Skip to contentSkip to conversation

oprofile: disable write access to oprofilefs while profiler is running

PATCH Applied Robert Richter 2 participants 7 messages Oct 5, 2010 – Oct 12, 2010

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 oprofile: disable write access to oprofilefs while profiler is running

Oprofile counters are setup when profiling is disabled. Thus, writing to oprofilefs has no immediate effect. Changes are updated only after oprofile is reenabled.

To keep userland and kernel states synchronized, we now allow configuration of oprofile only if profiling is disabled. In this case it checks if the profiler is running and then disables write access to oprofilefs by returning -EBUSY. The change should be backward compatible with current oprofile userland daemon.

Cc Maynard Johnson <[email protected]>
Cc William Cohen <[email protected]>
Cc Suravee Suthikulpanit <[email protected]>
---
 drivers/oprofile/oprof.c          |   19 +++++--------------
 drivers/oprofile/oprof.h          |    2 +-
 drivers/oprofile/oprofile_files.c |    7 +++++--
 drivers/oprofile/oprofilefs.c     |    8 ++++++--
 4 files changed, 17 insertions(+), 19 deletions(-)
drivers/oprofile/oprof.c +5 -14
index b4a6857..a7ab09e 100644
@@ -225,26 +225,17 @@ post_sync:
 	mutex_unlock(&start_mutex);
 }
 
-int oprofile_set_backtrace(unsigned long val)
+int oprofile_set_ulong(unsigned long *addr, unsigned long val)
 {
-	int err = 0;
+	int err = -EBUSY;
 
 	mutex_lock(&start_mutex);
-
 	if (oprofile_started) {
-		err = -EBUSY;
-		goto out;
-	}
-
-	if (!oprofile_ops.backtrace) {
-		err = -EINVAL;
-		goto out;
+		*addr = val;
+		err = 0;
 	}
-
-	oprofile_backtrace_depth = val;
-
-out:
 	mutex_unlock(&start_mutex);
+
 	return err;
 }
 
drivers/oprofile/oprof.h +1 -1
index 47e12cb..177b73d 100644
@@ -37,7 +37,7 @@ void oprofile_create_files(struct super_block *sb, struct dentry *root);
 int oprofile_timer_init(struct oprofile_operations *ops);
 void oprofile_timer_exit(void);
 
-int oprofile_set_backtrace(unsigned long depth);
+int oprofile_set_ulong(unsigned long *addr, unsigned long val);
 int oprofile_set_timeout(unsigned long time);
 
 #endif /* OPROF_H */
drivers/oprofile/oprofile_files.c +5 -2
index bbd7516..ccf099e 100644
@@ -79,14 +79,17 @@ static ssize_t depth_write(struct file *file, char const __user *buf, size_t cou
 	if (*offset)
 		return -EINVAL;
 
+	if (!oprofile_ops.backtrace)
+		return -EINVAL;
+
 	retval = oprofilefs_ulong_from_user(&val, buf, count);
 	if (retval)
 		return retval;
 
-	retval = oprofile_set_backtrace(val);
-
+	retval = oprofile_set_ulong(&oprofile_backtrace_depth, val);
 	if (retval)
 		return retval;
+
 	return count;
 }
 
drivers/oprofile/oprofilefs.c +6 -2
index 789a1a8..1944621 100644
@@ -91,16 +91,20 @@ static ssize_t ulong_read_file(struct file *file, char __user *buf, size_t count
 
 static ssize_t ulong_write_file(struct file *file, char const __user *buf, size_t count, loff_t *offset)
 {
-	unsigned long *value = file->private_data;
+	unsigned long value;
 	int retval;
 
 	if (*offset)
 		return -EINVAL;
 
-	retval = oprofilefs_ulong_from_user(value, buf, count);
+	retval = oprofilefs_ulong_from_user(&value, buf, count);
+	if (retval)
+		return retval;
 
+	retval = oprofile_set_ulong(file->private_data, value);
 	if (retval)
 		return retval;
+
 	return count;
 }
 
Signature · 1 line
1.7.2.2
#1 lore raw
Quoted 10 lines · Robert Richter

Oprofile counters are setup when profiling is disabled. Thus, writing to oprofilefs has no immediate effect. Changes are updated only after oprofile is reenabled.

To keep userland and kernel states synchronized, we now allow configuration of oprofile only if profiling is disabled. In this case it checks if the profiler is running and then disables write access to oprofilefs by returning -EBUSY. The change should be backward compatible with current oprofile userland daemon.

Quoted 4 lines
Cc Maynard Johnson <[email protected]>
Cc William Cohen <[email protected]>
Cc Suravee Suthikulpanit <[email protected]>

Maynard, please ack if you are fine with it.

Thanks,

-Robert

Signature · 2 lines
Advanced Micro Devices, Inc.
Operating System Research Center
#2 lore raw

PATCH 1/1 oprofile: disable write access to oprofilefs while profiler is running

In reply to Robert Richter
Quoted 28 lines · Robert Richter

Oprofile counters are setup when profiling is disabled. Thus, writing to oprofilefs has no immediate effect. Changes are updated only after oprofile is reenabled.

To keep userland and kernel states synchronized, we now allow configuration of oprofile only if profiling is disabled. In this case it checks if the profiler is running and then disables write access to oprofilefs by returning -EBUSY. The change should be backward compatible with current oprofile userland daemon.

Cc Maynard Johnson <[email protected]>
Cc William Cohen <[email protected]>
Cc Suravee Suthikulpanit <[email protected]>
---
 drivers/oprofile/oprof.c          |   19 +++++--------------
 drivers/oprofile/oprof.h          |    2 +-
 drivers/oprofile/oprofile_files.c |    7 +++++--
 drivers/oprofile/oprofilefs.c     |    8 ++++++--
 4 files changed, 17 insertions(+), 19 deletions(-)
drivers/oprofile/oprof.c +0 -0
index b4a6857..a7ab09e 100644
@@ -225,26 +225,17 @@ post_sync:
 	mutex_unlock(&start_mutex);
 }

drivers/oprofile/oprof.c post_sync: · line 233

Quoted 10 lines
drivers/oprofile/oprof.c +2 -3
 
-int oprofile_set_backtrace(unsigned long val)
+int oprofile_set_ulong(unsigned long *addr, unsigned long val)
 {
-	int err = 0;
+	int err = -EBUSY;
 
 	mutex_lock(&start_mutex);
-
 	if (oprofile_started) {

I did a late change which I never should do. It should be:

	if (!oprofile_started) ...

The concept remains the same.

Updated patch below.

Sorry,

-Robert

Quoted 8 lines
drivers/oprofile/oprof.c +1 -7
-		err = -EBUSY;
-		goto out;
-	}
-
-	if (!oprofile_ops.backtrace) {
-		err = -EINVAL;
-		goto out;
+		*addr = val;

drivers/oprofile/oprof.c post_sync: · line 240

Quoted 10 lines
drivers/oprofile/oprof.c +2 -4
+		err = 0;
 	}
-
-	oprofile_backtrace_depth = val;
-
-out:
 	mutex_unlock(&start_mutex);
+
 	return err;
 }

--

Quoted 1 line

From 2d5710a2850eec24c54a1338ae5986963928cf8a Mon Sep 17 00:00:00 2001

From: Robert Richter <[email protected]>
Date: Mon, 4 Oct 2010 21:09:36 +0200
Subject: [PATCH] oprofile: disable write access to oprofilefs while profiler is running

Oprofile counters are setup when profiling is disabled. Thus, writing to oprofilefs has no immediate effect. Changes are updated only after oprofile is reenabled.

To keep userland and kernel states synchronized, we now allow configuration of oprofile only if profiling is disabled. In this case it checks if the profiler is running and then disables write access to oprofilefs by returning -EBUSY. The change should be backward compatible with current oprofile userland daemon.

Cc Maynard Johnson <[email protected]>
Cc William Cohen <[email protected]>
Cc Suravee Suthikulpanit <[email protected]>
---
 drivers/oprofile/oprof.c          |   21 ++++++---------------
 drivers/oprofile/oprof.h          |    2 +-
 drivers/oprofile/oprofile_files.c |    7 +++++--
 drivers/oprofile/oprofilefs.c     |    8 ++++++--
 4 files changed, 18 insertions(+), 20 deletions(-)
drivers/oprofile/oprof.c +6 -15
index b4a6857..f9bda64 100644
@@ -225,26 +225,17 @@ post_sync:
 	mutex_unlock(&start_mutex);
 }
 
-int oprofile_set_backtrace(unsigned long val)
+int oprofile_set_ulong(unsigned long *addr, unsigned long val)
 {
-	int err = 0;
+	int err = -EBUSY;
 
 	mutex_lock(&start_mutex);
-
-	if (oprofile_started) {
-		err = -EBUSY;
-		goto out;
+	if (!oprofile_started) {
+		*addr = val;
+		err = 0;
 	}
-
-	if (!oprofile_ops.backtrace) {
-		err = -EINVAL;
-		goto out;
-	}
-
-	oprofile_backtrace_depth = val;
-
-out:
 	mutex_unlock(&start_mutex);
+
 	return err;
 }
 
drivers/oprofile/oprof.h +1 -1
index 47e12cb..177b73d 100644
@@ -37,7 +37,7 @@ void oprofile_create_files(struct super_block *sb, struct dentry *root);
 int oprofile_timer_init(struct oprofile_operations *ops);
 void oprofile_timer_exit(void);
 
-int oprofile_set_backtrace(unsigned long depth);
+int oprofile_set_ulong(unsigned long *addr, unsigned long val);
 int oprofile_set_timeout(unsigned long time);
 
 #endif /* OPROF_H */
drivers/oprofile/oprofile_files.c +5 -2
index bbd7516..ccf099e 100644
@@ -79,14 +79,17 @@ static ssize_t depth_write(struct file *file, char const __user *buf, size_t cou
 	if (*offset)
 		return -EINVAL;
 
+	if (!oprofile_ops.backtrace)
+		return -EINVAL;
+
 	retval = oprofilefs_ulong_from_user(&val, buf, count);
 	if (retval)
 		return retval;
 
-	retval = oprofile_set_backtrace(val);
-
+	retval = oprofile_set_ulong(&oprofile_backtrace_depth, val);
 	if (retval)
 		return retval;
+
 	return count;
 }
 
drivers/oprofile/oprofilefs.c +6 -2
index 789a1a8..1944621 100644
@@ -91,16 +91,20 @@ static ssize_t ulong_read_file(struct file *file, char __user *buf, size_t count
 
 static ssize_t ulong_write_file(struct file *file, char const __user *buf, size_t count, loff_t *offset)
 {
-	unsigned long *value = file->private_data;
+	unsigned long value;
 	int retval;
 
 	if (*offset)
 		return -EINVAL;
 
-	retval = oprofilefs_ulong_from_user(value, buf, count);
+	retval = oprofilefs_ulong_from_user(&value, buf, count);
+	if (retval)
+		return retval;
 
+	retval = oprofile_set_ulong(file->private_data, value);
 	if (retval)
 		return retval;
+
 	return count;
 }
 
Signature · 7 lines
1.7.2.2



-- 
Advanced Micro Devices, Inc.
Operating System Research Center
#3 lore raw
Quoted 4 lines · Robert Richter
Quoted 3 lines · Robert Richter

Oprofile counters are setup when profiling is disabled. Thus, writing to oprofilefs has no immediate effect. Changes are updated only after oprofile is reenabled.

Robert, I don't have any issue with this patch except I'm not sure what problem it's fixing. From my read of oprofile userland code, I don't see where profiling parameters are ever written to oprofilefs when profiling is active. Is your patch intended to guard against potential non-intentional cases occurring in userland code? Or is there some case existing right now?

-Maynard

Quoted 196 lines
Quoted 25 lines

To keep userland and kernel states synchronized, we now allow configuration of oprofile only if profiling is disabled. In this case it checks if the profiler is running and then disables write access to oprofilefs by returning -EBUSY. The change should be backward compatible with current oprofile userland daemon.

Cc Maynard Johnson<[email protected]>
Cc William Cohen<[email protected]>
Cc Suravee Suthikulpanit<[email protected]>
---
  drivers/oprofile/oprof.c          |   19 +++++--------------
  drivers/oprofile/oprof.h          |    2 +-
  drivers/oprofile/oprofile_files.c |    7 +++++--
  drivers/oprofile/oprofilefs.c     |    8 ++++++--
  4 files changed, 17 insertions(+), 19 deletions(-)
drivers/oprofile/oprof.c +0 -0
index b4a6857..a7ab09e 100644
@@ -225,26 +225,17 @@ post_sync:
  	mutex_unlock(&start_mutex);
  }

drivers/oprofile/oprof.c post_sync: · line 233

Quoted 10 lines
drivers/oprofile/oprof.c +2 -3
 
-int oprofile_set_backtrace(unsigned long val)
+int oprofile_set_ulong(unsigned long *addr, unsigned long val)
  {
-	int err = 0;
+	int err = -EBUSY;
 
  	mutex_lock(&start_mutex);
-
  	if (oprofile_started) {

I did a late change which I never should do. It should be:

	if (!oprofile_started) ...

The concept remains the same.

Updated patch below.

Sorry,

-Robert

Quoted 8 lines
drivers/oprofile/oprof.c +1 -7
-		err = -EBUSY;
-		goto out;
-	}
-
-	if (!oprofile_ops.backtrace) {
-		err = -EINVAL;
-		goto out;
+		*addr = val;

drivers/oprofile/oprof.c post_sync: · line 240

Quoted 10 lines
drivers/oprofile/oprof.c +2 -4
+		err = 0;
  	}
-
-	oprofile_backtrace_depth = val;
-
-out:
  	mutex_unlock(&start_mutex);
+
  	return err;
  }

--

From 2d5710a2850eec24c54a1338ae5986963928cf8a Mon Sep 17 00:00:00 2001
From: Robert Richter<[email protected]>
Date: Mon, 4 Oct 2010 21:09:36 +0200
Subject: [PATCH] oprofile: disable write access to oprofilefs while profiler is running

Oprofile counters are setup when profiling is disabled. Thus, writing to oprofilefs has no immediate effect. Changes are updated only after oprofile is reenabled.

To keep userland and kernel states synchronized, we now allow configuration of oprofile only if profiling is disabled. In this case it checks if the profiler is running and then disables write access to oprofilefs by returning -EBUSY. The change should be backward compatible with current oprofile userland daemon.

Cc Maynard Johnson<[email protected]>
Cc William Cohen<[email protected]>
Cc Suravee Suthikulpanit<[email protected]>
---
  drivers/oprofile/oprof.c          |   21 ++++++---------------
  drivers/oprofile/oprof.h          |    2 +-
  drivers/oprofile/oprofile_files.c |    7 +++++--
  drivers/oprofile/oprofilefs.c     |    8 ++++++--
  4 files changed, 18 insertions(+), 20 deletions(-)
drivers/oprofile/oprof.c +6 -15
index b4a6857..f9bda64 100644
@@ -225,26 +225,17 @@ post_sync:
  	mutex_unlock(&start_mutex);
  }
 
-int oprofile_set_backtrace(unsigned long val)
+int oprofile_set_ulong(unsigned long *addr, unsigned long val)
  {
-	int err = 0;
+	int err = -EBUSY;
 
  	mutex_lock(&start_mutex);
-
-	if (oprofile_started) {
-		err = -EBUSY;
-		goto out;
+	if (!oprofile_started) {
+		*addr = val;
+		err = 0;
  	}
-
-	if (!oprofile_ops.backtrace) {
-		err = -EINVAL;
-		goto out;
-	}
-
-	oprofile_backtrace_depth = val;
-
-out:
  	mutex_unlock(&start_mutex);
+
  	return err;
  }
 
drivers/oprofile/oprof.h +1 -1
index 47e12cb..177b73d 100644
@@ -37,7 +37,7 @@ void oprofile_create_files(struct super_block *sb, struct dentry *root);
  int oprofile_timer_init(struct oprofile_operations *ops);
  void oprofile_timer_exit(void);
 
-int oprofile_set_backtrace(unsigned long depth);
+int oprofile_set_ulong(unsigned long *addr, unsigned long val);
  int oprofile_set_timeout(unsigned long time);
 
  #endif /* OPROF_H */
drivers/oprofile/oprofile_files.c +5 -2
index bbd7516..ccf099e 100644
@@ -79,14 +79,17 @@ static ssize_t depth_write(struct file *file, char const __user *buf, size_t cou
  	if (*offset)
  		return -EINVAL;
 
+	if (!oprofile_ops.backtrace)
+		return -EINVAL;
+
  	retval = oprofilefs_ulong_from_user(&val, buf, count);
  	if (retval)
  		return retval;
 
-	retval = oprofile_set_backtrace(val);
-
+	retval = oprofile_set_ulong(&oprofile_backtrace_depth, val);
  	if (retval)
  		return retval;
+
  	return count;
  }
 
drivers/oprofile/oprofilefs.c +6 -2
index 789a1a8..1944621 100644
@@ -91,16 +91,20 @@ static ssize_t ulong_read_file(struct file *file, char __user *buf, size_t count
 
  static ssize_t ulong_write_file(struct file *file, char const __user *buf, size_t count, loff_t *offset)
  {
-	unsigned long *value = file->private_data;
+	unsigned long value;
  	int retval;
 
  	if (*offset)
  		return -EINVAL;
 
-	retval = oprofilefs_ulong_from_user(value, buf, count);
+	retval = oprofilefs_ulong_from_user(&value, buf, count);
+	if (retval)
+		return retval;
 
+	retval = oprofile_set_ulong(file->private_data, value);
  	if (retval)
  		return retval;
+
  	return count;
  }
 
#4 lore raw
Quoted 5 lines · Maynard Johnson

Robert, I don't have any issue with this patch except I'm not sure what problem it's fixing. From my read of oprofile userland code, I don't see where profiling parameters are ever written to oprofilefs when profiling is active. Is your patch intended to guard against potential non-intentional cases occurring in userland code? Or is there some case existing right now?

The kernel updates the counter configuration during profiling only when the counters are started. Since sampling data and its representation depends on the counter configuration, we do not want to change the configuration in-flight. Otherwise the daemon may not know which configuration was used for which sample.

Instead, userland is required to restart profiling which also flushes cpu buffers. Thus, the daemon always is aware of the configuration which was used when the kernel wrote a sample.

This may happen e.g. for IBS samples, where certain options may be enabled using oprofilefs. We want to know which options are set while a sample is written and make sure, the options do not change. There is another case, the latest patch I sent that implements IBS branch target address reporting extends the IBS sample size if its option is set. To always proper parse the sampling data by the daemon, we must ensure the sampling data is consistent over one profiling session and sample size does not change.

As we rely on not writing to the configuration while profiling and since this works with the current daemon, I added the check.

-Robert

Signature · 2 lines
Advanced Micro Devices, Inc.
Operating System Research Center
#5 lore raw
In reply to Robert Richter
Quoted 5 lines · Robert Richter
Quoted 4 lines · Robert Richter

Oprofile counters are setup when profiling is disabled. Thus, writing to oprofilefs has no immediate effect. Changes are updated only after oprofile is reenabled.

Quoted 7 lines
Quoted 7 lines

To keep userland and kernel states synchronized, we now allow configuration of oprofile only if profiling is disabled. In this case it checks if the profiler is running and then disables write access to oprofilefs by returning -EBUSY. The change should be backward compatible with current oprofile userland daemon.

Cc Maynard Johnson <[email protected]>

Ack. This patch looks fine, and tests OK on ppc64. A little paranoia in the kernel can't hurt.

-Maynard

Quoted 188 lines
Quoted 17 lines
Cc William Cohen <[email protected]>
Cc Suravee Suthikulpanit <[email protected]>
---
 drivers/oprofile/oprof.c          |   19 +++++--------------
 drivers/oprofile/oprof.h          |    2 +-
 drivers/oprofile/oprofile_files.c |    7 +++++--
 drivers/oprofile/oprofilefs.c     |    8 ++++++--
 4 files changed, 17 insertions(+), 19 deletions(-)
drivers/oprofile/oprof.c +0 -0
index b4a6857..a7ab09e 100644
@@ -225,26 +225,17 @@ post_sync:
 	mutex_unlock(&start_mutex);
 }

drivers/oprofile/oprof.c post_sync: · line 233

Quoted 10 lines
drivers/oprofile/oprof.c +2 -3
 
-int oprofile_set_backtrace(unsigned long val)
+int oprofile_set_ulong(unsigned long *addr, unsigned long val)
 {
-	int err = 0;
+	int err = -EBUSY;
 
 	mutex_lock(&start_mutex);
-
 	if (oprofile_started) {

I did a late change which I never should do. It should be:

	if (!oprofile_started) ...

The concept remains the same.

Updated patch below.

Sorry,

-Robert

Quoted 8 lines
drivers/oprofile/oprof.c +1 -7
-		err = -EBUSY;
-		goto out;
-	}
-
-	if (!oprofile_ops.backtrace) {
-		err = -EINVAL;
-		goto out;
+		*addr = val;

drivers/oprofile/oprof.c post_sync: · line 240

Quoted 10 lines
drivers/oprofile/oprof.c +2 -4
+		err = 0;
 	}
-
-	oprofile_backtrace_depth = val;
-
-out:
 	mutex_unlock(&start_mutex);
+
 	return err;
 }

--

From 2d5710a2850eec24c54a1338ae5986963928cf8a Mon Sep 17 00:00:00 2001
From: Robert Richter <[email protected]>
Date: Mon, 4 Oct 2010 21:09:36 +0200
Subject: [PATCH] oprofile: disable write access to oprofilefs while profiler is running

Oprofile counters are setup when profiling is disabled. Thus, writing to oprofilefs has no immediate effect. Changes are updated only after oprofile is reenabled.

To keep userland and kernel states synchronized, we now allow configuration of oprofile only if profiling is disabled. In this case it checks if the profiler is running and then disables write access to oprofilefs by returning -EBUSY. The change should be backward compatible with current oprofile userland daemon.

Cc Maynard Johnson <[email protected]>
Cc William Cohen <[email protected]>
Cc Suravee Suthikulpanit <[email protected]>
---
 drivers/oprofile/oprof.c          |   21 ++++++---------------
 drivers/oprofile/oprof.h          |    2 +-
 drivers/oprofile/oprofile_files.c |    7 +++++--
 drivers/oprofile/oprofilefs.c     |    8 ++++++--
 4 files changed, 18 insertions(+), 20 deletions(-)
drivers/oprofile/oprof.c +6 -15
index b4a6857..f9bda64 100644
@@ -225,26 +225,17 @@ post_sync:
 	mutex_unlock(&start_mutex);
 }
 
-int oprofile_set_backtrace(unsigned long val)
+int oprofile_set_ulong(unsigned long *addr, unsigned long val)
 {
-	int err = 0;
+	int err = -EBUSY;
 
 	mutex_lock(&start_mutex);
-
-	if (oprofile_started) {
-		err = -EBUSY;
-		goto out;
+	if (!oprofile_started) {
+		*addr = val;
+		err = 0;
 	}
-
-	if (!oprofile_ops.backtrace) {
-		err = -EINVAL;
-		goto out;
-	}
-
-	oprofile_backtrace_depth = val;
-
-out:
 	mutex_unlock(&start_mutex);
+
 	return err;
 }
 
drivers/oprofile/oprof.h +1 -1
index 47e12cb..177b73d 100644
@@ -37,7 +37,7 @@ void oprofile_create_files(struct super_block *sb, struct dentry *root);
 int oprofile_timer_init(struct oprofile_operations *ops);
 void oprofile_timer_exit(void);
 
-int oprofile_set_backtrace(unsigned long depth);
+int oprofile_set_ulong(unsigned long *addr, unsigned long val);
 int oprofile_set_timeout(unsigned long time);
 
 #endif /* OPROF_H */
drivers/oprofile/oprofile_files.c +5 -2
index bbd7516..ccf099e 100644
@@ -79,14 +79,17 @@ static ssize_t depth_write(struct file *file, char const __user *buf, size_t cou
 	if (*offset)
 		return -EINVAL;
 
+	if (!oprofile_ops.backtrace)
+		return -EINVAL;
+
 	retval = oprofilefs_ulong_from_user(&val, buf, count);
 	if (retval)
 		return retval;
 
-	retval = oprofile_set_backtrace(val);
-
+	retval = oprofile_set_ulong(&oprofile_backtrace_depth, val);
 	if (retval)
 		return retval;
+
 	return count;
 }
 
drivers/oprofile/oprofilefs.c +6 -2
index 789a1a8..1944621 100644
@@ -91,16 +91,20 @@ static ssize_t ulong_read_file(struct file *file, char __user *buf, size_t count
 
 static ssize_t ulong_write_file(struct file *file, char const __user *buf, size_t count, loff_t *offset)
 {
-	unsigned long *value = file->private_data;
+	unsigned long value;
 	int retval;
 
 	if (*offset)
 		return -EINVAL;
 
-	retval = oprofilefs_ulong_from_user(value, buf, count);
+	retval = oprofilefs_ulong_from_user(&value, buf, count);
+	if (retval)
+		return retval;
 
+	retval = oprofile_set_ulong(file->private_data, value);
 	if (retval)
 		return retval;
+
 	return count;
 }
 
author Applied
#6 lore raw
In reply to Robert Richter
Quoted 14 lines · Robert Richter

Oprofile counters are setup when profiling is disabled. Thus, writing to oprofilefs has no immediate effect. Changes are updated only after oprofile is reenabled.

To keep userland and kernel states synchronized, we now allow configuration of oprofile only if profiling is disabled. In this case it checks if the profiler is running and then disables write access to oprofilefs by returning -EBUSY. The change should be backward compatible with current oprofile userland daemon.

Cc Maynard Johnson <[email protected]>
Cc William Cohen <[email protected]>
Cc Suravee Suthikulpanit <[email protected]>
Quoted 6 lines
---
 drivers/oprofile/oprof.c          |   19 +++++--------------
 drivers/oprofile/oprof.h          |    2 +-
 drivers/oprofile/oprofile_files.c |    7 +++++--
 drivers/oprofile/oprofilefs.c     |    8 ++++++--
 4 files changed, 17 insertions(+), 19 deletions(-)

I applied the patch to oprofile/core. Maynard, thanks for review.

-Robert

Signature · 2 lines
Advanced Micro Devices, Inc.
Operating System Research Center