scsi probes asynchronous
> 
> This patch makes part of the scsi probe (which is mostly device spin up and the
> partition scan) asynchronous. Only the part that runs after getting the device
> number allocated is asynchronous, ensuring that device numbering remains stable.
..

Will this burn up the PSUs in loaded server boxes,
due to all drives spinning-up simultaneously now?

Just curious.

Re: [PATCH 1/6] fastboot: Asynchronous function calls to speed up kernel boot

[Zhang Rui]

[-- Attachment #1: Type: text/plain, Size: 7810 bytes --]

On Tue, 2009-01-06 at 12:10 +0800, Arjan van de Ven wrote:
> From 15815a54b95e6866ff9532dead9cca4d6a298b54 Mon Sep 17 00:00:00 2001
> From: Arjan van de Ven <arjan@linux.intel.com>
> Date: Sun, 4 Jan 2009 05:32:28 -0800
> Subject: [PATCH] fastboot: Asynchronous function calls to speed up kernel boot
> 
> Right now, most of the kernel boot is strictly synchronous, such that
> various hardware delays are done sequentially.
> 
> In order to make the kernel boot faster, this patch introduces
> infrastructure to allow doing some of the initialization steps
> asynchronously, which will hide significant portions of the hardware delays
> in practice.
> 
> In order to not change device order and other similar observables, this
> patch does NOT do full parallel initialization.
> 
> Rather, it operates more in the way an out of order CPU does; the work may
> be done out of order and asynchronous, but the observable effects
> (instruction retiring for the CPU) are still done in the original sequence.
> 
> Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>

> +
> +void async_synchronize_cookie(async_cookie_t cookie)
> +{
> +       ktime_t starttime, delta, endtime;
> +
> +       if (initcall_debug) {
> +               printk("async_waiting @ %i\n", task_pid_nr(current));
> +               starttime = ktime_get();
> +       }
> +
> +       wait_event(async_done, lowest_in_progress >= cookie);
> +

In some cases, we only want to wait for a specific cookie
rather than all the cookies smaller than it.

For example:
device          cookie
ACPI battery    1
ata port 0      2
ata port 1      3
ata port 0 and port 1 in the same host can not be probed in parallel.

In this case, ata port1 should only wait for cookie 2 rather than both 1 and 2.

how about the patch below? (it's just a prototype and I have not tested it yet)
If it's okay, I'll do some tricks in the libata-core so that port1 can get the cookie of port 0.

Introduces two new interfaces
async_synchronize_one_cookie
async_synchronize_one_cookie_special
users can use these to wait for a specific cookie.

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
---
 drivers/ata/libata-core.c |    4 +-
 include/linux/async.h     |    6 ++-
 kernel/async.c            |   82 +++++++++++++++++++++++++++++++++++++++++-----
 3 files changed, 80 insertions(+), 12 deletions(-)

Index: linux-2.6/kernel/async.c
===================================================================
--- linux-2.6.orig/kernel/async.c
+++ linux-2.6/kernel/async.c
@@ -87,6 +87,44 @@ extern int initcall_debug;
 /*
  * MUST be called with the lock held!
  */
+static int __cookie_is_done(struct list_head *running, async_cookie_t cookie)
+{
+	struct async_entry *entry;
+
+        if (!list_empty(running)) {
+		list_for_each_entry(entry, running, list) {
+			if (entry->cookie > cookie)
+				break;
+			if (entry->cookie == cookie)
+				return 0;
+		}
+	}
+
+	if (!list_empty(&async_pending)) {
+		list_for_each_entry(entry, &async_pending, list) {
+			if (entry->cookie > cookie)
+				break;
+			if (entry->cookie == cookie)
+				return 0;
+		}
+	}
+	return 1;
+}
+
+static int cookie_is_done(struct list_head *running, async_cookie_t cookie)
+{
+	unsigned long flags;
+	async_cookie_t ret;
+
+	spin_lock_irqsave(&async_lock, flags);
+	ret = __cookie_is_done(running, cookie);
+	spin_unlock_irqrestore(&async_lock, flags);
+	return ret;
+}
+
+/*
+ * MUST be called with the lock held!
+ */
 static async_cookie_t  __lowest_in_progress(struct list_head *running)
 {
 	struct async_entry *entry;
@@ -220,18 +258,19 @@ EXPORT_SYMBOL_GPL(async_schedule_special
 void async_synchronize_full(void)
 {
 	do {
-		async_synchronize_cookie(next_cookie);
+		async_synchronize_cookies(next_cookie);
 	} while (!list_empty(&async_running) || !list_empty(&async_pending));
 }
 EXPORT_SYMBOL_GPL(async_synchronize_full);
 
 void async_synchronize_full_special(struct list_head *list)
 {
-	async_synchronize_cookie_special(next_cookie, list);
+	async_synchronize_cookies_special(next_cookie, list);
 }
 EXPORT_SYMBOL_GPL(async_synchronize_full_special);
 
-void async_synchronize_cookie_special(async_cookie_t cookie, struct list_head *running)
+static void __async_synchronize_cookie_special(async_cookie_t cookie,
+						struct list_head *running, int type)
 {
 	ktime_t starttime, delta, endtime;
 
@@ -240,7 +279,10 @@ void async_synchronize_cookie_special(as
 		starttime = ktime_get();
 	}
 
-	wait_event(async_done, lowest_in_progress(running) >= cookie);
+	if (type)
+		wait_event(async_done, lowest_in_progress(running) >= cookie);
+	else
+		wait_event(async_done, cookie_is_done(running, cookie));
 
 	if (initcall_debug && system_state == SYSTEM_BOOTING) {
 		endtime = ktime_get();
@@ -250,13 +292,37 @@ void async_synchronize_cookie_special(as
 			task_pid_nr(current), ktime_to_ns(delta) >> 10);
 	}
 }
-EXPORT_SYMBOL_GPL(async_synchronize_cookie_special);
 
-void async_synchronize_cookie(async_cookie_t cookie)
+void async_synchronize_cookies_special(async_cookie_t cookie, struct list_head *running)
+{
+	__async_synchronize_cookie_special(cookie, running, 1);
+}
+EXPORT_SYMBOL_GPL(async_synchronize_cookies_special);
+
+void async_synchronize_cookies(async_cookie_t cookie)
+{
+	__async_synchronize_cookie_special(cookie, &async_running, 1);
+}
+EXPORT_SYMBOL_GPL(async_synchronize_cookies);
+
+
+/*
+ * sychronize a specific cookie
+ * block until the entry with a speicific cookie is done.
+ * Note that waiting for a cookie while holding it is not allowed.
+ * because it may cause a deadlock issue.
+ */
+void async_synchronize_one_cookie_special(async_cookie_t cookie, struct list_head *running)
+{
+	__async_synchronize_cookie_special(cookie, running, 0);
+}
+EXPORT_SYMBOL_GPL(async_synchronize_one_cookie_special);
+
+void async_synchronize_one_cookie(async_cookie_t cookie)
 {
-	async_synchronize_cookie_special(cookie, &async_running);
+	__async_synchronize_cookie_special(cookie, &async_running, 0);
 }
-EXPORT_SYMBOL_GPL(async_synchronize_cookie);
+EXPORT_SYMBOL_GPL(async_synchronize_one_cookie);
 
 
 static int async_thread(void *unused)
Index: linux-2.6/include/linux/async.h
===================================================================
--- linux-2.6.orig/include/linux/async.h
+++ linux-2.6/include/linux/async.h
@@ -20,6 +20,8 @@ extern async_cookie_t async_schedule(asy
 extern async_cookie_t async_schedule_special(async_func_ptr *ptr, void *data, struct list_head *list);
 extern void async_synchronize_full(void);
 extern void async_synchronize_full_special(struct list_head *list);
-extern void async_synchronize_cookie(async_cookie_t cookie);
-extern void async_synchronize_cookie_special(async_cookie_t cookie, struct list_head *list);
+extern void async_synchronize_cookies(async_cookie_t cookie);
+extern void async_synchronize_cookies_special(async_cookie_t cookie, struct list_head *list);
+extern void async_synchronize_one_cookie(async_cookie_t cookie);
+extern void async_synchronize_one_cookie_special(async_cookie_t cookie, struct list_head *list);
 
Index: linux-2.6/drivers/ata/libata-core.c
===================================================================
--- linux-2.6.orig/drivers/ata/libata-core.c
+++ linux-2.6/drivers/ata/libata-core.c
@@ -5929,7 +5929,7 @@ static void async_port_probe(void *data,
 	 * don't need to wait for port 0, only for later ports.
 	 */
 	if (!(ap->host->flags & ATA_HOST_PARALLEL_SCAN) && ap->port_no != 0)
-		async_synchronize_cookie(cookie);
+		async_synchronize_cookies(cookie);
 
 	/* probe */
 	if (ap->ops->error_handler) {
@@ -5969,7 +5969,7 @@ static void async_port_probe(void *data,
 	}
 
 	/* in order to keep device order, we need to synchronize at this point */
-	async_synchronize_cookie(cookie);
+	async_synchronize_cookies(cookie);
 
 	ata_scsi_scan_host(ap, 1);
 




[-- Attachment #2: patch-async-introduce-new-interface --]
[-- Type: text/plain, Size: 6271 bytes --]


In some cases, we only want to wait for a specific cookie
rather than all the cookies smaller than it.
Introduces two new interfaces
async_synchronize_one_cookie
async_synchronize_one_cookie_special
users can use these to wait for a specific cookie.

For example:
device          cookie
ACPI battery    1
ata port 0      2
ata port 1      3
ata port 0 and port 1 in the same host can not be porbed in parallel.

In this case, ata port1 should only wait for cookie 2 rather than both 1 and 2.
And using the new interface would be a better choice.

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
---
 drivers/ata/libata-core.c |    4 +-
 include/linux/async.h     |    6 ++-
 kernel/async.c            |   82 +++++++++++++++++++++++++++++++++++++++++-----
 3 files changed, 80 insertions(+), 12 deletions(-)

Index: linux-2.6/kernel/async.c
===================================================================
--- linux-2.6.orig/kernel/async.c
+++ linux-2.6/kernel/async.c
@@ -87,6 +87,44 @@ extern int initcall_debug;
 /*
  * MUST be called with the lock held!
  */
+static int __cookie_is_done(struct list_head *running, async_cookie_t cookie)
+{
+	struct async_entry *entry;
+
+        if (!list_empty(running)) {
+		list_for_each_entry(entry, running, list) {
+			if (entry->cookie > cookie)
+				break;
+			if (entry->cookie == cookie)
+				return 0;
+		}
+	}
+
+	if (!list_empty(&async_pending)) {
+		list_for_each_entry(entry, &async_pending, list) {
+			if (entry->cookie > cookie)
+				break;
+			if (entry->cookie == cookie)
+				return 0;
+		}
+	}
+	return 1;
+}
+
+static int cookie_is_done(struct list_head *running, async_cookie_t cookie)
+{
+	unsigned long flags;
+	async_cookie_t ret;
+
+	spin_lock_irqsave(&async_lock, flags);
+	ret = __cookie_is_done(running, cookie);
+	spin_unlock_irqrestore(&async_lock, flags);
+	return ret;
+}
+
+/*
+ * MUST be called with the lock held!
+ */
 static async_cookie_t  __lowest_in_progress(struct list_head *running)
 {
 	struct async_entry *entry;
@@ -220,18 +258,19 @@ EXPORT_SYMBOL_GPL(async_schedule_special
 void async_synchronize_full(void)
 {
 	do {
-		async_synchronize_cookie(next_cookie);
+		async_synchronize_cookies(next_cookie);
 	} while (!list_empty(&async_running) || !list_empty(&async_pending));
 }
 EXPORT_SYMBOL_GPL(async_synchronize_full);
 
 void async_synchronize_full_special(struct list_head *list)
 {
-	async_synchronize_cookie_special(next_cookie, list);
+	async_synchronize_cookies_special(next_cookie, list);
 }
 EXPORT_SYMBOL_GPL(async_synchronize_full_special);
 
-void async_synchronize_cookie_special(async_cookie_t cookie, struct list_head *running)
+static void __async_synchronize_cookie_special(async_cookie_t cookie,
+						struct list_head *running, int type)
 {
 	ktime_t starttime, delta, endtime;
 
@@ -240,7 +279,10 @@ void async_synchronize_cookie_special(as
 		starttime = ktime_get();
 	}
 
-	wait_event(async_done, lowest_in_progress(running) >= cookie);
+	if (type)
+		wait_event(async_done, lowest_in_progress(running) >= cookie);
+	else
+		wait_event(async_done, cookie_is_done(running, cookie));
 
 	if (initcall_debug && system_state == SYSTEM_BOOTING) {
 		endtime = ktime_get();
@@ -250,13 +292,37 @@ void async_synchronize_cookie_special(as
 			task_pid_nr(current), ktime_to_ns(delta) >> 10);
 	}
 }
-EXPORT_SYMBOL_GPL(async_synchronize_cookie_special);
 
-void async_synchronize_cookie(async_cookie_t cookie)
+void async_synchronize_cookies_special(async_cookie_t cookie, struct list_head *running)
+{
+	__async_synchronize_cookie_special(cookie, running, 1);
+}
+EXPORT_SYMBOL_GPL(async_synchronize_cookies_special);
+
+void async_synchronize_cookies(async_cookie_t cookie)
+{
+	__async_synchronize_cookie_special(cookie, &async_running, 1);
+}
+EXPORT_SYMBOL_GPL(async_synchronize_cookies);
+
+
+/*
+ * sychronize a specific cookie
+ * block until the entry with a speicific cookie is done.
+ * Note that waiting for a cookie while holding it is not allowed.
+ * because it may cause a deadlock issue.
+ */
+void async_synchronize_one_cookie_special(async_cookie_t cookie, struct list_head *running)
+{
+	__async_synchronize_cookie_special(cookie, running, 0);
+}
+EXPORT_SYMBOL_GPL(async_synchronize_one_cookie_special);
+
+void async_synchronize_one_cookie(async_cookie_t cookie)
 {
-	async_synchronize_cookie_special(cookie, &async_running);
+	__async_synchronize_cookie_special(cookie, &async_running, 0);
 }
-EXPORT_SYMBOL_GPL(async_synchronize_cookie);
+EXPORT_SYMBOL_GPL(async_synchronize_one_cookie);
 
 
 static int async_thread(void *unused)
Index: linux-2.6/include/linux/async.h
===================================================================
--- linux-2.6.orig/include/linux/async.h
+++ linux-2.6/include/linux/async.h
@@ -20,6 +20,8 @@ extern async_cookie_t async_schedule(asy
 extern async_cookie_t async_schedule_special(async_func_ptr *ptr, void *data, struct list_head *list);
 extern void async_synchronize_full(void);
 extern void async_synchronize_full_special(struct list_head *list);
-extern void async_synchronize_cookie(async_cookie_t cookie);
-extern void async_synchronize_cookie_special(async_cookie_t cookie, struct list_head *list);
+extern void async_synchronize_cookies(async_cookie_t cookie);
+extern void async_synchronize_cookies_special(async_cookie_t cookie, struct list_head *list);
+extern void async_synchronize_one_cookie(async_cookie_t cookie);
+extern void async_synchronize_one_cookie_special(async_cookie_t cookie, struct list_head *list);
 
Index: linux-2.6/drivers/ata/libata-core.c
===================================================================
--- linux-2.6.orig/drivers/ata/libata-core.c
+++ linux-2.6/drivers/ata/libata-core.c
@@ -5929,7 +5929,7 @@ static void async_port_probe(void *data,
 	 * don't need to wait for port 0, only for later ports.
 	 */
 	if (!(ap->host->flags & ATA_HOST_PARALLEL_SCAN) && ap->port_no != 0)
-		async_synchronize_cookie(cookie);
+		async_synchronize_cookies(cookie);
 
 	/* probe */
 	if (ap->ops->error_handler) {
@@ -5969,7 +5969,7 @@ static void async_port_probe(void *data,
 	}
 
 	/* in order to keep device order, we need to synchronize at this point */
-	async_synchronize_cookie(cookie);
+	async_synchronize_cookies(cookie);
 
 	ata_scsi_scan_host(ap, 1);
 

Re: [PATCH 1/6] fastboot: Asynchronous function calls to speed up kernel boot

[Arjan van de Ven]

On Wed, 14 Jan 2009 13:32:43 +0800
Zhang Rui <rui.zhang@intel.com> wrote:
> > +
> > +       wait_event(async_done, lowest_in_progress >= cookie);
> > +
> 
> In some cases, we only want to wait for a specific cookie
> rather than all the cookies smaller than it.
> 
> For example:
> device          cookie
> ACPI battery    1
> ata port 0      2
> ata port 1      3
> ata port 0 and port 1 in the same host can not be probed in parallel.
> 
> In this case, ata port1 should only wait for cookie 2 rather than
> both 1 and 2.
> 

I really don't want to go this way; ordering is really important and I
don't want to break this.

If you really have a fully separate domain (and there's only few of
those in our kernel), then use a special run list.



-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

Re: [PATCH 1/6] fastboot: Asynchronous function calls to speed up kernel boot

[Zhang Rui]

On Tue, 2009-01-06 at 12:10 +0800, Arjan van de Ven wrote:
> From 15815a54b95e6866ff9532dead9cca4d6a298b54 Mon Sep 17 00:00:00 2001
> From: Arjan van de Ven <arjan@linux.intel.com>
> Date: Sun, 4 Jan 2009 05:32:28 -0800
> Subject: [PATCH] fastboot: Asynchronous function calls to speed up kernel boot
> 
> Right now, most of the kernel boot is strictly synchronous, such that
> various hardware delays are done sequentially.
> 
> In order to make the kernel boot faster, this patch introduces
> infrastructure to allow doing some of the initialization steps
> asynchronously, which will hide significant portions of the hardware delays
> in practice.
> 
> In order to not change device order and other similar observables, this
> patch does NOT do full parallel initialization.
> 
> Rather, it operates more in the way an out of order CPU does; the work may
> be done out of order and asynchronous, but the observable effects
> (instruction retiring for the CPU) are still done in the original sequence.
> 
> Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
> ---
>  include/linux/async.h  |   21 ++++
>  init/do_mounts.c       |    2 +
>  init/main.c            |    5 +-
>  kernel/Makefile        |    3 +-
>  kernel/async.c         |  307 ++++++++++++++++++++++++++++++++++++++++++++++++
>  kernel/irq/autoprobe.c |    5 +
>  kernel/module.c        |    2 +
>  7 files changed, 343 insertions(+), 2 deletions(-)
>  create mode 100644 include/linux/async.h
>  create mode 100644 kernel/async.c
> +/*
> + * pick the first pending entry and run it
> + */
> +static void run_one_entry(void)
> +{
> +       unsigned long flags;
> +       struct async_entry *entry;
> +       ktime_t calltime, delta, rettime;
> +
> +       /* 1) pick one task from the pending queue */
> +
> +       spin_lock_irqsave(&async_lock, flags);
> +       if (list_empty(&async_pending))
> +               goto out;
> +       entry = list_first_entry(&async_pending, struct async_entry, list);
> +
> +       /* 2) move it to the running queue */
> +       list_del(&entry->list);
> +       list_add_tail(&entry->list, &async_running);
> +       spin_unlock_irqrestore(&async_lock, flags);
> +
another question,
we should move the entry to the proper run list, don't we?

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
---
 kernel/async.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-2.6/kernel/async.c
===================================================================
--- linux-2.6.orig/kernel/async.c
+++ linux-2.6/kernel/async.c
@@ -133,7 +133,7 @@ static void run_one_entry(void)
 
 	/* 2) move it to the running queue */
 	list_del(&entry->list);
-	list_add_tail(&entry->list, &async_running);
+	list_add_tail(&entry->list, entry->running);
 	spin_unlock_irqrestore(&async_lock, flags);
 
 	/* 3) run it (and print duration)*/


> +       /* 3) run it (and print duration)*/
> +       if (initcall_debug) {
> +               printk("calling  %lli_%pF @ %i\n", entry->cookie, entry->func, task_pid_nr(current));
> +               calltime = ktime_get();
> +       }
> +       entry->func(entry->data, entry->cookie);
> +       if (initcall_debug) {
> +               rettime = ktime_get();
> +               delta = ktime_sub(rettime, calltime);
> +               printk("initcall %lli_%pF returned 0 after %lld usecs\n", entry->cookie,
> +                       entry->func, ktime_to_ns(delta) >> 10);
> +       }
> +
> +       /* 4) remove it from the running queue */
> +       spin_lock_irqsave(&async_lock, flags);
> +       list_del(&entry->list);
> +
> +       /* 5) free the entry  */
> +       kfree(entry);
> +       atomic_dec(&entry_count);
> +
> +       /* 6) update the lowest_in_progress value */
> +       __recalc_lowest_in_progress();
> +
> +       spin_unlock_irqrestore(&async_lock, flags);
> +
> +       /* 7) wake up any waiters. */
> +       wake_up(&async_done);
> +       return;
> +
> +out:
> +       spin_unlock_irqrestore(&async_lock, flags);
> +}
> +
> +
> +async_cookie_t async_schedule(async_func_ptr *ptr, void *data)
> +{
> +       struct async_entry *entry;
> +       unsigned long flags;
> +       async_cookie_t newcookie;
> +
> +
> +       /* allow irq-off callers */
> +       entry = kzalloc(sizeof(struct async_entry), GFP_ATOMIC);
> +       if (!entry) {
> +               spin_lock_irqsave(&async_lock, flags);
> +               newcookie = next_cookie++;
> +               spin_unlock_irqrestore(&async_lock, flags);
> +
> +               /* low on memory.. run synchronously */
> +               ptr(data, newcookie);
> +               return newcookie;
> +       }
> +       entry->func = ptr;
> +       entry->data = data;
> +
> +       spin_lock_irqsave(&async_lock, flags);
> +       newcookie = entry->cookie = next_cookie++;
> +       list_add_tail(&entry->list, &async_pending);
> +       atomic_inc(&entry_count);
> +       spin_unlock_irqrestore(&async_lock, flags);
> +       wake_up(&async_new);
> +       return newcookie;
> +}
> +EXPORT_SYMBOL_GPL(async_schedule);
> +