Merge davem@nuts.ninka.net:/home/davem/src/BK/net-2.5

into kernel.bkbits.net:/home/davem/net-2.5

arch/i386/oprofile/nmi_int.c

@@ -58,7 +58,7 @@ static void nmi_save_registers(struct op_msrs * msrs)
 	unsigned int const nr_ctrls = model->num_controls; 
 	struct op_msr_group * counters = &msrs->counters;
 	struct op_msr_group * controls = &msrs->controls;
-	int i;
+	unsigned int i;
 
 	for (i = 0; i < nr_ctrs; ++i) {
 		rdmsr(counters->addrs[i],
@@ -108,7 +108,7 @@ static void nmi_restore_registers(struct op_msrs * msrs)
 	unsigned int const nr_ctrls = model->num_controls; 
 	struct op_msr_group * counters = &msrs->counters;
 	struct op_msr_group * controls = &msrs->controls;
-	int i;
+	unsigned int i;
 
 	for (i = 0; i < nr_ctrls; ++i) {
 		wrmsr(controls->addrs[i],
@@ -182,7 +182,7 @@ struct op_counter_config counter_config[OP_MAX_COUNTER];
 
 static int nmi_create_files(struct super_block * sb, struct dentry * root)
 {
-	int i;
+	unsigned int i;
 
 	for (i = 0; i < model->num_counters; ++i) {
 		struct dentry * dir;

arch/i386/oprofile/op_model_p4.c

@@ -389,7 +389,7 @@ static unsigned long reset_value[NUM_COUNTERS_NON_HT];
 
 static void p4_fill_in_addresses(struct op_msrs * const msrs)
 {
-	int i; 
+	unsigned int i; 
 	unsigned int addr, stag;
 
 	setup_num_counters();

drivers/base/platform.c

@@ -59,12 +59,9 @@ void platform_device_unregister(struct platform_device * pdev)
 
 static int platform_match(struct device * dev, struct device_driver * drv)
 {
-	char name[BUS_ID_SIZE];
+	struct platform_device *pdev = container_of(dev, struct platform_device, dev);
 
-	if (sscanf(dev->bus_id,"%s",name))
-		return (strcmp(name,drv->name) == 0);
-
-	return 0;
+	return (strncmp(pdev->name, drv->name, BUS_ID_SIZE) == 0);
 }
 
 struct bus_type platform_bus_type = {

drivers/cpufreq/userspace.c

@@ -511,7 +511,7 @@ static int cpufreq_governor_userspace(struct cpufreq_policy *policy,
 		cpu_min_freq[cpu] = policy->min;
 		cpu_max_freq[cpu] = policy->max;
 		cpu_cur_freq[cpu] = policy->cur;
-		device_create_file (policy->intf.dev, &dev_attr_scaling_setspeed);
+		device_create_file (policy->dev, &dev_attr_scaling_setspeed);
 		memcpy (&current_policy[cpu], policy, sizeof(struct cpufreq_policy));
 		up(&userspace_sem);
 		break;
@@ -520,7 +520,7 @@ static int cpufreq_governor_userspace(struct cpufreq_policy *policy,
 		cpu_is_managed[cpu] = 0;
 		cpu_min_freq[cpu] = 0;
 		cpu_max_freq[cpu] = 0;
-		device_remove_file (policy->intf.dev, &dev_attr_scaling_setspeed);
+		device_remove_file (policy->dev, &dev_attr_scaling_setspeed);
 		up(&userspace_sem);
 		module_put(THIS_MODULE);
 		break;

drivers/oprofile/buffer_sync.c

@@ -277,10 +277,7 @@ static void release_mm(struct mm_struct * mm)
  */
 static struct mm_struct * take_task_mm(struct task_struct * task)
 {
-	struct mm_struct * mm;
-	task_lock(task);
-	mm = task->mm;
-	task_unlock(task);
+	struct mm_struct * mm = task->mm;
  
 	/* if task->mm !NULL, mm_count must be at least 1. It cannot
 	 * drop to 0 without the task exiting, which will have to sleep
@@ -310,6 +307,32 @@ static inline int is_ctx_switch(unsigned long val)
 }
  
 
+/* compute number of filled slots in cpu_buffer queue */
+static unsigned long nr_filled_slots(struct oprofile_cpu_buffer * b)
+{
+	unsigned long head = b->head_pos;
+	unsigned long tail = b->tail_pos;
+
+	if (head >= tail)
+		return head - tail;
+
+	return head + (b->buffer_size - tail);
+}
+
+
+static void increment_tail(struct oprofile_cpu_buffer * b)
+{
+	unsigned long new_tail = b->tail_pos + 1;
+
+	rmb();
+
+	if (new_tail < (b->buffer_size))
+		b->tail_pos = new_tail;
+	else
+		b->tail_pos = 0;
+}
+
+
 /* Sync one of the CPU's buffers into the global event buffer.
  * Here we need to go through each batch of samples punctuated
  * by context switch notes, taking the task's mmap_sem and doing
@@ -322,10 +345,14 @@ static void sync_buffer(struct oprofile_cpu_buffer * cpu_buf)
 	struct task_struct * new;
 	unsigned long cookie;
 	int in_kernel = 1;
-	int i;
+	unsigned int i;
  
-	for (i=0; i < cpu_buf->pos; ++i) {
-		struct op_sample * s = &cpu_buf->buffer[i];
+	/* Remember, only we can modify tail_pos */
+
+	unsigned long const available_elements = nr_filled_slots(cpu_buf);
+  
+	for (i=0; i < available_elements; ++i) {
+		struct op_sample * s = &cpu_buf->buffer[cpu_buf->tail_pos];
  
 		if (is_ctx_switch(s->eip)) {
 			if (s->event <= 1) {
@@ -345,6 +372,8 @@ static void sync_buffer(struct oprofile_cpu_buffer * cpu_buf)
 		} else {
 			add_sample(mm, s, in_kernel);
 		}
+
+		increment_tail(cpu_buf);
 	}
 	release_mm(mm);
 
@@ -369,17 +398,8 @@ static void sync_cpu_buffers(void)
  
 		cpu_buf = &cpu_buffer[i];
  
-		/* We take a spin lock even though we might
-		 * sleep. It's OK because other users are try
-		 * lockers only, and this region is already
-		 * protected by buffer_sem. It's raw to prevent
-		 * the preempt bogometer firing. Fruity, huh ? */
-		if (cpu_buf->pos > 0) {
-			_raw_spin_lock(&cpu_buf->int_lock);
 		add_cpu_switch(i);
 		sync_buffer(cpu_buf);
-			_raw_spin_unlock(&cpu_buf->int_lock);
-		}
 	}
 
 	up(&buffer_sem);

drivers/oprofile/cpu_buffer.c

@@ -26,8 +26,6 @@
 
 struct oprofile_cpu_buffer cpu_buffer[NR_CPUS] __cacheline_aligned;
 
-static unsigned long buffer_size;
- 
 static void __free_cpu_buffers(int num)
 {
 	int i;
@@ -47,7 +45,7 @@ int alloc_cpu_buffers(void)
 {
 	int i;
  
-	buffer_size = fs_cpu_buffer_size;
+	unsigned long buffer_size = fs_cpu_buffer_size;
  
 	for (i=0; i < NR_CPUS; ++i) {
 		struct oprofile_cpu_buffer * b = &cpu_buffer[i];
@@ -59,12 +57,12 @@ int alloc_cpu_buffers(void)
 		if (!b->buffer)
 			goto fail;
  
-		spin_lock_init(&b->int_lock);
-		b->pos = 0;
 		b->last_task = 0;
 		b->last_is_kernel = -1;
+		b->buffer_size = buffer_size;
+		b->tail_pos = 0;
+		b->head_pos = 0;
 		b->sample_received = 0;
-		b->sample_lost_locked = 0;
 		b->sample_lost_overflow = 0;
 		b->sample_lost_task_exit = 0;
 	}
@@ -81,10 +79,40 @@ void free_cpu_buffers(void)
 }
 
 
-/* Note we can't use a semaphore here as this is supposed to
- * be safe from any context. Instead we trylock the CPU's int_lock.
- * int_lock is taken by the processing code in sync_cpu_buffers()
- * so we avoid disturbing that.
+/* compute number of available slots in cpu_buffer queue */
+static unsigned long nr_available_slots(struct oprofile_cpu_buffer const * b)
+{
+	unsigned long head = b->head_pos;
+	unsigned long tail = b->tail_pos;
+
+	if (tail == head)
+		return b->buffer_size;
+
+	if (tail > head)
+		return tail - head;
+
+	return tail + (b->buffer_size - head);
+}
+
+
+static void increment_head(struct oprofile_cpu_buffer * b)
+{
+	unsigned long new_head = b->head_pos + 1;
+
+	/* Ensure anything written to the slot before we
+	 * increment is visible */
+	wmb();
+
+	if (new_head < (b->buffer_size))
+		b->head_pos = new_head;
+	else
+		b->head_pos = 0;
+}
+
+
+/* This must be safe from any context. It's safe writing here
+ * because of the head/tail separation of the writer and reader
+ * of the CPU buffer.
  *
  * is_kernel is needed because on some architectures you cannot
  * tell if you are in kernel or user space simply by looking at
@@ -101,14 +129,10 @@ void oprofile_add_sample(unsigned long eip, unsigned int is_kernel,
 
 	cpu_buf->sample_received++;
  
-	if (!spin_trylock(&cpu_buf->int_lock)) {
-		cpu_buf->sample_lost_locked++;
-		return;
-	}
 
-	if (cpu_buf->pos > buffer_size - 2) {
+	if (nr_available_slots(cpu_buf) < 3) {
 		cpu_buf->sample_lost_overflow++;
-		goto out;
+		return;
 	}
 
 	task = current;
@@ -116,18 +140,18 @@ void oprofile_add_sample(unsigned long eip, unsigned int is_kernel,
 	/* notice a switch from user->kernel or vice versa */
 	if (cpu_buf->last_is_kernel != is_kernel) {
 		cpu_buf->last_is_kernel = is_kernel;
-		cpu_buf->buffer[cpu_buf->pos].eip = ~0UL;
-		cpu_buf->buffer[cpu_buf->pos].event = is_kernel;
-		cpu_buf->pos++;
+		cpu_buf->buffer[cpu_buf->head_pos].eip = ~0UL;
+		cpu_buf->buffer[cpu_buf->head_pos].event = is_kernel;
+		increment_head(cpu_buf);
 	}
 
 	/* notice a task switch */
 	if (cpu_buf->last_task != task) {
 		cpu_buf->last_task = task;
 		if (!(task->flags & PF_EXITING)) {
-			cpu_buf->buffer[cpu_buf->pos].eip = ~0UL;
-			cpu_buf->buffer[cpu_buf->pos].event = (unsigned long)task;
-			cpu_buf->pos++;
+			cpu_buf->buffer[cpu_buf->head_pos].eip = ~0UL;
+			cpu_buf->buffer[cpu_buf->head_pos].event = (unsigned long)task;
+			increment_head(cpu_buf);
 		}
 	}
  
@@ -138,23 +162,20 @@ void oprofile_add_sample(unsigned long eip, unsigned int is_kernel,
 	 */
 	if (task->flags & PF_EXITING) {
 		cpu_buf->sample_lost_task_exit++;
-		goto out;
+		return;
 	}
  
-	cpu_buf->buffer[cpu_buf->pos].eip = eip;
-	cpu_buf->buffer[cpu_buf->pos].event = event;
-	cpu_buf->pos++;
-out:
-	spin_unlock(&cpu_buf->int_lock);
+	cpu_buf->buffer[cpu_buf->head_pos].eip = eip;
+	cpu_buf->buffer[cpu_buf->head_pos].event = event;
+	increment_head(cpu_buf);
 }
 
+
 /* resets the cpu buffer to a sane state - should be called with 
  * cpu_buf->int_lock held
  */
 void cpu_buffer_reset(struct oprofile_cpu_buffer *cpu_buf)
 {
-	cpu_buf->pos = 0;
-
 	/* reset these to invalid values; the next sample
 	 * collected will populate the buffer with proper
 	 * values to initialize the buffer
@@ -162,4 +183,3 @@ void cpu_buffer_reset(struct oprofile_cpu_buffer *cpu_buf)
 	cpu_buf->last_is_kernel = -1;
 	cpu_buf->last_task = 0;
 }
-

drivers/oprofile/cpu_buffer.h

@@ -30,14 +30,13 @@ struct op_sample {
 };
  
 struct oprofile_cpu_buffer {
-	spinlock_t int_lock;
-	/* protected by int_lock */
-	unsigned long pos;
+	volatile unsigned long head_pos;
+	volatile unsigned long tail_pos;
+	unsigned long buffer_size;
 	struct task_struct * last_task;
 	int last_is_kernel;
 	struct op_sample * buffer;
 	unsigned long sample_received;
-	unsigned long sample_lost_locked;
 	unsigned long sample_lost_overflow;
 	unsigned long sample_lost_task_exit;
 } ____cacheline_aligned;

drivers/oprofile/oprofile_stats.c

@@ -27,7 +27,6 @@ void oprofile_reset_stats(void)
 
 		cpu_buf = &cpu_buffer[i]; 
 		cpu_buf->sample_received = 0;
-		cpu_buf->sample_lost_locked = 0;
 		cpu_buf->sample_lost_overflow = 0;
 		cpu_buf->sample_lost_task_exit = 0;
 	}
@@ -63,8 +62,6 @@ void oprofile_create_stats_files(struct super_block * sb, struct dentry * root)
 		 */
 		oprofilefs_create_ro_ulong(sb, cpudir, "sample_received",
 			&cpu_buf->sample_received);
-		oprofilefs_create_ro_ulong(sb, cpudir, "sample_lost_locked",
-			&cpu_buf->sample_lost_locked);
 		oprofilefs_create_ro_ulong(sb, cpudir, "sample_lost_overflow",
 			&cpu_buf->sample_lost_overflow);
 		oprofilefs_create_ro_ulong(sb, cpudir, "sample_lost_task_exit",

drivers/parport/ieee1284.c

@@ -170,7 +170,7 @@ int parport_wait_peripheral(struct parport *port,
 {
 	int ret;
 	int usec;
-	long deadline;
+	unsigned long deadline;
 	unsigned char status;
 
 	usec = port->physport->spintime; /* usecs of fast polling */

drivers/parport/ieee1284_ops.c

@@ -58,7 +58,7 @@ size_t parport_ieee1284_write_compat (struct parport *port,
 	parport_write_control (port, ctl);
 	parport_data_forward (port);
 	while (count < len) {
-		long expire = jiffies + dev->timeout;
+		unsigned long expire = jiffies + dev->timeout;
 		long wait = (HZ + 99) / 100;
 		unsigned char mask = (PARPORT_STATUS_ERROR
 				      | PARPORT_STATUS_BUSY);
@@ -431,7 +431,7 @@ size_t parport_ieee1284_ecp_write_data (struct parport *port,
 			      | PARPORT_CONTROL_INIT,
 			      PARPORT_CONTROL_INIT);
 	for (written = 0; written < len; written++, buf++) {
-		long expire = jiffies + port->cad->timeout;
+		unsigned long expire = jiffies + port->cad->timeout;
 		unsigned char byte;
 
 		byte = *buf;
@@ -520,7 +520,7 @@ size_t parport_ieee1284_ecp_read_data (struct parport *port,
 	parport_write_control (port,
 			       ctl | PARPORT_CONTROL_AUTOFD);
 	while (count < len) {
-		long expire = jiffies + dev->timeout;
+		unsigned long expire = jiffies + dev->timeout;
 		unsigned char byte;
 		int command;
 
@@ -668,7 +668,7 @@ size_t parport_ieee1284_ecp_write_addr (struct parport *port,
 			      PARPORT_CONTROL_AUTOFD
 			      | PARPORT_CONTROL_INIT);
 	for (written = 0; written < len; written++, buf++) {
-		long expire = jiffies + port->cad->timeout;
+		unsigned long expire = jiffies + port->cad->timeout;
 		unsigned char byte;
 
 		byte = *buf;

fs/filesystems.c

@@ -125,9 +125,7 @@ int register_filesystem(struct file_system_type * fs)
 	if (!res) {
 		/* we implicitly possess reference to @fs during registration,
 		 * so it cannot be unregister from under us. */
-		if (register_fs_subsys(fs))
-			printk(KERN_WARNING "Failed to register '%s' in sysfs\n",
-			       fs->name);
+		register_fs_subsys(fs);
 	}
 	return res;
 }

fs/sysfs/mount.c

@@ -33,7 +33,7 @@ static int sysfs_fill_super(struct super_block *sb, void *data, int silent)
 	sb->s_op = &sysfs_ops;
 	sysfs_sb = sb;
 
-	inode = sysfs_new_inode(S_IFDIR | S_IRUGO | S_IWUSR);
+	inode = sysfs_new_inode(S_IFDIR | S_IRWXU | S_IRUGO | S_IXUGO);
 	if (inode) {
 		inode->i_op = &simple_dir_inode_operations;
 		inode->i_fop = &simple_dir_operations;

include/linux/jiffies.h

 #ifndef _LINUX_JIFFIES_H
 #define _LINUX_JIFFIES_H
 
+#include <linux/kernel.h>
 #include <linux/types.h>
 #include <linux/spinlock.h>
 #include <linux/seqlock.h>
@@ -37,10 +38,16 @@ static inline u64 get_jiffies_64(void)
  * good compiler would generate better code (and a really good compiler
  * wouldn't care). Gcc is currently neither.
  */
-#define time_after(a,b)		((long)(b) - (long)(a) < 0)
+#define time_after(a,b)		\
+	(typecheck(unsigned long, a) && \
+	 typecheck(unsigned long, b) && \
+	 ((long)(b) - (long)(a) < 0))
 #define time_before(a,b)	time_after(b,a)
 
-#define time_after_eq(a,b)	((long)(a) - (long)(b) >= 0)
+#define time_after_eq(a,b)	\
+	(typecheck(unsigned long, a) && \
+	 typecheck(unsigned long, b) && \
+	 ((long)(a) - (long)(b) >= 0))
 #define time_before_eq(a,b)	time_after_eq(b,a)
 
 #endif

include/linux/kernel.h

@@ -188,6 +188,17 @@ extern void dump_stack(void);
         const typeof( ((type *)0)->member ) *__mptr = (ptr);	\
         (type *)( (char *)__mptr - offsetof(type,member) );})
 
+/*
+ * Check at compile time that something is of a particular type.
+ * Always evaluates to 1 so you may use it easily in conparisons.
+ */
+#define typecheck(type,x) \
+({	type __dummy; \
+	typeof(x) __dummy2; \
+	(void)(&__dummy == &__dummy2); \
+	1; \
+})
+
 #endif /* __KERNEL__ */
 
 #define SI_LOAD_SHIFT	16

sound/pci/ac97/ac97_codec.c

@@ -1808,7 +1808,7 @@ static void snd_ac97_get_name(ac97_t *ac97, unsigned int id, char *name)
  */
 static int ac97_reset_wait(ac97_t *ac97, int timeout, int with_modem)
 {
-	signed long end_time;
+	unsigned long end_time;
 	end_time = jiffies + timeout;
 	do {
 		unsigned short ext_mid;
@@ -1866,7 +1866,7 @@ int snd_ac97_mixer(snd_card_t * card, ac97_t * _ac97, ac97_t ** rac97)
 	int err;
 	ac97_t *ac97;
 	char name[64];
-	signed long end_time;
+	unsigned long end_time;
 	static snd_device_ops_t ops = {
 		.dev_free =	snd_ac97_dev_free,
 	};

sound/pci/intel8x0.c

@@ -1741,7 +1741,7 @@ static void do_delay(intel8x0_t *chip)
 
 static int snd_intel8x0_ich_chip_init(intel8x0_t *chip)
 {
-	signed long end_time;
+	unsigned long end_time;
 	unsigned int cnt, status, nstatus;
 	
 	/* put logic to right state */