Re: [PATCH 08/10] oprofile: set values to default when creating oprofilefs

[Robert Richter <robert.richter@amd.com>]

On 16.12.08 01:32:25, Andrew Morton wrote:
> > +unsigned long fs_buffer_size;
> > +unsigned long fs_cpu_buffer_size;
> > +unsigned long fs_buffer_watershed;
> 
> These were poorly chosen kernel-wide identifiers.

Right, didn't notice that. Will change this.

> >  static ssize_t depth_read(struct file *file, char __user *buf, size_t count, loff_t *offset)
> >  {
> > @@ -120,6 +124,11 @@ static const struct file_operations dump_fops = {
> >  
> >  void oprofile_create_files(struct super_block *sb, struct dentry *root)
> >  {
> > +	/* reinitialize default values */
> > +	fs_buffer_size =	FS_BUFFER_SIZE_DEFAULT;
> > +	fs_cpu_buffer_size =	FS_CPU_BUFFER_SIZE_DEFAULT;
> > +	fs_buffer_watershed =	FS_BUFFER_WATERSHED_DEFAULT;
> > +
> >  	oprofilefs_create_file(sb, root, "enable", &enable_fops);
> >  	oprofilefs_create_file_perm(sb, root, "dump", &dump_fops, 0666);
> >  	oprofilefs_create_file(sb, root, "buffer", &event_buffer_fops);
> 
> afacit oprofile_create_files() only gets run when the module is loaded,
> so this patch is a no-op?

This code runs when oprofilefs is mounted. This happens typically in
an opcontrol --init/--deinit sequence. Before, the previous values
would have been used after mounting again. Now, default values are
restored.

-Robert

-- 
Advanced Micro Devices, Inc.
Operating System Research Center
email: robert.richter@amd.com