Re: [RFC PATCH] usb: common: change usb_debug_root as static variable

From: Greg Kroah-Hartman
Date: Fri Nov 01 2019 - 05:02:27 EST


On Fri, Nov 01, 2019 at 10:29:09AM +0800, Chunfeng Yun wrote:
> Try to avoid using extern global variable, and provide two
> functions for the usage cases

That is 3 different things all in one patch, not generally considered a
good thing at all.

Also, who is going to use these new functions? Why are they needed?

> Signed-off-by: Chunfeng Yun <chunfeng.yun@xxxxxxxxxxxx>
> ---
> NOTE:
> Prepared but not send out patches for drivers using usb_debug_root,
> because I'm not sure whether this patch is needed, and many drivers
> will be modified.
> ---
> drivers/usb/common/common.c | 16 ++++++++++++++--
> include/linux/usb.h | 5 ++++-
> 2 files changed, 18 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/usb/common/common.c b/drivers/usb/common/common.c
> index 1433260d99b4..639ee6d243a2 100644
> --- a/drivers/usb/common/common.c
> +++ b/drivers/usb/common/common.c
> @@ -293,8 +293,20 @@ struct device *usb_of_get_companion_dev(struct device *dev)
> EXPORT_SYMBOL_GPL(usb_of_get_companion_dev);
> #endif
>
> -struct dentry *usb_debug_root;
> -EXPORT_SYMBOL_GPL(usb_debug_root);
> +static struct dentry *usb_debug_root;

Doesn't this break things as-is? You can't do that in a single patch
either :(

> +
> +struct dentry *usb_debugfs_create_dir(const char *name)
> +{
> + return debugfs_create_dir(name, usb_debug_root);
> +}
> +EXPORT_SYMBOL_GPL(usb_debugfs_create_dir);
> +
> +struct dentry *usb_debugfs_create_file(const char *name, umode_t mode,
> + void *data, const struct file_operations *fops)
> +{
> + return debugfs_create_file(name, mode, usb_debug_root, data, fops);

I doubt many people want to create a file in the usb "root" debugfs
directory, right? They _should_ be just creating a new subdirectory in
there instead.

thanks,

greg k-h