Re: [PATCH v2 3/3] usb: gadget: f_fs: add dev name as prefix for pr_vdebug()

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Mon, Mar 27, 2023 at 09:54:11AM +0800, Linyu Yuan wrote:
> 
> On 3/26/2023 2:29 PM, Greg Kroah-Hartman wrote:
> > On Sun, Mar 26, 2023 at 10:55:18AM +0800, Linyu Yuan wrote:
> > > On 3/25/2023 5:00 PM, Greg Kroah-Hartman wrote:
> > > > On Fri, Mar 24, 2023 at 02:10:27PM +0800, Linyu Yuan wrote:
> > > > > when multiple instances in use, the debug message is hard to understand
> > > > > as there is no instance name show.
> > > > > 
> > > > > this change will show each instance name for debug messages.
> > > > > 
> > > > > Signed-off-by: Linyu Yuan <quic_linyyuan@quicinc.com>
> > > > > ---
> > > > > v2: split to several changes according to v1 comments
> > > > > v1: https://lore.kernel.org/linux-usb/1679481369-30094-1-git-send-email-quic_linyyuan@quicinc.com/
> > > > > 
> > > > >    drivers/usb/gadget/function/f_fs.c | 136 +++++++++++++++++++------------------
> > > > >    1 file changed, 69 insertions(+), 67 deletions(-)
> > > > > 
> > > > > diff --git a/drivers/usb/gadget/function/f_fs.c b/drivers/usb/gadget/function/f_fs.c
> > > > > index a4051c8..df67ab5 100644
> > > > > --- a/drivers/usb/gadget/function/f_fs.c
> > > > > +++ b/drivers/usb/gadget/function/f_fs.c
> > > > > @@ -317,12 +317,12 @@ static int __ffs_ep0_queue_wait(struct ffs_data *ffs, char *data, size_t len)
> > > > >    static int __ffs_ep0_stall(struct ffs_data *ffs)
> > > > >    {
> > > > >    	if (ffs->ev.can_stall) {
> > > > > -		pr_vdebug("ep0 stall\n");
> > > > > +		pr_vdebug("%s: ep0 stall\n", ffs->dev_name);
> > > > Again, no, please use dev_dbg() instead.  Do NOT roll your own debugging
> > > > macros.  You have access to a struct device pointer for this device that
> > > > the driver is controlling, so please always use that instead.
> > > 
> > > thanks for your suggestion, i didn't know dev_dbg can accept NULL dev
> > > pointer
> > > 
> > > as this driver have no real struct device.
> > That is not true, you have access to a struct usb_gadget, which is a
> > struct device.  Use that please.
> 
> but this is not good, as for gadget/udc, from my view, there are two layer,
> 
> one configfs driver, one lower UDC, only after bind, there will be a
> reference to usb_gadget.
> 
> but configuration to driver can happen before bind, if so the usb_gadget
> will be NULL.

But is that where this debug message is being called?  This is a
function while the driver is bound to a device, so you have a usb_gadget
pointer here.

If you have messages that could happen before the device is bound, sure,
don't use the device there.

thanks,

greg k-h