Re: [PATCH RESEND] tracing: Make tracing work when debugfs is not compiled or initialized.

[Steven Rostedt]

On Wed,  4 Nov 2015 09:11:18 +0800
Jiaxing Wang <hello.wjx@xxxxxxxxx> wrote:

> Currently tracing_init_dentry() returns -ENODEV when debugfs is not
> initialized, which causes tracefs not populated with tracing files and
> directories, so we will get an empty directory even after we manually
> mount tracefs.
> 
> We can make tracing_init_dentry() return NULL as long as tracefs
> is initialized and get a populated tracefs.
> 
> We also need to make global_trace.dir not NULL in order to pass the checks
> in tracing_get_dentry() and add_tracer_options().
> 
> Also added stub debugfs_create_automount() for when debugfs is not
> configured in.
> 
> Signed-off-by: Jiaxing Wang <hello.wjx@xxxxxxxxx>
> ---
>  include/linux/debugfs.h |  8 ++++++++
>  kernel/trace/Kconfig    |  1 -
>  kernel/trace/trace.c    | 29 +++++++++++++++++------------
>  3 files changed, 25 insertions(+), 13 deletions(-)
> 
> diff --git a/include/linux/debugfs.h b/include/linux/debugfs.h
> index 9beb636..b42ef88 100644
> --- a/include/linux/debugfs.h
> +++ b/include/linux/debugfs.h
> @@ -160,6 +160,14 @@ static inline struct dentry *debugfs_create_symlink(const char *name,
>  	return ERR_PTR(-ENODEV);
>  }
>  
> +static inline struct dentry *debugfs_create_automount(const char *name,
> +					struct dentry *parent,
> +					struct vfsmount *(*f)(void *),
> +					void *data)
> +{
> +	return ERR_PTR(-ENODEV);
> +}

This part needs an Acked-by from Greg KH.

> +
>  static inline void debugfs_remove(struct dentry *dentry)
>  { }
>  
> diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
> index 1153c43..59f6377f 100644
> --- a/kernel/trace/Kconfig
> +++ b/kernel/trace/Kconfig
> @@ -95,7 +95,6 @@ config RING_BUFFER_ALLOW_SWAP
>  
>  config TRACING
>  	bool
> -	select DEBUG_FS
>  	select RING_BUFFER
>  	select STACKTRACE if STACKTRACE_SUPPORT
>  	select TRACEPOINTS
> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> index 6e79408..6dd064e 100644
> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -6706,6 +6706,8 @@ static struct vfsmount *trace_automount(void *ingore)
>  	return mnt;
>  }
>  
> +#define TRACE_TOP_DIR_ENTRY	((struct dentry *)1)
> +
>  /**
>   * tracing_init_dentry - initialize top level trace array
>   *
> @@ -6716,27 +6718,30 @@ static struct vfsmount *trace_automount(void *ingore)
>  struct dentry *tracing_init_dentry(void)
>  {
>  	struct trace_array *tr = &global_trace;
> +	struct dentry *traced;
>  
>  	/* The top level trace array uses  NULL as parent */
>  	if (tr->dir)
>  		return NULL;
>  
> -	if (WARN_ON(!debugfs_initialized()))
> +	if (WARN_ON(!tracefs_initialized()))
>  		return ERR_PTR(-ENODEV);
>  
> -	/*
> -	 * As there may still be users that expect the tracing
> -	 * files to exist in debugfs/tracing, we must automount
> -	 * the tracefs file system there, so older tools still
> -	 * work with the newer kerenl.
> -	 */
> -	tr->dir = debugfs_create_automount("tracing", NULL,
> -					   trace_automount, NULL);
> -	if (!tr->dir) {
> -		pr_warn_once("Could not create debugfs directory 'tracing'\n");
> -		return ERR_PTR(-ENOMEM);
> +	if (debugfs_initialized()) {
> +		/*
> +		 * As there may still be users that expect the tracing
> +		 * files to exist in debugfs/tracing, we must automount
> +		 * the tracefs file system there, so older tools still
> +		 * work with the newer kerenl.
> +		 */
> +		traced = debugfs_create_automount("tracing", NULL,
> +						   trace_automount, NULL);
> +		if (!traced)
> +			pr_warn_once("Could not create debugfs directory 'tracing'\n");

This should return a warning, and please keep the tr->dir instead of
this new traced variable.

>  	}
>  
> +	tr->dir = TRACE_TOP_DIR_ENTRY;
> +

Also, no need to add this, because if debugfs is not initialize, then
tr->dir would be ERR_PTR(-ENODEV), which still works as tr->dir is not
NULL.

-- Steve

>  	return NULL;
>  }
>  

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/